Blame for log.jsonl

{"event": "start", "run": "pilot_B_s2", "condition": "B", "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": "parity_a", "p3_objective": "none", "freeze_in_p3": true, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 2.2}
{"event": "probe_schedule", "n": 40, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 200, 256, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000, 2200, 2400, 2600, 2800, 3000, 3200, 3400, 3600, 3800, 4000, 4200, 4400, 4600, 4800, 5000, 5200, 5400, 5600, 5800, 6000], "t": 2.2}
{"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": 4.0}
{"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.64089, "bpb": 8.13808, "chunk_loss": 0.065103, "F": 0.62439, "G": 0.5091, "chunks": 639.38, "bytes_per_chunk": 1.602, "boundary_entropy": 0.69066, "router_noise": 0.0, "grad_norm": 7.8729, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.005, "alpha": 0.03, "t": 4.3}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10922.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.627593, "cps": 85.328125, "mean_p": 0.513204, "boundary_entropy": 0.686301, "bpb": 8.10084}, "de": {"chunks": 16283.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.793016, "cps": 127.210938, "mean_p": 0.524587, "boundary_entropy": 0.686399, "bpb": 8.104479}, "vi": {"chunks": 5097.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.276485, "cps": 39.820312, "mean_p": 0.481627, "boundary_entropy": 0.687278, "bpb": 7.854704}, "tr": {"chunks": 8206.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.453295, "cps": 64.109375, "mean_p": 0.498086, "boundary_entropy": 0.686828, "bpb": 8.081333}, "fi": {"chunks": 13521.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.718782, "cps": 105.632812, "mean_p": 0.517085, "boundary_entropy": 0.686669, "bpb": 8.0429}, "ru": {"chunks": 33469.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.766174, "cps": 261.476562, "mean_p": 0.530332, "boundary_entropy": 0.688547, "bpb": 8.195792}, "el": {"chunks": 20680.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.00765, "cps": 161.5625, "mean_p": 0.503626, "boundary_entropy": 0.690112, "bpb": 8.279734}, "hi": {"chunks": 885.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.0526, "cps": 6.914062, "mean_p": 0.462026, "boundary_entropy": 0.687437, "bpb": 8.016759}, "ta": {"chunks": 1278.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.068361, "cps": 9.984375, "mean_p": 0.477597, "boundary_entropy": 0.690061, "bpb": 8.085538}, "th": {"chunks": 3634.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.225925, "cps": 28.390625, "mean_p": 0.484856, "boundary_entropy": 0.690398, "bpb": 7.718775}, "zh": {"chunks": 11707.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.982221, "cps": 91.460938, "mean_p": 0.520268, "boundary_entropy": 0.685623, "bpb": 8.126843}, "my": {"chunks": 1619.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.085444, "cps": 12.648438, "mean_p": 0.474597, "boundary_entropy": 0.689682, "bpb": 7.800614}}, "bpb": {"en": 8.10084, "de": 8.10448, "vi": 7.8547, "tr": 8.08133, "fi": 8.0429, "ru": 8.19579, "el": 8.27973, "hi": 8.01676, "ta": 8.08554, "th": 7.71877, "zh": 8.12684, "my": 7.80061}, "gini_cpc": 0.489171, "gini_cps": 0.463253, "bpic": 3.014752, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2635835509223954, "vi": 0.4405482086230508, "tr": 0.7222755425947089, "fi": 1.1452989890055154, "ru": 2.8142033145562273, "el": 1.605578844938961, "hi": 0.08381276064532156, "ta": 0.10892494073686158, "th": 0.35998616123133165, "zh": 3.1584508548843684, "my": 0.13614616755648523}, "t": 5.6}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10902.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.626444, "cps": 85.171875, "mean_p": 0.513135, "boundary_entropy": 0.686312, "bpb": 8.052601}, "de": {"chunks": 16259.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.791847, "cps": 127.023438, "mean_p": 0.524504, "boundary_entropy": 0.686413, "bpb": 8.060202}, "vi": {"chunks": 5089.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.276051, "cps": 39.757812, "mean_p": 0.481574, "boundary_entropy": 0.687273, "bpb": 7.831341}, "tr": {"chunks": 8250.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.455726, "cps": 64.453125, "mean_p": 0.498254, "boundary_entropy": 0.686833, "bpb": 8.048573}, "fi": {"chunks": 13501.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.717718, "cps": 105.476562, "mean_p": 0.516945, "boundary_entropy": 0.686683, "bpb": 7.998133}, "ru": {"chunks": 33462.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.765805, "cps": 261.421875, "mean_p": 0.530286, "boundary_entropy": 0.688551, "bpb": 8.105187}, "el": {"chunks": 20636.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.005506, "cps": 161.21875, "mean_p": 0.503586, "boundary_entropy": 0.690113, "bpb": 8.273486}, "hi": {"chunks": 906.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.053848, "cps": 7.078125, "mean_p": 0.462267, "boundary_entropy": 0.687476, "bpb": 7.990564}, "ta": {"chunks": 1319.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.070554, "cps": 10.304688, "mean_p": 0.477815, "boundary_entropy": 0.690081, "bpb": 8.070662}, "th": {"chunks": 3700.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.230028, "cps": 28.90625, "mean_p": 0.484975, "boundary_entropy": 0.690405, "bpb": 7.702812}, "zh": {"chunks": 11682.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.977988, "cps": 91.265625, "mean_p": 0.520179, "boundary_entropy": 0.685628, "bpb": 8.118739}, "my": {"chunks": 1614.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.08518, "cps": 12.609375, "mean_p": 0.474587, "boundary_entropy": 0.68968, "bpb": 7.800129}}, "bpb": {"en": 8.0526, "de": 8.0602, "vi": 7.83134, "tr": 8.04857, "fi": 7.99813, "ru": 8.10519, "el": 8.27349, "hi": 7.99056, "ta": 8.07066, "th": 7.70281, "zh": 8.11874, "my": 7.80013}, "gini_cpc": 0.487895, "gini_cps": 0.462038, "bpic": 3.014303, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2640357772999535, "vi": 0.44066367440436505, "tr": 0.7274804715425409, "fi": 1.1457028576734998, "ru": 2.8187763761484548, "el": 1.6051019285422041, "hi": 0.0859589431496, "ta": 0.1126256391340116, "th": 0.3671965577749877, "zh": 3.1574879496257737, "my": 0.13597469609204185}, "t": 7.1}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10855.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.623743, "cps": 84.804688, "mean_p": 0.512767, "boundary_entropy": 0.686354, "bpb": 7.898108}, "de": {"chunks": 16228.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.790338, "cps": 126.78125, "mean_p": 0.524111, "boundary_entropy": 0.686473, "bpb": 7.921947}, "vi": {"chunks": 5064.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.274695, "cps": 39.5625, "mean_p": 0.481477, "boundary_entropy": 0.687264, "bpb": 7.752798}, "tr": {"chunks": 8386.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.463238, "cps": 65.515625, "mean_p": 0.498731, "boundary_entropy": 0.686848, "bpb": 7.941202}, "fi": {"chunks": 13397.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.71219, "cps": 104.664062, "mean_p": 0.516362, "boundary_entropy": 0.686734, "bpb": 7.85634}, "ru": {"chunks": 33403.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.762691, "cps": 260.960938, "mean_p": 0.530055, "boundary_entropy": 0.68857, "bpb": 7.817035}, "el": {"chunks": 20288.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.988549, "cps": 158.5, "mean_p": 0.503177, "boundary_entropy": 0.690117, "bpb": 8.218061}, "hi": {"chunks": 994.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.059079, "cps": 7.765625, "mean_p": 0.463105, "boundary_entropy": 0.687608, "bpb": 7.902834}, "ta": {"chunks": 1471.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.078684, "cps": 11.492188, "mean_p": 0.478672, "boundary_entropy": 0.69016, "bpb": 7.989569}, "th": {"chunks": 3938.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.244824, "cps": 30.765625, "mean_p": 0.485307, "boundary_entropy": 0.690423, "bpb": 7.620447}, "zh": {"chunks": 11614.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.966475, "cps": 90.734375, "mean_p": 0.519763, "boundary_entropy": 0.685653, "bpb": 8.08797}, "my": {"chunks": 1601.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.084494, "cps": 12.507812, "mean_p": 0.474498, "boundary_entropy": 0.689669, "bpb": 7.794014}}, "bpb": {"en": 7.89811, "de": 7.92195, "vi": 7.7528, "tr": 7.9412, "fi": 7.85634, "ru": 7.81704, "el": 8.21806, "hi": 7.90283, "ta": 7.98957, "th": 7.62045, "zh": 8.08797, "my": 7.79401}, "gini_cpc": 0.483149, "gini_cps": 0.456919, "bpic": 3.016221, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2670883102580173, "vi": 0.44039750254763677, "tr": 0.7426746479845497, "fi": 1.1417998098420465, "ru": 2.8259895504302945, "el": 1.5848664920392537, "hi": 0.09471649180778915, "ta": 0.12614832712421967, "th": 0.3925083851092201, "zh": 3.1527001793646288, "my": 0.13546348623938853}, "t": 8.7}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10130.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.582084, "cps": 79.140625, "mean_p": 0.50941, "boundary_entropy": 0.686568, "bpb": 7.422695}, "de": {"chunks": 15619.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.760678, "cps": 122.023438, "mean_p": 0.520563, "boundary_entropy": 0.686863, "bpb": 7.489079}, "vi": {"chunks": 4865.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.2639, "cps": 38.007812, "mean_p": 0.480766, "boundary_entropy": 0.687199, "bpb": 7.511586}, "tr": {"chunks": 8659.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.478319, "cps": 67.648438, "mean_p": 0.499755, "boundary_entropy": 0.686899, "bpb": 7.589519}, "fi": {"chunks": 12507.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.664877, "cps": 97.710938, "mean_p": 0.512532, "boundary_entropy": 0.686991, "bpb": 7.42635}, "ru": {"chunks": 33186.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.75124, "cps": 259.265625, "mean_p": 0.529047, "boundary_entropy": 0.68866, "bpb": 7.199649}, "el": {"chunks": 18825.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.917264, "cps": 147.070312, "mean_p": 0.501354, "boundary_entropy": 0.69013, "bpb": 7.947597}, "hi": {"chunks": 1278.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.075958, "cps": 9.984375, "mean_p": 0.465336, "boundary_entropy": 0.687947, "bpb": 7.609082}, "ta": {"chunks": 2097.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.112169, "cps": 16.382812, "mean_p": 0.480874, "boundary_entropy": 0.690348, "bpb": 7.734842}, "th": {"chunks": 4620.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.287224, "cps": 36.09375, "mean_p": 0.486153, "boundary_entropy": 0.690464, "bpb": 7.339848}, "zh": {"chunks": 11282.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.910261, "cps": 88.140625, "mean_p": 0.518094, "boundary_entropy": 0.685748, "bpb": 7.997335}, "my": {"chunks": 1467.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.077422, "cps": 11.460938, "mean_p": 0.473998, "boundary_entropy": 0.68962, "bpb": 7.66307}}, "bpb": {"en": 7.42269, "de": 7.48908, "vi": 7.51159, "tr": 7.58952, "fi": 7.42635, "ru": 7.19965, "el": 7.9476, "hi": 7.60908, "ta": 7.73484, "th": 7.33985, "zh": 7.99733, "my": 7.66307}, "gini_cpc": 0.471553, "gini_cps": 0.442995, "bpic": 3.081712, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.3068191578923178, "vi": 0.45337166871355855, "tr": 0.821735147332591, "fi": 1.1422362564005655, "ru": 3.0085717232703093, "el": 1.5758280058407492, "hi": 0.13049397358852013, "ta": 0.19270262727891252, "th": 0.49344140104657486, "zh": 3.2817638561885465, "my": 0.13300911769541873}, "t": 10.5}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4827.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.277366, "cps": 37.710938, "mean_p": 0.48872, "boundary_entropy": 0.686417, "bpb": 6.857526}, "de": {"chunks": 8526.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.415234, "cps": 66.609375, "mean_p": 0.497539, "boundary_entropy": 0.687717, "bpb": 6.930862}, "vi": {"chunks": 3519.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.190887, "cps": 27.492188, "mean_p": 0.475364, "boundary_entropy": 0.686685, "bpb": 7.167673}, "tr": {"chunks": 7554.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.417279, "cps": 59.015625, "mean_p": 0.495275, "boundary_entropy": 0.686954, "bpb": 7.150655}, "fi": {"chunks": 6362.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.338206, "cps": 49.703125, "mean_p": 0.492198, "boundary_entropy": 0.687148, "bpb": 6.916919}, "ru": {"chunks": 25773.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.360053, "cps": 201.351562, "mean_p": 0.511522, "boundary_entropy": 0.689919, "bpb": 6.367917}, "el": {"chunks": 14724.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.717439, "cps": 115.03125, "mean_p": 0.496062, "boundary_entropy": 0.690088, "bpb": 7.322722}, "hi": {"chunks": 3463.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.205825, "cps": 27.054688, "mean_p": 0.475509, "boundary_entropy": 0.689261, "bpb": 6.574795}, "ta": {"chunks": 6940.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.371222, "cps": 54.21875, "mean_p": 0.490183, "boundary_entropy": 0.690934, "bpb": 6.749366}, "th": {"chunks": 6897.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.428785, "cps": 53.882812, "mean_p": 0.489675, "boundary_entropy": 0.690643, "bpb": 6.489059}, "zh": {"chunks": 9452.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.600406, "cps": 73.84375, "mean_p": 0.509541, "boundary_entropy": 0.686084, "bpb": 7.74738}, "my": {"chunks": 1206.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.063648, "cps": 9.421875, "mean_p": 0.473249, "boundary_entropy": 0.689569, "bpb": 7.216403}}, "bpb": {"en": 6.85753, "de": 6.93086, "vi": 7.16767, "tr": 7.15065, "fi": 6.91692, "ru": 6.36792, "el": 7.32272, "hi": 6.5748, "ta": 6.74937, "th": 6.48906, "zh": 7.74738, "my": 7.2164}, "gini_cpc": 0.418163, "gini_cps": 0.365459, "bpic": 3.867084, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.4970618474461082, "vi": 0.6882131177302612, "tr": 1.5044343830879767, "fi": 1.2193506179803884, "ru": 4.903459366918603, "el": 2.586614959759408, "hi": 0.7420689158402264, "ta": 1.3383842656918967, "th": 1.5459163204797077, "zh": 5.770016986654606, "my": 0.22947255601113278}, "t": 12.5}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5784.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.332356, "cps": 45.1875, "mean_p": 0.496389, "boundary_entropy": 0.687223, "bpb": 6.378213}, "de": {"chunks": 7940.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.386695, "cps": 62.03125, "mean_p": 0.498177, "boundary_entropy": 0.688173, "bpb": 6.459351}, "vi": {"chunks": 11220.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.608625, "cps": 87.65625, "mean_p": 0.499904, "boundary_entropy": 0.688483, "bpb": 6.771814}, "tr": {"chunks": 11095.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.612882, "cps": 86.679688, "mean_p": 0.506386, "boundary_entropy": 0.687558, "bpb": 6.734067}, "fi": {"chunks": 5799.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.308277, "cps": 45.304688, "mean_p": 0.494583, "boundary_entropy": 0.687785, "bpb": 6.467618}, "ru": {"chunks": 228.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.012032, "cps": 1.78125, "mean_p": 0.462503, "boundary_entropy": 0.687137, "bpb": 6.192282}, "el": {"chunks": 12884.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.627783, "cps": 100.65625, "mean_p": 0.495534, "boundary_entropy": 0.690206, "bpb": 6.2473}, "hi": {"chunks": 14922.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.886895, "cps": 116.578125, "mean_p": 0.49853, "boundary_entropy": 0.690707, "bpb": 5.728922}, "ta": {"chunks": 16835.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.900508, "cps": 131.523438, "mean_p": 0.49778, "boundary_entropy": 0.691142, "bpb": 5.925473}, "th": {"chunks": 3362.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.209015, "cps": 26.265625, "mean_p": 0.483367, "boundary_entropy": 0.690305, "bpb": 5.672709}, "zh": {"chunks": 3547.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.600576, "cps": 27.710938, "mean_p": 0.483364, "boundary_entropy": 0.685424, "bpb": 7.347754}, "my": {"chunks": 1968.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.103863, "cps": 15.375, "mean_p": 0.482495, "boundary_entropy": 0.690587, "bpb": 5.910207}}, "bpb": {"en": 6.37821, "de": 6.45935, "vi": 6.77181, "tr": 6.73407, "fi": 6.46762, "ru": 6.19228, "el": 6.2473, "hi": 5.72892, "ta": 5.92547, "th": 5.67271, "zh": 7.34775, "my": 5.91021}, "gini_cpc": 0.332731, "gini_cps": 0.368224, "bpic": 4.015118, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1634934190543993, "vi": 1.8312412007198107, "tr": 1.8440495695103836, "fi": 0.9275494264744452, "ru": 0.036201075115777484, "el": 1.888885822464544, "hi": 2.668503634558828, "ta": 2.7094646370709548, "th": 0.6288868008002095, "zh": 1.8070225897287095, "my": 0.31250541994381653}, "t": 14.9}
{"event": "train", "step": 50, "phase": 1, "objective": "global", "lm_loss": 4.26344, "bpb": 6.15085, "chunk_loss": 0.029955, "F": 0.1636, "G": 0.48516, "chunks": 167.53, "bytes_per_chunk": 6.112, "boundary_entropy": 0.69148, "router_noise": 0.0, "grad_norm": 1.5514, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 16.2}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2984.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.171465, "cps": 23.3125, "mean_p": 0.487159, "boundary_entropy": 0.686885, "bpb": 5.798923}, "de": {"chunks": 4611.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.224565, "cps": 36.023438, "mean_p": 0.489878, "boundary_entropy": 0.687949, "bpb": 5.925155}, "vi": {"chunks": 3807.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.206509, "cps": 29.742188, "mean_p": 0.481383, "boundary_entropy": 0.68772, "bpb": 6.240455}, "tr": {"chunks": 2996.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.165497, "cps": 23.40625, "mean_p": 0.482768, "boundary_entropy": 0.686984, "bpb": 6.225466}, "fi": {"chunks": 1371.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.072883, "cps": 10.710938, "mean_p": 0.47853, "boundary_entropy": 0.686839, "bpb": 5.91473}, "ru": {"chunks": 7816.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.412454, "cps": 61.0625, "mean_p": 0.494167, "boundary_entropy": 0.690247, "bpb": 5.485012}, "el": {"chunks": 7523.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.366564, "cps": 58.773438, "mean_p": 0.491954, "boundary_entropy": 0.690328, "bpb": 5.660218}, "hi": {"chunks": 12722.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.756137, "cps": 99.390625, "mean_p": 0.497155, "boundary_entropy": 0.690745, "bpb": 5.197064}, "ta": {"chunks": 9992.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.534474, "cps": 78.0625, "mean_p": 0.492377, "boundary_entropy": 0.691029, "bpb": 5.249779}, "th": {"chunks": 7496.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.466024, "cps": 58.5625, "mean_p": 0.484489, "boundary_entropy": 0.690229, "bpb": 5.04117}, "zh": {"chunks": 4159.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.704199, "cps": 32.492188, "mean_p": 0.49186, "boundary_entropy": 0.686699, "bpb": 6.955329}, "my": {"chunks": 7819.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.412656, "cps": 61.085938, "mean_p": 0.493888, "boundary_entropy": 0.691268, "bpb": 5.092866}}, "bpb": {"en": 5.79892, "de": 5.92515, "vi": 6.24045, "tr": 6.22547, "fi": 5.91473, "ru": 5.48501, "el": 5.66022, "hi": 5.19706, "ta": 5.24978, "th": 5.04117, "zh": 6.95533, "my": 5.09287}, "gini_cpc": 0.311806, "gini_cps": 0.289597, "bpic": 5.236043, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 1.3096885070511617, "vi": 1.2043841633272763, "tr": 0.9651983237414622, "fi": 0.4250606213665127, "ru": 2.4054738376000055, "el": 2.137841572242429, "hi": 4.409868302068686, "ta": 3.1171109248433475, "th": 2.7179021301880675, "zh": 4.106962894552638, "my": 2.40665112603932}, "t": 18.0}
{"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.40333, "bpb": 4.90996, "chunk_loss": 0.03185, "F": 0.1936, "G": 0.47421, "chunks": 198.25, "bytes_per_chunk": 5.165, "boundary_entropy": 0.68907, "router_noise": 0.0, "grad_norm": 1.6357, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 19.9}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3093.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.177728, "cps": 24.164062, "mean_p": 0.427349, "boundary_entropy": 0.664123, "bpb": 4.466943}, "de": {"chunks": 3566.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.173672, "cps": 27.859375, "mean_p": 0.433066, "boundary_entropy": 0.668789, "bpb": 4.652957}, "vi": {"chunks": 6163.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.33431, "cps": 48.148438, "mean_p": 0.464256, "boundary_entropy": 0.677711, "bpb": 5.03644}, "tr": {"chunks": 3707.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204773, "cps": 28.960938, "mean_p": 0.445793, "boundary_entropy": 0.673791, "bpb": 5.005885}, "fi": {"chunks": 2343.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.124555, "cps": 18.304688, "mean_p": 0.433794, "boundary_entropy": 0.671244, "bpb": 4.716292}, "ru": {"chunks": 6720.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.354617, "cps": 52.5, "mean_p": 0.481292, "boundary_entropy": 0.687204, "bpb": 3.753573}, "el": {"chunks": 2897.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.141159, "cps": 22.632812, "mean_p": 0.468556, "boundary_entropy": 0.687575, "bpb": 4.421478}, "hi": {"chunks": 9212.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.547519, "cps": 71.96875, "mean_p": 0.489735, "boundary_entropy": 0.68982, "bpb": 3.60914}, "ta": {"chunks": 6771.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.362182, "cps": 52.898438, "mean_p": 0.480963, "boundary_entropy": 0.689876, "bpb": 3.66532}, "th": {"chunks": 7479.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.464967, "cps": 58.429688, "mean_p": 0.470704, "boundary_entropy": 0.687677, "bpb": 3.404458}, "zh": {"chunks": 3381.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.572469, "cps": 26.414062, "mean_p": 0.48703, "boundary_entropy": 0.686258, "bpb": 6.40163}, "my": {"chunks": 5594.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.295229, "cps": 43.703125, "mean_p": 0.490588, "boundary_entropy": 0.691054, "bpb": 4.269001}}, "bpb": {"en": 4.46694, "de": 4.65296, "vi": 5.03644, "tr": 5.00588, "fi": 4.71629, "ru": 3.75357, "el": 4.42148, "hi": 3.60914, "ta": 3.66532, "th": 3.40446, "zh": 6.40163, "my": 4.269}, "gini_cpc": 0.267201, "gini_cps": 0.23104, "bpic": 6.299133, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9771767649181324, "vi": 1.881019188976815, "tr": 1.1521691294126672, "fi": 0.7008169632261619, "ru": 1.995281881719467, "el": 0.7942401732994512, "hi": 3.0806549419698124, "ta": 2.0378468596794375, "th": 2.6161742585386136, "zh": 3.221038592655778, "my": 1.6611287167677449}, "t": 22.2}
{"event": "train", "step": 150, "phase": 1, "objective": "global", "lm_loss": 2.69976, "bpb": 3.89493, "chunk_loss": 0.029233, "F": 0.15445, "G": 0.434, "chunks": 158.16, "bytes_per_chunk": 6.475, "boundary_entropy": 0.66837, "router_noise": 0.0, "grad_norm": 1.8398, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.069, "alpha": 0.03, "t": 23.3}
{"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3571.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.205195, "cps": 27.898438, "mean_p": 0.357025, "boundary_entropy": 0.595504, "bpb": 3.876998}, "de": {"chunks": 4726.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.230166, "cps": 36.921875, "mean_p": 0.374402, "boundary_entropy": 0.613628, "bpb": 4.10121}, "vi": {"chunks": 4978.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.27003, "cps": 38.890625, "mean_p": 0.384122, "boundary_entropy": 0.624064, "bpb": 3.631007}, "tr": {"chunks": 4737.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.261669, "cps": 37.007812, "mean_p": 0.392862, "boundary_entropy": 0.62662, "bpb": 4.221434}, "fi": {"chunks": 4100.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.217958, "cps": 32.03125, "mean_p": 0.388186, "boundary_entropy": 0.625976, "bpb": 4.109701}, "ru": {"chunks": 6255.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.330079, "cps": 48.867188, "mean_p": 0.48923, "boundary_entropy": 0.683334, "bpb": 2.854165}, "el": {"chunks": 6994.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.340788, "cps": 54.640625, "mean_p": 0.472616, "boundary_entropy": 0.687172, "bpb": 3.043674}, "hi": {"chunks": 13589.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.807667, "cps": 106.164062, "mean_p": 0.451772, "boundary_entropy": 0.663516, "bpb": 2.460094}, "ta": {"chunks": 10059.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.538058, "cps": 78.585938, "mean_p": 0.46186, "boundary_entropy": 0.681375, "bpb": 2.321479}, "th": {"chunks": 13126.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.81604, "cps": 102.546875, "mean_p": 0.465399, "boundary_entropy": 0.684306, "bpb": 2.349698}, "zh": {"chunks": 3423.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.57958, "cps": 26.742188, "mean_p": 0.470954, "boundary_entropy": 0.682064, "bpb": 5.352389}, "my": {"chunks": 11780.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.621701, "cps": 92.03125, "mean_p": 0.461842, "boundary_entropy": 0.685928, "bpb": 2.543923}}, "bpb": {"en": 3.877, "de": 4.10121, "vi": 3.63101, "tr": 4.22143, "fi": 4.1097, "ru": 2.85416, "el": 3.04367, "hi": 2.46009, "ta": 2.32148, "th": 2.3497, "zh": 5.35239, "my": 2.54392}, "gini_cpc": 0.274191, "gini_cps": 0.273167, "bpic": 4.394204, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.121697056283972, "vi": 1.3159700954187734, "tr": 1.2752258371319423, "fi": 1.0621998684284195, "ru": 1.6086159468423809, "el": 1.6608065647310508, "hi": 3.936105186114161, "ta": 2.62218669021775, "th": 3.976908552619724, "zh": 2.824539981032062, "my": 3.029815509463415}, "t": 26.5}
{"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.87875, "de": 4.10259, "vi": 3.63823, "tr": 4.23074, "fi": 4.11158, "ru": 2.83876, "el": 3.03552, "hi": 2.45511, "ta": 2.33806, "th": 2.3837, "zh": 5.35967, "my": 2.56011}, "bytes_per_chunk": {"en": 4.761, "de": 4.524, "vi": 4.816, "tr": 4.143, "fi": 4.757, "ru": 4.865, "el": 5.103, "hi": 3.12, "ta": 4.948, "th": 3.86, "zh": 5.678, "my": 4.412}, "bpb_mean": 3.41107, "t": 27.9}
{"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.52154, "bpb": 3.63782, "chunk_loss": 0.032632, "F": 0.21625, "G": 0.41544, "chunks": 221.44, "bytes_per_chunk": 4.624, "boundary_entropy": 0.64155, "router_noise": 0.0, "grad_norm": 1.2787, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 28.0}
{"event": "train", "step": 250, "phase": 1, "objective": "global", "lm_loss": 2.47841, "bpb": 3.57559, "chunk_loss": 0.031137, "F": 0.19144, "G": 0.40579, "chunks": 196.03, "bytes_per_chunk": 5.224, "boundary_entropy": 0.64378, "router_noise": 0.0, "grad_norm": 1.4762, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.045, "alpha": 0.03, "t": 30.2}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2410.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.138482, "cps": 18.828125, "mean_p": 0.349514, "boundary_entropy": 0.596026, "bpb": 3.647878}, "de": {"chunks": 3948.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.192276, "cps": 30.84375, "mean_p": 0.357817, "boundary_entropy": 0.601509, "bpb": 3.876854}, "vi": {"chunks": 2951.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.160076, "cps": 23.054688, "mean_p": 0.381119, "boundary_entropy": 0.630071, "bpb": 3.261717}, "tr": {"chunks": 3554.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196321, "cps": 27.765625, "mean_p": 0.360602, "boundary_entropy": 0.604053, "bpb": 3.875046}, "fi": {"chunks": 2967.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157727, "cps": 23.179688, "mean_p": 0.340028, "boundary_entropy": 0.591503, "bpb": 3.912374}, "ru": {"chunks": 4732.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.24971, "cps": 36.96875, "mean_p": 0.472308, "boundary_entropy": 0.681855, "bpb": 2.65579}, "el": {"chunks": 5440.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.265068, "cps": 42.5, "mean_p": 0.462298, "boundary_entropy": 0.684057, "bpb": 2.74678}, "hi": {"chunks": 10641.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.632452, "cps": 83.132812, "mean_p": 0.394554, "boundary_entropy": 0.617825, "bpb": 2.103636}, "ta": {"chunks": 14920.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.798074, "cps": 116.5625, "mean_p": 0.386711, "boundary_entropy": 0.61155, "bpb": 1.941335}, "th": {"chunks": 8801.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.547156, "cps": 68.757812, "mean_p": 0.389299, "boundary_entropy": 0.644851, "bpb": 2.064194}, "zh": {"chunks": 4094.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.693193, "cps": 31.984375, "mean_p": 0.415315, "boundary_entropy": 0.645979, "bpb": 4.980676}, "my": {"chunks": 12171.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.642337, "cps": 95.085938, "mean_p": 0.417715, "boundary_entropy": 0.663536, "bpb": 2.253245}}, "bpb": {"en": 3.64788, "de": 3.87685, "vi": 3.26172, "tr": 3.87505, "fi": 3.91237, "ru": 2.65579, "el": 2.74678, "hi": 2.10364, "ta": 1.94133, "th": 2.06419, "zh": 4.98068, "my": 2.25324}, "gini_cpc": 0.333829, "gini_cps": 0.332399, "bpic": 5.0083, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.3884550181642625, "vi": 1.155934285203029, "tr": 1.4176660845512181, "fi": 1.138971194350617, "ru": 1.803194604714306, "el": 1.9141022429445114, "hi": 4.5670361359614775, "ta": 5.763024040643703, "th": 3.951100107442696, "zh": 5.0056614484461255, "my": 4.638418861141696}, "t": 31.5}
{"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.31848, "bpb": 3.34486, "chunk_loss": 0.032338, "F": 0.21802, "G": 0.38534, "chunks": 223.25, "bytes_per_chunk": 4.587, "boundary_entropy": 0.6155, "router_noise": 0.0, "grad_norm": 1.7269, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.066, "alpha": 0.03, "t": 33.5}
{"event": "train", "step": 350, "phase": 1, "objective": "global", "lm_loss": 2.09305, "bpb": 3.01964, "chunk_loss": 0.031542, "F": 0.1958, "G": 0.36968, "chunks": 200.5, "bytes_per_chunk": 5.107, "boundary_entropy": 0.59279, "router_noise": 0.0, "grad_norm": 0.9203, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 35.6}
{"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2997.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.172212, "cps": 23.414062, "mean_p": 0.332503, "boundary_entropy": 0.556356, "bpb": 3.211345}, "de": {"chunks": 3932.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.191497, "cps": 30.71875, "mean_p": 0.362919, "boundary_entropy": 0.586988, "bpb": 3.516883}, "vi": {"chunks": 4388.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.238025, "cps": 34.28125, "mean_p": 0.375008, "boundary_entropy": 0.608562, "bpb": 2.798211}, "tr": {"chunks": 3818.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210904, "cps": 29.828125, "mean_p": 0.352709, "boundary_entropy": 0.559617, "bpb": 3.448812}, "fi": {"chunks": 4131.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.219606, "cps": 32.273438, "mean_p": 0.354608, "boundary_entropy": 0.55846, "bpb": 3.594341}, "ru": {"chunks": 6133.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.323641, "cps": 47.914062, "mean_p": 0.328707, "boundary_entropy": 0.532732, "bpb": 2.504748}, "el": {"chunks": 7325.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.356917, "cps": 57.226562, "mean_p": 0.317212, "boundary_entropy": 0.498972, "bpb": 2.548584}, "hi": {"chunks": 8195.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.487073, "cps": 64.023438, "mean_p": 0.351999, "boundary_entropy": 0.573467, "bpb": 1.783133}, "ta": {"chunks": 10492.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.56122, "cps": 81.96875, "mean_p": 0.279507, "boundary_entropy": 0.443932, "bpb": 1.583469}, "th": {"chunks": 7685.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.477774, "cps": 60.039062, "mean_p": 0.279225, "boundary_entropy": 0.512887, "bpb": 1.896794}, "zh": {"chunks": 2720.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.460549, "cps": 21.25, "mean_p": 0.334484, "boundary_entropy": 0.572466, "bpb": 4.145905}, "my": {"chunks": 9593.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.50628, "cps": 74.945312, "mean_p": 0.428441, "boundary_entropy": 0.670758, "bpb": 2.016529}}, "bpb": {"en": 3.21135, "de": 3.51688, "vi": 2.79821, "tr": 3.44881, "fi": 3.59434, "ru": 2.50475, "el": 2.54858, "hi": 1.78313, "ta": 1.58347, "th": 1.89679, "zh": 4.1459, "my": 2.01653}, "gini_cpc": 0.218692, "gini_cps": 0.239917, "bpic": 5.374407, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1119838486516516, "vi": 1.3821680644571883, "tr": 1.224680350681566, "fi": 1.2752070022284259, "ru": 1.879321696366551, "el": 2.0725459172567295, "hi": 2.828337698768605, "ta": 3.2588936623170377, "th": 2.774343196164769, "zh": 2.6743167142760775, "my": 2.939872160528694}, "t": 38.6}
{"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 3.21814, "de": 3.51506, "vi": 2.80467, "tr": 3.45728, "fi": 3.58667, "ru": 2.4909, "el": 2.54222, "hi": 1.77932, "ta": 1.60076, "th": 1.9267, "zh": 4.15317, "my": 2.03248}, "bytes_per_chunk": {"en": 5.919, "de": 5.592, "vi": 5.687, "tr": 4.764, "fi": 4.858, "ru": 5.864, "el": 5.012, "hi": 4.709, "ta": 4.775, "th": 5.667, "zh": 6.45, "my": 4.883}, "bpb_mean": 2.75895, "t": 40.0}
{"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.22697, "bpb": 3.21284, "chunk_loss": 0.030172, "F": 0.17007, "G": 0.33393, "chunks": 174.16, "bytes_per_chunk": 5.88, "boundary_entropy": 0.56086, "router_noise": 0.0, "grad_norm": 0.8055, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 40.0}
{"event": "train", "step": 450, "phase": 1, "objective": "global", "lm_loss": 2.17496, "bpb": 3.13781, "chunk_loss": 0.030666, "F": 0.18341, "G": 0.32641, "chunks": 187.81, "bytes_per_chunk": 5.452, "boundary_entropy": 0.53817, "router_noise": 0.0, "grad_norm": 1.2743, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 42.1}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 2.05747, "bpb": 2.96831, "chunk_loss": 0.030329, "F": 0.17584, "G": 0.30893, "chunks": 180.06, "bytes_per_chunk": 5.687, "boundary_entropy": 0.52017, "router_noise": 0.0, "grad_norm": 0.8166, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 44.1}
{"event": "train", "step": 550, "phase": 1, "objective": "global", "lm_loss": 1.90738, "bpb": 2.75177, "chunk_loss": 0.030197, "F": 0.17261, "G": 0.31807, "chunks": 176.75, "bytes_per_chunk": 5.793, "boundary_entropy": 0.51813, "router_noise": 0.0, "grad_norm": 0.7479, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 46.1}
{"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3393.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194966, "cps": 26.507812, "mean_p": 0.290497, "boundary_entropy": 0.436523, "bpb": 2.890352}, "de": {"chunks": 3901.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.189987, "cps": 30.476562, "mean_p": 0.285183, "boundary_entropy": 0.453366, "bpb": 3.212867}, "vi": {"chunks": 4439.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.240792, "cps": 34.679688, "mean_p": 0.339354, "boundary_entropy": 0.542032, "bpb": 2.582228}, "tr": {"chunks": 3725.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205767, "cps": 29.101562, "mean_p": 0.291995, "boundary_entropy": 0.45804, "bpb": 3.153803}, "fi": {"chunks": 3700.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.196693, "cps": 28.90625, "mean_p": 0.290607, "boundary_entropy": 0.442819, "bpb": 3.362949}, "ru": {"chunks": 5365.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.283113, "cps": 41.914062, "mean_p": 0.341484, "boundary_entropy": 0.572857, "bpb": 2.160143}, "el": {"chunks": 7143.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.348049, "cps": 55.804688, "mean_p": 0.363224, "boundary_entropy": 0.596308, "bpb": 2.216286}, "hi": {"chunks": 6685.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.397325, "cps": 52.226562, "mean_p": 0.321486, "boundary_entropy": 0.555696, "bpb": 1.6319}, "ta": {"chunks": 7257.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.388179, "cps": 56.695312, "mean_p": 0.26847, "boundary_entropy": 0.484208, "bpb": 1.360352}, "th": {"chunks": 6442.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.400497, "cps": 50.328125, "mean_p": 0.342506, "boundary_entropy": 0.599431, "bpb": 1.754275}, "zh": {"chunks": 3163.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.535557, "cps": 24.710938, "mean_p": 0.390144, "boundary_entropy": 0.592568, "bpb": 3.835012}, "my": {"chunks": 9081.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.479259, "cps": 70.945312, "mean_p": 0.317578, "boundary_entropy": 0.552418, "bpb": 1.661812}}, "bpb": {"en": 2.89035, "de": 3.21287, "vi": 2.58223, "tr": 3.1538, "fi": 3.36295, "ru": 2.16014, "el": 2.21629, "hi": 1.6319, "ta": 1.36035, "th": 1.75428, "zh": 3.83501, "my": 1.66181}, "gini_cpc": 0.199719, "gini_cps": 0.190951, "bpic": 5.969157, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9744595219969596, "vi": 1.2350435264102109, "tr": 1.055397318019059, "fi": 1.0088581370241587, "ru": 1.4521142006630132, "el": 1.7851719963278565, "hi": 2.0379175025585843, "ta": 1.9910029987166598, "th": 2.0541867131043166, "zh": 2.746919989951623, "my": 2.4581623362333542}, "t": 49.1}
{"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.90116, "de": 3.21262, "vi": 2.5856, "tr": 3.16354, "fi": 3.35501, "ru": 2.14523, "el": 2.20499, "hi": 1.6247, "ta": 1.37304, "th": 1.78112, "zh": 3.84555, "my": 1.67483}, "bytes_per_chunk": {"en": 5.323, "de": 5.417, "vi": 6.018, "tr": 5.254, "fi": 5.453, "ru": 7.22, "el": 5.187, "hi": 6.496, "ta": 7.272, "th": 7.64, "zh": 5.254, "my": 6.813}, "bpb_mean": 2.48895, "t": 50.4}
{"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.90408, "bpb": 2.74701, "chunk_loss": 0.030303, "F": 0.17395, "G": 0.31824, "chunks": 178.12, "bytes_per_chunk": 5.749, "boundary_entropy": 0.50763, "router_noise": 0.0, "grad_norm": 1.1161, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 50.4}
{"event": "train", "step": 650, "phase": 1, "objective": "global", "lm_loss": 1.93671, "bpb": 2.79409, "chunk_loss": 0.030395, "F": 0.17993, "G": 0.31889, "chunks": 184.25, "bytes_per_chunk": 5.558, "boundary_entropy": 0.50113, "router_noise": 0.0, "grad_norm": 0.7916, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 52.4}
{"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.77509, "bpb": 2.56091, "chunk_loss": 0.030221, "F": 0.17593, "G": 0.30555, "chunks": 180.16, "bytes_per_chunk": 5.684, "boundary_entropy": 0.47442, "router_noise": 0.0, "grad_norm": 0.8429, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 54.5}
{"event": "train", "step": 750, "phase": 1, "objective": "global", "lm_loss": 1.85773, "bpb": 2.68014, "chunk_loss": 0.030646, "F": 0.18527, "G": 0.30664, "chunks": 189.72, "bytes_per_chunk": 5.397, "boundary_entropy": 0.46927, "router_noise": 0.0, "grad_norm": 0.8483, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 56.5}
{"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3086.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.177326, "cps": 24.109375, "mean_p": 0.25096, "boundary_entropy": 0.357351, "bpb": 2.681516}, "de": {"chunks": 3581.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174402, "cps": 27.976562, "mean_p": 0.252968, "boundary_entropy": 0.389603, "bpb": 3.041525}, "vi": {"chunks": 4513.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.244806, "cps": 35.257812, "mean_p": 0.330015, "boundary_entropy": 0.508389, "bpb": 2.421464}, "tr": {"chunks": 3365.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185881, "cps": 26.289062, "mean_p": 0.259018, "boundary_entropy": 0.402631, "bpb": 3.03595}, "fi": {"chunks": 3305.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.175695, "cps": 25.820312, "mean_p": 0.238554, "boundary_entropy": 0.362948, "bpb": 3.227507}, "ru": {"chunks": 6723.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.354776, "cps": 52.523438, "mean_p": 0.36588, "boundary_entropy": 0.595884, "bpb": 1.991873}, "el": {"chunks": 7441.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.362569, "cps": 58.132812, "mean_p": 0.372141, "boundary_entropy": 0.608435, "bpb": 2.099828}, "hi": {"chunks": 7962.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.473224, "cps": 62.203125, "mean_p": 0.276658, "boundary_entropy": 0.468265, "bpb": 1.526191}, "ta": {"chunks": 9296.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.497245, "cps": 72.625, "mean_p": 0.328787, "boundary_entropy": 0.52408, "bpb": 1.27045}, "th": {"chunks": 8828.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.548834, "cps": 68.96875, "mean_p": 0.407945, "boundary_entropy": 0.632651, "bpb": 1.585552}, "zh": {"chunks": 3099.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.524721, "cps": 24.210938, "mean_p": 0.364251, "boundary_entropy": 0.540179, "bpb": 3.713446}, "my": {"chunks": 8498.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.448491, "cps": 66.390625, "mean_p": 0.381015, "boundary_entropy": 0.622879, "bpb": 1.517124}}, "bpb": {"en": 2.68152, "de": 3.04153, "vi": 2.42146, "tr": 3.03595, "fi": 3.22751, "ru": 1.99187, "el": 2.09983, "hi": 1.52619, "ta": 1.27045, "th": 1.58555, "zh": 3.71345, "my": 1.51712}, "gini_cpc": 0.231027, "gini_cps": 0.230757, "bpic": 5.506421, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9835130171089524, "vi": 1.3805444362239612, "tr": 1.0482447973859097, "fi": 0.9908040630520063, "ru": 2.0007005678893703, "el": 2.0446484981568664, "hi": 2.668672613413017, "ta": 2.8041345215353677, "th": 3.0950627448839416, "zh": 2.959077447739801, "my": 2.52919054242802}, "t": 59.5}
{"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.68939, "de": 3.04721, "vi": 2.42769, "tr": 3.04698, "fi": 3.22552, "ru": 1.97666, "el": 2.08889, "hi": 1.52208, "ta": 1.28259, "th": 1.61251, "zh": 3.7252, "my": 1.52727}, "bytes_per_chunk": {"en": 5.608, "de": 5.768, "vi": 5.596, "tr": 5.787, "fi": 5.602, "ru": 5.555, "el": 4.989, "hi": 5.328, "ta": 5.458, "th": 5.692, "zh": 5.619, "my": 6.115}, "bpb_mean": 2.34767, "t": 60.8}
{"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.83722, "bpb": 2.65054, "chunk_loss": 0.030346, "F": 0.17731, "G": 0.29916, "chunks": 181.56, "bytes_per_chunk": 5.64, "boundary_entropy": 0.45813, "router_noise": 0.0, "grad_norm": 0.9262, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 60.8}
{"event": "train", "step": 850, "phase": 1, "objective": "global", "lm_loss": 1.63446, "bpb": 2.35803, "chunk_loss": 0.029834, "F": 0.16428, "G": 0.29876, "chunks": 168.22, "bytes_per_chunk": 6.087, "boundary_entropy": 0.46715, "router_noise": 0.0, "grad_norm": 0.8286, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 62.9}
{"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.72348, "bpb": 2.48645, "chunk_loss": 0.030028, "F": 0.17108, "G": 0.29436, "chunks": 175.19, "bytes_per_chunk": 5.845, "boundary_entropy": 0.44436, "router_noise": 0.0, "grad_norm": 0.861, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 64.9}
{"event": "train", "step": 950, "phase": 1, "objective": "global", "lm_loss": 1.87142, "bpb": 2.69989, "chunk_loss": 0.030563, "F": 0.18521, "G": 0.28822, "chunks": 189.66, "bytes_per_chunk": 5.399, "boundary_entropy": 0.41958, "router_noise": 0.0, "grad_norm": 0.8247, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 66.9}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3375.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193932, "cps": 26.367188, "mean_p": 0.2548, "boundary_entropy": 0.320475, "bpb": 2.542984}, "de": {"chunks": 4143.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.201773, "cps": 32.367188, "mean_p": 0.263161, "boundary_entropy": 0.355064, "bpb": 2.897384}, "vi": {"chunks": 4512.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.244752, "cps": 35.25, "mean_p": 0.308459, "boundary_entropy": 0.463663, "bpb": 2.293407}, "tr": {"chunks": 4293.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.237143, "cps": 33.539062, "mean_p": 0.274633, "boundary_entropy": 0.361021, "bpb": 2.908095}, "fi": {"chunks": 4075.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.216629, "cps": 31.835938, "mean_p": 0.263406, "boundary_entropy": 0.342188, "bpb": 3.127147}, "ru": {"chunks": 6107.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.322269, "cps": 47.710938, "mean_p": 0.346205, "boundary_entropy": 0.579299, "bpb": 1.911002}, "el": {"chunks": 7177.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.349705, "cps": 56.070312, "mean_p": 0.38079, "boundary_entropy": 0.615873, "bpb": 1.989757}, "hi": {"chunks": 8098.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.481308, "cps": 63.265625, "mean_p": 0.290111, "boundary_entropy": 0.476803, "bpb": 1.459367}, "ta": {"chunks": 9072.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.485263, "cps": 70.875, "mean_p": 0.345612, "boundary_entropy": 0.545243, "bpb": 1.24263}, "th": {"chunks": 11314.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.703388, "cps": 88.390625, "mean_p": 0.442664, "boundary_entropy": 0.639916, "bpb": 1.530698}, "zh": {"chunks": 2844.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.481544, "cps": 22.21875, "mean_p": 0.340081, "boundary_entropy": 0.502435, "bpb": 3.618559}, "my": {"chunks": 16055.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.847319, "cps": 125.429688, "mean_p": 0.459532, "boundary_entropy": 0.652461, "bpb": 1.414505}}, "bpb": {"en": 2.54298, "de": 2.89738, "vi": 2.29341, "tr": 2.90809, "fi": 3.12715, "ru": 1.911, "el": 1.98976, "hi": 1.45937, "ta": 1.24263, "th": 1.5307, "zh": 3.61856, "my": 1.4145}, "gini_cpc": 0.271796, "gini_cps": 0.289657, "bpic": 4.734238, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.040430006980633, "vi": 1.2620492179730587, "tr": 1.2228147820803181, "fi": 1.1170331779868754, "ru": 1.661762861330988, "el": 1.803235481059191, "hi": 2.4818357866930825, "ta": 2.50223396630115, "th": 3.626982433599282, "zh": 2.4830558753809684, "my": 4.36915323810194}, "t": 69.9}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.55213, "de": 2.90044, "vi": 2.29762, "tr": 2.91846, "fi": 3.12392, "ru": 1.89456, "el": 1.97945, "hi": 1.45618, "ta": 1.2512, "th": 1.55572, "zh": 3.62661, "my": 1.42201}, "bytes_per_chunk": {"en": 5.261, "de": 5.032, "vi": 5.782, "tr": 4.589, "fi": 4.654, "ru": 5.932, "el": 5.358, "hi": 5.172, "ta": 5.648, "th": 4.167, "zh": 5.779, "my": 3.309}, "bpb_mean": 2.24819, "t": 71.2}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.543, "bpb": 2.22608, "chunk_loss": 0.030883, "F": 0.18976, "G": 0.31591, "chunks": 194.31, "bytes_per_chunk": 5.27, "boundary_entropy": 0.46784, "router_noise": 0.0, "grad_norm": 0.7526, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 71.3}
{"event": "train", "step": 1050, "phase": 1, "objective": "global", "lm_loss": 1.76876, "bpb": 2.55178, "chunk_loss": 0.030323, "F": 0.17984, "G": 0.28434, "chunks": 184.16, "bytes_per_chunk": 5.56, "boundary_entropy": 0.41806, "router_noise": 0.0, "grad_norm": 0.9583, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 73.3}
{"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.58719, "bpb": 2.28983, "chunk_loss": 0.030089, "F": 0.17175, "G": 0.28422, "chunks": 175.88, "bytes_per_chunk": 5.822, "boundary_entropy": 0.41874, "router_noise": 0.0, "grad_norm": 0.8269, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 75.3}
{"event": "train", "step": 1150, "phase": 1, "objective": "global", "lm_loss": 1.61171, "bpb": 2.3252, "chunk_loss": 0.029863, "F": 0.16934, "G": 0.29091, "chunks": 173.41, "bytes_per_chunk": 5.905, "boundary_entropy": 0.42521, "router_noise": 0.0, "grad_norm": 0.984, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 77.4}
{"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3207.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184279, "cps": 25.054688, "mean_p": 0.23852, "boundary_entropy": 0.300331, "bpb": 2.459775}, "de": {"chunks": 3748.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.182535, "cps": 29.28125, "mean_p": 0.240066, "boundary_entropy": 0.333523, "bpb": 2.798222}, "vi": {"chunks": 4515.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.244915, "cps": 35.273438, "mean_p": 0.297602, "boundary_entropy": 0.445162, "bpb": 2.247722}, "tr": {"chunks": 3978.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.219743, "cps": 31.078125, "mean_p": 0.25687, "boundary_entropy": 0.341583, "bpb": 2.822566}, "fi": {"chunks": 3570.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189783, "cps": 27.890625, "mean_p": 0.231087, "boundary_entropy": 0.30572, "bpb": 3.042068}, "ru": {"chunks": 6523.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.344222, "cps": 50.960938, "mean_p": 0.340696, "boundary_entropy": 0.561752, "bpb": 1.830406}, "el": {"chunks": 6367.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.310237, "cps": 49.742188, "mean_p": 0.381008, "boundary_entropy": 0.616568, "bpb": 1.914678}, "hi": {"chunks": 6956.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.413432, "cps": 54.34375, "mean_p": 0.293387, "boundary_entropy": 0.519858, "bpb": 1.425755}, "ta": {"chunks": 8019.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.428938, "cps": 62.648438, "mean_p": 0.40219, "boundary_entropy": 0.609351, "bpb": 1.169076}, "th": {"chunks": 7805.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.485235, "cps": 60.976562, "mean_p": 0.426082, "boundary_entropy": 0.646746, "bpb": 1.459327}, "zh": {"chunks": 2837.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.480359, "cps": 22.164062, "mean_p": 0.316509, "boundary_entropy": 0.472743, "bpb": 3.579431}, "my": {"chunks": 9653.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.509447, "cps": 75.414062, "mean_p": 0.437386, "boundary_entropy": 0.658469, "bpb": 1.366959}}, "bpb": {"en": 2.45978, "de": 2.79822, "vi": 2.24772, "tr": 2.82257, "fi": 3.04207, "ru": 1.83041, "el": 1.91468, "hi": 1.42575, "ta": 1.16908, "th": 1.45933, "zh": 3.57943, "my": 1.36696}, "gini_cpc": 0.20711, "gini_cps": 0.21517, "bpic": 5.712897, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9905407239444847, "vi": 1.3290452663683465, "tr": 1.192447830099327, "fi": 1.0298678315471228, "ru": 1.8679417303671966, "el": 1.6835234301226845, "hi": 2.243518529072038, "ta": 2.327661933689618, "th": 2.6331585040068646, "zh": 2.6067000089015404, "my": 2.764547716926222}, "t": 80.3}
{"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.46796, "de": 2.80103, "vi": 2.25266, "tr": 2.83123, "fi": 3.0439, "ru": 1.81369, "el": 1.90461, "hi": 1.42252, "ta": 1.18005, "th": 1.48617, "zh": 3.58638, "my": 1.37477}, "bytes_per_chunk": {"en": 5.58, "de": 5.507, "vi": 5.967, "tr": 4.955, "fi": 5.187, "ru": 5.567, "el": 6.193, "hi": 6.114, "ta": 6.616, "th": 6.286, "zh": 5.803, "my": 5.641}, "bpb_mean": 2.18041, "t": 81.6}
{"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.69267, "bpb": 2.44201, "chunk_loss": 0.030227, "F": 0.17743, "G": 0.27761, "chunks": 181.69, "bytes_per_chunk": 5.636, "boundary_entropy": 0.40618, "router_noise": 0.0, "grad_norm": 0.7948, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 81.6}
{"event": "train", "step": 1250, "phase": 1, "objective": "global", "lm_loss": 1.60079, "bpb": 2.30945, "chunk_loss": 0.030463, "F": 0.18027, "G": 0.29779, "chunks": 184.59, "bytes_per_chunk": 5.547, "boundary_entropy": 0.43949, "router_noise": 0.0, "grad_norm": 0.9406, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 83.7}
{"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.716, "bpb": 2.47566, "chunk_loss": 0.03015, "F": 0.17706, "G": 0.29466, "chunks": 181.31, "bytes_per_chunk": 5.648, "boundary_entropy": 0.42019, "router_noise": 0.0, "grad_norm": 0.8914, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 85.7}
{"event": "train", "step": 1350, "phase": 1, "objective": "global", "lm_loss": 1.50462, "bpb": 2.17071, "chunk_loss": 0.03007, "F": 0.17059, "G": 0.28402, "chunks": 174.69, "bytes_per_chunk": 5.862, "boundary_entropy": 0.42336, "router_noise": 0.0, "grad_norm": 0.7863, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 87.7}
{"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3361.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193128, "cps": 26.257812, "mean_p": 0.24003, "boundary_entropy": 0.274771, "bpb": 2.397381}, "de": {"chunks": 3998.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.194711, "cps": 31.234375, "mean_p": 0.243375, "boundary_entropy": 0.303095, "bpb": 2.728631}, "vi": {"chunks": 4389.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.23808, "cps": 34.289062, "mean_p": 0.29839, "boundary_entropy": 0.438546, "bpb": 2.198585}, "tr": {"chunks": 3991.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.220461, "cps": 31.179688, "mean_p": 0.247748, "boundary_entropy": 0.310519, "bpb": 2.713516}, "fi": {"chunks": 3514.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186806, "cps": 27.453125, "mean_p": 0.222277, "boundary_entropy": 0.275249, "bpb": 2.964964}, "ru": {"chunks": 6223.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.328391, "cps": 48.617188, "mean_p": 0.331611, "boundary_entropy": 0.551058, "bpb": 1.807458}, "el": {"chunks": 5123.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.249622, "cps": 40.023438, "mean_p": 0.350803, "boundary_entropy": 0.599117, "bpb": 1.860995}, "hi": {"chunks": 6310.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.375037, "cps": 49.296875, "mean_p": 0.336243, "boundary_entropy": 0.567159, "bpb": 1.452209}, "ta": {"chunks": 8650.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.462691, "cps": 67.578125, "mean_p": 0.439466, "boundary_entropy": 0.630621, "bpb": 1.147519}, "th": {"chunks": 10133.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.629966, "cps": 79.164062, "mean_p": 0.452019, "boundary_entropy": 0.648382, "bpb": 1.406365}, "zh": {"chunks": 2887.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.488825, "cps": 22.554688, "mean_p": 0.302163, "boundary_entropy": 0.435743, "bpb": 3.540177}, "my": {"chunks": 7871.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.4154, "cps": 61.492188, "mean_p": 0.419708, "boundary_entropy": 0.650604, "bpb": 1.330662}}, "bpb": {"en": 2.39738, "de": 2.72863, "vi": 2.19858, "tr": 2.71352, "fi": 2.96496, "ru": 1.80746, "el": 1.86099, "hi": 1.45221, "ta": 1.14752, "th": 1.40637, "zh": 3.54018, "my": 1.33066}, "gini_cpc": 0.227549, "gini_cps": 0.221733, "bpic": 5.775485, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0081983685799911, "vi": 1.232758616098264, "tr": 1.141528566502405, "fi": 0.9672650180894777, "ru": 1.7003808076343656, "el": 1.2925254954843952, "hi": 1.9419135587697887, "ta": 2.3957761969077356, "th": 3.2619145888627035, "zh": 2.531097931865818, "my": 2.1509095313206905}, "t": 90.7}
{"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.40597, "de": 2.7332, "vi": 2.20252, "tr": 2.72314, "fi": 2.96805, "ru": 1.78897, "el": 1.85187, "hi": 1.44895, "ta": 1.15488, "th": 1.42781, "zh": 3.54431, "my": 1.33867}, "bytes_per_chunk": {"en": 5.45, "de": 5.252, "vi": 5.944, "tr": 4.964, "fi": 5.431, "ru": 5.62, "el": 7.662, "hi": 6.814, "ta": 6.283, "th": 4.892, "zh": 5.584, "my": 6.539}, "bpb_mean": 2.13236, "t": 91.9}
{"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.70085, "bpb": 2.45381, "chunk_loss": 0.030374, "F": 0.18091, "G": 0.29626, "chunks": 185.25, "bytes_per_chunk": 5.528, "boundary_entropy": 0.42026, "router_noise": 0.0, "grad_norm": 0.7333, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 92.0}
{"event": "train", "step": 1450, "phase": 1, "objective": "global", "lm_loss": 1.70349, "bpb": 2.45761, "chunk_loss": 0.029948, "F": 0.16922, "G": 0.29502, "chunks": 173.28, "bytes_per_chunk": 5.909, "boundary_entropy": 0.43209, "router_noise": 0.0, "grad_norm": 0.8039, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 94.0}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.67621, "bpb": 2.41827, "chunk_loss": 0.030392, "F": 0.17953, "G": 0.27108, "chunks": 183.84, "bytes_per_chunk": 5.57, "boundary_entropy": 0.38335, "router_noise": 0.0, "grad_norm": 0.8966, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 96.1}
{"event": "train", "step": 1550, "phase": 1, "objective": "global", "lm_loss": 1.57079, "bpb": 2.26617, "chunk_loss": 0.030359, "F": 0.18289, "G": 0.28567, "chunks": 187.28, "bytes_per_chunk": 5.468, "boundary_entropy": 0.40339, "router_noise": 0.0, "grad_norm": 0.8461, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 98.1}
{"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3142.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.180544, "cps": 24.546875, "mean_p": 0.222063, "boundary_entropy": 0.256719, "bpb": 2.346986}, "de": {"chunks": 3623.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176448, "cps": 28.304688, "mean_p": 0.225428, "boundary_entropy": 0.287022, "bpb": 2.650366}, "vi": {"chunks": 3981.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.215948, "cps": 31.101562, "mean_p": 0.272802, "boundary_entropy": 0.413913, "bpb": 2.15517}, "tr": {"chunks": 3890.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.214882, "cps": 30.390625, "mean_p": 0.238746, "boundary_entropy": 0.306382, "bpb": 2.657389}, "fi": {"chunks": 3076.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.163521, "cps": 24.03125, "mean_p": 0.201007, "boundary_entropy": 0.261975, "bpb": 2.897092}, "ru": {"chunks": 5781.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.305066, "cps": 45.164062, "mean_p": 0.323582, "boundary_entropy": 0.541637, "bpb": 1.75656}, "el": {"chunks": 6170.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.300638, "cps": 48.203125, "mean_p": 0.364266, "boundary_entropy": 0.599868, "bpb": 1.822233}, "hi": {"chunks": 6477.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.384963, "cps": 50.601562, "mean_p": 0.356881, "boundary_entropy": 0.578922, "bpb": 1.372174}, "ta": {"chunks": 8154.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.436159, "cps": 63.703125, "mean_p": 0.426535, "boundary_entropy": 0.626186, "bpb": 1.11253}, "th": {"chunks": 7438.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.462418, "cps": 58.109375, "mean_p": 0.415701, "boundary_entropy": 0.638187, "bpb": 1.362799}, "zh": {"chunks": 2838.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.480528, "cps": 22.171875, "mean_p": 0.292067, "boundary_entropy": 0.426781, "bpb": 3.489746}, "my": {"chunks": 7686.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.405636, "cps": 60.046875, "mean_p": 0.373728, "boundary_entropy": 0.619879, "bpb": 1.309291}}, "bpb": {"en": 2.34699, "de": 2.65037, "vi": 2.15517, "tr": 2.65739, "fi": 2.89709, "ru": 1.75656, "el": 1.82223, "hi": 1.37217, "ta": 1.11253, "th": 1.3628, "zh": 3.48975, "my": 1.30929}, "gini_cpc": 0.2083, "gini_cps": 0.20473, "bpic": 6.164562, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9773134290694679, "vi": 1.1960985809207674, "tr": 1.1901918975346286, "fi": 0.9057167243146091, "ru": 1.6897081334007378, "el": 1.6651841114871568, "hi": 2.1322433255807574, "ta": 2.4158122387094654, "th": 2.5612563507797037, "zh": 2.66156384244313, "my": 2.2467509994521504}, "t": 101.0}
{"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.35521, "de": 2.65694, "vi": 2.15779, "tr": 2.66575, "fi": 2.90143, "ru": 1.73866, "el": 1.81316, "hi": 1.36852, "ta": 1.12134, "th": 1.38398, "zh": 3.49998, "my": 1.31735}, "bytes_per_chunk": {"en": 5.778, "de": 5.832, "vi": 6.462, "tr": 5.103, "fi": 6.124, "ru": 6.099, "el": 6.254, "hi": 6.583, "ta": 6.408, "th": 6.632, "zh": 5.674, "my": 7.617}, "bpb_mean": 2.08167, "t": 102.3}
{"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.78382, "bpb": 2.57351, "chunk_loss": 0.030173, "F": 0.17294, "G": 0.25111, "chunks": 177.09, "bytes_per_chunk": 5.782, "boundary_entropy": 0.34024, "router_noise": 0.0, "grad_norm": 0.8973, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 102.3}
{"event": "train", "step": 1650, "phase": 1, "objective": "global", "lm_loss": 1.57061, "bpb": 2.26591, "chunk_loss": 0.030042, "F": 0.17203, "G": 0.28361, "chunks": 176.16, "bytes_per_chunk": 5.813, "boundary_entropy": 0.40498, "router_noise": 0.0, "grad_norm": 0.7483, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 104.4}
{"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.56268, "bpb": 2.25447, "chunk_loss": 0.030516, "F": 0.18085, "G": 0.29572, "chunks": 185.19, "bytes_per_chunk": 5.53, "boundary_entropy": 0.42091, "router_noise": 0.0, "grad_norm": 0.8055, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 106.4}
{"event": "train", "step": 1750, "phase": 1, "objective": "global", "lm_loss": 1.51304, "bpb": 2.18285, "chunk_loss": 0.030044, "F": 0.17175, "G": 0.28981, "chunks": 175.88, "bytes_per_chunk": 5.822, "boundary_entropy": 0.41027, "router_noise": 0.0, "grad_norm": 0.8631, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 108.4}
{"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3360.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19307, "cps": 26.25, "mean_p": 0.239611, "boundary_entropy": 0.262791, "bpb": 2.315434}, "de": {"chunks": 3940.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.191886, "cps": 30.78125, "mean_p": 0.240075, "boundary_entropy": 0.289253, "bpb": 2.599726}, "vi": {"chunks": 4241.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.230052, "cps": 33.132812, "mean_p": 0.281734, "boundary_entropy": 0.415115, "bpb": 2.137541}, "tr": {"chunks": 4165.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.230072, "cps": 32.539062, "mean_p": 0.250881, "boundary_entropy": 0.294767, "bpb": 2.600386}, "fi": {"chunks": 3500.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186061, "cps": 27.34375, "mean_p": 0.221477, "boundary_entropy": 0.263784, "bpb": 2.824079}, "ru": {"chunks": 5983.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.315726, "cps": 46.742188, "mean_p": 0.315714, "boundary_entropy": 0.526849, "bpb": 1.710042}, "el": {"chunks": 6832.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.332895, "cps": 53.375, "mean_p": 0.360663, "boundary_entropy": 0.591765, "bpb": 1.774427}, "hi": {"chunks": 6603.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.392452, "cps": 51.585938, "mean_p": 0.367977, "boundary_entropy": 0.583067, "bpb": 1.350851}, "ta": {"chunks": 8330.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.445574, "cps": 65.078125, "mean_p": 0.416858, "boundary_entropy": 0.617923, "bpb": 1.110406}, "th": {"chunks": 9744.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.605782, "cps": 76.125, "mean_p": 0.432342, "boundary_entropy": 0.633828, "bpb": 1.323167}, "zh": {"chunks": 2929.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.495936, "cps": 22.882812, "mean_p": 0.294119, "boundary_entropy": 0.420623, "bpb": 3.466366}, "my": {"chunks": 8609.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.454349, "cps": 67.257812, "mean_p": 0.39324, "boundary_entropy": 0.632244, "bpb": 1.284165}}, "bpb": {"en": 2.31543, "de": 2.59973, "vi": 2.13754, "tr": 2.60039, "fi": 2.82408, "ru": 1.71004, "el": 1.77443, "hi": 1.35085, "ta": 1.11041, "th": 1.32317, "zh": 3.46637, "my": 1.28417}, "gini_cpc": 0.22019, "gini_cps": 0.219254, "bpic": 5.624319, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9938678851465585, "vi": 1.1915436959975203, "tr": 1.1916515908965366, "fi": 0.9636981021742598, "ru": 1.6352894364869959, "el": 1.7242167324465234, "hi": 2.0326896094247506, "ta": 2.3078329767317465, "th": 3.1376251165682314, "zh": 2.568684539935175, "my": 2.353283092448506}, "t": 111.3}
{"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.32376, "de": 2.60586, "vi": 2.14351, "tr": 2.60591, "fi": 2.82749, "ru": 1.69127, "el": 1.76485, "hi": 1.34836, "ta": 1.1166, "th": 1.34232, "zh": 3.47099, "my": 1.29151}, "bytes_per_chunk": {"en": 5.371, "de": 5.347, "vi": 6.242, "tr": 4.748, "fi": 5.309, "ru": 5.873, "el": 5.573, "hi": 6.537, "ta": 6.267, "th": 5.167, "zh": 5.546, "my": 5.718}, "bpb_mean": 2.04437, "t": 112.6}
{"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.45066, "bpb": 2.09286, "chunk_loss": 0.030432, "F": 0.1824, "G": 0.30347, "chunks": 186.78, "bytes_per_chunk": 5.482, "boundary_entropy": 0.43442, "router_noise": 0.0, "grad_norm": 0.7007, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 112.6}
{"event": "train", "step": 1850, "phase": 1, "objective": "global", "lm_loss": 1.44156, "bpb": 2.07973, "chunk_loss": 0.029967, "F": 0.16977, "G": 0.29495, "chunks": 173.84, "bytes_per_chunk": 5.89, "boundary_entropy": 0.42922, "router_noise": 0.0, "grad_norm": 0.7919, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 114.6}
{"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.55304, "bpb": 2.24056, "chunk_loss": 0.030483, "F": 0.18365, "G": 0.27065, "chunks": 188.06, "bytes_per_chunk": 5.445, "boundary_entropy": 0.36055, "router_noise": 0.0, "grad_norm": 0.7707, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 116.6}
{"event": "train", "step": 1950, "phase": 1, "objective": "global", "lm_loss": 1.61101, "bpb": 2.3242, "chunk_loss": 0.030139, "F": 0.17615, "G": 0.28728, "chunks": 180.38, "bytes_per_chunk": 5.677, "boundary_entropy": 0.40135, "router_noise": 0.0, "grad_norm": 0.7541, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 118.7}
{"event": "probe", "step": 2000, "phase": 1, "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": 0.231722, "boundary_entropy": 0.242018, "bpb": 2.275517}, "de": {"chunks": 3938.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.191789, "cps": 30.765625, "mean_p": 0.235032, "boundary_entropy": 0.272851, "bpb": 2.560917}, "vi": {"chunks": 4018.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.217955, "cps": 31.390625, "mean_p": 0.279028, "boundary_entropy": 0.407617, "bpb": 2.101683}, "tr": {"chunks": 3881.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.214384, "cps": 30.320312, "mean_p": 0.231998, "boundary_entropy": 0.279374, "bpb": 2.553571}, "fi": {"chunks": 3374.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.179363, "cps": 26.359375, "mean_p": 0.217493, "boundary_entropy": 0.257569, "bpb": 2.80452}, "ru": {"chunks": 6274.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.331082, "cps": 49.015625, "mean_p": 0.315466, "boundary_entropy": 0.520846, "bpb": 1.675962}, "el": {"chunks": 7018.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.341958, "cps": 54.828125, "mean_p": 0.361308, "boundary_entropy": 0.59122, "bpb": 1.766237}, "hi": {"chunks": 7451.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.442853, "cps": 58.210938, "mean_p": 0.391904, "boundary_entropy": 0.600789, "bpb": 1.308664}, "ta": {"chunks": 8219.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.439636, "cps": 64.210938, "mean_p": 0.409399, "boundary_entropy": 0.614353, "bpb": 1.085058}, "th": {"chunks": 8208.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.510289, "cps": 64.125, "mean_p": 0.419934, "boundary_entropy": 0.633349, "bpb": 1.305409}, "zh": {"chunks": 2938.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.49746, "cps": 22.953125, "mean_p": 0.292195, "boundary_entropy": 0.409584, "bpb": 3.423893}, "my": {"chunks": 10449.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.551457, "cps": 81.632812, "mean_p": 0.40624, "boundary_entropy": 0.632474, "bpb": 1.261443}}, "bpb": {"en": 2.27552, "de": 2.56092, "vi": 2.10168, "tr": 2.55357, "fi": 2.80452, "ru": 1.67596, "el": 1.76624, "hi": 1.30866, "ta": 1.08506, "th": 1.30541, "zh": 3.42389, "my": 1.26144}, "gini_cpc": 0.221817, "gini_cps": 0.229278, "bpic": 5.556246, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0102000512020513, "vi": 1.1480237481209747, "tr": 1.129216389428317, "fi": 0.9447508174535455, "ru": 1.743891181713059, "el": 1.8011778611583131, "hi": 2.3326177284321945, "ta": 2.315674922468296, "th": 2.6878211318351886, "zh": 2.620248194268725, "my": 2.9046608731888117}, "t": 121.6}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.28143, "de": 2.56706, "vi": 2.10476, "tr": 2.55964, "fi": 2.80658, "ru": 1.65573, "el": 1.75801, "hi": 1.30581, "ta": 1.09265, "th": 1.32463, "zh": 3.42686, "my": 1.26807}, "bytes_per_chunk": {"en": 5.586, "de": 5.458, "vi": 6.423, "tr": 5.223, "fi": 5.482, "ru": 5.681, "el": 5.647, "hi": 5.858, "ta": 6.197, "th": 5.721, "zh": 5.472, "my": 4.742}, "bpb_mean": 2.0126, "t": 122.9}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.42189, "bpb": 2.05135, "chunk_loss": 0.03108, "F": 0.18958, "G": 0.31333, "chunks": 194.12, "bytes_per_chunk": 5.275, "boundary_entropy": 0.45369, "router_noise": 0.0, "grad_norm": 1.2207, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 123.0}
{"event": "train", "step": 2050, "phase": 1, "objective": "global", "lm_loss": 1.5841, "bpb": 2.28538, "chunk_loss": 0.030199, "F": 0.17715, "G": 0.27246, "chunks": 181.41, "bytes_per_chunk": 5.645, "boundary_entropy": 0.37052, "router_noise": 0.0, "grad_norm": 0.8627, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 125.0}
{"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.64132, "bpb": 2.36793, "chunk_loss": 0.030347, "F": 0.18259, "G": 0.27977, "chunks": 186.97, "bytes_per_chunk": 5.477, "boundary_entropy": 0.38622, "router_noise": 0.0, "grad_norm": 0.8686, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 127.0}
{"event": "train", "step": 2150, "phase": 1, "objective": "global", "lm_loss": 1.37735, "bpb": 1.98709, "chunk_loss": 0.030061, "F": 0.1741, "G": 0.28151, "chunks": 178.28, "bytes_per_chunk": 5.744, "boundary_entropy": 0.40663, "router_noise": 0.0, "grad_norm": 0.685, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 129.0}
{"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3206.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184221, "cps": 25.046875, "mean_p": 0.221684, "boundary_entropy": 0.230662, "bpb": 2.242369}, "de": {"chunks": 3710.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.180685, "cps": 28.984375, "mean_p": 0.2238, "boundary_entropy": 0.25902, "bpb": 2.508792}, "vi": {"chunks": 4036.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.218931, "cps": 31.53125, "mean_p": 0.27306, "boundary_entropy": 0.403551, "bpb": 2.083079}, "tr": {"chunks": 3863.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.21339, "cps": 30.179688, "mean_p": 0.229584, "boundary_entropy": 0.267524, "bpb": 2.507235}, "fi": {"chunks": 3384.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.179895, "cps": 26.4375, "mean_p": 0.217265, "boundary_entropy": 0.252265, "bpb": 2.726603}, "ru": {"chunks": 6529.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.344538, "cps": 51.007812, "mean_p": 0.313791, "boundary_entropy": 0.503516, "bpb": 1.639371}, "el": {"chunks": 6700.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.326463, "cps": 52.34375, "mean_p": 0.350576, "boundary_entropy": 0.573539, "bpb": 1.731257}, "hi": {"chunks": 6441.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.382823, "cps": 50.320312, "mean_p": 0.329786, "boundary_entropy": 0.551776, "bpb": 1.332266}, "ta": {"chunks": 7631.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.408184, "cps": 59.617188, "mean_p": 0.395978, "boundary_entropy": 0.607188, "bpb": 1.080086}, "th": {"chunks": 7216.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.448617, "cps": 56.375, "mean_p": 0.405259, "boundary_entropy": 0.624325, "bpb": 1.269497}, "zh": {"chunks": 2916.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.493735, "cps": 22.78125, "mean_p": 0.290638, "boundary_entropy": 0.405069, "bpb": 3.414655}, "my": {"chunks": 7057.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.37244, "cps": 55.132812, "mean_p": 0.35591, "boundary_entropy": 0.607522, "bpb": 1.244821}}, "bpb": {"en": 2.24237, "de": 2.50879, "vi": 2.08308, "tr": 2.50724, "fi": 2.7266, "ru": 1.63937, "el": 1.73126, "hi": 1.33227, "ta": 1.08009, "th": 1.2695, "zh": 3.41466, "my": 1.24482}, "gini_cpc": 0.1944, "gini_cps": 0.1848, "bpic": 6.121983, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9808037205844166, "vi": 1.1884163491256983, "tr": 1.1583365464459987, "fi": 0.9765153470515866, "ru": 1.8702430798453424, "el": 1.7721258461171736, "hi": 2.078063567660562, "ta": 2.2157287160657746, "th": 2.4352079980592807, "zh": 2.680122712683782, "my": 2.0217029441709395}, "t": 131.8}
{"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.2499, "de": 2.51483, "vi": 2.08803, "tr": 2.51158, "fi": 2.733, "ru": 1.61785, "el": 1.72436, "hi": 1.32845, "ta": 1.08543, "th": 1.28695, "zh": 3.41796, "my": 1.25336}, "bytes_per_chunk": {"en": 5.717, "de": 5.66, "vi": 6.325, "tr": 5.204, "fi": 5.531, "ru": 5.295, "el": 5.797, "hi": 6.514, "ta": 6.774, "th": 6.35, "zh": 5.588, "my": 7.533}, "bpb_mean": 1.98431, "t": 133.1}
{"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.28636, "bpb": 1.85582, "chunk_loss": 0.029997, "F": 0.17114, "G": 0.29676, "chunks": 175.25, "bytes_per_chunk": 5.843, "boundary_entropy": 0.43731, "router_noise": 0.0, "grad_norm": 0.7786, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.083, "alpha": 0.03, "t": 133.1}
{"event": "train", "step": 2250, "phase": 1, "objective": "global", "lm_loss": 1.52739, "bpb": 2.20356, "chunk_loss": 0.030202, "F": 0.17596, "G": 0.26546, "chunks": 180.19, "bytes_per_chunk": 5.683, "boundary_entropy": 0.36408, "router_noise": 0.0, "grad_norm": 0.7973, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 135.1}
{"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.48103, "bpb": 2.13667, "chunk_loss": 0.030148, "F": 0.17703, "G": 0.27166, "chunks": 181.28, "bytes_per_chunk": 5.649, "boundary_entropy": 0.36789, "router_noise": 0.0, "grad_norm": 0.8001, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 137.1}
{"event": "train", "step": 2350, "phase": 1, "objective": "global", "lm_loss": 1.46056, "bpb": 2.10715, "chunk_loss": 0.030291, "F": 0.17981, "G": 0.29121, "chunks": 184.12, "bytes_per_chunk": 5.561, "boundary_entropy": 0.41296, "router_noise": 0.0, "grad_norm": 0.6954, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.039, "alpha": 0.03, "t": 139.1}
{"event": "probe", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3318.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190657, "cps": 25.921875, "mean_p": 0.232368, "boundary_entropy": 0.233602, "bpb": 2.215499}, "de": {"chunks": 3782.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.184191, "cps": 29.546875, "mean_p": 0.224484, "boundary_entropy": 0.253049, "bpb": 2.469309}, "vi": {"chunks": 4050.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.219691, "cps": 31.640625, "mean_p": 0.277095, "boundary_entropy": 0.403027, "bpb": 2.057003}, "tr": {"chunks": 3756.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207479, "cps": 29.34375, "mean_p": 0.225376, "boundary_entropy": 0.257934, "bpb": 2.474278}, "fi": {"chunks": 3395.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18048, "cps": 26.523438, "mean_p": 0.220761, "boundary_entropy": 0.25712, "bpb": 2.676612}, "ru": {"chunks": 6440.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.339842, "cps": 50.3125, "mean_p": 0.305473, "boundary_entropy": 0.490787, "bpb": 1.602483}, "el": {"chunks": 7154.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.348585, "cps": 55.890625, "mean_p": 0.352308, "boundary_entropy": 0.576936, "bpb": 1.716906}, "hi": {"chunks": 6495.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.386033, "cps": 50.742188, "mean_p": 0.351777, "boundary_entropy": 0.564018, "bpb": 1.277792}, "ta": {"chunks": 7697.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.411714, "cps": 60.132812, "mean_p": 0.401082, "boundary_entropy": 0.606, "bpb": 1.051203}, "th": {"chunks": 6426.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.399503, "cps": 50.203125, "mean_p": 0.387213, "boundary_entropy": 0.613378, "bpb": 1.247893}, "zh": {"chunks": 2830.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.479174, "cps": 22.109375, "mean_p": 0.275838, "boundary_entropy": 0.384518, "bpb": 3.380735}, "my": {"chunks": 8210.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.433291, "cps": 64.140625, "mean_p": 0.362133, "boundary_entropy": 0.606451, "bpb": 1.227308}}, "bpb": {"en": 2.2155, "de": 2.46931, "vi": 2.057, "tr": 2.47428, "fi": 2.67661, "ru": 1.60248, "el": 1.71691, "hi": 1.27779, "ta": 1.0512, "th": 1.24789, "zh": 3.38073, "my": 1.22731}, "gini_cpc": 0.188302, "gini_cps": 0.195308, "bpic": 6.038755, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9660883741243421, "vi": 1.1522842340837829, "tr": 1.0882352030409874, "fi": 0.9466199079585133, "ru": 1.7824788751210714, "el": 1.8283352361074363, "hi": 2.0247519272158616, "ta": 2.1594529970251495, "th": 2.095402194827322, "zh": 2.513279167475501, "my": 2.2726236380828695}, "t": 141.9}
{"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.225, "de": 2.47774, "vi": 2.06278, "tr": 2.47992, "fi": 2.68646, "ru": 1.57921, "el": 1.70827, "hi": 1.27712, "ta": 1.05855, "th": 1.26789, "zh": 3.38183, "my": 1.23401}, "bytes_per_chunk": {"en": 5.441, "de": 5.542, "vi": 6.246, "tr": 5.386, "fi": 5.494, "ru": 5.551, "el": 5.413, "hi": 6.459, "ta": 6.583, "th": 7.386, "zh": 5.615, "my": 6.71}, "bpb_mean": 1.95323, "t": 143.2}
{"event": "phase_transition", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 143.2}
{"event": "train", "step": 2400, "phase": 2, "objective": "parity_a", "lm_loss": 1.31745, "bpb": 1.90068, "chunk_loss": 0.037025, "F": 0.17371, "G": 0.28524, "chunks": 177.88, "bytes_per_chunk": 5.757, "boundary_entropy": 0.40152, "router_noise": 0.0, "grad_norm": 0.7295, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 10.8309, "alpha": 0.03, "t": 143.3}
{"event": "train", "step": 2450, "phase": 2, "objective": "parity_a", "lm_loss": 1.48031, "bpb": 2.13564, "chunk_loss": 0.030891, "F": 0.14197, "G": 0.22338, "chunks": 145.38, "bytes_per_chunk": 7.044, "boundary_entropy": 0.32524, "router_noise": 0.0, "grad_norm": 0.7826, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.039, "n_target_mean": 9.4427, "alpha": 0.03, "t": 145.2}
{"event": "train", "step": 2500, "phase": 2, "objective": "parity_a", "lm_loss": 1.47531, "bpb": 2.12842, "chunk_loss": 0.03155, "F": 0.14728, "G": 0.22168, "chunks": 150.81, "bytes_per_chunk": 6.79, "boundary_entropy": 0.32189, "router_noise": 0.0, "grad_norm": 0.7544, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.039, "n_target_mean": 9.5095, "alpha": 0.03, "t": 147.2}
{"event": "train", "step": 2550, "phase": 2, "objective": "parity_a", "lm_loss": 1.52785, "bpb": 2.20423, "chunk_loss": 0.030272, "F": 0.15805, "G": 0.21196, "chunks": 161.84, "bytes_per_chunk": 6.327, "boundary_entropy": 0.25712, "router_noise": 0.0, "grad_norm": 0.7359, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.039, "n_target_mean": 7.5354, "alpha": 0.03, "t": 149.1}
{"event": "probe", "step": 2600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3106.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.178475, "cps": 24.265625, "mean_p": 0.208232, "boundary_entropy": 0.187617, "bpb": 2.196509}, "de": {"chunks": 3553.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.173039, "cps": 27.757812, "mean_p": 0.202808, "boundary_entropy": 0.206686, "bpb": 2.449322}, "vi": {"chunks": 3420.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.185517, "cps": 26.71875, "mean_p": 0.212486, "boundary_entropy": 0.320083, "bpb": 2.052707}, "tr": {"chunks": 3625.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.200243, "cps": 28.320312, "mean_p": 0.211689, "boundary_entropy": 0.220271, "bpb": 2.422862}, "fi": {"chunks": 3231.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.171761, "cps": 25.242188, "mean_p": 0.199943, "boundary_entropy": 0.210537, "bpb": 2.643728}, "ru": {"chunks": 3407.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.179789, "cps": 26.617188, "mean_p": 0.211124, "boundary_entropy": 0.408627, "bpb": 1.619462}, "el": {"chunks": 2688.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.130975, "cps": 21.0, "mean_p": 0.234325, "boundary_entropy": 0.471075, "bpb": 1.678724}, "hi": {"chunks": 3199.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.190134, "cps": 24.992188, "mean_p": 0.14707, "boundary_entropy": 0.319159, "bpb": 1.347768}, "ta": {"chunks": 3315.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.17732, "cps": 25.898438, "mean_p": 0.150847, "boundary_entropy": 0.332662, "bpb": 1.091388}, "th": {"chunks": 3023.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.187939, "cps": 23.617188, "mean_p": 0.176204, "boundary_entropy": 0.382625, "bpb": 1.251037}, "zh": {"chunks": 2835.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.48002, "cps": 22.148438, "mean_p": 0.287083, "boundary_entropy": 0.3971, "bpb": 3.360726}, "my": {"chunks": 2785.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.146981, "cps": 21.757812, "mean_p": 0.188624, "boundary_entropy": 0.414838, "bpb": 1.217006}}, "bpb": {"en": 2.19651, "de": 2.44932, "vi": 2.05271, "tr": 2.42286, "fi": 2.64373, "ru": 1.61946, "el": 1.67872, "hi": 1.34777, "ta": 1.09139, "th": 1.25104, "zh": 3.36073, "my": 1.21701}, "gini_cpc": 0.157768, "gini_cps": 0.05214, "bpic": 10.050043, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9695394146510624, "vi": 1.0394548570524411, "tr": 1.1219671160507847, "fi": 0.9623825580811436, "ru": 1.0073620552271798, "el": 0.7338564032833735, "hi": 1.065324306404036, "ta": 0.9935294205781044, "th": 1.053027591467656, "zh": 2.689566516332611, "my": 0.8235396097442814}, "t": 152.0}
{"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.20417, "de": 2.45618, "vi": 2.05661, "tr": 2.43019, "fi": 2.65215, "ru": 1.59787, "el": 1.67002, "hi": 1.34402, "ta": 1.09646, "th": 1.27151, "zh": 3.36112, "my": 1.2255}, "bytes_per_chunk": {"en": 5.839, "de": 5.986, "vi": 7.459, "tr": 5.623, "fi": 5.685, "ru": 10.748, "el": 14.079, "hi": 13.799, "ta": 15.01, "th": 15.89, "zh": 5.695, "my": 20.965}, "bpb_mean": 1.94715, "t": 153.2}
{"event": "train", "step": 2600, "phase": 2, "objective": "parity_a", "lm_loss": 1.50578, "bpb": 2.17238, "chunk_loss": 0.030277, "F": 0.13947, "G": 0.20545, "chunks": 142.81, "bytes_per_chunk": 7.17, "boundary_entropy": 0.2806, "router_noise": 0.0, "grad_norm": 0.7378, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.083, "n_target_mean": 9.2882, "alpha": 0.03, "t": 153.3}
{"event": "train", "step": 2650, "phase": 2, "objective": "parity_a", "lm_loss": 1.45139, "bpb": 2.09391, "chunk_loss": 0.030313, "F": 0.15192, "G": 0.21047, "chunks": 155.56, "bytes_per_chunk": 6.583, "boundary_entropy": 0.27712, "router_noise": 0.0, "grad_norm": 0.7973, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.0661, "alpha": 0.03, "t": 155.2}
{"event": "train", "step": 2700, "phase": 2, "objective": "parity_a", "lm_loss": 1.50198, "bpb": 2.1669, "chunk_loss": 0.030493, "F": 0.15536, "G": 0.20519, "chunks": 159.09, "bytes_per_chunk": 6.436, "boundary_entropy": 0.25219, "router_noise": 0.0, "grad_norm": 0.8667, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.2992, "alpha": 0.03, "t": 157.2}
{"event": "train", "step": 2750, "phase": 2, "objective": "parity_a", "lm_loss": 1.53197, "bpb": 2.21016, "chunk_loss": 0.030257, "F": 0.14664, "G": 0.20641, "chunks": 150.16, "bytes_per_chunk": 6.82, "boundary_entropy": 0.26645, "router_noise": 0.0, "grad_norm": 0.7677, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.0733, "alpha": 0.03, "t": 159.3}
{"event": "probe", "step": 2800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3201.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.183934, "cps": 25.007812, "mean_p": 0.212692, "boundary_entropy": 0.185922, "bpb": 2.17688}, "de": {"chunks": 3648.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177665, "cps": 28.5, "mean_p": 0.203216, "boundary_entropy": 0.198769, "bpb": 2.413189}, "vi": {"chunks": 3537.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.191863, "cps": 27.632812, "mean_p": 0.200287, "boundary_entropy": 0.292617, "bpb": 2.039412}, "tr": {"chunks": 3331.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184003, "cps": 26.023438, "mean_p": 0.202318, "boundary_entropy": 0.213671, "bpb": 2.400343}, "fi": {"chunks": 3261.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.173356, "cps": 25.476562, "mean_p": 0.198492, "boundary_entropy": 0.207829, "bpb": 2.620602}, "ru": {"chunks": 3416.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.180264, "cps": 26.6875, "mean_p": 0.202005, "boundary_entropy": 0.388991, "bpb": 1.586093}, "el": {"chunks": 3298.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.160698, "cps": 25.765625, "mean_p": 0.234066, "boundary_entropy": 0.459981, "bpb": 1.677504}, "hi": {"chunks": 2552.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.151679, "cps": 19.9375, "mean_p": 0.127943, "boundary_entropy": 0.291051, "bpb": 1.316809}, "ta": {"chunks": 2374.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.126986, "cps": 18.546875, "mean_p": 0.11463, "boundary_entropy": 0.270927, "bpb": 1.090983}, "th": {"chunks": 2444.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.151943, "cps": 19.09375, "mean_p": 0.150577, "boundary_entropy": 0.343759, "bpb": 1.226288}, "zh": {"chunks": 3054.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.517101, "cps": 23.859375, "mean_p": 0.285972, "boundary_entropy": 0.381677, "bpb": 3.31985}, "my": {"chunks": 2430.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.128246, "cps": 18.984375, "mean_p": 0.208346, "boundary_entropy": 0.442369, "bpb": 1.21291}}, "bpb": {"en": 2.17688, "de": 2.41319, "vi": 2.03941, "tr": 2.40034, "fi": 2.6206, "ru": 1.58609, "el": 1.6775, "hi": 1.31681, "ta": 1.09098, "th": 1.22629, "zh": 3.31985, "my": 1.21291}, "gini_cpc": 0.190208, "gini_cps": 0.080875, "bpic": 10.501313, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9659193550912238, "vi": 1.04311061255522, "tr": 1.0003743030159453, "fi": 0.9424913248180726, "ru": 0.9800474291098017, "el": 0.8736716677076274, "hi": 0.8246393284416599, "ta": 0.690388726644079, "th": 0.8260732944330713, "zh": 2.8113443003175966, "my": 0.6972384738065744}, "t": 162.2}
{"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.18621, "de": 2.41947, "vi": 2.04322, "tr": 2.40683, "fi": 2.62647, "ru": 1.56305, "el": 1.66867, "hi": 1.31349, "ta": 1.09479, "th": 1.2448, "zh": 3.31853, "my": 1.22165}, "bytes_per_chunk": {"en": 5.655, "de": 5.796, "vi": 7.208, "tr": 5.947, "fi": 5.716, "ru": 10.804, "el": 11.786, "hi": 16.746, "ta": 20.648, "th": 19.03, "zh": 5.392, "my": 22.348}, "bpb_mean": 1.9256, "t": 163.4}
{"event": "train", "step": 2800, "phase": 2, "objective": "parity_a", "lm_loss": 1.41174, "bpb": 2.03671, "chunk_loss": 0.030309, "F": 0.1442, "G": 0.20497, "chunks": 147.66, "bytes_per_chunk": 6.935, "boundary_entropy": 0.27563, "router_noise": 0.0, "grad_norm": 0.7253, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 8.7824, "alpha": 0.03, "t": 163.4}
{"event": "train", "step": 2850, "phase": 2, "objective": "parity_a", "lm_loss": 1.33403, "bpb": 1.92461, "chunk_loss": 0.030124, "F": 0.12836, "G": 0.19167, "chunks": 131.44, "bytes_per_chunk": 7.791, "boundary_entropy": 0.27461, "router_noise": 0.0, "grad_norm": 0.7537, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.986, "alpha": 0.03, "t": 165.5}
{"event": "train", "step": 2900, "phase": 2, "objective": "parity_a", "lm_loss": 1.4103, "bpb": 2.03464, "chunk_loss": 0.030158, "F": 0.1463, "G": 0.20231, "chunks": 149.81, "bytes_per_chunk": 6.835, "boundary_entropy": 0.25908, "router_noise": 0.0, "grad_norm": 0.813, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.5507, "alpha": 0.03, "t": 167.5}
{"event": "train", "step": 2950, "phase": 2, "objective": "parity_a", "lm_loss": 1.30737, "bpb": 1.88614, "chunk_loss": 0.030317, "F": 0.13486, "G": 0.2019, "chunks": 138.09, "bytes_per_chunk": 7.415, "boundary_entropy": 0.28794, "router_noise": 0.0, "grad_norm": 0.6804, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.6732, "alpha": 0.03, "t": 169.5}
{"event": "probe", "step": 3000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3261.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187381, "cps": 25.476562, "mean_p": 0.215887, "boundary_entropy": 0.180527, "bpb": 2.159534}, "de": {"chunks": 3779.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.184045, "cps": 29.523438, "mean_p": 0.206071, "boundary_entropy": 0.189389, "bpb": 2.396133}, "vi": {"chunks": 3516.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.190724, "cps": 27.46875, "mean_p": 0.193538, "boundary_entropy": 0.272987, "bpb": 2.020547}, "tr": {"chunks": 3562.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196763, "cps": 27.828125, "mean_p": 0.210337, "boundary_entropy": 0.208981, "bpb": 2.379932}, "fi": {"chunks": 3398.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.180639, "cps": 26.546875, "mean_p": 0.205616, "boundary_entropy": 0.201839, "bpb": 2.593827}, "ru": {"chunks": 3471.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.183166, "cps": 27.117188, "mean_p": 0.195685, "boundary_entropy": 0.374632, "bpb": 1.560313}, "el": {"chunks": 2864.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.139551, "cps": 22.375, "mean_p": 0.214406, "boundary_entropy": 0.438219, "bpb": 1.63859}, "hi": {"chunks": 3160.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.187816, "cps": 24.6875, "mean_p": 0.140544, "boundary_entropy": 0.297588, "bpb": 1.280718}, "ta": {"chunks": 2547.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.13624, "cps": 19.898438, "mean_p": 0.119147, "boundary_entropy": 0.275886, "bpb": 1.062805}, "th": {"chunks": 2585.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.160709, "cps": 20.195312, "mean_p": 0.149697, "boundary_entropy": 0.337356, "bpb": 1.208574}, "zh": {"chunks": 3174.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.53742, "cps": 24.796875, "mean_p": 0.292923, "boundary_entropy": 0.385397, "bpb": 3.307965}, "my": {"chunks": 3968.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.209415, "cps": 31.0, "mean_p": 0.207699, "boundary_entropy": 0.429747, "bpb": 1.190684}}, "bpb": {"en": 2.15953, "de": 2.39613, "vi": 2.02055, "tr": 2.37993, "fi": 2.59383, "ru": 1.56031, "el": 1.63859, "hi": 1.28072, "ta": 1.0628, "th": 1.20857, "zh": 3.30797, "my": 1.19068}, "gini_cpc": 0.179168, "gini_cps": 0.072273, "bpic": 9.769149, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9821951971675188, "vi": 1.0178389023918295, "tr": 1.0500662136691907, "fi": 0.9640172662191381, "ru": 0.9775043995601653, "el": 0.7447413880475431, "hi": 1.0023175417708403, "ta": 0.7270709567445064, "th": 0.8576553549608094, "zh": 2.868050547195857, "my": 1.11758768830163}, "t": 172.4}
{"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.16773, "de": 2.40172, "vi": 2.0219, "tr": 2.38711, "fi": 2.60238, "ru": 1.53496, "el": 1.63089, "hi": 1.27831, "ta": 1.06742, "th": 1.22814, "zh": 3.30492, "my": 1.1987}, "bytes_per_chunk": {"en": 5.557, "de": 5.622, "vi": 7.481, "tr": 5.551, "fi": 5.384, "ru": 10.74, "el": 13.307, "hi": 13.621, "ta": 19.441, "th": 17.549, "zh": 5.278, "my": 15.028}, "bpb_mean": 1.90201, "t": 173.6}
{"event": "train", "step": 3000, "phase": 2, "objective": "parity_a", "lm_loss": 1.51577, "bpb": 2.18679, "chunk_loss": 0.030755, "F": 0.15192, "G": 0.21484, "chunks": 155.56, "bytes_per_chunk": 6.583, "boundary_entropy": 0.28709, "router_noise": 0.0, "grad_norm": 0.7016, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 9.0106, "alpha": 0.03, "t": 173.7}
{"event": "train", "step": 3050, "phase": 2, "objective": "parity_a", "lm_loss": 1.48009, "bpb": 2.13532, "chunk_loss": 0.030365, "F": 0.14334, "G": 0.21144, "chunks": 146.78, "bytes_per_chunk": 6.976, "boundary_entropy": 0.28939, "router_noise": 0.0, "grad_norm": 0.8104, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.0983, "alpha": 0.03, "t": 175.7}
{"event": "train", "step": 3100, "phase": 2, "objective": "parity_a", "lm_loss": 1.41907, "bpb": 2.04729, "chunk_loss": 0.0303, "F": 0.13394, "G": 0.20391, "chunks": 137.16, "bytes_per_chunk": 7.466, "boundary_entropy": 0.29804, "router_noise": 0.0, "grad_norm": 0.7573, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.8865, "alpha": 0.03, "t": 177.7}
{"event": "train", "step": 3150, "phase": 2, "objective": "parity_a", "lm_loss": 1.33079, "bpb": 1.91993, "chunk_loss": 0.030303, "F": 0.13068, "G": 0.19754, "chunks": 133.81, "bytes_per_chunk": 7.652, "boundary_entropy": 0.28509, "router_noise": 0.0, "grad_norm": 0.7644, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.6709, "alpha": 0.03, "t": 179.7}
{"event": "probe", "step": 3200, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3166.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.181923, "cps": 24.734375, "mean_p": 0.205389, "boundary_entropy": 0.162492, "bpb": 2.138747}, "de": {"chunks": 3586.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174646, "cps": 28.015625, "mean_p": 0.193373, "boundary_entropy": 0.172293, "bpb": 2.371177}, "vi": {"chunks": 3524.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.191158, "cps": 27.53125, "mean_p": 0.181479, "boundary_entropy": 0.259358, "bpb": 2.008569}, "tr": {"chunks": 3425.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.189195, "cps": 26.757812, "mean_p": 0.197893, "boundary_entropy": 0.190911, "bpb": 2.347323}, "fi": {"chunks": 3134.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.166605, "cps": 24.484375, "mean_p": 0.185416, "boundary_entropy": 0.17457, "bpb": 2.557263}, "ru": {"chunks": 3467.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.182955, "cps": 27.085938, "mean_p": 0.193341, "boundary_entropy": 0.367663, "bpb": 1.543167}, "el": {"chunks": 2714.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.132242, "cps": 21.203125, "mean_p": 0.212942, "boundary_entropy": 0.436321, "bpb": 1.626345}, "hi": {"chunks": 3058.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.181753, "cps": 23.890625, "mean_p": 0.14236, "boundary_entropy": 0.300239, "bpb": 1.280287}, "ta": {"chunks": 2923.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.156352, "cps": 22.835938, "mean_p": 0.127365, "boundary_entropy": 0.281727, "bpb": 1.055125}, "th": {"chunks": 2808.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.174573, "cps": 21.9375, "mean_p": 0.141197, "boundary_entropy": 0.311706, "bpb": 1.187776}, "zh": {"chunks": 3106.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.525906, "cps": 24.265625, "mean_p": 0.283091, "boundary_entropy": 0.3651, "bpb": 3.273868}, "my": {"chunks": 3077.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.162392, "cps": 24.039062, "mean_p": 0.16147, "boundary_entropy": 0.366647, "bpb": 1.175344}}, "bpb": {"en": 2.13875, "de": 2.37118, "vi": 2.00857, "tr": 2.34732, "fi": 2.55726, "ru": 1.54317, "el": 1.62635, "hi": 1.28029, "ta": 1.05513, "th": 1.18778, "zh": 3.27387, "my": 1.17534}, "gini_cpc": 0.170203, "gini_cps": 0.048204, "bpic": 10.10269, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9599996788061829, "vi": 1.0507658929009251, "tr": 1.0399758014162588, "fi": 0.9157993234761733, "ru": 1.005675423405344, "el": 0.7269126184759999, "hi": 0.9990693090310402, "ta": 0.8594419646967808, "th": 0.9595978094107209, "zh": 2.8908211164208955, "my": 0.8926420261670367}, "t": 182.6}
{"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.14917, "de": 2.37663, "vi": 2.0106, "tr": 2.35428, "fi": 2.56659, "ru": 1.51997, "el": 1.61798, "hi": 1.27527, "ta": 1.05905, "th": 1.20728, "zh": 3.27248, "my": 1.18424}, "bytes_per_chunk": {"en": 5.696, "de": 5.895, "vi": 7.232, "tr": 5.9, "fi": 5.87, "ru": 10.659, "el": 13.918, "hi": 14.265, "ta": 17.374, "th": 16.126, "zh": 5.344, "my": 18.346}, "bpb_mean": 1.88279, "t": 183.9}
{"event": "train", "step": 3200, "phase": 2, "objective": "parity_a", "lm_loss": 1.40583, "bpb": 2.02818, "chunk_loss": 0.030203, "F": 0.13766, "G": 0.19169, "chunks": 140.97, "bytes_per_chunk": 7.264, "boundary_entropy": 0.2547, "router_noise": 0.0, "grad_norm": 0.7386, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 9.2811, "alpha": 0.03, "t": 183.9}
{"event": "train", "step": 3250, "phase": 2, "objective": "parity_a", "lm_loss": 1.38648, "bpb": 2.00027, "chunk_loss": 0.030141, "F": 0.13806, "G": 0.18909, "chunks": 141.38, "bytes_per_chunk": 7.243, "boundary_entropy": 0.23947, "router_noise": 0.0, "grad_norm": 0.7318, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.2351, "alpha": 0.03, "t": 185.9}
{"event": "train", "step": 3300, "phase": 2, "objective": "parity_a", "lm_loss": 1.28944, "bpb": 1.86027, "chunk_loss": 0.030091, "F": 0.11591, "G": 0.17476, "chunks": 118.69, "bytes_per_chunk": 8.628, "boundary_entropy": 0.26352, "router_noise": 0.0, "grad_norm": 0.7975, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.7221, "alpha": 0.03, "t": 187.9}
{"event": "train", "step": 3350, "phase": 2, "objective": "parity_a", "lm_loss": 1.36052, "bpb": 1.96281, "chunk_loss": 0.030017, "F": 0.12805, "G": 0.18691, "chunks": 131.12, "bytes_per_chunk": 7.809, "boundary_entropy": 0.26006, "router_noise": 0.0, "grad_norm": 0.699, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.0553, "alpha": 0.03, "t": 189.9}
{"event": "probe", "step": 3400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3167.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.18198, "cps": 24.742188, "mean_p": 0.206021, "boundary_entropy": 0.159582, "bpb": 2.123949}, "de": {"chunks": 3614.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176009, "cps": 28.234375, "mean_p": 0.193627, "boundary_entropy": 0.165424, "bpb": 2.350103}, "vi": {"chunks": 3618.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.196257, "cps": 28.265625, "mean_p": 0.194726, "boundary_entropy": 0.267115, "bpb": 1.997663}, "tr": {"chunks": 3357.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185439, "cps": 26.226562, "mean_p": 0.19897, "boundary_entropy": 0.192835, "bpb": 2.331103}, "fi": {"chunks": 3215.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.170911, "cps": 25.117188, "mean_p": 0.188625, "boundary_entropy": 0.171324, "bpb": 2.551754}, "ru": {"chunks": 3432.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.181108, "cps": 26.8125, "mean_p": 0.194356, "boundary_entropy": 0.366257, "bpb": 1.52875}, "el": {"chunks": 2840.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.138381, "cps": 22.1875, "mean_p": 0.2103, "boundary_entropy": 0.427498, "bpb": 1.598222}, "hi": {"chunks": 3154.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.187459, "cps": 24.640625, "mean_p": 0.145058, "boundary_entropy": 0.301898, "bpb": 1.27191}, "ta": {"chunks": 2713.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.145119, "cps": 21.195312, "mean_p": 0.125973, "boundary_entropy": 0.280554, "bpb": 1.043991}, "th": {"chunks": 2789.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.173391, "cps": 21.789062, "mean_p": 0.142838, "boundary_entropy": 0.313412, "bpb": 1.176993}, "zh": {"chunks": 3157.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.534541, "cps": 24.664062, "mean_p": 0.28795, "boundary_entropy": 0.37056, "bpb": 3.244345}, "my": {"chunks": 4087.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.215696, "cps": 31.929688, "mean_p": 0.177119, "boundary_entropy": 0.377089, "bpb": 1.189504}}, "bpb": {"en": 2.12395, "de": 2.3501, "vi": 1.99766, "tr": 2.3311, "fi": 2.55175, "ru": 1.52875, "el": 1.59822, "hi": 1.27191, "ta": 1.04399, "th": 1.17699, "zh": 3.24434, "my": 1.1895}, "gini_cpc": 0.176631, "gini_cps": 0.063934, "bpic": 9.804588, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9671900006291135, "vi": 1.078453632006018, "tr": 1.0190062474664108, "fi": 0.9391720311179778, "ru": 0.9952086032327107, "el": 0.7604200365679046, "hi": 1.0301077934550724, "ta": 0.7974443421574774, "th": 0.9528038555043712, "zh": 2.937360132444397, "my": 1.1852700715680806}, "t": 192.8}
{"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 2.13264, "de": 2.3551, "vi": 2.00048, "tr": 2.33698, "fi": 2.55847, "ru": 1.50466, "el": 1.58883, "hi": 1.26883, "ta": 1.04825, "th": 1.19497, "zh": 3.2431, "my": 1.19819}, "bytes_per_chunk": {"en": 5.722, "de": 5.85, "vi": 7.038, "tr": 5.953, "fi": 5.765, "ru": 10.663, "el": 13.238, "hi": 13.53, "ta": 18.104, "th": 16.304, "zh": 5.221, "my": 14.014}, "bpb_mean": 1.86921, "t": 194.1}
{"event": "train", "step": 3400, "phase": 2, "objective": "parity_a", "lm_loss": 1.31399, "bpb": 1.89569, "chunk_loss": 0.030415, "F": 0.12866, "G": 0.19125, "chunks": 131.75, "bytes_per_chunk": 7.772, "boundary_entropy": 0.2734, "router_noise": 0.0, "grad_norm": 0.7424, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 10.5619, "alpha": 0.03, "t": 194.1}
{"event": "train", "step": 3450, "phase": 2, "objective": "parity_a", "lm_loss": 1.22319, "bpb": 1.76469, "chunk_loss": 0.030238, "F": 0.12195, "G": 0.17951, "chunks": 124.88, "bytes_per_chunk": 8.2, "boundary_entropy": 0.25496, "router_noise": 0.0, "grad_norm": 0.7106, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.9543, "alpha": 0.03, "t": 196.1}
{"event": "train", "step": 3500, "phase": 2, "objective": "parity_a", "lm_loss": 1.33349, "bpb": 1.92382, "chunk_loss": 0.030129, "F": 0.12903, "G": 0.1843, "chunks": 132.12, "bytes_per_chunk": 7.75, "boundary_entropy": 0.24651, "router_noise": 0.0, "grad_norm": 0.7459, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.0046, "alpha": 0.03, "t": 198.1}
{"event": "train", "step": 3550, "phase": 2, "objective": "parity_a", "lm_loss": 1.37685, "bpb": 1.98637, "chunk_loss": 0.030047, "F": 0.12949, "G": 0.19151, "chunks": 132.59, "bytes_per_chunk": 7.723, "boundary_entropy": 0.26378, "router_noise": 0.0, "grad_norm": 0.7856, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.7442, "alpha": 0.03, "t": 200.1}
{"event": "probe", "step": 3600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3126.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.179624, "cps": 24.421875, "mean_p": 0.199386, "boundary_entropy": 0.150433, "bpb": 2.112035}, "de": {"chunks": 3472.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.169094, "cps": 27.125, "mean_p": 0.187622, "boundary_entropy": 0.161592, "bpb": 2.32607}, "vi": {"chunks": 3561.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.193165, "cps": 27.820312, "mean_p": 0.187575, "boundary_entropy": 0.254946, "bpb": 1.979961}, "tr": {"chunks": 3340.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.1845, "cps": 26.09375, "mean_p": 0.195366, "boundary_entropy": 0.183487, "bpb": 2.30805}, "fi": {"chunks": 3241.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.172293, "cps": 25.320312, "mean_p": 0.188008, "boundary_entropy": 0.163157, "bpb": 2.502121}, "ru": {"chunks": 3316.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.174987, "cps": 25.90625, "mean_p": 0.186696, "boundary_entropy": 0.35601, "bpb": 1.515687}, "el": {"chunks": 2770.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.134971, "cps": 21.640625, "mean_p": 0.200826, "boundary_entropy": 0.41036, "bpb": 1.597367}, "hi": {"chunks": 3490.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.207429, "cps": 27.265625, "mean_p": 0.148962, "boundary_entropy": 0.307708, "bpb": 1.260625}, "ta": {"chunks": 3336.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.178443, "cps": 26.0625, "mean_p": 0.133334, "boundary_entropy": 0.278924, "bpb": 1.038032}, "th": {"chunks": 2813.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.174883, "cps": 21.976562, "mean_p": 0.135896, "boundary_entropy": 0.296873, "bpb": 1.160582}, "zh": {"chunks": 3114.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.52726, "cps": 24.328125, "mean_p": 0.288102, "boundary_entropy": 0.371114, "bpb": 3.232315}, "my": {"chunks": 3917.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.206724, "cps": 30.601562, "mean_p": 0.174283, "boundary_entropy": 0.371275, "bpb": 1.170321}}, "bpb": {"en": 2.11203, "de": 2.32607, "vi": 1.97996, "tr": 2.30805, "fi": 2.50212, "ru": 1.51569, "el": 1.59737, "hi": 1.26063, "ta": 1.03803, "th": 1.16058, "zh": 3.23231, "my": 1.17032}, "gini_cpc": 0.167186, "gini_cps": 0.050697, "bpic": 9.716959, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9413745561776731, "vi": 1.075385009425137, "tr": 1.0271433574333604, "fi": 0.9591848135575511, "ru": 0.9741827923771517, "el": 0.7514049823600374, "hi": 1.1547966107012175, "ta": 0.9934264501526935, "th": 0.9736072818525863, "zh": 2.9353528374475872, "my": 1.1508675425257204}, "t": 203.0}
{"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 2.11971, "de": 2.33307, "vi": 1.98374, "tr": 2.31534, "fi": 2.5122, "ru": 1.49306, "el": 1.5889, "hi": 1.25575, "ta": 1.04353, "th": 1.17858, "zh": 3.23239, "my": 1.17811}, "bytes_per_chunk": {"en": 5.763, "de": 6.024, "vi": 7.135, "tr": 6.008, "fi": 5.707, "ru": 11.146, "el": 13.693, "hi": 12.126, "ta": 14.82, "th": 15.853, "zh": 5.312, "my": 14.868}, "bpb_mean": 1.85286, "t": 204.3}
{"event": "train", "step": 3600, "phase": 2, "objective": "parity_a", "lm_loss": 1.32218, "bpb": 1.9075, "chunk_loss": 0.030181, "F": 0.12424, "G": 0.18199, "chunks": 127.22, "bytes_per_chunk": 8.049, "boundary_entropy": 0.2619, "router_noise": 0.0, "grad_norm": 0.7585, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 10.3328, "alpha": 0.03, "t": 204.3}
{"event": "train", "step": 3650, "phase": 2, "objective": "parity_a", "lm_loss": 1.31503, "bpb": 1.89719, "chunk_loss": 0.030288, "F": 0.13974, "G": 0.18824, "chunks": 143.09, "bytes_per_chunk": 7.156, "boundary_entropy": 0.23183, "router_noise": 0.0, "grad_norm": 0.7321, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.3376, "alpha": 0.03, "t": 206.3}
{"event": "train", "step": 3700, "phase": 2, "objective": "parity_a", "lm_loss": 1.53333, "bpb": 2.21212, "chunk_loss": 0.030224, "F": 0.15009, "G": 0.20317, "chunks": 153.69, "bytes_per_chunk": 6.663, "boundary_entropy": 0.24221, "router_noise": 0.0, "grad_norm": 0.8211, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.3595, "alpha": 0.03, "t": 208.3}
{"event": "train", "step": 3750, "phase": 2, "objective": "parity_a", "lm_loss": 1.42754, "bpb": 2.05951, "chunk_loss": 0.030308, "F": 0.151, "G": 0.19496, "chunks": 154.62, "bytes_per_chunk": 6.622, "boundary_entropy": 0.20878, "router_noise": 0.0, "grad_norm": 0.741, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.2469, "alpha": 0.03, "t": 210.3}
{"event": "probe", "step": 3800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3132.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.179969, "cps": 24.46875, "mean_p": 0.199723, "boundary_entropy": 0.142888, "bpb": 2.0993}, "de": {"chunks": 3448.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.167925, "cps": 26.9375, "mean_p": 0.185519, "boundary_entropy": 0.155242, "bpb": 2.314095}, "vi": {"chunks": 3466.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.188012, "cps": 27.078125, "mean_p": 0.175586, "boundary_entropy": 0.234267, "bpb": 1.971638}, "tr": {"chunks": 3339.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184445, "cps": 26.085938, "mean_p": 0.193357, "boundary_entropy": 0.176593, "bpb": 2.28021}, "fi": {"chunks": 3174.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.168731, "cps": 24.796875, "mean_p": 0.182182, "boundary_entropy": 0.155564, "bpb": 2.483886}, "ru": {"chunks": 3154.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.166438, "cps": 24.640625, "mean_p": 0.182688, "boundary_entropy": 0.354646, "bpb": 1.493835}, "el": {"chunks": 3076.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.149881, "cps": 24.03125, "mean_p": 0.206874, "boundary_entropy": 0.413369, "bpb": 1.565859}, "hi": {"chunks": 3009.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.178841, "cps": 23.507812, "mean_p": 0.143111, "boundary_entropy": 0.296982, "bpb": 1.240182}, "ta": {"chunks": 2757.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.147473, "cps": 21.539062, "mean_p": 0.121209, "boundary_entropy": 0.265106, "bpb": 1.0246}, "th": {"chunks": 2989.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.185825, "cps": 23.351562, "mean_p": 0.134741, "boundary_entropy": 0.287844, "bpb": 1.15036}, "zh": {"chunks": 3063.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.518625, "cps": 23.929688, "mean_p": 0.28572, "boundary_entropy": 0.368, "bpb": 3.221453}, "my": {"chunks": 2802.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.147878, "cps": 21.890625, "mean_p": 0.157046, "boundary_entropy": 0.359309, "bpb": 1.157868}}, "bpb": {"en": 2.0993, "de": 2.31409, "vi": 1.97164, "tr": 2.28021, "fi": 2.48389, "ru": 1.49384, "el": 1.56586, "hi": 1.24018, "ta": 1.0246, "th": 1.15036, "zh": 3.22145, "my": 1.15787}, "gini_cpc": 0.168893, "gini_cps": 0.038353, "bpic": 10.259055, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9330764251472212, "vi": 1.0446908315121888, "tr": 1.0248687110347494, "fi": 0.937556406529073, "ru": 0.9248149496052326, "el": 0.832813684599822, "hi": 0.9937324725453025, "ta": 0.8194334031165749, "th": 1.0325407884597748, "zh": 2.8817474728602517, "my": 0.8216883351653285}, "t": 213.2}
{"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 2.10665, "de": 2.3225, "vi": 1.97536, "tr": 2.29083, "fi": 2.49428, "ru": 1.46989, "el": 1.5584, "hi": 1.236, "ta": 1.02909, "th": 1.1684, "zh": 3.21835, "my": 1.16428}, "bytes_per_chunk": {"en": 5.755, "de": 6.076, "vi": 7.484, "tr": 5.99, "fi": 5.823, "ru": 11.741, "el": 12.251, "hi": 13.859, "ta": 18.007, "th": 15.034, "zh": 5.352, "my": 19.482}, "bpb_mean": 1.83617, "t": 214.5}
{"event": "train", "step": 3800, "phase": 2, "objective": "parity_a", "lm_loss": 1.37903, "bpb": 1.98951, "chunk_loss": 0.030168, "F": 0.13284, "G": 0.18573, "chunks": 136.03, "bytes_per_chunk": 7.528, "boundary_entropy": 0.24466, "router_noise": 0.0, "grad_norm": 0.7587, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 9.6644, "alpha": 0.03, "t": 214.5}
{"event": "train", "step": 3850, "phase": 2, "objective": "parity_a", "lm_loss": 1.53095, "bpb": 2.20869, "chunk_loss": 0.030342, "F": 0.15305, "G": 0.20428, "chunks": 156.72, "bytes_per_chunk": 6.534, "boundary_entropy": 0.2439, "router_noise": 0.0, "grad_norm": 0.7316, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.3671, "alpha": 0.03, "t": 216.5}
{"event": "train", "step": 3900, "phase": 2, "objective": "parity_a", "lm_loss": 1.46716, "bpb": 2.11667, "chunk_loss": 0.030069, "F": 0.14758, "G": 0.19562, "chunks": 151.12, "bytes_per_chunk": 6.776, "boundary_entropy": 0.22131, "router_noise": 0.0, "grad_norm": 0.7125, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.3124, "alpha": 0.03, "t": 218.5}
{"event": "train", "step": 3950, "phase": 2, "objective": "parity_a", "lm_loss": 1.42769, "bpb": 2.05972, "chunk_loss": 0.030211, "F": 0.14487, "G": 0.19955, "chunks": 148.34, "bytes_per_chunk": 6.903, "boundary_entropy": 0.24609, "router_noise": 0.0, "grad_norm": 0.819, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.5064, "alpha": 0.03, "t": 220.5}
{"event": "probe", "step": 4000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3188.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.183187, "cps": 24.90625, "mean_p": 0.201915, "boundary_entropy": 0.14186, "bpb": 2.087387}, "de": {"chunks": 3556.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.173185, "cps": 27.78125, "mean_p": 0.190873, "boundary_entropy": 0.157269, "bpb": 2.29569}, "vi": {"chunks": 3264.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.177055, "cps": 25.5, "mean_p": 0.176385, "boundary_entropy": 0.235952, "bpb": 1.966223}, "tr": {"chunks": 3438.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.189913, "cps": 26.859375, "mean_p": 0.198262, "boundary_entropy": 0.175954, "bpb": 2.263062}, "fi": {"chunks": 3283.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.174526, "cps": 25.648438, "mean_p": 0.190929, "boundary_entropy": 0.162025, "bpb": 2.459967}, "ru": {"chunks": 3445.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.181794, "cps": 26.914062, "mean_p": 0.193107, "boundary_entropy": 0.3609, "bpb": 1.478137}, "el": {"chunks": 3224.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.157092, "cps": 25.1875, "mean_p": 0.213763, "boundary_entropy": 0.421008, "bpb": 1.558425}, "hi": {"chunks": 2967.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.176345, "cps": 23.179688, "mean_p": 0.14039, "boundary_entropy": 0.291778, "bpb": 1.23505}, "ta": {"chunks": 2405.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.128644, "cps": 18.789062, "mean_p": 0.114693, "boundary_entropy": 0.25901, "bpb": 1.02041}, "th": {"chunks": 2696.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.16761, "cps": 21.0625, "mean_p": 0.13142, "boundary_entropy": 0.288338, "bpb": 1.143078}, "zh": {"chunks": 3248.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.549949, "cps": 25.375, "mean_p": 0.290875, "boundary_entropy": 0.36429, "bpb": 3.185539}, "my": {"chunks": 2573.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.135793, "cps": 20.101562, "mean_p": 0.146605, "boundary_entropy": 0.344894, "bpb": 1.155716}}, "bpb": {"en": 2.08739, "de": 2.29569, "vi": 1.96622, "tr": 2.26306, "fi": 2.45997, "ru": 1.47814, "el": 1.55842, "hi": 1.23505, "ta": 1.02041, "th": 1.14308, "zh": 3.18554, "my": 1.15572}, "gini_cpc": 0.187483, "gini_cps": 0.061538, "bpic": 10.292622, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9453990305167782, "vi": 0.966524385077074, "tr": 1.0367191681837882, "fi": 0.952718956919878, "ru": 0.9923978607111762, "el": 0.8575510745327136, "hi": 0.9626497023570932, "ta": 0.7022559356882239, "th": 0.9149653132726119, "zh": 3.002122333964014, "my": 0.7412798886419386}, "t": 223.4}
{"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 2.09491, "de": 2.30284, "vi": 1.96759, "tr": 2.27121, "fi": 2.47087, "ru": 1.4537, "el": 1.55133, "hi": 1.22973, "ta": 1.02457, "th": 1.1625, "zh": 3.18444, "my": 1.16308}, "bytes_per_chunk": {"en": 5.716, "de": 5.891, "vi": 8.153, "tr": 5.857, "fi": 5.623, "ru": 10.683, "el": 11.787, "hi": 14.048, "ta": 20.741, "th": 16.287, "zh": 5.117, "my": 21.535}, "bpb_mean": 1.82306, "t": 224.6}
{"event": "train", "step": 4000, "phase": 2, "objective": "parity_a", "lm_loss": 1.31466, "bpb": 1.89666, "chunk_loss": 0.030188, "F": 0.12823, "G": 0.18382, "chunks": 131.31, "bytes_per_chunk": 7.798, "boundary_entropy": 0.2437, "router_noise": 0.0, "grad_norm": 0.7568, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.083, "n_target_mean": 10.021, "alpha": 0.03, "t": 224.7}
{"event": "train", "step": 4050, "phase": 2, "objective": "parity_a", "lm_loss": 1.46234, "bpb": 2.1097, "chunk_loss": 0.030188, "F": 0.14734, "G": 0.19764, "chunks": 150.88, "bytes_per_chunk": 6.787, "boundary_entropy": 0.23208, "router_noise": 0.0, "grad_norm": 0.7472, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.039, "n_target_mean": 8.3995, "alpha": 0.03, "t": 226.7}
{"event": "train", "step": 4100, "phase": 2, "objective": "parity_a", "lm_loss": 1.33629, "bpb": 1.92786, "chunk_loss": 0.030063, "F": 0.1257, "G": 0.17834, "chunks": 128.72, "bytes_per_chunk": 7.955, "boundary_entropy": 0.23721, "router_noise": 0.0, "grad_norm": 0.7626, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.039, "n_target_mean": 9.8635, "alpha": 0.03, "t": 228.6}
{"event": "train", "step": 4150, "phase": 2, "objective": "parity_a", "lm_loss": 1.48803, "bpb": 2.14677, "chunk_loss": 0.030214, "F": 0.15149, "G": 0.205, "chunks": 155.12, "bytes_per_chunk": 6.601, "boundary_entropy": 0.23718, "router_noise": 0.0, "grad_norm": 0.7206, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.039, "n_target_mean": 8.0388, "alpha": 0.03, "t": 230.6}
{"event": "probe", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3250.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186749, "cps": 25.390625, "mean_p": 0.206872, "boundary_entropy": 0.141992, "bpb": 2.073535}, "de": {"chunks": 3657.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178104, "cps": 28.570312, "mean_p": 0.194137, "boundary_entropy": 0.151687, "bpb": 2.278441}, "vi": {"chunks": 3500.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.189856, "cps": 27.34375, "mean_p": 0.179896, "boundary_entropy": 0.233793, "bpb": 1.958033}, "tr": {"chunks": 3437.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.189858, "cps": 26.851562, "mean_p": 0.197799, "boundary_entropy": 0.169549, "bpb": 2.249962}, "fi": {"chunks": 3348.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177981, "cps": 26.15625, "mean_p": 0.192637, "boundary_entropy": 0.157151, "bpb": 2.440224}, "ru": {"chunks": 3291.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.173668, "cps": 25.710938, "mean_p": 0.187679, "boundary_entropy": 0.35473, "bpb": 1.467024}, "el": {"chunks": 2831.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.137943, "cps": 22.117188, "mean_p": 0.200893, "boundary_entropy": 0.405543, "bpb": 1.539723}, "hi": {"chunks": 2876.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.170936, "cps": 22.46875, "mean_p": 0.140717, "boundary_entropy": 0.299096, "bpb": 1.228797}, "ta": {"chunks": 2930.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.156726, "cps": 22.890625, "mean_p": 0.128025, "boundary_entropy": 0.27481, "bpb": 1.007691}, "th": {"chunks": 2840.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.176562, "cps": 22.1875, "mean_p": 0.136459, "boundary_entropy": 0.294692, "bpb": 1.135476}, "zh": {"chunks": 3229.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.546732, "cps": 25.226562, "mean_p": 0.285621, "boundary_entropy": 0.353027, "bpb": 3.171651}, "my": {"chunks": 2434.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.128457, "cps": 19.015625, "mean_p": 0.160812, "boundary_entropy": 0.372237, "bpb": 1.14467}}, "bpb": {"en": 2.07353, "de": 2.27844, "vi": 1.95803, "tr": 2.24996, "fi": 2.44022, "ru": 1.46702, "el": 1.53972, "hi": 1.2288, "ta": 1.00769, "th": 1.13548, "zh": 3.17165, "my": 1.14467}, "gini_cpc": 0.186237, "gini_cps": 0.060063, "bpic": 10.200702, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9537033593202686, "vi": 1.0166364148463416, "tr": 1.0166459617827899, "fi": 0.9530470142265368, "ru": 0.9299496326364928, "el": 0.7386518390248838, "hi": 0.9153234061035547, "ta": 0.8392336906206924, "th": 0.9454488414911169, "zh": 2.9276244236630284, "my": 0.6878566765723193}, "t": 233.4}
{"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 2.08172, "de": 2.28488, "vi": 1.96031, "tr": 2.25654, "fi": 2.44995, "ru": 1.44223, "el": 1.53325, "hi": 1.22239, "ta": 1.01364, "th": 1.15436, "zh": 3.16928, "my": 1.15192}, "bytes_per_chunk": {"en": 5.605, "de": 5.751, "vi": 7.588, "tr": 5.766, "fi": 5.567, "ru": 11.088, "el": 12.974, "hi": 14.478, "ta": 16.722, "th": 15.693, "zh": 5.145, "my": 22.064}, "bpb_mean": 1.81004, "t": 234.7}
{"event": "phase_transition", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 131584, "t": 234.7}
{"event": "train", "step": 4200, "phase": 3, "objective": "none", "lm_loss": 1.30022, "bpb": 1.87582, "chunk_loss": 0.0, "F": 0.13318, "G": 0.19205, "chunks": 136.38, "bytes_per_chunk": 7.509, "boundary_entropy": 0.25857, "router_noise": 0.0, "grad_norm": 0.692, "lr": 9.12e-05, "router_trainable": false, "sec_per_step": 0.083, "alpha": 0.0, "t": 234.7}
{"event": "train", "step": 4250, "phase": 3, "objective": "none", "lm_loss": 1.30885, "bpb": 1.88827, "chunk_loss": 0.0, "F": 0.13861, "G": 0.19354, "chunks": 141.94, "bytes_per_chunk": 7.214, "boundary_entropy": 0.24648, "router_noise": 0.0, "grad_norm": 0.7357, "lr": 8.81e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 236.7}
{"event": "train", "step": 4300, "phase": 3, "objective": "none", "lm_loss": 1.3503, "bpb": 1.94808, "chunk_loss": 0.0, "F": 0.15103, "G": 0.19447, "chunks": 154.66, "bytes_per_chunk": 6.621, "boundary_entropy": 0.21477, "router_noise": 0.0, "grad_norm": 0.756, "lr": 8.5e-05, "router_trainable": false, "sec_per_step": 0.038, "alpha": 0.0, "t": 238.6}
{"event": "train", "step": 4350, "phase": 3, "objective": "none", "lm_loss": 1.47908, "bpb": 2.13387, "chunk_loss": 0.0, "F": 0.15619, "G": 0.21773, "chunks": 159.94, "bytes_per_chunk": 6.403, "boundary_entropy": 0.26928, "router_noise": 0.0, "grad_norm": 0.7705, "lr": 8.21e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 240.5}
{"event": "probe", "step": 4400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3228.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185485, "cps": 25.21875, "mean_p": 0.206495, "boundary_entropy": 0.145185, "bpb": 2.060195}, "de": {"chunks": 3685.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.179467, "cps": 28.789062, "mean_p": 0.19671, "boundary_entropy": 0.156795, "bpb": 2.265764}, "vi": {"chunks": 3560.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.193111, "cps": 27.8125, "mean_p": 0.181314, "boundary_entropy": 0.235601, "bpb": 1.943848}, "tr": {"chunks": 3511.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193946, "cps": 27.429688, "mean_p": 0.20333, "boundary_entropy": 0.177774, "bpb": 2.234455}, "fi": {"chunks": 3331.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177077, "cps": 26.023438, "mean_p": 0.192674, "boundary_entropy": 0.158862, "bpb": 2.420721}, "ru": {"chunks": 3753.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.198047, "cps": 29.320312, "mean_p": 0.201425, "boundary_entropy": 0.362655, "bpb": 1.449218}, "el": {"chunks": 3217.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.156751, "cps": 25.132812, "mean_p": 0.215426, "boundary_entropy": 0.42265, "bpb": 1.528511}, "hi": {"chunks": 2766.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.164398, "cps": 21.609375, "mean_p": 0.143449, "boundary_entropy": 0.308978, "bpb": 1.216536}, "ta": {"chunks": 2967.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.158706, "cps": 23.179688, "mean_p": 0.130406, "boundary_entropy": 0.281573, "bpb": 1.004011}, "th": {"chunks": 2646.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.164501, "cps": 20.671875, "mean_p": 0.137605, "boundary_entropy": 0.301277, "bpb": 1.130113}, "zh": {"chunks": 3268.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.553336, "cps": 25.53125, "mean_p": 0.290646, "boundary_entropy": 0.359606, "bpb": 3.159369}, "my": {"chunks": 3046.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.160756, "cps": 23.796875, "mean_p": 0.161457, "boundary_entropy": 0.363767, "bpb": 1.142222}}, "bpb": {"en": 2.0602, "de": 2.26576, "vi": 1.94385, "tr": 2.23446, "fi": 2.42072, "ru": 1.44922, "el": 1.52851, "hi": 1.21654, "ta": 1.00401, "th": 1.13011, "zh": 3.15937, "my": 1.14222}, "gini_cpc": 0.172922, "gini_cps": 0.058379, "bpic": 9.846093, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9675550392450126, "vi": 1.0411119950232053, "tr": 1.045612754525067, "fi": 0.9546701496929246, "ru": 1.067726309697796, "el": 0.8450858108782476, "hi": 0.8863141788695128, "ta": 0.8556234346897119, "th": 0.8868687837498974, "zh": 2.983178218357741, "my": 0.8666766921482625}, "t": 243.4}
{"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 2.06935, "de": 2.2713, "vi": 1.94537, "tr": 2.24205, "fi": 2.4306, "ru": 1.42456, "el": 1.52078, "hi": 1.21118, "ta": 1.009, "th": 1.14965, "zh": 3.1559, "my": 1.14895}, "bytes_per_chunk": {"en": 5.608, "de": 5.729, "vi": 7.248, "tr": 5.655, "fi": 5.545, "ru": 9.553, "el": 11.292, "hi": 15.094, "ta": 16.75, "th": 16.246, "zh": 5.084, "my": 18.895}, "bpb_mean": 1.79822, "t": 244.6}
{"event": "train", "step": 4400, "phase": 3, "objective": "none", "lm_loss": 1.44814, "bpb": 2.08923, "chunk_loss": 0.0, "F": 0.16138, "G": 0.20283, "chunks": 165.25, "bytes_per_chunk": 6.197, "boundary_entropy": 0.21162, "router_noise": 0.0, "grad_norm": 0.8303, "lr": 7.92e-05, "router_trainable": false, "sec_per_step": 0.083, "alpha": 0.0, "t": 244.7}
{"event": "train", "step": 4450, "phase": 3, "objective": "none", "lm_loss": 1.38045, "bpb": 1.99157, "chunk_loss": 0.0, "F": 0.14789, "G": 0.20317, "chunks": 151.44, "bytes_per_chunk": 6.762, "boundary_entropy": 0.24714, "router_noise": 0.0, "grad_norm": 0.7649, "lr": 7.63e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 246.6}
{"event": "train", "step": 4500, "phase": 3, "objective": "none", "lm_loss": 1.30988, "bpb": 1.88976, "chunk_loss": 0.0, "F": 0.15366, "G": 0.19946, "chunks": 157.34, "bytes_per_chunk": 6.508, "boundary_entropy": 0.21887, "router_noise": 0.0, "grad_norm": 0.7362, "lr": 7.36e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 248.6}
{"event": "train", "step": 4550, "phase": 3, "objective": "none", "lm_loss": 1.36453, "bpb": 1.96861, "chunk_loss": 0.0, "F": 0.1492, "G": 0.20324, "chunks": 152.78, "bytes_per_chunk": 6.702, "boundary_entropy": 0.24954, "router_noise": 0.0, "grad_norm": 0.7442, "lr": 7.09e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 250.6}
{"event": "probe", "step": 4600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3301.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.18968, "cps": 25.789062, "mean_p": 0.210979, "boundary_entropy": 0.14796, "bpb": 2.055383}, "de": {"chunks": 3718.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181074, "cps": 29.046875, "mean_p": 0.200384, "boundary_entropy": 0.158735, "bpb": 2.248548}, "vi": {"chunks": 3550.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.192568, "cps": 27.734375, "mean_p": 0.185738, "boundary_entropy": 0.243979, "bpb": 1.93627}, "tr": {"chunks": 3524.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194664, "cps": 27.53125, "mean_p": 0.202036, "boundary_entropy": 0.174207, "bpb": 2.220517}, "fi": {"chunks": 3337.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177396, "cps": 26.070312, "mean_p": 0.194156, "boundary_entropy": 0.160269, "bpb": 2.409072}, "ru": {"chunks": 3714.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.195989, "cps": 29.015625, "mean_p": 0.200581, "boundary_entropy": 0.366363, "bpb": 1.43927}, "el": {"chunks": 3181.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.154997, "cps": 24.851562, "mean_p": 0.213654, "boundary_entropy": 0.420623, "bpb": 1.516113}, "hi": {"chunks": 2954.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.175572, "cps": 23.078125, "mean_p": 0.141574, "boundary_entropy": 0.299733, "bpb": 1.21201}, "ta": {"chunks": 3350.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.179192, "cps": 26.171875, "mean_p": 0.13027, "boundary_entropy": 0.277165, "bpb": 0.996805}, "th": {"chunks": 2992.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.186012, "cps": 23.375, "mean_p": 0.139817, "boundary_entropy": 0.298371, "bpb": 1.116301}, "zh": {"chunks": 3218.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.54487, "cps": 25.140625, "mean_p": 0.28724, "boundary_entropy": 0.358885, "bpb": 3.145297}, "my": {"chunks": 2856.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.150728, "cps": 22.3125, "mean_p": 0.165636, "boundary_entropy": 0.375037, "bpb": 1.137066}}, "bpb": {"en": 2.05538, "de": 2.24855, "vi": 1.93627, "tr": 2.22052, "fi": 2.40907, "ru": 1.43927, "el": 1.51611, "hi": 1.21201, "ta": 0.9968, "th": 1.1163, "zh": 3.1453, "my": 1.13707}, "gini_cpc": 0.163525, "gini_cps": 0.046973, "bpic": 9.668245, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9546310899354971, "vi": 1.0152285139818815, "tr": 1.0262754597070454, "fi": 0.9352396680634004, "ru": 1.0332639585509789, "el": 0.817149312813165, "hi": 0.9256227366825218, "ta": 0.9447087403065147, "th": 0.9806614869864041, "zh": 2.8725737862481107, "my": 0.7946454900054275}, "t": 253.5}
{"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 2.063, "de": 2.25643, "vi": 1.93808, "tr": 2.22778, "fi": 2.4183, "ru": 1.41599, "el": 1.50756, "hi": 1.20743, "ta": 1.00124, "th": 1.13609, "zh": 3.14285, "my": 1.14409}, "bytes_per_chunk": {"en": 5.475, "de": 5.628, "vi": 7.274, "tr": 5.666, "fi": 5.507, "ru": 9.797, "el": 11.9, "hi": 14.091, "ta": 14.973, "th": 14.317, "zh": 5.127, "my": 19.035}, "bpb_mean": 1.78824, "t": 254.8}
{"event": "train", "step": 4600, "phase": 3, "objective": "none", "lm_loss": 1.40809, "bpb": 2.03145, "chunk_loss": 0.0, "F": 0.14456, "G": 0.20428, "chunks": 148.03, "bytes_per_chunk": 6.917, "boundary_entropy": 0.25849, "router_noise": 0.0, "grad_norm": 0.7301, "lr": 6.82e-05, "router_trainable": false, "sec_per_step": 0.084, "alpha": 0.0, "t": 254.8}
{"event": "train", "step": 4650, "phase": 3, "objective": "none", "lm_loss": 1.51331, "bpb": 2.18325, "chunk_loss": 0.0, "F": 0.15796, "G": 0.21181, "chunks": 161.75, "bytes_per_chunk": 6.331, "boundary_entropy": 0.25258, "router_noise": 0.0, "grad_norm": 0.7728, "lr": 6.57e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 256.7}
{"event": "train", "step": 4700, "phase": 3, "objective": "none", "lm_loss": 1.45266, "bpb": 2.09575, "chunk_loss": 0.0, "F": 0.14621, "G": 0.2079, "chunks": 149.72, "bytes_per_chunk": 6.839, "boundary_entropy": 0.26785, "router_noise": 0.0, "grad_norm": 0.8009, "lr": 6.32e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 258.7}
{"event": "train", "step": 4750, "phase": 3, "objective": "none", "lm_loss": 1.46247, "bpb": 2.10989, "chunk_loss": 0.0, "F": 0.15967, "G": 0.20279, "chunks": 163.5, "bytes_per_chunk": 6.263, "boundary_entropy": 0.21245, "router_noise": 0.0, "grad_norm": 0.7852, "lr": 6.08e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 260.6}
{"event": "probe", "step": 4800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3248.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186634, "cps": 25.375, "mean_p": 0.208421, "boundary_entropy": 0.148293, "bpb": 2.04305}, "de": {"chunks": 3670.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178737, "cps": 28.671875, "mean_p": 0.198723, "boundary_entropy": 0.160616, "bpb": 2.243564}, "vi": {"chunks": 3513.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.190561, "cps": 27.445312, "mean_p": 0.185659, "boundary_entropy": 0.246259, "bpb": 1.933029}, "tr": {"chunks": 3526.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194774, "cps": 27.546875, "mean_p": 0.204351, "boundary_entropy": 0.178774, "bpb": 2.210017}, "fi": {"chunks": 3311.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.176014, "cps": 25.867188, "mean_p": 0.193025, "boundary_entropy": 0.161577, "bpb": 2.389788}, "ru": {"chunks": 3714.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.195989, "cps": 29.015625, "mean_p": 0.203641, "boundary_entropy": 0.371474, "bpb": 1.432719}, "el": {"chunks": 3224.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.157092, "cps": 25.1875, "mean_p": 0.213432, "boundary_entropy": 0.420811, "bpb": 1.508266}, "hi": {"chunks": 3372.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.200416, "cps": 26.34375, "mean_p": 0.154917, "boundary_entropy": 0.316775, "bpb": 1.196523}, "ta": {"chunks": 2970.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.158866, "cps": 23.203125, "mean_p": 0.132202, "boundary_entropy": 0.283185, "bpb": 0.990988}, "th": {"chunks": 3349.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.208206, "cps": 26.164062, "mean_p": 0.146459, "boundary_entropy": 0.304535, "bpb": 1.110249}, "zh": {"chunks": 3322.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.562479, "cps": 25.953125, "mean_p": 0.294058, "boundary_entropy": 0.362289, "bpb": 3.13447}, "my": {"chunks": 3131.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.165242, "cps": 24.460938, "mean_p": 0.168687, "boundary_entropy": 0.373088, "bpb": 1.131811}}, "bpb": {"en": 2.04305, "de": 2.24356, "vi": 1.93303, "tr": 2.21002, "fi": 2.38979, "ru": 1.43272, "el": 1.50827, "hi": 1.19652, "ta": 0.99099, "th": 1.11025, "zh": 3.13447, "my": 1.13181}, "gini_cpc": 0.171567, "gini_cps": 0.034457, "bpic": 9.511301, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9576829525173822, "vi": 1.0210408257744892, "tr": 1.0436139030925664, "fi": 0.943094901369155, "ru": 1.0501244849682208, "el": 0.8417096137962717, "hi": 1.073842510924542, "ta": 0.8512146302388484, "th": 1.1155837560925563, "zh": 3.0137990046024834, "my": 0.885376093477508}, "t": 263.5}
{"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 2.05185, "de": 2.24977, "vi": 1.93416, "tr": 2.21745, "fi": 2.39891, "ru": 1.40774, "el": 1.50052, "hi": 1.19177, "ta": 0.99635, "th": 1.12888, "zh": 3.13183, "my": 1.13865}, "bytes_per_chunk": {"en": 5.585, "de": 5.694, "vi": 7.515, "tr": 5.659, "fi": 5.593, "ru": 9.785, "el": 11.645, "hi": 12.412, "ta": 16.749, "th": 13.097, "zh": 5.058, "my": 17.811}, "bpb_mean": 1.77899, "t": 264.7}
{"event": "train", "step": 4800, "phase": 3, "objective": "none", "lm_loss": 1.3524, "bpb": 1.95109, "chunk_loss": 0.0, "F": 0.14713, "G": 0.20244, "chunks": 150.66, "bytes_per_chunk": 6.797, "boundary_entropy": 0.24742, "router_noise": 0.0, "grad_norm": 0.7442, "lr": 5.85e-05, "router_trainable": false, "sec_per_step": 0.083, "alpha": 0.0, "t": 264.8}
{"event": "train", "step": 4850, "phase": 3, "objective": "none", "lm_loss": 1.28454, "bpb": 1.85321, "chunk_loss": 0.0, "F": 0.14685, "G": 0.19261, "chunks": 150.38, "bytes_per_chunk": 6.81, "boundary_entropy": 0.21345, "router_noise": 0.0, "grad_norm": 0.7366, "lr": 5.62e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 266.7}
{"event": "train", "step": 4900, "phase": 3, "objective": "none", "lm_loss": 1.38404, "bpb": 1.99675, "chunk_loss": 0.0, "F": 0.14783, "G": 0.19796, "chunks": 151.38, "bytes_per_chunk": 6.765, "boundary_entropy": 0.23717, "router_noise": 0.0, "grad_norm": 0.7977, "lr": 5.41e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 268.7}
{"event": "train", "step": 4950, "phase": 3, "objective": "none", "lm_loss": 1.26281, "bpb": 1.82186, "chunk_loss": 0.0, "F": 0.14182, "G": 0.19561, "chunks": 145.22, "bytes_per_chunk": 7.051, "boundary_entropy": 0.25442, "router_noise": 0.0, "grad_norm": 0.7586, "lr": 5.2e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 270.6}
{"event": "probe", "step": 5000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3287.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188875, "cps": 25.679688, "mean_p": 0.21337, "boundary_entropy": 0.151298, "bpb": 2.037022}, "de": {"chunks": 3791.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18463, "cps": 29.617188, "mean_p": 0.204592, "boundary_entropy": 0.163423, "bpb": 2.23567}, "vi": {"chunks": 3557.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.192948, "cps": 27.789062, "mean_p": 0.187656, "boundary_entropy": 0.245977, "bpb": 1.924306}, "tr": {"chunks": 3555.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196376, "cps": 27.773438, "mean_p": 0.206452, "boundary_entropy": 0.179864, "bpb": 2.200449}, "fi": {"chunks": 3371.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.179204, "cps": 26.335938, "mean_p": 0.197333, "boundary_entropy": 0.162638, "bpb": 2.375876}, "ru": {"chunks": 3811.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.201108, "cps": 29.773438, "mean_p": 0.201438, "boundary_entropy": 0.365339, "bpb": 1.425368}, "el": {"chunks": 3153.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.153633, "cps": 24.632812, "mean_p": 0.210663, "boundary_entropy": 0.416261, "bpb": 1.500728}, "hi": {"chunks": 3454.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.20529, "cps": 26.984375, "mean_p": 0.156852, "boundary_entropy": 0.321229, "bpb": 1.194648}, "ta": {"chunks": 2828.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.15127, "cps": 22.09375, "mean_p": 0.132299, "boundary_entropy": 0.28729, "bpb": 0.989357}, "th": {"chunks": 3543.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.220267, "cps": 27.679688, "mean_p": 0.149957, "boundary_entropy": 0.308595, "bpb": 1.111836}, "zh": {"chunks": 3338.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.565188, "cps": 26.078125, "mean_p": 0.29355, "boundary_entropy": 0.360043, "bpb": 3.125366}, "my": {"chunks": 2888.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.152417, "cps": 22.5625, "mean_p": 0.158815, "boundary_entropy": 0.360506, "bpb": 1.12867}}, "bpb": {"en": 2.03702, "de": 2.23567, "vi": 1.92431, "tr": 2.20045, "fi": 2.37588, "ru": 1.42537, "el": 1.50073, "hi": 1.19465, "ta": 0.98936, "th": 1.11184, "zh": 3.12537, "my": 1.12867}, "gini_cpc": 0.182482, "gini_cps": 0.048892, "bpic": 9.458325, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9775203186761281, "vi": 1.021562963599237, "tr": 1.0397129893958885, "fi": 0.9487925921302868, "ru": 1.0647659405044099, "el": 0.8134063185929502, "hi": 1.0869052248241644, "ta": 0.8009001071025256, "th": 1.1662039369303572, "zh": 2.992383875112283, "my": 0.8069715595221971}, "t": 273.5}
{"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 2.04537, "de": 2.24261, "vi": 1.9261, "tr": 2.2079, "fi": 2.38719, "ru": 1.40168, "el": 1.49324, "hi": 1.18887, "ta": 0.99492, "th": 1.13141, "zh": 3.12093, "my": 1.13552}, "bytes_per_chunk": {"en": 5.476, "de": 5.568, "vi": 7.305, "tr": 5.555, "fi": 5.426, "ru": 9.425, "el": 11.74, "hi": 11.879, "ta": 17.415, "th": 12.275, "zh": 4.98, "my": 18.492}, "bpb_mean": 1.77298, "t": 274.7}
{"event": "train", "step": 5000, "phase": 3, "objective": "none", "lm_loss": 1.37113, "bpb": 1.97812, "chunk_loss": 0.0, "F": 0.14026, "G": 0.21228, "chunks": 143.62, "bytes_per_chunk": 7.13, "boundary_entropy": 0.29219, "router_noise": 0.0, "grad_norm": 0.7544, "lr": 5e-05, "router_trainable": false, "sec_per_step": 0.083, "alpha": 0.0, "t": 274.8}
{"event": "train", "step": 5050, "phase": 3, "objective": "none", "lm_loss": 1.39254, "bpb": 2.00901, "chunk_loss": 0.0, "F": 0.14178, "G": 0.20541, "chunks": 145.19, "bytes_per_chunk": 7.053, "boundary_entropy": 0.26961, "router_noise": 0.0, "grad_norm": 0.9195, "lr": 4.81e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 276.7}
{"event": "train", "step": 5100, "phase": 3, "objective": "none", "lm_loss": 1.47513, "bpb": 2.12817, "chunk_loss": 0.0, "F": 0.15771, "G": 0.20894, "chunks": 161.5, "bytes_per_chunk": 6.341, "boundary_entropy": 0.24503, "router_noise": 0.0, "grad_norm": 0.7515, "lr": 4.63e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 278.7}
{"event": "train", "step": 5150, "phase": 3, "objective": "none", "lm_loss": 1.3392, "bpb": 1.93206, "chunk_loss": 0.0, "F": 0.15311, "G": 0.20188, "chunks": 156.78, "bytes_per_chunk": 6.531, "boundary_entropy": 0.23597, "router_noise": 0.0, "grad_norm": 0.7638, "lr": 4.45e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 280.6}
{"event": "probe", "step": 5200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3250.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186749, "cps": 25.390625, "mean_p": 0.209427, "boundary_entropy": 0.148528, "bpb": 2.029719}, "de": {"chunks": 3714.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18088, "cps": 29.015625, "mean_p": 0.200345, "boundary_entropy": 0.160188, "bpb": 2.22894}, "vi": {"chunks": 3578.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.194087, "cps": 27.953125, "mean_p": 0.185214, "boundary_entropy": 0.241621, "bpb": 1.921395}, "tr": {"chunks": 3525.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194719, "cps": 27.539062, "mean_p": 0.203724, "boundary_entropy": 0.17764, "bpb": 2.189357}, "fi": {"chunks": 3335.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.17729, "cps": 26.054688, "mean_p": 0.193382, "boundary_entropy": 0.160333, "bpb": 2.363891}, "ru": {"chunks": 3972.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.209604, "cps": 31.03125, "mean_p": 0.207679, "boundary_entropy": 0.372099, "bpb": 1.416543}, "el": {"chunks": 3348.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.163134, "cps": 26.15625, "mean_p": 0.220971, "boundary_entropy": 0.429457, "bpb": 1.494463}, "hi": {"chunks": 3843.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.22841, "cps": 30.023438, "mean_p": 0.162206, "boundary_entropy": 0.322965, "bpb": 1.183921}, "ta": {"chunks": 3688.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.197272, "cps": 28.8125, "mean_p": 0.14501, "boundary_entropy": 0.297341, "bpb": 0.981124}, "th": {"chunks": 3696.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.229779, "cps": 28.875, "mean_p": 0.155454, "boundary_entropy": 0.31308, "bpb": 1.098589}, "zh": {"chunks": 3344.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.566204, "cps": 26.125, "mean_p": 0.298741, "boundary_entropy": 0.369503, "bpb": 3.118379}, "my": {"chunks": 2900.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.15305, "cps": 22.65625, "mean_p": 0.160215, "boundary_entropy": 0.361211, "bpb": 1.121743}}, "bpb": {"en": 2.02972, "de": 2.22894, "vi": 1.9214, "tr": 2.18936, "fi": 2.36389, "ru": 1.41654, "el": 1.49446, "hi": 1.18392, "ta": 0.98112, "th": 1.09859, "zh": 3.11838, "my": 1.12174}, "gini_cpc": 0.176482, "gini_cps": 0.044666, "bpic": 9.095845, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9685683009339561, "vi": 1.0392928835200599, "tr": 1.0426758845750173, "fi": 0.9493464135141876, "ru": 1.1223822366551655, "el": 0.8735451632127555, "hi": 1.2230833969596524, "ta": 1.0563460242351925, "th": 1.230415112025059, "zh": 3.0318910104457, "my": 0.8195498611584742}, "t": 283.5}
{"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 2.03862, "de": 2.23471, "vi": 1.92156, "tr": 2.19704, "fi": 2.37333, "ru": 1.3939, "el": 1.48644, "hi": 1.17844, "ta": 0.987, "th": 1.11782, "zh": 3.11528, "my": 1.12883}, "bytes_per_chunk": {"en": 5.538, "de": 5.632, "vi": 7.205, "tr": 5.63, "fi": 5.552, "ru": 9.104, "el": 11.144, "hi": 10.948, "ta": 13.546, "th": 11.886, "zh": 4.981, "my": 18.633}, "bpb_mean": 1.76441, "t": 284.7}
{"event": "train", "step": 5200, "phase": 3, "objective": "none", "lm_loss": 1.25067, "bpb": 1.80434, "chunk_loss": 0.0, "F": 0.13812, "G": 0.19893, "chunks": 141.44, "bytes_per_chunk": 7.24, "boundary_entropy": 0.26582, "router_noise": 0.0, "grad_norm": 0.7481, "lr": 4.29e-05, "router_trainable": false, "sec_per_step": 0.083, "alpha": 0.0, "t": 284.8}
{"event": "train", "step": 5250, "phase": 3, "objective": "none", "lm_loss": 1.3039, "bpb": 1.88113, "chunk_loss": 0.0, "F": 0.14484, "G": 0.20668, "chunks": 148.31, "bytes_per_chunk": 6.904, "boundary_entropy": 0.26979, "router_noise": 0.0, "grad_norm": 0.7532, "lr": 4.14e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 286.7}
{"event": "train", "step": 5300, "phase": 3, "objective": "none", "lm_loss": 1.23346, "bpb": 1.77951, "chunk_loss": 0.0, "F": 0.13824, "G": 0.19418, "chunks": 141.56, "bytes_per_chunk": 7.234, "boundary_entropy": 0.25703, "router_noise": 0.0, "grad_norm": 0.7222, "lr": 3.99e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 288.7}
{"event": "train", "step": 5350, "phase": 3, "objective": "none", "lm_loss": 1.45227, "bpb": 2.09518, "chunk_loss": 0.0, "F": 0.15033, "G": 0.2186, "chunks": 153.94, "bytes_per_chunk": 6.652, "boundary_entropy": 0.28664, "router_noise": 0.0, "grad_norm": 0.7904, "lr": 3.86e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 290.6}
{"event": "probe", "step": 5400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3260.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187324, "cps": 25.46875, "mean_p": 0.209967, "boundary_entropy": 0.148105, "bpb": 2.028988}, "de": {"chunks": 3717.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181026, "cps": 29.039062, "mean_p": 0.201049, "boundary_entropy": 0.159646, "bpb": 2.223114}, "vi": {"chunks": 3545.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.192297, "cps": 27.695312, "mean_p": 0.187779, "boundary_entropy": 0.25011, "bpb": 1.917264}, "tr": {"chunks": 3574.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197426, "cps": 27.921875, "mean_p": 0.206429, "boundary_entropy": 0.177901, "bpb": 2.184012}, "fi": {"chunks": 3359.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.178566, "cps": 26.242188, "mean_p": 0.195422, "boundary_entropy": 0.160532, "bpb": 2.35322}, "ru": {"chunks": 3803.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.200686, "cps": 29.710938, "mean_p": 0.204381, "boundary_entropy": 0.371074, "bpb": 1.408751}, "el": {"chunks": 3341.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.162793, "cps": 26.101562, "mean_p": 0.218288, "boundary_entropy": 0.42542, "bpb": 1.485684}, "hi": {"chunks": 3608.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.214443, "cps": 28.1875, "mean_p": 0.158951, "boundary_entropy": 0.321477, "bpb": 1.180101}, "ta": {"chunks": 3331.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.178176, "cps": 26.023438, "mean_p": 0.140565, "boundary_entropy": 0.293201, "bpb": 0.977302}, "th": {"chunks": 3420.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.21262, "cps": 26.71875, "mean_p": 0.151892, "boundary_entropy": 0.31359, "bpb": 1.096038}, "zh": {"chunks": 3330.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.563833, "cps": 26.015625, "mean_p": 0.297667, "boundary_entropy": 0.366965, "bpb": 3.106752}, "my": {"chunks": 2671.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.140965, "cps": 20.867188, "mean_p": 0.162732, "boundary_entropy": 0.373594, "bpb": 1.124132}}, "bpb": {"en": 2.02899, "de": 2.22311, "vi": 1.91726, "tr": 2.18401, "fi": 2.35322, "ru": 1.40875, "el": 1.48568, "hi": 1.1801, "ta": 0.9773, "th": 1.09604, "zh": 3.10675, "my": 1.12413}, "gini_cpc": 0.176308, "gini_cps": 0.041503, "bpic": 9.369882, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9663771979805664, "vi": 1.02654884264228, "tr": 1.05392696665197, "fi": 0.9532452295077316, "ru": 1.0713308998494586, "el": 0.8690447706583276, "hi": 1.1447693050984968, "ta": 0.9511646104647252, "th": 1.1350410301103608, "zh": 3.0099363442397142, "my": 0.7525182417232309}, "t": 293.5}
{"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 2.03665, "de": 2.22944, "vi": 1.91761, "tr": 2.19102, "fi": 2.36384, "ru": 1.38591, "el": 1.47882, "hi": 1.17357, "ta": 0.98328, "th": 1.11547, "zh": 3.10366, "my": 1.13139}, "bytes_per_chunk": {"en": 5.538, "de": 5.655, "vi": 7.41, "tr": 5.552, "fi": 5.466, "ru": 9.485, "el": 11.229, "hi": 11.586, "ta": 15.15, "th": 12.687, "zh": 4.978, "my": 20.746}, "bpb_mean": 1.75922, "t": 294.7}
{"event": "train", "step": 5400, "phase": 3, "objective": "none", "lm_loss": 1.35663, "bpb": 1.95721, "chunk_loss": 0.0, "F": 0.15811, "G": 0.21076, "chunks": 161.91, "bytes_per_chunk": 6.325, "boundary_entropy": 0.24835, "router_noise": 0.0, "grad_norm": 0.8295, "lr": 3.73e-05, "router_trainable": false, "sec_per_step": 0.084, "alpha": 0.0, "t": 294.8}
{"event": "train", "step": 5450, "phase": 3, "objective": "none", "lm_loss": 1.21498, "bpb": 1.75284, "chunk_loss": 0.0, "F": 0.13773, "G": 0.20281, "chunks": 141.03, "bytes_per_chunk": 7.261, "boundary_entropy": 0.27954, "router_noise": 0.0, "grad_norm": 0.7581, "lr": 3.62e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 296.8}
{"event": "train", "step": 5500, "phase": 3, "objective": "none", "lm_loss": 1.31448, "bpb": 1.8964, "chunk_loss": 0.0, "F": 0.15134, "G": 0.20583, "chunks": 154.97, "bytes_per_chunk": 6.608, "boundary_entropy": 0.25192, "router_noise": 0.0, "grad_norm": 0.7955, "lr": 3.51e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 298.7}
{"event": "train", "step": 5550, "phase": 3, "objective": "none", "lm_loss": 1.38908, "bpb": 2.00402, "chunk_loss": 0.0, "F": 0.1528, "G": 0.20622, "chunks": 156.47, "bytes_per_chunk": 6.544, "boundary_entropy": 0.23792, "router_noise": 0.0, "grad_norm": 0.8256, "lr": 3.41e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 300.7}
{"event": "probe", "step": 5600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3271.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187956, "cps": 25.554688, "mean_p": 0.211374, "boundary_entropy": 0.150292, "bpb": 2.023352}, "de": {"chunks": 3733.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181805, "cps": 29.164062, "mean_p": 0.20186, "boundary_entropy": 0.159998, "bpb": 2.2153}, "vi": {"chunks": 3571.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.193708, "cps": 27.898438, "mean_p": 0.191504, "boundary_entropy": 0.256291, "bpb": 1.91165}, "tr": {"chunks": 3531.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195051, "cps": 27.585938, "mean_p": 0.205086, "boundary_entropy": 0.179693, "bpb": 2.1753}, "fi": {"chunks": 3335.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.17729, "cps": 26.054688, "mean_p": 0.194853, "boundary_entropy": 0.162363, "bpb": 2.352056}, "ru": {"chunks": 3813.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.201214, "cps": 29.789062, "mean_p": 0.202213, "boundary_entropy": 0.368303, "bpb": 1.403893}, "el": {"chunks": 3138.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.152902, "cps": 24.515625, "mean_p": 0.217705, "boundary_entropy": 0.426887, "bpb": 1.48325}, "hi": {"chunks": 3515.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.208915, "cps": 27.460938, "mean_p": 0.157078, "boundary_entropy": 0.318411, "bpb": 1.177105}, "ta": {"chunks": 2850.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.152447, "cps": 22.265625, "mean_p": 0.132289, "boundary_entropy": 0.288895, "bpb": 0.978835}, "th": {"chunks": 3108.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.193224, "cps": 24.28125, "mean_p": 0.146149, "boundary_entropy": 0.309959, "bpb": 1.091557}, "zh": {"chunks": 3356.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.568236, "cps": 26.21875, "mean_p": 0.298135, "boundary_entropy": 0.364754, "bpb": 3.10123}, "my": {"chunks": 3303.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.174319, "cps": 25.804688, "mean_p": 0.176827, "boundary_entropy": 0.38498, "bpb": 1.118453}}, "bpb": {"en": 2.02335, "de": 2.2153, "vi": 1.91165, "tr": 2.1753, "fi": 2.35206, "ru": 1.40389, "el": 1.48325, "hi": 1.1771, "ta": 0.97884, "th": 1.09156, "zh": 3.10123, "my": 1.11845}, "gini_cpc": 0.173808, "gini_cps": 0.043608, "bpic": 9.470462, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9672732084327481, "vi": 1.0306003469103313, "tr": 1.037745221376919, "fi": 0.9432515572978019, "ru": 1.0705357305625756, "el": 0.8134964651734102, "hi": 1.111511172749131, "ta": 0.8110786443045529, "th": 1.0280246325909046, "zh": 3.0232362448356564, "my": 0.9274463015708505}, "t": 303.6}
{"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 2.03034, "de": 2.222, "vi": 1.91257, "tr": 2.1831, "fi": 2.36148, "ru": 1.3809, "el": 1.47636, "hi": 1.17199, "ta": 0.98397, "th": 1.1102, "zh": 3.0968, "my": 1.1252}, "bytes_per_chunk": {"en": 5.529, "de": 5.638, "vi": 7.21, "tr": 5.64, "fi": 5.51, "ru": 9.534, "el": 11.79, "hi": 11.676, "ta": 17.204, "th": 13.863, "zh": 4.899, "my": 16.413}, "bpb_mean": 1.75458, "t": 304.9}
{"event": "train", "step": 5600, "phase": 3, "objective": "none", "lm_loss": 1.3363, "bpb": 1.92788, "chunk_loss": 0.0, "F": 0.14545, "G": 0.20212, "chunks": 148.94, "bytes_per_chunk": 6.875, "boundary_entropy": 0.2534, "router_noise": 0.0, "grad_norm": 0.7737, "lr": 3.33e-05, "router_trainable": false, "sec_per_step": 0.084, "alpha": 0.0, "t": 304.9}
{"event": "train", "step": 5650, "phase": 3, "objective": "none", "lm_loss": 1.31983, "bpb": 1.90411, "chunk_loss": 0.0, "F": 0.1445, "G": 0.2053, "chunks": 147.97, "bytes_per_chunk": 6.92, "boundary_entropy": 0.26552, "router_noise": 0.0, "grad_norm": 0.8295, "lr": 3.25e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 306.9}
{"event": "train", "step": 5700, "phase": 3, "objective": "none", "lm_loss": 1.29171, "bpb": 1.86354, "chunk_loss": 0.0, "F": 0.14972, "G": 0.19723, "chunks": 153.31, "bytes_per_chunk": 6.679, "boundary_entropy": 0.22786, "router_noise": 0.0, "grad_norm": 0.7681, "lr": 3.18e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 308.9}
{"event": "train", "step": 5750, "phase": 3, "objective": "none", "lm_loss": 1.20382, "bpb": 1.73675, "chunk_loss": 0.0, "F": 0.14233, "G": 0.19846, "chunks": 145.75, "bytes_per_chunk": 7.026, "boundary_entropy": 0.25131, "router_noise": 0.0, "grad_norm": 0.7475, "lr": 3.13e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 310.9}
{"event": "probe", "step": 5800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3274.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188128, "cps": 25.578125, "mean_p": 0.211147, "boundary_entropy": 0.150746, "bpb": 2.017304}, "de": {"chunks": 3701.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.180246, "cps": 28.914062, "mean_p": 0.199872, "boundary_entropy": 0.158195, "bpb": 2.210803}, "vi": {"chunks": 3593.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.194901, "cps": 28.070312, "mean_p": 0.191377, "boundary_entropy": 0.256351, "bpb": 1.908559}, "tr": {"chunks": 3540.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195548, "cps": 27.65625, "mean_p": 0.206083, "boundary_entropy": 0.178878, "bpb": 2.169685}, "fi": {"chunks": 3369.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.179097, "cps": 26.320312, "mean_p": 0.197951, "boundary_entropy": 0.165572, "bpb": 2.340456}, "ru": {"chunks": 4087.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.215673, "cps": 31.929688, "mean_p": 0.212454, "boundary_entropy": 0.377365, "bpb": 1.398866}, "el": {"chunks": 3636.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.177167, "cps": 28.40625, "mean_p": 0.231296, "boundary_entropy": 0.441686, "bpb": 1.476448}, "hi": {"chunks": 3428.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.203744, "cps": 26.78125, "mean_p": 0.160022, "boundary_entropy": 0.326731, "bpb": 1.173535}, "ta": {"chunks": 3080.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.16475, "cps": 24.0625, "mean_p": 0.138479, "boundary_entropy": 0.296153, "bpb": 0.972992}, "th": {"chunks": 3453.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.214672, "cps": 26.976562, "mean_p": 0.150194, "boundary_entropy": 0.311614, "bpb": 1.08902}, "zh": {"chunks": 3332.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.564172, "cps": 26.03125, "mean_p": 0.297636, "boundary_entropy": 0.366551, "bpb": 3.093909}, "my": {"chunks": 3506.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.185033, "cps": 27.390625, "mean_p": 0.175821, "boundary_entropy": 0.378006, "bpb": 1.115177}}, "bpb": {"en": 2.0173, "de": 2.2108, "vi": 1.90856, "tr": 2.16969, "fi": 2.34046, "ru": 1.39887, "el": 1.47645, "hi": 1.17353, "ta": 0.97299, "th": 1.08902, "zh": 3.09391, "my": 1.11518}, "gini_cpc": 0.160995, "gini_cps": 0.03682, "bpic": 9.13786, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9581028301923559, "vi": 1.0359994393284266, "tr": 1.0394369610588106, "fi": 0.9519947922956865, "ru": 1.1464123831643893, "el": 0.9417344861915877, "hi": 1.0830068049316466, "ta": 0.8757309365872737, "th": 1.1410927819508039, "zh": 2.9988655501037327, "my": 0.9835444245118524}, "t": 313.8}
{"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 2.02525, "de": 2.21759, "vi": 1.90951, "tr": 2.17639, "fi": 2.35066, "ru": 1.37627, "el": 1.46869, "hi": 1.16757, "ta": 0.97947, "th": 1.10788, "zh": 3.09057, "my": 1.12232}, "bytes_per_chunk": {"en": 5.517, "de": 5.679, "vi": 7.212, "tr": 5.577, "fi": 5.422, "ru": 8.836, "el": 10.341, "hi": 12.23, "ta": 15.87, "th": 12.431, "zh": 4.938, "my": 16.194}, "bpb_mean": 1.74935, "t": 315.1}
{"event": "train", "step": 5800, "phase": 3, "objective": "none", "lm_loss": 1.29316, "bpb": 1.86564, "chunk_loss": 0.0, "F": 0.14719, "G": 0.20031, "chunks": 150.72, "bytes_per_chunk": 6.794, "boundary_entropy": 0.24858, "router_noise": 0.0, "grad_norm": 0.7987, "lr": 3.08e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 315.1}
{"event": "train", "step": 5850, "phase": 3, "objective": "none", "lm_loss": 1.28134, "bpb": 1.84858, "chunk_loss": 0.0, "F": 0.14438, "G": 0.20651, "chunks": 147.84, "bytes_per_chunk": 6.926, "boundary_entropy": 0.27056, "router_noise": 0.0, "grad_norm": 0.7541, "lr": 3.05e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 317.1}
{"event": "train", "step": 5900, "phase": 3, "objective": "none", "lm_loss": 1.35891, "bpb": 1.9605, "chunk_loss": 0.0, "F": 0.1463, "G": 0.20749, "chunks": 149.81, "bytes_per_chunk": 6.835, "boundary_entropy": 0.27166, "router_noise": 0.0, "grad_norm": 0.8368, "lr": 3.02e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 319.1}
{"event": "train", "step": 5950, "phase": 3, "objective": "none", "lm_loss": 1.39412, "bpb": 2.0113, "chunk_loss": 0.0, "F": 0.15527, "G": 0.21862, "chunks": 159.0, "bytes_per_chunk": 6.44, "boundary_entropy": 0.27603, "router_noise": 0.0, "grad_norm": 0.7855, "lr": 3.01e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 321.1}
{"event": "probe", "step": 6000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3264.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187554, "cps": 25.5, "mean_p": 0.211714, "boundary_entropy": 0.151535, "bpb": 2.013746}, "de": {"chunks": 3759.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.183071, "cps": 29.367188, "mean_p": 0.20365, "boundary_entropy": 0.163728, "bpb": 2.20705}, "vi": {"chunks": 3608.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.195715, "cps": 28.1875, "mean_p": 0.190464, "boundary_entropy": 0.253208, "bpb": 1.904504}, "tr": {"chunks": 3555.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196376, "cps": 27.773438, "mean_p": 0.206584, "boundary_entropy": 0.179838, "bpb": 2.165635}, "fi": {"chunks": 3360.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.178619, "cps": 26.25, "mean_p": 0.196732, "boundary_entropy": 0.164138, "bpb": 2.341374}, "ru": {"chunks": 4079.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.215251, "cps": 31.867188, "mean_p": 0.213091, "boundary_entropy": 0.377455, "bpb": 1.395942}, "el": {"chunks": 3271.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.159382, "cps": 25.554688, "mean_p": 0.220017, "boundary_entropy": 0.428675, "bpb": 1.473141}, "hi": {"chunks": 3488.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.207311, "cps": 27.25, "mean_p": 0.15794, "boundary_entropy": 0.32086, "bpb": 1.171004}, "ta": {"chunks": 3171.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.169618, "cps": 24.773438, "mean_p": 0.137445, "boundary_entropy": 0.293257, "bpb": 0.969614}, "th": {"chunks": 3264.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.202922, "cps": 25.5, "mean_p": 0.146132, "boundary_entropy": 0.30671, "bpb": 1.08654}, "zh": {"chunks": 3333.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.564341, "cps": 26.039062, "mean_p": 0.298487, "boundary_entropy": 0.364217, "bpb": 3.089451}, "my": {"chunks": 3163.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.166931, "cps": 24.710938, "mean_p": 0.164833, "boundary_entropy": 0.367127, "bpb": 1.113552}}, "bpb": {"en": 2.01375, "de": 2.20705, "vi": 1.9045, "tr": 2.16564, "fi": 2.34137, "ru": 1.39594, "el": 1.47314, "hi": 1.171, "ta": 0.96961, "th": 1.08654, "zh": 3.08945, "my": 1.11355}, "gini_cpc": 0.168846, "gini_cps": 0.040157, "bpic": 9.289144, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9760990467998429, "vi": 1.04351178225562, "tr": 1.0470393983285189, "fi": 0.952360477442798, "ru": 1.1476737835635573, "el": 0.8497940169709165, "hi": 1.10533869417009, "ta": 0.9043670747919387, "th": 1.0819396953683555, "zh": 3.00895602801737, "my": 0.89004053469868}, "t": 324.0}
{"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 2.02244, "de": 2.21562, "vi": 1.90581, "tr": 2.17226, "fi": 2.35179, "ru": 1.37368, "el": 1.4658, "hi": 1.16471, "ta": 0.97609, "th": 1.10487, "zh": 3.08598, "my": 1.12053}, "bytes_per_chunk": {"en": 5.518, "de": 5.618, "vi": 7.239, "tr": 5.603, "fi": 5.462, "ru": 8.818, "el": 11.468, "hi": 11.917, "ta": 15.711, "th": 13.097, "zh": 4.952, "my": 18.256}, "bpb_mean": 1.74663, "t": 325.3}
{"event": "probe", "step": 6000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3264.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187554, "cps": 25.5, "mean_p": 0.211714, "boundary_entropy": 0.151535, "bpb": 2.013746}, "de": {"chunks": 3759.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.183071, "cps": 29.367188, "mean_p": 0.20365, "boundary_entropy": 0.163728, "bpb": 2.20705}, "vi": {"chunks": 3608.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.195715, "cps": 28.1875, "mean_p": 0.190464, "boundary_entropy": 0.253208, "bpb": 1.904504}, "tr": {"chunks": 3555.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196376, "cps": 27.773438, "mean_p": 0.206584, "boundary_entropy": 0.179838, "bpb": 2.165635}, "fi": {"chunks": 3360.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.178619, "cps": 26.25, "mean_p": 0.196732, "boundary_entropy": 0.164138, "bpb": 2.341374}, "ru": {"chunks": 4079.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.215251, "cps": 31.867188, "mean_p": 0.213091, "boundary_entropy": 0.377455, "bpb": 1.395942}, "el": {"chunks": 3271.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.159382, "cps": 25.554688, "mean_p": 0.220017, "boundary_entropy": 0.428675, "bpb": 1.473141}, "hi": {"chunks": 3488.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.207311, "cps": 27.25, "mean_p": 0.15794, "boundary_entropy": 0.32086, "bpb": 1.171004}, "ta": {"chunks": 3171.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.169618, "cps": 24.773438, "mean_p": 0.137445, "boundary_entropy": 0.293257, "bpb": 0.969614}, "th": {"chunks": 3264.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.202922, "cps": 25.5, "mean_p": 0.146132, "boundary_entropy": 0.30671, "bpb": 1.08654}, "zh": {"chunks": 3333.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.564341, "cps": 26.039062, "mean_p": 0.298487, "boundary_entropy": 0.364217, "bpb": 3.089451}, "my": {"chunks": 3163.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.166931, "cps": 24.710938, "mean_p": 0.164833, "boundary_entropy": 0.367127, "bpb": 1.113552}}, "bpb": {"en": 2.01375, "de": 2.20705, "vi": 1.9045, "tr": 2.16564, "fi": 2.34137, "ru": 1.39594, "el": 1.47314, "hi": 1.171, "ta": 0.96961, "th": 1.08654, "zh": 3.08945, "my": 1.11355}, "gini_cpc": 0.168846, "gini_cps": 0.040157, "bpic": 9.289144, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9760990467998429, "vi": 1.04351178225562, "tr": 1.0470393983285189, "fi": 0.952360477442798, "ru": 1.1476737835635573, "el": 0.8497940169709165, "hi": 1.10533869417009, "ta": 0.9043670747919387, "th": 1.0819396953683555, "zh": 3.00895602801737, "my": 0.89004053469868}, "t": 326.2}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_B_s2/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_B_s2/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_B_s2/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_B_s2/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_B_s2/th0.70"], "t": 328.2}
{"event": "final", "run": "pilot_B_s2", "condition": "B", "step": 6000, "params": 22516992, "per_lang": {"en": {"cpc": 0.18755387002240992, "cps": 25.5, "bpb": 2.031368200658346, "bytes_per_chunk": 5.507979030749997}, "de": {"cpc": 0.18307115375249597, "cps": 29.3671875, "bpb": 2.20476016335365, "bytes_per_chunk": 5.634173338348289}, "vi": {"cpc": 0.19571467317602387, "cps": 28.1875, "bpb": 1.9099308209274897, "bytes_per_chunk": 7.275108927927178}, "tr": {"cpc": 0.19637629122244932, "cps": 27.7734375, "bpb": 2.1655820640326437, "bytes_per_chunk": 5.618655692729766}, "fi": {"cpc": 0.17861889320078678, "cps": 26.25, "bpb": 2.367652628209442, "bytes_per_chunk": 5.483782568221992}, "ru": {"cpc": 0.21525065963060686, "cps": 31.8671875, "bpb": 1.3740306504389745, "bytes_per_chunk": 8.823871955837555}, "el": {"cpc": 0.15938215660478489, "cps": 25.5546875, "bpb": 1.461245057364279, "bytes_per_chunk": 11.419162328752206}, "hi": {"cpc": 0.20731054977711738, "cps": 27.25, "bpb": 1.173552301643864, "bytes_per_chunk": 11.89832970225127}, "ta": {"cpc": 0.16961754479807434, "cps": 24.7734375, "bpb": 0.975765793053311, "bytes_per_chunk": 15.744384384384384}, "th": {"cpc": 0.20292197699720235, "cps": 25.5, "bpb": 1.1010206340641642, "bytes_per_chunk": 13.037449644402447}, "zh": {"cpc": 0.5643413477819167, "cps": 26.0390625, "bpb": 3.0928095428727036, "bytes_per_chunk": 4.946859903381642}, "my": {"cpc": 0.16693054675955246, "cps": 24.7109375, "bpb": 1.1309327542366832, "bytes_per_chunk": 18.21329813103592}}, "gini_cpc": 0.16884574026563004, "gini_cps": 0.04015692444229302, "bpic": 9.289144378555005, "seed": 2, "mean_boundary_entropy": 0.2723730802536011, "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": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 131584}], "t": 330.6}