| {"event": "start", "run": "exp32_local_s0", "condition": "A", "params": 82470528, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.184, "vi": 1.39, "tr": 1.12, "fi": 1.104, "ru": 1.971, "el": 2.174, "hi": 2.552, "ta": 3.192, "th": 2.762, "zh": 0.915, "my": 3.524}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "global", "p3_objective": "global", "freeze_in_p3": false, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 2.1} |
| {"event": "probe_schedule", "n": 21, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 11000], "t": 2.1} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8633.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.496064, "cps": 67.445312, "mean_p": 0.503405, "boundary_entropy": 0.687423, "bpb": 8.245545}, "de": {"chunks": 12185.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.593435, "cps": 95.195312, "mean_p": 0.507242, "boundary_entropy": 0.68817, "bpb": 8.205783}, "vi": {"chunks": 6146.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.333388, "cps": 48.015625, "mean_p": 0.490614, "boundary_entropy": 0.688472, "bpb": 8.129058}, "tr": {"chunks": 11394.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.629398, "cps": 89.015625, "mean_p": 0.506619, "boundary_entropy": 0.687865, "bpb": 8.138517}, "fi": {"chunks": 7257.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.385785, "cps": 56.695312, "mean_p": 0.49745, "boundary_entropy": 0.687937, "bpb": 8.227952}, "ru": {"chunks": 10390.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.548285, "cps": 81.171875, "mean_p": 0.49637, "boundary_entropy": 0.69018, "bpb": 8.065313}, "el": {"chunks": 36558.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.781319, "cps": 285.609375, "mean_p": 0.529623, "boundary_entropy": 0.688846, "bpb": 8.137822}, "hi": {"chunks": 42514.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.526835, "cps": 332.140625, "mean_p": 0.520835, "boundary_entropy": 0.690193, "bpb": 7.879441}, "ta": {"chunks": 49746.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.660925, "cps": 388.640625, "mean_p": 0.521454, "boundary_entropy": 0.690378, "bpb": 7.812101}, "th": {"chunks": 45763.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.845073, "cps": 357.523438, "mean_p": 0.525569, "boundary_entropy": 0.689911, "bpb": 8.064733}, "zh": {"chunks": 6153.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.041822, "cps": 48.070312, "mean_p": 0.496515, "boundary_entropy": 0.686535, "bpb": 8.032213}, "my": {"chunks": 53337.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.814915, "cps": 416.695312, "mean_p": 0.521486, "boundary_entropy": 0.69057, "bpb": 8.245367}}, "bpb": {"en": 8.24554, "de": 8.20578, "vi": 8.12906, "tr": 8.13852, "fi": 8.22795, "ru": 8.06531, "el": 8.13782, "hi": 7.87944, "ta": 7.8121, "th": 8.06473, "zh": 8.03221, "my": 8.24537}, "gini_cpc": 0.39279, "gini_cps": 0.41072, "bpic": 1.323036, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1962873370015168, "vi": 0.6720657963081282, "tr": 1.2687850214788603, "fi": 0.7776919775720443, "ru": 1.1052708405220564, "el": 3.5909053880998, "hi": 5.093769334952837, "ta": 5.364077888056945, "th": 5.735295526368824, "zh": 2.1001767762124004, "my": 5.674499837031958}, "t": 10.0} |
| {"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.6369, "bpb": 8.13233, "chunk_loss": 0.05651, "F": 0.52254, "G": 0.50189, "chunks": 535.08, "bytes_per_chunk": 1.914, "boundary_entropy": 0.6917, "router_noise": 0.0, "grad_norm": 9.4523, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.005, "alpha": 0.03, "t": 10.9} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8613.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.494915, "cps": 67.289062, "mean_p": 0.503352, "boundary_entropy": 0.687421, "bpb": 8.206543}, "de": {"chunks": 12144.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.591438, "cps": 94.875, "mean_p": 0.507156, "boundary_entropy": 0.68817, "bpb": 8.16824}, "vi": {"chunks": 6125.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.332248, "cps": 47.851562, "mean_p": 0.49057, "boundary_entropy": 0.688469, "bpb": 8.108242}, "tr": {"chunks": 11368.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.627962, "cps": 88.8125, "mean_p": 0.506543, "boundary_entropy": 0.687865, "bpb": 8.111621}, "fi": {"chunks": 7207.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.383127, "cps": 56.304688, "mean_p": 0.497319, "boundary_entropy": 0.687933, "bpb": 8.198516}, "ru": {"chunks": 10418.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.549763, "cps": 81.390625, "mean_p": 0.496388, "boundary_entropy": 0.69018, "bpb": 8.020677}, "el": {"chunks": 36544.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.780636, "cps": 285.5, "mean_p": 0.529451, "boundary_entropy": 0.688865, "bpb": 8.11969}, "hi": {"chunks": 42494.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.525646, "cps": 331.984375, "mean_p": 0.520744, "boundary_entropy": 0.6902, "bpb": 7.856369}, "ta": {"chunks": 49690.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.65793, "cps": 388.203125, "mean_p": 0.521216, "boundary_entropy": 0.690398, "bpb": 7.778495}, "th": {"chunks": 45762.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.845011, "cps": 357.515625, "mean_p": 0.525441, "boundary_entropy": 0.689923, "bpb": 8.032258}, "zh": {"chunks": 6147.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.040806, "cps": 48.023438, "mean_p": 0.496489, "boundary_entropy": 0.686535, "bpb": 8.027051}, "my": {"chunks": 53333.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.814703, "cps": 416.664062, "mean_p": 0.521364, "boundary_entropy": 0.69058, "bpb": 8.217099}}, "bpb": {"en": 8.20654, "de": 8.16824, "vi": 8.10824, "tr": 8.11162, "fi": 8.19852, "ru": 8.02068, "el": 8.11969, "hi": 7.85637, "ta": 7.77849, "th": 8.03226, "zh": 8.02705, "my": 8.2171}, "gini_cpc": 0.393158, "gini_cps": 0.41103, "bpic": 1.32409, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.195030595548181, "vi": 0.6713246963239948, "tr": 1.2688292637472331, "fi": 0.7741271721206426, "ru": 1.1108228679295915, "el": 3.597865384799136, "hi": 5.103195587581507, "ta": 5.370481195088549, "th": 5.748487674381364, "zh": 2.1030008269577616, "my": 5.687249883767758}, "t": 17.2} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8546.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.491065, "cps": 66.765625, "mean_p": 0.503192, "boundary_entropy": 0.687419, "bpb": 8.129079}, "de": {"chunks": 12017.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.585253, "cps": 93.882812, "mean_p": 0.50693, "boundary_entropy": 0.688173, "bpb": 8.094989}, "vi": {"chunks": 6094.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.330567, "cps": 47.609375, "mean_p": 0.490467, "boundary_entropy": 0.688462, "bpb": 8.066491}, "tr": {"chunks": 11282.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.623212, "cps": 88.140625, "mean_p": 0.506348, "boundary_entropy": 0.687866, "bpb": 8.059256}, "fi": {"chunks": 7079.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.376322, "cps": 55.304688, "mean_p": 0.497015, "boundary_entropy": 0.687926, "bpb": 8.140369}, "ru": {"chunks": 10441.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.550976, "cps": 81.570312, "mean_p": 0.496394, "boundary_entropy": 0.69018, "bpb": 7.939832}, "el": {"chunks": 36511.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.779028, "cps": 285.242188, "mean_p": 0.529122, "boundary_entropy": 0.688901, "bpb": 8.084667}, "hi": {"chunks": 42450.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.523031, "cps": 331.640625, "mean_p": 0.520503, "boundary_entropy": 0.690218, "bpb": 7.804458}, "ta": {"chunks": 49577.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.651886, "cps": 387.320312, "mean_p": 0.52074, "boundary_entropy": 0.690436, "bpb": 7.712279}, "th": {"chunks": 45758.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.844762, "cps": 357.484375, "mean_p": 0.525174, "boundary_entropy": 0.689949, "bpb": 7.95274}, "zh": {"chunks": 6131.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.038097, "cps": 47.898438, "mean_p": 0.496435, "boundary_entropy": 0.686534, "bpb": 8.015536}, "my": {"chunks": 53324.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.814228, "cps": 416.59375, "mean_p": 0.521144, "boundary_entropy": 0.690597, "bpb": 8.175922}}, "bpb": {"en": 8.12908, "de": 8.09499, "vi": 8.06649, "tr": 8.05926, "fi": 8.14037, "ru": 7.93983, "el": 8.08467, "hi": 7.80446, "ta": 7.71228, "th": 7.95274, "zh": 8.01554, "my": 8.17592}, "gini_cpc": 0.394275, "gini_cps": 0.411973, "bpic": 1.326998, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1918041290641157, "vi": 0.673163468765311, "tr": 1.269102722248055, "fi": 0.7663395870208192, "ru": 1.122003245512258, "el": 3.622797960412216, "hi": 5.137878777868151, "ta": 5.400276608248225, "th": 5.793048979717437, "zh": 2.113971389192838, "my": 5.730870243322024}, "t": 24.1} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8357.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.480205, "cps": 65.289062, "mean_p": 0.502711, "boundary_entropy": 0.687411, "bpb": 7.871371}, "de": {"chunks": 11669.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.568305, "cps": 91.164062, "mean_p": 0.506245, "boundary_entropy": 0.688178, "bpb": 7.852891}, "vi": {"chunks": 5942.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.322322, "cps": 46.421875, "mean_p": 0.490065, "boundary_entropy": 0.688437, "bpb": 7.929478}, "tr": {"chunks": 11058.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.610838, "cps": 86.390625, "mean_p": 0.505741, "boundary_entropy": 0.687869, "bpb": 7.879046}, "fi": {"chunks": 6625.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.352188, "cps": 51.757812, "mean_p": 0.496057, "boundary_entropy": 0.687897, "bpb": 7.943051}, "ru": {"chunks": 10518.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.55504, "cps": 82.171875, "mean_p": 0.496435, "boundary_entropy": 0.690181, "bpb": 7.699061}, "el": {"chunks": 36309.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.769186, "cps": 283.664062, "mean_p": 0.527599, "boundary_entropy": 0.689064, "bpb": 7.938534}, "hi": {"chunks": 42302.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.514235, "cps": 330.484375, "mean_p": 0.519762, "boundary_entropy": 0.690272, "bpb": 7.625288}, "ta": {"chunks": 49083.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.625461, "cps": 383.460938, "mean_p": 0.519047, "boundary_entropy": 0.690564, "bpb": 7.461311}, "th": {"chunks": 45735.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.843332, "cps": 357.304688, "mean_p": 0.524223, "boundary_entropy": 0.690039, "bpb": 7.682418}, "zh": {"chunks": 6070.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.027768, "cps": 47.421875, "mean_p": 0.49626, "boundary_entropy": 0.686531, "bpb": 7.978496}, "my": {"chunks": 53254.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.810534, "cps": 416.046875, "mean_p": 0.520349, "boundary_entropy": 0.690659, "bpb": 8.004102}}, "bpb": {"en": 7.87137, "de": 7.85289, "vi": 7.92948, "tr": 7.87905, "fi": 7.94305, "ru": 7.69906, "el": 7.93853, "hi": 7.62529, "ta": 7.46131, "th": 7.68242, "zh": 7.9785, "my": 8.0041}, "gini_cpc": 0.397914, "gini_cps": 0.415045, "bpic": 1.33758, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1834637251109785, "vi": 0.6712174268040523, "tr": 1.2720370243079615, "fi": 0.7334115029111274, "ru": 1.1558398675466162, "el": 3.6842336161733944, "hi": 5.2357577715543435, "ta": 5.467381109433109, "th": 5.9210855531496644, "zh": 2.1402719831384927, "my": 5.852785069503308}, "t": 31.5} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 7622.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.43797, "cps": 59.546875, "mean_p": 0.50084, "boundary_entropy": 0.687375, "bpb": 7.138292}, "de": {"chunks": 10432.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.50806, "cps": 81.5, "mean_p": 0.503725, "boundary_entropy": 0.688184, "bpb": 7.175691}, "vi": {"chunks": 5433.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.294711, "cps": 42.445312, "mean_p": 0.488581, "boundary_entropy": 0.688344, "bpb": 7.517611}, "tr": {"chunks": 10190.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.56289, "cps": 79.609375, "mean_p": 0.503618, "boundary_entropy": 0.687864, "bpb": 7.34975}, "fi": {"chunks": 5137.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.273085, "cps": 40.132812, "mean_p": 0.492573, "boundary_entropy": 0.687763, "bpb": 7.342846}, "ru": {"chunks": 11405.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.601847, "cps": 89.101562, "mean_p": 0.497257, "boundary_entropy": 0.690202, "bpb": 6.950884}, "el": {"chunks": 35086.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.709594, "cps": 274.109375, "mean_p": 0.522062, "boundary_entropy": 0.689577, "bpb": 7.465033}, "hi": {"chunks": 41259.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.452244, "cps": 322.335938, "mean_p": 0.516708, "boundary_entropy": 0.690473, "bpb": 7.072854}, "ta": {"chunks": 45355.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.42605, "cps": 354.335938, "mean_p": 0.513179, "boundary_entropy": 0.69092, "bpb": 6.79261}, "th": {"chunks": 45531.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.83065, "cps": 355.710938, "mean_p": 0.520785, "boundary_entropy": 0.690332, "bpb": 6.941397}, "zh": {"chunks": 5919.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.002201, "cps": 46.242188, "mean_p": 0.495753, "boundary_entropy": 0.686526, "bpb": 7.864665}, "my": {"chunks": 53056.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.800084, "cps": 414.5, "mean_p": 0.518252, "boundary_entropy": 0.690808, "bpb": 7.564684}}, "bpb": {"en": 7.13829, "de": 7.17569, "vi": 7.51761, "tr": 7.34975, "fi": 7.34285, "ru": 6.95088, "el": 7.46503, "hi": 7.07285, "ta": 6.79261, "th": 6.9414, "zh": 7.86466, "my": 7.56468}, "gini_cpc": 0.409931, "gini_cps": 0.424434, "bpic": 1.388373, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1600330080295316, "vi": 0.672901875624904, "tr": 1.285223945146213, "fi": 0.623523545635582, "ru": 1.374172493317151, "el": 3.903446124975936, "hi": 5.599107432422684, "ta": 5.539299885629703, "th": 6.463106306719575, "zh": 2.28828478579349, "my": 6.3933179661089055}, "t": 39.7} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4482.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.257542, "cps": 35.015625, "mean_p": 0.492179, "boundary_entropy": 0.687135, "bpb": 6.302509}, "de": {"chunks": 6180.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.300979, "cps": 48.28125, "mean_p": 0.494532, "boundary_entropy": 0.688088, "bpb": 6.384438}, "vi": {"chunks": 3267.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.177217, "cps": 25.523438, "mean_p": 0.48073, "boundary_entropy": 0.687747, "bpb": 6.992915}, "tr": {"chunks": 6586.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.363807, "cps": 51.453125, "mean_p": 0.494469, "boundary_entropy": 0.687718, "bpb": 6.698177}, "fi": {"chunks": 2550.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.135559, "cps": 19.921875, "mean_p": 0.48353, "boundary_entropy": 0.687227, "bpb": 6.561361}, "ru": {"chunks": 12670.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.668602, "cps": 98.984375, "mean_p": 0.498267, "boundary_entropy": 0.69027, "bpb": 5.83176}, "el": {"chunks": 18318.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.89256, "cps": 143.109375, "mean_p": 0.502086, "boundary_entropy": 0.690405, "bpb": 6.499584}, "hi": {"chunks": 27673.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.644755, "cps": 216.195312, "mean_p": 0.504328, "boundary_entropy": 0.690909, "bpb": 5.786875}, "ta": {"chunks": 8379.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.448195, "cps": 65.460938, "mean_p": 0.490976, "boundary_entropy": 0.691014, "bpb": 5.836925}, "th": {"chunks": 26191.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.628287, "cps": 204.617188, "mean_p": 0.502751, "boundary_entropy": 0.691089, "bpb": 5.315367}, "zh": {"chunks": 5146.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.871317, "cps": 40.203125, "mean_p": 0.493216, "boundary_entropy": 0.686467, "bpb": 7.471781}, "my": {"chunks": 49531.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.614049, "cps": 386.960938, "mean_p": 0.51026, "boundary_entropy": 0.691217, "bpb": 6.203846}}, "bpb": {"en": 6.30251, "de": 6.38444, "vi": 6.99291, "tr": 6.69818, "fi": 6.56136, "ru": 5.83176, "el": 6.49958, "hi": 5.78688, "ta": 5.83692, "th": 5.31537, "zh": 7.47178, "my": 6.20385}, "gini_cpc": 0.456263, "gini_cps": 0.47548, "bpic": 2.244688, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1686604206726092, "vi": 0.6881106198594215, "tr": 1.4126137945880264, "fi": 0.5263571963682704, "ru": 2.59608954728324, "el": 3.4656881191383473, "hi": 6.38636061830283, "ta": 1.7402794381591624, "th": 6.322419134625748, "zh": 3.3832072844941115, "my": 10.14999873535377}, "t": 50.0} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5938.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.341206, "cps": 46.390625, "mean_p": 0.498285, "boundary_entropy": 0.687566, "bpb": 5.763945}, "de": {"chunks": 8188.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.398773, "cps": 63.96875, "mean_p": 0.499978, "boundary_entropy": 0.688519, "bpb": 5.883469}, "vi": {"chunks": 3196.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.173366, "cps": 24.96875, "mean_p": 0.483574, "boundary_entropy": 0.68816, "bpb": 6.405878}, "tr": {"chunks": 6029.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.333039, "cps": 47.101562, "mean_p": 0.495035, "boundary_entropy": 0.687997, "bpb": 6.238512}, "fi": {"chunks": 4715.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.250651, "cps": 36.835938, "mean_p": 0.49352, "boundary_entropy": 0.688011, "bpb": 5.937385}, "ru": {"chunks": 1755.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.092612, "cps": 13.710938, "mean_p": 0.487432, "boundary_entropy": 0.690004, "bpb": 5.197644}, "el": {"chunks": 489.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.023827, "cps": 3.820312, "mean_p": 0.467249, "boundary_entropy": 0.688109, "bpb": 6.146488}, "hi": {"chunks": 6259.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.372006, "cps": 48.898438, "mean_p": 0.492787, "boundary_entropy": 0.690789, "bpb": 4.915677}, "ta": {"chunks": 4920.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.263172, "cps": 38.4375, "mean_p": 0.489196, "boundary_entropy": 0.690956, "bpb": 5.143984}, "th": {"chunks": 2632.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.163631, "cps": 20.5625, "mean_p": 0.488514, "boundary_entropy": 0.690762, "bpb": 4.817305}, "zh": {"chunks": 2640.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.447003, "cps": 20.625, "mean_p": 0.48398, "boundary_entropy": 0.686056, "bpb": 7.056543}, "my": {"chunks": 4465.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.235645, "cps": 34.882812, "mean_p": 0.490914, "boundary_entropy": 0.691177, "bpb": 4.729786}}, "bpb": {"en": 5.76394, "de": 5.88347, "vi": 6.40588, "tr": 6.23851, "fi": 5.93739, "ru": 5.19764, "el": 6.14649, "hi": 4.91568, "ta": 5.14398, "th": 4.81731, "zh": 7.05654, "my": 4.72979}, "gini_cpc": 0.269605, "gini_cps": 0.278042, "bpic": 7.491918, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1687169797906998, "vi": 0.5080980821348674, "tr": 0.976064818793444, "fi": 0.7346047641767547, "ru": 0.2714262417896096, "el": 0.06983159165196236, "hi": 1.0902693559050256, "ta": 0.7713004064195133, "th": 0.47956639915956756, "zh": 1.3100697271740256, "my": 0.6906245527192431}, "t": 59.2} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5097.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.292881, "cps": 39.820312, "mean_p": 0.495927, "boundary_entropy": 0.687506, "bpb": 5.360606}, "de": {"chunks": 4907.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.238981, "cps": 38.335938, "mean_p": 0.4934, "boundary_entropy": 0.688322, "bpb": 5.501292}, "vi": {"chunks": 3400.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.184432, "cps": 26.5625, "mean_p": 0.485693, "boundary_entropy": 0.688406, "bpb": 5.897547}, "tr": {"chunks": 4184.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.231122, "cps": 32.6875, "mean_p": 0.491161, "boundary_entropy": 0.687896, "bpb": 5.831534}, "fi": {"chunks": 5580.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.296635, "cps": 43.59375, "mean_p": 0.495344, "boundary_entropy": 0.688081, "bpb": 5.537399}, "ru": {"chunks": 6516.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.343852, "cps": 50.90625, "mean_p": 0.495467, "boundary_entropy": 0.690387, "bpb": 4.811055}, "el": {"chunks": 1143.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.055694, "cps": 8.929688, "mean_p": 0.487489, "boundary_entropy": 0.690248, "bpb": 5.109541}, "hi": {"chunks": 9449.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.561605, "cps": 73.820312, "mean_p": 0.495861, "boundary_entropy": 0.690912, "bpb": 4.414246}, "ta": {"chunks": 15194.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.812731, "cps": 118.703125, "mean_p": 0.494333, "boundary_entropy": 0.691117, "bpb": 4.350063}, "th": {"chunks": 8091.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.503015, "cps": 63.210938, "mean_p": 0.49373, "boundary_entropy": 0.690945, "bpb": 4.391138}, "zh": {"chunks": 1817.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.307653, "cps": 14.195312, "mean_p": 0.4847, "boundary_entropy": 0.686394, "bpb": 6.68753}, "my": {"chunks": 22127.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.167775, "cps": 172.867188, "mean_p": 0.500231, "boundary_entropy": 0.691411, "bpb": 4.272447}}, "bpb": {"en": 5.36061, "de": 5.50129, "vi": 5.89755, "tr": 5.83153, "fi": 5.5374, "ru": 4.81106, "el": 5.10954, "hi": 4.41425, "ta": 4.35006, "th": 4.39114, "zh": 6.68753, "my": 4.27245}, "gini_cpc": 0.36168, "gini_cps": 0.398458, "bpic": 4.385818, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8159680190951174, "vi": 0.6297167736793113, "tr": 0.7891337376012358, "fi": 1.0128189118554198, "ru": 1.1740358211643975, "el": 0.1901581185833881, "hi": 1.9175215907967889, "ta": 2.7749562420573373, "th": 1.7174757848195734, "zh": 1.0504393233843854, "my": 3.9872057449631675}, "t": 73.1} |
| {"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.26542, "bpb": 4.711, "chunk_loss": 0.032162, "F": 0.19778, "G": 0.47216, "chunks": 202.53, "bytes_per_chunk": 5.056, "boundary_entropy": 0.68806, "router_noise": 0.0, "grad_norm": 1.9245, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 0.731, "alpha": 0.03, "t": 84.0} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3642.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.209274, "cps": 28.453125, "mean_p": 0.454347, "boundary_entropy": 0.65861, "bpb": 4.303812}, "de": {"chunks": 4136.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.201432, "cps": 32.3125, "mean_p": 0.45371, "boundary_entropy": 0.663924, "bpb": 4.546935}, "vi": {"chunks": 6904.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.374505, "cps": 53.9375, "mean_p": 0.489322, "boundary_entropy": 0.68186, "bpb": 4.698088}, "tr": {"chunks": 3501.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193393, "cps": 27.351562, "mean_p": 0.469064, "boundary_entropy": 0.675174, "bpb": 4.85856}, "fi": {"chunks": 3905.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.207591, "cps": 30.507812, "mean_p": 0.465749, "boundary_entropy": 0.674774, "bpb": 4.578552}, "ru": {"chunks": 4017.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.211979, "cps": 31.382812, "mean_p": 0.474003, "boundary_entropy": 0.687943, "bpb": 3.218386}, "el": {"chunks": 6541.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.318716, "cps": 51.101562, "mean_p": 0.490029, "boundary_entropy": 0.690133, "bpb": 3.903639}, "hi": {"chunks": 10565.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.627935, "cps": 82.539062, "mean_p": 0.475916, "boundary_entropy": 0.688758, "bpb": 2.886505}, "ta": {"chunks": 13855.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.741107, "cps": 108.242188, "mean_p": 0.441847, "boundary_entropy": 0.678917, "bpb": 2.913737}, "th": {"chunks": 7280.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.452596, "cps": 56.875, "mean_p": 0.474147, "boundary_entropy": 0.688282, "bpb": 2.765651}, "zh": {"chunks": 2094.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.354555, "cps": 16.359375, "mean_p": 0.485031, "boundary_entropy": 0.686373, "bpb": 6.216205}, "my": {"chunks": 9375.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.494775, "cps": 73.242188, "mean_p": 0.47632, "boundary_entropy": 0.689426, "bpb": 2.860307}}, "bpb": {"en": 4.30381, "de": 4.54693, "vi": 4.69809, "tr": 4.85856, "fi": 4.57855, "ru": 3.21839, "el": 3.90364, "hi": 2.8865, "ta": 2.91374, "th": 2.76565, "zh": 6.21621, "my": 2.86031}, "gini_cpc": 0.258549, "gini_cps": 0.286903, "bpic": 5.062072, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9625256277695302, "vi": 1.789541686382852, "tr": 0.9241144008370189, "fi": 0.9919581124192379, "ru": 1.0129238479828562, "el": 1.522956443532113, "hi": 3.0005343797351847, "ta": 3.5413205479621053, "th": 2.1626910989241153, "zh": 1.6942106734223314, "my": 2.364242821505071}, "t": 93.9} |
| {"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.4349, "bpb": 3.51282, "chunk_loss": 0.031171, "F": 0.1881, "G": 0.39809, "chunks": 192.61, "bytes_per_chunk": 5.316, "boundary_entropy": 0.58671, "router_noise": 0.0, "grad_norm": 2.3973, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.272, "alpha": 0.03, "t": 111.2} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2903.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.16681, "cps": 22.679688, "mean_p": 0.332443, "boundary_entropy": 0.457051, "bpb": 3.475955}, "de": {"chunks": 3250.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.158282, "cps": 25.390625, "mean_p": 0.332933, "boundary_entropy": 0.488987, "bpb": 3.701775}, "vi": {"chunks": 6875.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.372932, "cps": 53.710938, "mean_p": 0.33189, "boundary_entropy": 0.421791, "bpb": 3.047568}, "tr": {"chunks": 3666.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202508, "cps": 28.640625, "mean_p": 0.34433, "boundary_entropy": 0.511695, "bpb": 3.669172}, "fi": {"chunks": 2930.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.15576, "cps": 22.890625, "mean_p": 0.340053, "boundary_entropy": 0.519996, "bpb": 3.760697}, "ru": {"chunks": 9471.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.499789, "cps": 73.992188, "mean_p": 0.295812, "boundary_entropy": 0.48003, "bpb": 2.601375}, "el": {"chunks": 10905.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.531355, "cps": 85.195312, "mean_p": 0.374326, "boundary_entropy": 0.56242, "bpb": 2.675292}, "hi": {"chunks": 10153.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.603447, "cps": 79.320312, "mean_p": 0.409401, "boundary_entropy": 0.639944, "bpb": 2.050878}, "ta": {"chunks": 12745.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.681733, "cps": 99.570312, "mean_p": 0.402576, "boundary_entropy": 0.64771, "bpb": 1.902684}, "th": {"chunks": 7270.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.451974, "cps": 56.796875, "mean_p": 0.444242, "boundary_entropy": 0.669999, "bpb": 2.066414}, "zh": {"chunks": 4218.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.714189, "cps": 32.953125, "mean_p": 0.374542, "boundary_entropy": 0.602445, "bpb": 4.872947}, "my": {"chunks": 9829.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.518735, "cps": 76.789062, "mean_p": 0.294155, "boundary_entropy": 0.550102, "bpb": 2.132625}}, "bpb": {"en": 3.47596, "de": 3.70177, "vi": 3.04757, "tr": 3.66917, "fi": 3.7607, "ru": 2.60138, "el": 2.67529, "hi": 2.05088, "ta": 1.90268, "th": 2.06641, "zh": 4.87295, "my": 2.13262}, "gini_cpc": 0.264288, "gini_cps": 0.272445, "bpic": 4.557157, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9488728888722169, "vi": 2.2356645730252396, "tr": 1.214000857686379, "fi": 0.9337547439875155, "ru": 2.99615106563404, "el": 3.1853848422484377, "hi": 3.617565453373796, "ta": 4.086875940697143, "th": 2.7095079522314656, "zh": 4.281443498959878, "my": 3.1097325777509104}, "t": 127.1} |
| {"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.0991, "bpb": 3.02836, "chunk_loss": 0.030783, "F": 0.19588, "G": 0.31068, "chunks": 200.58, "bytes_per_chunk": 5.105, "boundary_entropy": 0.44294, "router_noise": 0.0, "grad_norm": 1.2151, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.258, "alpha": 0.03, "t": 137.0} |
| {"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.0188, "bpb": 2.91251, "chunk_loss": 0.031516, "F": 0.20018, "G": 0.32485, "chunks": 204.98, "bytes_per_chunk": 4.996, "boundary_entropy": 0.44598, "router_noise": 0.0, "grad_norm": 1.1238, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.222, "alpha": 0.03, "t": 159.1} |
| {"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.9036, "bpb": 2.74631, "chunk_loss": 0.030985, "F": 0.19302, "G": 0.28856, "chunks": 197.66, "bytes_per_chunk": 5.181, "boundary_entropy": 0.37104, "router_noise": 0.0, "grad_norm": 0.8714, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.214, "alpha": 0.03, "t": 180.5} |
| {"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.78299, "bpb": 2.57232, "chunk_loss": 0.030201, "F": 0.17719, "G": 0.26403, "chunks": 181.44, "bytes_per_chunk": 5.644, "boundary_entropy": 0.33447, "router_noise": 0.0, "grad_norm": 0.8695, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.195, "alpha": 0.03, "t": 200.0} |
| {"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.66425, "bpb": 2.40101, "chunk_loss": 0.030252, "F": 0.17838, "G": 0.25096, "chunks": 182.66, "bytes_per_chunk": 5.606, "boundary_entropy": 0.30586, "router_noise": 0.0, "grad_norm": 0.7175, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.208, "alpha": 0.03, "t": 220.9} |
| {"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.56035, "bpb": 2.25111, "chunk_loss": 0.030286, "F": 0.18176, "G": 0.23729, "chunks": 186.12, "bytes_per_chunk": 5.502, "boundary_entropy": 0.26019, "router_noise": 0.0, "grad_norm": 0.66, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.196, "alpha": 0.03, "t": 240.4} |
| {"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.54578, "bpb": 2.23009, "chunk_loss": 0.030235, "F": 0.18062, "G": 0.24179, "chunks": 184.95, "bytes_per_chunk": 5.537, "boundary_entropy": 0.26943, "router_noise": 0.0, "grad_norm": 0.6532, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.207, "alpha": 0.03, "t": 261.1} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2935.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.168649, "cps": 22.929688, "mean_p": 0.210377, "boundary_entropy": 0.189219, "bpb": 2.294282}, "de": {"chunks": 3324.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.161886, "cps": 25.96875, "mean_p": 0.224716, "boundary_entropy": 0.249341, "bpb": 2.584911}, "vi": {"chunks": 4527.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.245566, "cps": 35.367188, "mean_p": 0.235109, "boundary_entropy": 0.21809, "bpb": 2.091356}, "tr": {"chunks": 3751.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207203, "cps": 29.304688, "mean_p": 0.223765, "boundary_entropy": 0.197305, "bpb": 2.579423}, "fi": {"chunks": 3092.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.164372, "cps": 24.15625, "mean_p": 0.229129, "boundary_entropy": 0.276879, "bpb": 2.820111}, "ru": {"chunks": 6446.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.340158, "cps": 50.359375, "mean_p": 0.20653, "boundary_entropy": 0.186231, "bpb": 1.573155}, "el": {"chunks": 7126.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.34722, "cps": 55.671875, "mean_p": 0.220561, "boundary_entropy": 0.198875, "bpb": 1.679536}, "hi": {"chunks": 7028.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.417712, "cps": 54.90625, "mean_p": 0.314448, "boundary_entropy": 0.506466, "bpb": 1.381624}, "ta": {"chunks": 8171.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.437069, "cps": 63.835938, "mean_p": 0.313705, "boundary_entropy": 0.531908, "bpb": 1.117139}, "th": {"chunks": 7093.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.44097, "cps": 55.414062, "mean_p": 0.291957, "boundary_entropy": 0.473759, "bpb": 1.306998}, "zh": {"chunks": 3001.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.508127, "cps": 23.445312, "mean_p": 0.245714, "boundary_entropy": 0.29276, "bpb": 3.442287}, "my": {"chunks": 8818.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.465379, "cps": 68.890625, "mean_p": 0.267413, "boundary_entropy": 0.447265, "bpb": 1.250324}}, "bpb": {"en": 2.29428, "de": 2.58491, "vi": 2.09136, "tr": 2.57942, "fi": 2.82011, "ru": 1.57316, "el": 1.67954, "hi": 1.38162, "ta": 1.11714, "th": 1.307, "zh": 3.44229, "my": 1.25032}, "gini_cpc": 0.215505, "gini_cps": 0.217005, "bpic": 5.876118, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.959896973924304, "vi": 1.4560737320545643, "tr": 1.2286057041944631, "fi": 0.9746386203222822, "ru": 2.0169591456041855, "el": 2.058832368321377, "hi": 2.4768100118718848, "ta": 2.5915867783431055, "th": 2.6147183166091126, "zh": 3.012926709245528, "my": 2.7594512948375884}, "t": 285.2} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.30402, "de": 2.57829, "vi": 2.09462, "tr": 2.57008, "fi": 2.83174, "ru": 1.57366, "el": 1.67623, "hi": 1.3894, "ta": 1.1162, "th": 1.30449, "zh": 3.44893, "my": 1.26047}, "bytes_per_chunk": {"en": 5.548, "de": 6.14, "vi": 5.38, "tr": 5.211, "fi": 6.147, "ru": 5.447, "el": 5.326, "hi": 5.929, "ta": 6.044, "th": 6.462, "zh": 5.364, "my": 6.058}, "bpb_mean": 2.01234, "t": 291.3} |
| {"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.45859, "bpb": 2.1043, "chunk_loss": 0.030136, "F": 0.17674, "G": 0.23029, "chunks": 180.98, "bytes_per_chunk": 5.658, "boundary_entropy": 0.24284, "router_noise": 0.0, "grad_norm": 0.6524, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.303, "alpha": 0.03, "t": 291.5} |
| {"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.38599, "bpb": 1.99956, "chunk_loss": 0.030181, "F": 0.17802, "G": 0.2386, "chunks": 182.3, "bytes_per_chunk": 5.617, "boundary_entropy": 0.26543, "router_noise": 0.0, "grad_norm": 0.5814, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.195, "alpha": 0.03, "t": 310.9} |
| {"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.42382, "bpb": 2.05414, "chunk_loss": 0.030088, "F": 0.17427, "G": 0.2289, "chunks": 178.45, "bytes_per_chunk": 5.738, "boundary_entropy": 0.24473, "router_noise": 0.0, "grad_norm": 0.5515, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.209, "alpha": 0.03, "t": 331.8} |
| {"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.44652, "bpb": 2.08689, "chunk_loss": 0.030067, "F": 0.17329, "G": 0.22165, "chunks": 177.45, "bytes_per_chunk": 5.771, "boundary_entropy": 0.2239, "router_noise": 0.0, "grad_norm": 0.6853, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.204, "alpha": 0.03, "t": 352.2} |
| {"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.40248, "bpb": 2.02335, "chunk_loss": 0.030177, "F": 0.18071, "G": 0.2233, "chunks": 185.05, "bytes_per_chunk": 5.534, "boundary_entropy": 0.21235, "router_noise": 0.0, "grad_norm": 0.5402, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.198, "alpha": 0.03, "t": 372.0} |
| {"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.40628, "bpb": 2.02883, "chunk_loss": 0.030302, "F": 0.1786, "G": 0.22404, "chunks": 182.89, "bytes_per_chunk": 5.599, "boundary_entropy": 0.2285, "router_noise": 0.0, "grad_norm": 0.5579, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.209, "alpha": 0.03, "t": 392.9} |
| {"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.38401, "bpb": 1.9967, "chunk_loss": 0.030156, "F": 0.17792, "G": 0.22005, "chunks": 182.19, "bytes_per_chunk": 5.621, "boundary_entropy": 0.21338, "router_noise": 0.0, "grad_norm": 0.5215, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 412.2} |
| {"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.32769, "bpb": 1.91545, "chunk_loss": 0.030095, "F": 0.17523, "G": 0.21677, "chunks": 179.44, "bytes_per_chunk": 5.707, "boundary_entropy": 0.20868, "router_noise": 0.0, "grad_norm": 0.484, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.207, "alpha": 0.03, "t": 432.9} |
| {"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.34467, "bpb": 1.93995, "chunk_loss": 0.030107, "F": 0.17628, "G": 0.22067, "chunks": 180.52, "bytes_per_chunk": 5.673, "boundary_entropy": 0.22118, "router_noise": 0.0, "grad_norm": 0.5053, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.201, "alpha": 0.03, "t": 453.0} |
| {"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.34164, "bpb": 1.93557, "chunk_loss": 0.030154, "F": 0.17844, "G": 0.22131, "chunks": 182.72, "bytes_per_chunk": 5.604, "boundary_entropy": 0.21007, "router_noise": 0.0, "grad_norm": 0.4783, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.201, "alpha": 0.03, "t": 473.2} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2929.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.168304, "cps": 22.882812, "mean_p": 0.192323, "boundary_entropy": 0.138048, "bpb": 2.063898}, "de": {"chunks": 3289.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.160181, "cps": 25.695312, "mean_p": 0.214513, "boundary_entropy": 0.220986, "bpb": 2.268997}, "vi": {"chunks": 4495.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.24383, "cps": 35.117188, "mean_p": 0.209386, "boundary_entropy": 0.142482, "bpb": 1.929974}, "tr": {"chunks": 3708.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204828, "cps": 28.96875, "mean_p": 0.207828, "boundary_entropy": 0.147402, "bpb": 2.265376}, "fi": {"chunks": 2971.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.15794, "cps": 23.210938, "mean_p": 0.205789, "boundary_entropy": 0.225588, "bpb": 2.464541}, "ru": {"chunks": 6373.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.336306, "cps": 49.789062, "mean_p": 0.20378, "boundary_entropy": 0.168544, "bpb": 1.352409}, "el": {"chunks": 7107.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.346294, "cps": 55.523438, "mean_p": 0.214324, "boundary_entropy": 0.17567, "bpb": 1.393798}, "hi": {"chunks": 7045.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.418722, "cps": 55.039062, "mean_p": 0.281347, "boundary_entropy": 0.444616, "bpb": 1.182118}, "ta": {"chunks": 8097.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.43311, "cps": 63.257812, "mean_p": 0.269413, "boundary_entropy": 0.459115, "bpb": 1.032508}, "th": {"chunks": 7438.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.462418, "cps": 58.109375, "mean_p": 0.288092, "boundary_entropy": 0.428926, "bpb": 1.124967}, "zh": {"chunks": 2823.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.477988, "cps": 22.054688, "mean_p": 0.225277, "boundary_entropy": 0.243983, "bpb": 3.143514}, "my": {"chunks": 10356.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.546548, "cps": 80.90625, "mean_p": 0.303653, "boundary_entropy": 0.479541, "bpb": 1.136098}}, "bpb": {"en": 2.0639, "de": 2.269, "vi": 1.92997, "tr": 2.26538, "fi": 2.46454, "ru": 1.35241, "el": 1.3938, "hi": 1.18212, "ta": 1.03251, "th": 1.12497, "zh": 3.14351, "my": 1.1361}, "gini_cpc": 0.228946, "gini_cps": 0.239657, "bpic": 5.759796, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9517353814792755, "vi": 1.4487428401098856, "tr": 1.217009371679491, "fi": 0.93841624414385, "ru": 1.998202291523848, "el": 2.0575491524826535, "hi": 2.4878871275968093, "ta": 2.573377019158326, "th": 2.7475136408842022, "zh": 2.8400251371868155, "my": 3.2473822626306217}, "t": 497.1} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.07286, "de": 2.26406, "vi": 1.93769, "tr": 2.25536, "fi": 2.48109, "ru": 1.35225, "el": 1.39106, "hi": 1.18946, "ta": 1.03157, "th": 1.12029, "zh": 3.15066, "my": 1.1452}, "bytes_per_chunk": {"en": 5.585, "de": 6.189, "vi": 5.437, "tr": 5.281, "fi": 6.316, "ru": 5.545, "el": 5.318, "hi": 5.843, "ta": 6.141, "th": 6.043, "zh": 5.587, "my": 5.178}, "bpb_mean": 1.78263, "t": 503.3} |
| {"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.33555, "bpb": 1.92679, "chunk_loss": 0.03014, "F": 0.17857, "G": 0.21728, "chunks": 182.86, "bytes_per_chunk": 5.6, "boundary_entropy": 0.20134, "router_noise": 0.0, "grad_norm": 0.5238, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.303, "alpha": 0.03, "t": 503.5} |
| {"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.27255, "bpb": 1.8359, "chunk_loss": 0.030144, "F": 0.17793, "G": 0.21572, "chunks": 182.2, "bytes_per_chunk": 5.62, "boundary_entropy": 0.19988, "router_noise": 0.0, "grad_norm": 0.4753, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 522.8} |
| {"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.23962, "bpb": 1.78839, "chunk_loss": 0.030101, "F": 0.17636, "G": 0.2236, "chunks": 180.59, "bytes_per_chunk": 5.67, "boundary_entropy": 0.22688, "router_noise": 0.0, "grad_norm": 0.4633, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.205, "alpha": 0.03, "t": 543.3} |
| {"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.31446, "bpb": 1.89636, "chunk_loss": 0.030162, "F": 0.17966, "G": 0.21639, "chunks": 183.97, "bytes_per_chunk": 5.566, "boundary_entropy": 0.19665, "router_noise": 0.0, "grad_norm": 0.4267, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.205, "alpha": 0.03, "t": 563.9} |
| {"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.33564, "bpb": 1.92692, "chunk_loss": 0.030093, "F": 0.17601, "G": 0.21536, "chunks": 180.23, "bytes_per_chunk": 5.681, "boundary_entropy": 0.19645, "router_noise": 0.0, "grad_norm": 0.4578, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 583.1} |
| {"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.24199, "bpb": 1.79181, "chunk_loss": 0.030097, "F": 0.17543, "G": 0.21614, "chunks": 179.64, "bytes_per_chunk": 5.7, "boundary_entropy": 0.20146, "router_noise": 0.0, "grad_norm": 0.4116, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.205, "alpha": 0.03, "t": 603.6} |
| {"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.26011, "bpb": 1.81795, "chunk_loss": 0.030141, "F": 0.17737, "G": 0.21925, "chunks": 181.62, "bytes_per_chunk": 5.638, "boundary_entropy": 0.2031, "router_noise": 0.0, "grad_norm": 0.454, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 622.7} |
| {"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.26, "bpb": 1.8178, "chunk_loss": 0.030004, "F": 0.17024, "G": 0.21465, "chunks": 174.33, "bytes_per_chunk": 5.874, "boundary_entropy": 0.20738, "router_noise": 0.0, "grad_norm": 0.4212, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.206, "alpha": 0.03, "t": 643.3} |
| {"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.34178, "bpb": 1.93578, "chunk_loss": 0.030057, "F": 0.17345, "G": 0.21881, "chunks": 177.61, "bytes_per_chunk": 5.765, "boundary_entropy": 0.20449, "router_noise": 0.0, "grad_norm": 0.4498, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 662.5} |
| {"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.26215, "bpb": 1.82089, "chunk_loss": 0.03014, "F": 0.17656, "G": 0.21912, "chunks": 180.8, "bytes_per_chunk": 5.664, "boundary_entropy": 0.2065, "router_noise": 0.0, "grad_norm": 0.4039, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.206, "alpha": 0.03, "t": 683.1} |
| {"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2916.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.167557, "cps": 22.78125, "mean_p": 0.187789, "boundary_entropy": 0.12292, "bpb": 1.963674}, "de": {"chunks": 3324.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.161886, "cps": 25.96875, "mean_p": 0.227346, "boundary_entropy": 0.242845, "bpb": 2.13985}, "vi": {"chunks": 4453.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.241551, "cps": 34.789062, "mean_p": 0.205364, "boundary_entropy": 0.135379, "bpb": 1.805219}, "tr": {"chunks": 3690.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203834, "cps": 28.828125, "mean_p": 0.20206, "boundary_entropy": 0.141725, "bpb": 2.123334}, "fi": {"chunks": 3083.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.163893, "cps": 24.085938, "mean_p": 0.225264, "boundary_entropy": 0.254991, "bpb": 2.295674}, "ru": {"chunks": 6339.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.334512, "cps": 49.523438, "mean_p": 0.196876, "boundary_entropy": 0.137521, "bpb": 1.258781}, "el": {"chunks": 7027.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.342396, "cps": 54.898438, "mean_p": 0.200469, "boundary_entropy": 0.13191, "bpb": 1.295455}, "hi": {"chunks": 6808.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.404636, "cps": 53.1875, "mean_p": 0.268436, "boundary_entropy": 0.408414, "bpb": 1.107359}, "ta": {"chunks": 8841.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.472907, "cps": 69.070312, "mean_p": 0.276045, "boundary_entropy": 0.45896, "bpb": 0.939247}, "th": {"chunks": 8072.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.501834, "cps": 63.0625, "mean_p": 0.317171, "boundary_entropy": 0.458628, "bpb": 1.045122}, "zh": {"chunks": 2754.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.466305, "cps": 21.515625, "mean_p": 0.209135, "boundary_entropy": 0.199783, "bpb": 2.966321}, "my": {"chunks": 8278.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.43688, "cps": 64.671875, "mean_p": 0.298245, "boundary_entropy": 0.499359, "bpb": 1.061981}}, "bpb": {"en": 1.96367, "de": 2.13985, "vi": 1.80522, "tr": 2.12333, "fi": 2.29567, "ru": 1.25878, "el": 1.29545, "hi": 1.10736, "ta": 0.93925, "th": 1.04512, "zh": 2.96632, "my": 1.06198}, "gini_cpc": 0.219635, "gini_cps": 0.228575, "bpic": 5.851658, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.966151446662494, "vi": 1.4416045811052292, "tr": 1.2165008459821476, "fi": 0.9781337445278167, "ru": 1.9964026515521678, "el": 2.043457967162808, "hi": 2.414910694085203, "ta": 2.8223607355237914, "th": 2.994999046986751, "zh": 2.7829608307935434, "my": 2.6073458788227866}, "t": 706.9} |
| {"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 1.97266, "de": 2.13375, "vi": 1.81071, "tr": 2.11285, "fi": 2.311, "ru": 1.25834, "el": 1.29293, "hi": 1.11439, "ta": 0.94034, "th": 1.04135, "zh": 2.97456, "my": 1.07208}, "bytes_per_chunk": {"en": 5.583, "de": 6.14, "vi": 5.449, "tr": 5.317, "fi": 6.148, "ru": 5.581, "el": 5.381, "hi": 6.011, "ta": 5.646, "th": 5.643, "zh": 5.727, "my": 6.423}, "bpb_mean": 1.66958, "t": 713.1} |
| {"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.31321, "bpb": 1.89457, "chunk_loss": 0.030111, "F": 0.17838, "G": 0.2131, "chunks": 182.66, "bytes_per_chunk": 5.606, "boundary_entropy": 0.17752, "router_noise": 0.0, "grad_norm": 0.4225, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.302, "alpha": 0.03, "t": 713.3} |
| {"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.31291, "bpb": 1.89413, "chunk_loss": 0.030155, "F": 0.17804, "G": 0.21814, "chunks": 182.31, "bytes_per_chunk": 5.617, "boundary_entropy": 0.19413, "router_noise": 0.0, "grad_norm": 0.4256, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 732.5} |
| {"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.19241, "bpb": 1.72029, "chunk_loss": 0.030094, "F": 0.17628, "G": 0.2188, "chunks": 180.52, "bytes_per_chunk": 5.673, "boundary_entropy": 0.20241, "router_noise": 0.0, "grad_norm": 0.3981, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.206, "alpha": 0.03, "t": 753.1} |
| {"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.26997, "bpb": 1.83218, "chunk_loss": 0.030028, "F": 0.17268, "G": 0.21674, "chunks": 176.83, "bytes_per_chunk": 5.791, "boundary_entropy": 0.20561, "router_noise": 0.0, "grad_norm": 0.4022, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 772.1} |
| {"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.22588, "bpb": 1.76857, "chunk_loss": 0.030168, "F": 0.17915, "G": 0.2206, "chunks": 183.45, "bytes_per_chunk": 5.582, "boundary_entropy": 0.19657, "router_noise": 0.0, "grad_norm": 0.426, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 791.2} |
| {"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.30414, "bpb": 1.88148, "chunk_loss": 0.030083, "F": 0.17558, "G": 0.21343, "chunks": 179.8, "bytes_per_chunk": 5.695, "boundary_entropy": 0.17961, "router_noise": 0.0, "grad_norm": 0.3996, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 810.3} |
| {"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.21797, "bpb": 1.75716, "chunk_loss": 0.029995, "F": 0.17047, "G": 0.21482, "chunks": 174.56, "bytes_per_chunk": 5.866, "boundary_entropy": 0.1984, "router_noise": 0.0, "grad_norm": 0.4048, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 829.5} |
| {"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.12891, "bpb": 1.62867, "chunk_loss": 0.03005, "F": 0.17163, "G": 0.22216, "chunks": 175.75, "bytes_per_chunk": 5.826, "boundary_entropy": 0.22576, "router_noise": 0.0, "grad_norm": 0.4051, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 848.6} |
| {"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.2213, "bpb": 1.76196, "chunk_loss": 0.030081, "F": 0.1754, "G": 0.2174, "chunks": 179.61, "bytes_per_chunk": 5.701, "boundary_entropy": 0.19345, "router_noise": 0.0, "grad_norm": 0.3941, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 867.7} |
| {"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.23454, "bpb": 1.78106, "chunk_loss": 0.030049, "F": 0.17325, "G": 0.21031, "chunks": 177.41, "bytes_per_chunk": 5.772, "boundary_entropy": 0.17372, "router_noise": 0.0, "grad_norm": 0.4012, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 886.8} |
| {"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2909.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.167155, "cps": 22.726562, "mean_p": 0.185789, "boundary_entropy": 0.1142, "bpb": 1.891926}, "de": {"chunks": 3312.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.161301, "cps": 25.875, "mean_p": 0.239216, "boundary_entropy": 0.267153, "bpb": 2.062146}, "vi": {"chunks": 4385.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.237863, "cps": 34.257812, "mean_p": 0.19963, "boundary_entropy": 0.123848, "bpb": 1.717328}, "tr": {"chunks": 3399.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.187759, "cps": 26.554688, "mean_p": 0.188435, "boundary_entropy": 0.144983, "bpb": 2.052872}, "fi": {"chunks": 3008.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.159906, "cps": 23.5, "mean_p": 0.231683, "boundary_entropy": 0.273988, "bpb": 2.197363}, "ru": {"chunks": 6275.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.331135, "cps": 49.023438, "mean_p": 0.194565, "boundary_entropy": 0.134051, "bpb": 1.211807}, "el": {"chunks": 6919.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.337134, "cps": 54.054688, "mean_p": 0.199093, "boundary_entropy": 0.140721, "bpb": 1.230108}, "hi": {"chunks": 6200.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.368499, "cps": 48.4375, "mean_p": 0.261956, "boundary_entropy": 0.409207, "bpb": 1.053397}, "ta": {"chunks": 7492.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.400749, "cps": 58.53125, "mean_p": 0.261805, "boundary_entropy": 0.449626, "bpb": 0.90463}, "th": {"chunks": 7834.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.487038, "cps": 61.203125, "mean_p": 0.321043, "boundary_entropy": 0.455985, "bpb": 0.995376}, "zh": {"chunks": 2710.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.458855, "cps": 21.171875, "mean_p": 0.204825, "boundary_entropy": 0.186649, "bpb": 2.872043}, "my": {"chunks": 9420.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.49715, "cps": 73.59375, "mean_p": 0.322649, "boundary_entropy": 0.516236, "bpb": 1.026699}}, "bpb": {"en": 1.89193, "de": 2.06215, "vi": 1.71733, "tr": 2.05287, "fi": 2.19736, "ru": 1.21181, "el": 1.23011, "hi": 1.0534, "ta": 0.90463, "th": 0.99538, "zh": 2.87204, "my": 1.0267}, "gini_cpc": 0.222821, "gini_cps": 0.232269, "bpic": 6.009442, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9649800168245711, "vi": 1.4230064044668198, "tr": 1.1232618582027478, "fi": 0.9566351789916007, "ru": 1.9810020036081273, "el": 2.016893119453617, "hi": 2.204535087858197, "ta": 2.3974673319266016, "th": 2.9136870311157437, "zh": 2.745087847574647, "my": 2.9741846350051193}, "t": 909.1} |
| {"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.8994, "de": 2.05776, "vi": 1.72668, "tr": 2.04298, "fi": 2.21584, "ru": 1.21002, "el": 1.22665, "hi": 1.06247, "ta": 0.90708, "th": 0.9936, "zh": 2.87943, "my": 1.03608}, "bytes_per_chunk": {"en": 5.594, "de": 6.159, "vi": 5.55, "tr": 5.756, "fi": 6.165, "ru": 5.632, "el": 5.514, "hi": 6.604, "ta": 6.704, "th": 5.874, "zh": 5.792, "my": 5.73}, "bpb_mean": 1.60483, "t": 915.2} |
| {"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.31014, "bpb": 1.89013, "chunk_loss": 0.030114, "F": 0.17142, "G": 0.2139, "chunks": 175.53, "bytes_per_chunk": 5.834, "boundary_entropy": 0.19826, "router_noise": 0.0, "grad_norm": 0.4298, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.286, "alpha": 0.03, "t": 915.4} |
| {"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.28384, "bpb": 1.8522, "chunk_loss": 0.030017, "F": 0.17146, "G": 0.21028, "chunks": 175.58, "bytes_per_chunk": 5.832, "boundary_entropy": 0.18445, "router_noise": 0.0, "grad_norm": 0.4104, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 934.5} |
| {"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.22305, "bpb": 1.76449, "chunk_loss": 0.030084, "F": 0.17728, "G": 0.21316, "chunks": 181.53, "bytes_per_chunk": 5.641, "boundary_entropy": 0.17453, "router_noise": 0.0, "grad_norm": 0.4009, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 953.6} |
| {"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.19886, "bpb": 1.72959, "chunk_loss": 0.029971, "F": 0.16853, "G": 0.21929, "chunks": 172.58, "bytes_per_chunk": 5.934, "boundary_entropy": 0.20857, "router_noise": 0.0, "grad_norm": 0.3945, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 972.8} |
| {"event": "phase_transition", "step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 991.6} |
| {"event": "train", "step": 4400, "phase": 2, "objective": "global", "lm_loss": 1.1528, "bpb": 1.66314, "chunk_loss": 0.030088, "F": 0.177, "G": 0.21468, "chunks": 181.25, "bytes_per_chunk": 5.65, "boundary_entropy": 0.1837, "router_noise": 0.0, "grad_norm": 0.3724, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 991.8} |
| {"event": "train", "step": 4500, "phase": 2, "objective": "global", "lm_loss": 1.24256, "bpb": 1.79263, "chunk_loss": 0.030029, "F": 0.17166, "G": 0.22037, "chunks": 175.78, "bytes_per_chunk": 5.825, "boundary_entropy": 0.20197, "router_noise": 0.0, "grad_norm": 0.4102, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1010.9} |
| {"event": "train", "step": 4600, "phase": 2, "objective": "global", "lm_loss": 1.18217, "bpb": 1.70551, "chunk_loss": 0.03013, "F": 0.17828, "G": 0.21912, "chunks": 182.56, "bytes_per_chunk": 5.609, "boundary_entropy": 0.19472, "router_noise": 0.0, "grad_norm": 0.374, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1030.2} |
| {"event": "train", "step": 4700, "phase": 2, "objective": "global", "lm_loss": 1.11185, "bpb": 1.60405, "chunk_loss": 0.029981, "F": 0.1694, "G": 0.22004, "chunks": 173.47, "bytes_per_chunk": 5.903, "boundary_entropy": 0.21302, "router_noise": 0.0, "grad_norm": 0.3767, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 1049.5} |
| {"event": "train", "step": 4800, "phase": 2, "objective": "global", "lm_loss": 1.24114, "bpb": 1.79059, "chunk_loss": 0.030062, "F": 0.17451, "G": 0.21248, "chunks": 178.7, "bytes_per_chunk": 5.73, "boundary_entropy": 0.17618, "router_noise": 0.0, "grad_norm": 0.3945, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1068.6} |
| {"event": "train", "step": 4900, "phase": 2, "objective": "global", "lm_loss": 1.2348, "bpb": 1.78144, "chunk_loss": 0.030097, "F": 0.17545, "G": 0.20935, "chunks": 179.66, "bytes_per_chunk": 5.7, "boundary_entropy": 0.1619, "router_noise": 0.0, "grad_norm": 0.3959, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1087.7} |
| {"event": "probe", "step": 5000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2914.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.167442, "cps": 22.765625, "mean_p": 0.183232, "boundary_entropy": 0.100889, "bpb": 1.827635}, "de": {"chunks": 3327.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.162032, "cps": 25.992188, "mean_p": 0.245554, "boundary_entropy": 0.273284, "bpb": 1.994826}, "vi": {"chunks": 4344.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.235639, "cps": 33.9375, "mean_p": 0.194934, "boundary_entropy": 0.113883, "bpb": 1.646584}, "tr": {"chunks": 3353.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185218, "cps": 26.195312, "mean_p": 0.182899, "boundary_entropy": 0.132973, "bpb": 1.962563}, "fi": {"chunks": 3041.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.161661, "cps": 23.757812, "mean_p": 0.240332, "boundary_entropy": 0.288492, "bpb": 2.114381}, "ru": {"chunks": 6313.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.33314, "cps": 49.320312, "mean_p": 0.190588, "boundary_entropy": 0.116197, "bpb": 1.172572}, "el": {"chunks": 6865.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.334503, "cps": 53.632812, "mean_p": 0.196133, "boundary_entropy": 0.12335, "bpb": 1.181922}, "hi": {"chunks": 7124.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.423418, "cps": 55.65625, "mean_p": 0.278597, "boundary_entropy": 0.426601, "bpb": 1.010927}, "ta": {"chunks": 8638.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.462049, "cps": 67.484375, "mean_p": 0.27892, "boundary_entropy": 0.459263, "bpb": 0.860466}, "th": {"chunks": 7815.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.485856, "cps": 61.054688, "mean_p": 0.324746, "boundary_entropy": 0.452689, "bpb": 0.962741}, "zh": {"chunks": 2726.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.461565, "cps": 21.296875, "mean_p": 0.206561, "boundary_entropy": 0.181895, "bpb": 2.794716}, "my": {"chunks": 9229.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.48707, "cps": 72.101562, "mean_p": 0.325184, "boundary_entropy": 0.525646, "bpb": 0.995408}}, "bpb": {"en": 1.82763, "de": 1.99483, "vi": 1.64658, "tr": 1.96256, "fi": 2.11438, "ru": 1.17257, "el": 1.18192, "hi": 1.01093, "ta": 0.86047, "th": 0.96274, "zh": 2.79472, "my": 0.99541}, "gini_cpc": 0.224438, "gini_cps": 0.238443, "bpic": 5.842394, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9676871330985317, "vi": 1.407282371291367, "tr": 1.1061590446886498, "fi": 0.9654707246271266, "ru": 1.9895788143128526, "el": 1.9977183976020878, "hi": 2.528735529967029, "ta": 2.759448562270879, "th": 2.901633053982052, "zh": 2.756557027844838, "my": 2.9088802479176112}, "t": 1110.1} |
| {"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.83114, "de": 1.9918, "vi": 1.65728, "tr": 1.95555, "fi": 2.12995, "ru": 1.17039, "el": 1.17883, "hi": 1.01779, "ta": 0.86436, "th": 0.95988, "zh": 2.80228, "my": 1.00354}, "bytes_per_chunk": {"en": 5.597, "de": 6.119, "vi": 5.574, "tr": 5.884, "fi": 6.129, "ru": 5.611, "el": 5.535, "hi": 5.916, "ta": 5.86, "th": 5.752, "zh": 5.774, "my": 5.94}, "bpb_mean": 1.5469, "t": 1116.1} |
| {"event": "train", "step": 5000, "phase": 2, "objective": "global", "lm_loss": 1.19996, "bpb": 1.73118, "chunk_loss": 0.030071, "F": 0.17464, "G": 0.2154, "chunks": 178.83, "bytes_per_chunk": 5.726, "boundary_entropy": 0.19007, "router_noise": 0.0, "grad_norm": 0.3982, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.286, "alpha": 0.03, "t": 1116.3} |
| {"event": "train", "step": 5100, "phase": 2, "objective": "global", "lm_loss": 1.20343, "bpb": 1.73618, "chunk_loss": 0.03003, "F": 0.17186, "G": 0.21077, "chunks": 175.98, "bytes_per_chunk": 5.819, "boundary_entropy": 0.17967, "router_noise": 0.0, "grad_norm": 0.3811, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 1135.2} |
| {"event": "train", "step": 5200, "phase": 2, "objective": "global", "lm_loss": 1.13341, "bpb": 1.63516, "chunk_loss": 0.030126, "F": 0.17566, "G": 0.22811, "chunks": 179.88, "bytes_per_chunk": 5.693, "boundary_entropy": 0.21674, "router_noise": 0.0, "grad_norm": 0.3985, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1154.2} |
| {"event": "train", "step": 5300, "phase": 2, "objective": "global", "lm_loss": 1.15997, "bpb": 1.67348, "chunk_loss": 0.030076, "F": 0.17546, "G": 0.21551, "chunks": 179.67, "bytes_per_chunk": 5.699, "boundary_entropy": 0.18125, "router_noise": 0.0, "grad_norm": 0.3993, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1173.2} |
| {"event": "train", "step": 5400, "phase": 2, "objective": "global", "lm_loss": 1.21332, "bpb": 1.75046, "chunk_loss": 0.030155, "F": 0.17731, "G": 0.22206, "chunks": 181.56, "bytes_per_chunk": 5.64, "boundary_entropy": 0.18987, "router_noise": 0.0, "grad_norm": 0.401, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 1192.5} |
| {"event": "train", "step": 5500, "phase": 2, "objective": "global", "lm_loss": 1.16859, "bpb": 1.68592, "chunk_loss": 0.030096, "F": 0.17616, "G": 0.22077, "chunks": 180.39, "bytes_per_chunk": 5.677, "boundary_entropy": 0.19465, "router_noise": 0.0, "grad_norm": 0.425, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1211.5} |
| {"event": "train", "step": 5600, "phase": 2, "objective": "global", "lm_loss": 1.13027, "bpb": 1.63064, "chunk_loss": 0.030037, "F": 0.17216, "G": 0.20912, "chunks": 176.3, "bytes_per_chunk": 5.808, "boundary_entropy": 0.17002, "router_noise": 0.0, "grad_norm": 0.3855, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1230.5} |
| {"event": "train", "step": 5700, "phase": 2, "objective": "global", "lm_loss": 1.11743, "bpb": 1.61211, "chunk_loss": 0.030057, "F": 0.17513, "G": 0.21808, "chunks": 179.33, "bytes_per_chunk": 5.71, "boundary_entropy": 0.1907, "router_noise": 0.0, "grad_norm": 0.39, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1249.6} |
| {"event": "train", "step": 5800, "phase": 2, "objective": "global", "lm_loss": 1.10081, "bpb": 1.58814, "chunk_loss": 0.030074, "F": 0.17357, "G": 0.21656, "chunks": 177.73, "bytes_per_chunk": 5.761, "boundary_entropy": 0.19439, "router_noise": 0.0, "grad_norm": 0.3853, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1268.8} |
| {"event": "train", "step": 5900, "phase": 2, "objective": "global", "lm_loss": 1.15721, "bpb": 1.66951, "chunk_loss": 0.030058, "F": 0.17448, "G": 0.21985, "chunks": 178.67, "bytes_per_chunk": 5.731, "boundary_entropy": 0.18708, "router_noise": 0.0, "grad_norm": 0.4057, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1287.9} |
| {"event": "probe", "step": 6000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2912.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.167327, "cps": 22.75, "mean_p": 0.185294, "boundary_entropy": 0.10199, "bpb": 1.78224}, "de": {"chunks": 3377.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.164467, "cps": 26.382812, "mean_p": 0.26856, "boundary_entropy": 0.312995, "bpb": 1.933136}, "vi": {"chunks": 4377.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.237429, "cps": 34.195312, "mean_p": 0.198377, "boundary_entropy": 0.113262, "bpb": 1.582784}, "tr": {"chunks": 3489.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.19273, "cps": 27.257812, "mean_p": 0.191639, "boundary_entropy": 0.131595, "bpb": 1.889481}, "fi": {"chunks": 3082.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.16384, "cps": 24.078125, "mean_p": 0.252646, "boundary_entropy": 0.306432, "bpb": 2.051723}, "ru": {"chunks": 6294.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.332137, "cps": 49.171875, "mean_p": 0.189842, "boundary_entropy": 0.108521, "bpb": 1.140068}, "el": {"chunks": 6856.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.334064, "cps": 53.5625, "mean_p": 0.194445, "boundary_entropy": 0.113055, "bpb": 1.147231}, "hi": {"chunks": 6901.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.410163, "cps": 53.914062, "mean_p": 0.288745, "boundary_entropy": 0.438169, "bpb": 0.982646}, "ta": {"chunks": 8457.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.452367, "cps": 66.070312, "mean_p": 0.278713, "boundary_entropy": 0.462231, "bpb": 0.843566}, "th": {"chunks": 7645.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.475288, "cps": 59.726562, "mean_p": 0.3236, "boundary_entropy": 0.454982, "bpb": 0.937561}, "zh": {"chunks": 2787.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.471893, "cps": 21.773438, "mean_p": 0.202728, "boundary_entropy": 0.160296, "bpb": 2.718448}, "my": {"chunks": 8202.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.432869, "cps": 64.078125, "mean_p": 0.306232, "boundary_entropy": 0.504054, "bpb": 0.96476}}, "bpb": {"en": 1.78224, "de": 1.93314, "vi": 1.58278, "tr": 1.88948, "fi": 2.05172, "ru": 1.14007, "el": 1.14723, "hi": 0.98265, "ta": 0.84357, "th": 0.93756, "zh": 2.71845, "my": 0.96476}, "gini_cpc": 0.21534, "gini_cps": 0.222069, "bpic": 5.961276, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9829046802440242, "vi": 1.4189469348796038, "tr": 1.1518161604870298, "fi": 0.9791596321651874, "ru": 1.9849532097190408, "el": 1.9964696537200557, "hi": 2.451261838087618, "ta": 2.703482798626888, "th": 2.8404632455669914, "zh": 2.820176410746382, "my": 2.5869565737460762}, "t": 1310.2} |
| {"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.78323, "de": 1.93245, "vi": 1.5956, "tr": 1.88246, "fi": 2.0686, "ru": 1.13833, "el": 1.14447, "hi": 0.98862, "ta": 0.84777, "th": 0.93408, "zh": 2.72553, "my": 0.97415}, "bytes_per_chunk": {"en": 5.604, "de": 6.061, "vi": 5.537, "tr": 5.585, "fi": 6.112, "ru": 5.642, "el": 5.555, "hi": 6.023, "ta": 5.991, "th": 5.851, "zh": 5.7, "my": 6.611}, "bpb_mean": 1.50127, "t": 1316.4} |
| {"event": "train", "step": 6000, "phase": 2, "objective": "global", "lm_loss": 1.1389, "bpb": 1.64309, "chunk_loss": 0.030077, "F": 0.17281, "G": 0.22303, "chunks": 176.95, "bytes_per_chunk": 5.787, "boundary_entropy": 0.20121, "router_noise": 0.0, "grad_norm": 0.459, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.287, "alpha": 0.03, "t": 1316.6} |
| {"event": "train", "step": 6100, "phase": 2, "objective": "global", "lm_loss": 1.07053, "bpb": 1.54444, "chunk_loss": 0.030148, "F": 0.17729, "G": 0.21777, "chunks": 181.55, "bytes_per_chunk": 5.64, "boundary_entropy": 0.1786, "router_noise": 0.0, "grad_norm": 0.3788, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 1335.8} |
| {"event": "train", "step": 6200, "phase": 2, "objective": "global", "lm_loss": 1.18953, "bpb": 1.71613, "chunk_loss": 0.030096, "F": 0.17714, "G": 0.20848, "chunks": 181.39, "bytes_per_chunk": 5.645, "boundary_entropy": 0.15142, "router_noise": 0.0, "grad_norm": 0.3871, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 1355.1} |
| {"event": "train", "step": 6300, "phase": 2, "objective": "global", "lm_loss": 1.11071, "bpb": 1.60241, "chunk_loss": 0.030051, "F": 0.17427, "G": 0.20714, "chunks": 178.45, "bytes_per_chunk": 5.738, "boundary_entropy": 0.14955, "router_noise": 0.0, "grad_norm": 0.3996, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1374.2} |
| {"event": "train", "step": 6400, "phase": 2, "objective": "global", "lm_loss": 1.01573, "bpb": 1.46539, "chunk_loss": 0.03016, "F": 0.17644, "G": 0.229, "chunks": 180.67, "bytes_per_chunk": 5.668, "boundary_entropy": 0.21438, "router_noise": 0.0, "grad_norm": 0.3873, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1393.4} |
| {"event": "train", "step": 6500, "phase": 2, "objective": "global", "lm_loss": 1.08002, "bpb": 1.55814, "chunk_loss": 0.030016, "F": 0.16936, "G": 0.22035, "chunks": 173.42, "bytes_per_chunk": 5.905, "boundary_entropy": 0.20602, "router_noise": 0.0, "grad_norm": 0.4002, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1412.6} |
| {"event": "train", "step": 6600, "phase": 2, "objective": "global", "lm_loss": 1.13088, "bpb": 1.63152, "chunk_loss": 0.030106, "F": 0.17844, "G": 0.21087, "chunks": 182.72, "bytes_per_chunk": 5.604, "boundary_entropy": 0.1524, "router_noise": 0.0, "grad_norm": 0.3897, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1431.7} |
| {"event": "train", "step": 6700, "phase": 2, "objective": "global", "lm_loss": 1.11725, "bpb": 1.61185, "chunk_loss": 0.030087, "F": 0.17467, "G": 0.21946, "chunks": 178.86, "bytes_per_chunk": 5.725, "boundary_entropy": 0.18626, "router_noise": 0.0, "grad_norm": 0.3839, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1450.8} |
| {"event": "train", "step": 6800, "phase": 2, "objective": "global", "lm_loss": 1.12429, "bpb": 1.62201, "chunk_loss": 0.030062, "F": 0.17422, "G": 0.22137, "chunks": 178.41, "bytes_per_chunk": 5.74, "boundary_entropy": 0.19027, "router_noise": 0.0, "grad_norm": 0.3931, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1470.0} |
| {"event": "train", "step": 6900, "phase": 2, "objective": "global", "lm_loss": 1.10311, "bpb": 1.59145, "chunk_loss": 0.030077, "F": 0.17506, "G": 0.21108, "chunks": 179.27, "bytes_per_chunk": 5.712, "boundary_entropy": 0.16254, "router_noise": 0.0, "grad_norm": 0.3845, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1488.9} |
| {"event": "probe", "step": 7000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2911.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.16727, "cps": 22.742188, "mean_p": 0.181801, "boundary_entropy": 0.091633, "bpb": 1.743446}, "de": {"chunks": 3365.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.163883, "cps": 26.289062, "mean_p": 0.26678, "boundary_entropy": 0.307212, "bpb": 1.895089}, "vi": {"chunks": 4346.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.235747, "cps": 33.953125, "mean_p": 0.193293, "boundary_entropy": 0.096359, "bpb": 1.537177}, "tr": {"chunks": 3341.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184555, "cps": 26.101562, "mean_p": 0.18373, "boundary_entropy": 0.128066, "bpb": 1.8465}, "fi": {"chunks": 3079.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.163681, "cps": 24.054688, "mean_p": 0.252657, "boundary_entropy": 0.305692, "bpb": 1.997338}, "ru": {"chunks": 6306.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.33277, "cps": 49.265625, "mean_p": 0.189451, "boundary_entropy": 0.097843, "bpb": 1.115146}, "el": {"chunks": 6854.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.333967, "cps": 53.546875, "mean_p": 0.194315, "boundary_entropy": 0.106319, "bpb": 1.110115}, "hi": {"chunks": 7213.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.428707, "cps": 56.351562, "mean_p": 0.289473, "boundary_entropy": 0.428135, "bpb": 0.95518}, "ta": {"chunks": 9230.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.493715, "cps": 72.109375, "mean_p": 0.286354, "boundary_entropy": 0.449956, "bpb": 0.825604}, "th": {"chunks": 7634.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.474604, "cps": 59.640625, "mean_p": 0.324171, "boundary_entropy": 0.442771, "bpb": 0.906727}, "zh": {"chunks": 2799.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.473925, "cps": 21.867188, "mean_p": 0.199132, "boundary_entropy": 0.143796, "bpb": 2.669233}, "my": {"chunks": 9877.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.521269, "cps": 77.164062, "mean_p": 0.314415, "boundary_entropy": 0.496846, "bpb": 0.931815}}, "bpb": {"en": 1.74345, "de": 1.89509, "vi": 1.53718, "tr": 1.8465, "fi": 1.99734, "ru": 1.11515, "el": 1.11011, "hi": 0.95518, "ta": 0.8256, "th": 0.90673, "zh": 2.66923, "my": 0.93181}, "gini_cpc": 0.232343, "gini_cps": 0.246933, "bpic": 5.731924, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9797484298678956, "vi": 1.4093812672618298, "tr": 1.1033361276923854, "fi": 0.9785425621198091, "ru": 1.9894208574699372, "el": 1.9965728892986703, "hi": 2.562965581697338, "ta": 2.951604380980959, "th": 2.8373506129996655, "zh": 2.8332922298187464, "my": 3.1163310902347}, "t": 1511.4} |
| {"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.74437, "de": 1.89367, "vi": 1.5501, "tr": 1.83998, "fi": 2.01399, "ru": 1.11336, "el": 1.10818, "hi": 0.96145, "ta": 0.82965, "th": 0.90506, "zh": 2.67717, "my": 0.94022}, "bytes_per_chunk": {"en": 5.607, "de": 6.068, "vi": 5.565, "tr": 5.874, "fi": 6.123, "ru": 5.627, "el": 5.55, "hi": 5.949, "ta": 5.536, "th": 5.908, "zh": 5.667, "my": 5.597}, "bpb_mean": 1.46477, "t": 1517.6} |
| {"event": "train", "step": 7000, "phase": 2, "objective": "global", "lm_loss": 1.0975, "bpb": 1.58336, "chunk_loss": 0.030146, "F": 0.17671, "G": 0.2232, "chunks": 180.95, "bytes_per_chunk": 5.659, "boundary_entropy": 0.18601, "router_noise": 0.0, "grad_norm": 0.3928, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.289, "alpha": 0.03, "t": 1517.8} |
| {"event": "train", "step": 7100, "phase": 2, "objective": "global", "lm_loss": 1.08448, "bpb": 1.56458, "chunk_loss": 0.030022, "F": 0.17183, "G": 0.22543, "chunks": 175.95, "bytes_per_chunk": 5.82, "boundary_entropy": 0.20754, "router_noise": 0.0, "grad_norm": 0.4034, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1536.9} |
| {"event": "train", "step": 7200, "phase": 2, "objective": "global", "lm_loss": 1.13508, "bpb": 1.63758, "chunk_loss": 0.030109, "F": 0.17734, "G": 0.20676, "chunks": 181.59, "bytes_per_chunk": 5.639, "boundary_entropy": 0.15027, "router_noise": 0.0, "grad_norm": 0.3964, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1555.9} |
| {"event": "train", "step": 7300, "phase": 2, "objective": "global", "lm_loss": 1.14303, "bpb": 1.64904, "chunk_loss": 0.030094, "F": 0.17596, "G": 0.21531, "chunks": 180.19, "bytes_per_chunk": 5.683, "boundary_entropy": 0.17287, "router_noise": 0.0, "grad_norm": 0.3955, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1574.9} |
| {"event": "train", "step": 7400, "phase": 2, "objective": "global", "lm_loss": 1.10628, "bpb": 1.59602, "chunk_loss": 0.030081, "F": 0.17519, "G": 0.21232, "chunks": 179.39, "bytes_per_chunk": 5.708, "boundary_entropy": 0.16876, "router_noise": 0.0, "grad_norm": 0.387, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1594.1} |
| {"event": "train", "step": 7500, "phase": 2, "objective": "global", "lm_loss": 1.08311, "bpb": 1.56259, "chunk_loss": 0.030082, "F": 0.17656, "G": 0.219, "chunks": 180.8, "bytes_per_chunk": 5.664, "boundary_entropy": 0.17709, "router_noise": 0.0, "grad_norm": 0.3756, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1613.1} |
| {"event": "train", "step": 7600, "phase": 2, "objective": "global", "lm_loss": 1.14424, "bpb": 1.65079, "chunk_loss": 0.030068, "F": 0.17519, "G": 0.21373, "chunks": 179.39, "bytes_per_chunk": 5.708, "boundary_entropy": 0.16082, "router_noise": 0.0, "grad_norm": 0.3842, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1632.2} |
| {"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1651.0} |
| {"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.10054, "bpb": 1.58774, "chunk_loss": 0.030046, "F": 0.17372, "G": 0.21095, "chunks": 177.89, "bytes_per_chunk": 5.756, "boundary_entropy": 0.15875, "router_noise": 0.0, "grad_norm": 0.3918, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1651.2} |
| {"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.14094, "bpb": 1.64603, "chunk_loss": 0.030023, "F": 0.17194, "G": 0.21311, "chunks": 176.06, "bytes_per_chunk": 5.816, "boundary_entropy": 0.16704, "router_noise": 0.0, "grad_norm": 0.414, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1670.3} |
| {"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.11737, "bpb": 1.61202, "chunk_loss": 0.030059, "F": 0.17392, "G": 0.21038, "chunks": 178.09, "bytes_per_chunk": 5.75, "boundary_entropy": 0.15451, "router_noise": 0.0, "grad_norm": 0.3996, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1689.4} |
| {"event": "probe", "step": 8000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2909.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.167155, "cps": 22.726562, "mean_p": 0.179391, "boundary_entropy": 0.080213, "bpb": 1.714137}, "de": {"chunks": 3357.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.163493, "cps": 26.226562, "mean_p": 0.255952, "boundary_entropy": 0.284578, "bpb": 1.857737}, "vi": {"chunks": 4333.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.235042, "cps": 33.851562, "mean_p": 0.190671, "boundary_entropy": 0.08661, "bpb": 1.496549}, "tr": {"chunks": 3282.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181296, "cps": 25.640625, "mean_p": 0.181032, "boundary_entropy": 0.117419, "bpb": 1.805889}, "fi": {"chunks": 3040.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.161608, "cps": 23.75, "mean_p": 0.249468, "boundary_entropy": 0.296857, "bpb": 1.955334}, "ru": {"chunks": 6298.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.332348, "cps": 49.203125, "mean_p": 0.187871, "boundary_entropy": 0.099128, "bpb": 1.092652}, "el": {"chunks": 6814.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.332018, "cps": 53.234375, "mean_p": 0.191815, "boundary_entropy": 0.098469, "bpb": 1.088488}, "hi": {"chunks": 7320.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.435067, "cps": 57.1875, "mean_p": 0.289747, "boundary_entropy": 0.421113, "bpb": 0.938517}, "ta": {"chunks": 8817.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.471623, "cps": 68.882812, "mean_p": 0.280784, "boundary_entropy": 0.440591, "bpb": 0.80719}, "th": {"chunks": 7221.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.448928, "cps": 56.414062, "mean_p": 0.306238, "boundary_entropy": 0.424599, "bpb": 0.889621}, "zh": {"chunks": 2784.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.471385, "cps": 21.75, "mean_p": 0.196045, "boundary_entropy": 0.132762, "bpb": 2.62199}, "my": {"chunks": 8550.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.451235, "cps": 66.796875, "mean_p": 0.306267, "boundary_entropy": 0.494398, "bpb": 0.908779}}, "bpb": {"en": 1.71414, "de": 1.85774, "vi": 1.49655, "tr": 1.80589, "fi": 1.95533, "ru": 1.09265, "el": 1.08849, "hi": 0.93852, "ta": 0.80719, "th": 0.88962, "zh": 2.62199, "my": 0.90878}, "gini_cpc": 0.218727, "gini_cps": 0.230621, "bpic": 5.929409, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9780911583575138, "vi": 1.406131528062652, "tr": 1.0845970634367221, "fi": 0.966812148980873, "ru": 1.9882630468086033, "el": 1.9862855493506206, "hi": 2.602773684374516, "ta": 2.821472165723018, "th": 2.6856949261790635, "zh": 2.8200459659217043, "my": 2.699498792918659}, "t": 1711.6} |
| {"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.71381, "de": 1.85787, "vi": 1.51114, "tr": 1.79835, "fi": 1.97071, "ru": 1.09021, "el": 1.08718, "hi": 0.94518, "ta": 0.81127, "th": 0.88818, "zh": 2.62911, "my": 0.91792}, "bytes_per_chunk": {"en": 5.607, "de": 6.104, "vi": 5.581, "tr": 5.929, "fi": 6.183, "ru": 5.641, "el": 5.571, "hi": 5.844, "ta": 5.715, "th": 6.24, "zh": 5.68, "my": 6.435}, "bpb_mean": 1.43508, "t": 1717.8} |
| {"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.1159, "bpb": 1.60991, "chunk_loss": 0.030058, "F": 0.17563, "G": 0.21858, "chunks": 179.84, "bytes_per_chunk": 5.694, "boundary_entropy": 0.1717, "router_noise": 0.0, "grad_norm": 0.3984, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.286, "alpha": 0.03, "t": 1718.0} |
| {"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.14398, "bpb": 1.65041, "chunk_loss": 0.03011, "F": 0.17633, "G": 0.21242, "chunks": 180.56, "bytes_per_chunk": 5.671, "boundary_entropy": 0.15674, "router_noise": 0.0, "grad_norm": 0.4187, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1737.1} |
| {"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.09447, "bpb": 1.57899, "chunk_loss": 0.030043, "F": 0.17505, "G": 0.2109, "chunks": 179.25, "bytes_per_chunk": 5.713, "boundary_entropy": 0.15808, "router_noise": 0.0, "grad_norm": 0.4041, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1756.2} |
| {"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.02813, "bpb": 1.48328, "chunk_loss": 0.03004, "F": 0.17395, "G": 0.21369, "chunks": 178.12, "bytes_per_chunk": 5.749, "boundary_entropy": 0.16603, "router_noise": 0.0, "grad_norm": 0.3968, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1775.3} |
| {"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.08096, "bpb": 1.55949, "chunk_loss": 0.03007, "F": 0.17407, "G": 0.2079, "chunks": 178.25, "bytes_per_chunk": 5.745, "boundary_entropy": 0.14961, "router_noise": 0.0, "grad_norm": 0.4073, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1794.3} |
| {"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.08399, "bpb": 1.56387, "chunk_loss": 0.030079, "F": 0.17505, "G": 0.20533, "chunks": 179.25, "bytes_per_chunk": 5.713, "boundary_entropy": 0.14356, "router_noise": 0.0, "grad_norm": 0.3904, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 1813.7} |
| {"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.14688, "bpb": 1.6546, "chunk_loss": 0.030066, "F": 0.17268, "G": 0.21481, "chunks": 176.83, "bytes_per_chunk": 5.791, "boundary_entropy": 0.17253, "router_noise": 0.0, "grad_norm": 0.418, "lr": 6.33e-05, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1832.9} |
| {"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 1.01392, "bpb": 1.46278, "chunk_loss": 0.030052, "F": 0.17398, "G": 0.21943, "chunks": 178.16, "bytes_per_chunk": 5.748, "boundary_entropy": 0.18137, "router_noise": 0.0, "grad_norm": 0.4061, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1851.9} |
| {"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.03493, "bpb": 1.49309, "chunk_loss": 0.030065, "F": 0.17461, "G": 0.21468, "chunks": 178.8, "bytes_per_chunk": 5.727, "boundary_entropy": 0.16701, "router_noise": 0.0, "grad_norm": 0.4005, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 1871.2} |
| {"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.09073, "bpb": 1.57359, "chunk_loss": 0.030058, "F": 0.17441, "G": 0.19887, "chunks": 178.59, "bytes_per_chunk": 5.734, "boundary_entropy": 0.12295, "router_noise": 0.0, "grad_norm": 0.3964, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1890.4} |
| {"event": "probe", "step": 9000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2913.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.167385, "cps": 22.757812, "mean_p": 0.179784, "boundary_entropy": 0.080321, "bpb": 1.692013}, "de": {"chunks": 3339.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.162616, "cps": 26.085938, "mean_p": 0.256761, "boundary_entropy": 0.28714, "bpb": 1.834766}, "vi": {"chunks": 4307.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.233632, "cps": 33.648438, "mean_p": 0.188727, "boundary_entropy": 0.083347, "bpb": 1.46958}, "tr": {"chunks": 3292.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181848, "cps": 25.71875, "mean_p": 0.181575, "boundary_entropy": 0.118526, "bpb": 1.778891}, "fi": {"chunks": 3061.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.162724, "cps": 23.914062, "mean_p": 0.253786, "boundary_entropy": 0.301689, "bpb": 1.922497}, "ru": {"chunks": 6295.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.33219, "cps": 49.179688, "mean_p": 0.186959, "boundary_entropy": 0.089506, "bpb": 1.074617}, "el": {"chunks": 6852.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.333869, "cps": 53.53125, "mean_p": 0.191162, "boundary_entropy": 0.093164, "bpb": 1.068576}, "hi": {"chunks": 7039.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.418366, "cps": 54.992188, "mean_p": 0.287852, "boundary_entropy": 0.419894, "bpb": 0.917894}, "ta": {"chunks": 8677.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.464135, "cps": 67.789062, "mean_p": 0.267102, "boundary_entropy": 0.419837, "bpb": 0.794098}, "th": {"chunks": 7647.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.475412, "cps": 59.742188, "mean_p": 0.323254, "boundary_entropy": 0.437431, "bpb": 0.870023}, "zh": {"chunks": 2803.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.474602, "cps": 21.898438, "mean_p": 0.196181, "boundary_entropy": 0.124994, "bpb": 2.59345}, "my": {"chunks": 8434.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.445113, "cps": 65.890625, "mean_p": 0.303131, "boundary_entropy": 0.487436, "bpb": 0.892375}}, "bpb": {"en": 1.69201, "de": 1.83477, "vi": 1.46958, "tr": 1.77889, "fi": 1.9225, "ru": 1.07462, "el": 1.06858, "hi": 0.91789, "ta": 0.7941, "th": 0.87002, "zh": 2.59345, "my": 0.89238}, "gini_cpc": 0.220608, "gini_cps": 0.228979, "bpic": 5.935461, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9715108326035962, "vi": 1.3957748394797094, "tr": 1.0864078892193332, "fi": 0.9721540319688269, "ru": 1.9845870617294685, "el": 1.994619886668059, "hi": 2.499421652896862, "ta": 2.772858854906765, "th": 2.8402309819711005, "zh": 2.8353931818189744, "my": 2.659217475677919}, "t": 1912.9} |
| {"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.69162, "de": 1.83424, "vi": 1.48456, "tr": 1.77125, "fi": 1.9389, "ru": 1.07218, "el": 1.06729, "hi": 0.92407, "ta": 0.7992, "th": 0.86937, "zh": 2.5997, "my": 0.90196}, "bytes_per_chunk": {"en": 5.604, "de": 6.125, "vi": 5.611, "tr": 5.916, "fi": 6.145, "ru": 5.642, "el": 5.546, "hi": 6.041, "ta": 5.866, "th": 5.939, "zh": 5.66, "my": 6.411}, "bpb_mean": 1.41286, "t": 1918.9} |
| {"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.09445, "bpb": 1.57896, "chunk_loss": 0.030043, "F": 0.17146, "G": 0.21431, "chunks": 175.58, "bytes_per_chunk": 5.832, "boundary_entropy": 0.17512, "router_noise": 0.0, "grad_norm": 0.4311, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.288, "alpha": 0.03, "t": 1919.1} |
| {"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.06586, "bpb": 1.5377, "chunk_loss": 0.030078, "F": 0.17616, "G": 0.21148, "chunks": 180.39, "bytes_per_chunk": 5.677, "boundary_entropy": 0.15365, "router_noise": 0.0, "grad_norm": 0.3993, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 1938.3} |
| {"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 0.9901, "bpb": 1.42841, "chunk_loss": 0.030034, "F": 0.17209, "G": 0.20946, "chunks": 176.22, "bytes_per_chunk": 5.811, "boundary_entropy": 0.16023, "router_noise": 0.0, "grad_norm": 0.3957, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 1957.5} |
| {"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 1.06908, "bpb": 1.54236, "chunk_loss": 0.030055, "F": 0.17603, "G": 0.21355, "chunks": 180.25, "bytes_per_chunk": 5.681, "boundary_entropy": 0.15098, "router_noise": 0.0, "grad_norm": 0.4057, "lr": 4.71e-05, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1976.5} |
| {"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.11779, "bpb": 1.61264, "chunk_loss": 0.030071, "F": 0.17319, "G": 0.21612, "chunks": 177.34, "bytes_per_chunk": 5.774, "boundary_entropy": 0.16807, "router_noise": 0.0, "grad_norm": 0.4247, "lr": 4.52e-05, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1995.7} |
| {"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.12819, "bpb": 1.62764, "chunk_loss": 0.030039, "F": 0.17462, "G": 0.20716, "chunks": 178.81, "bytes_per_chunk": 5.727, "boundary_entropy": 0.14312, "router_noise": 0.0, "grad_norm": 0.4308, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 2014.9} |
| {"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 0.98452, "bpb": 1.42036, "chunk_loss": 0.030044, "F": 0.17253, "G": 0.22467, "chunks": 176.67, "bytes_per_chunk": 5.796, "boundary_entropy": 0.19532, "router_noise": 0.0, "grad_norm": 0.4129, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 2033.9} |
| {"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 0.99351, "bpb": 1.43333, "chunk_loss": 0.030015, "F": 0.17371, "G": 0.21067, "chunks": 177.88, "bytes_per_chunk": 5.757, "boundary_entropy": 0.15718, "router_noise": 0.0, "grad_norm": 0.3942, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 2053.1} |
| {"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 1.01324, "bpb": 1.46179, "chunk_loss": 0.030109, "F": 0.1758, "G": 0.22236, "chunks": 180.02, "bytes_per_chunk": 5.688, "boundary_entropy": 0.18757, "router_noise": 0.0, "grad_norm": 0.3997, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 2072.2} |
| {"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.09224, "bpb": 1.57577, "chunk_loss": 0.030047, "F": 0.17288, "G": 0.21872, "chunks": 177.03, "bytes_per_chunk": 5.784, "boundary_entropy": 0.17329, "router_noise": 0.0, "grad_norm": 0.4222, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 2091.3} |
| {"event": "probe", "step": 10000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2911.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.16727, "cps": 22.742188, "mean_p": 0.178792, "boundary_entropy": 0.075809, "bpb": 1.677847}, "de": {"chunks": 3341.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.162714, "cps": 26.101562, "mean_p": 0.254043, "boundary_entropy": 0.279859, "bpb": 1.816238}, "vi": {"chunks": 4303.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.233415, "cps": 33.617188, "mean_p": 0.187062, "boundary_entropy": 0.077515, "bpb": 1.451529}, "tr": {"chunks": 3316.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183174, "cps": 25.90625, "mean_p": 0.180951, "boundary_entropy": 0.112691, "bpb": 1.760836}, "fi": {"chunks": 3041.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.161661, "cps": 23.757812, "mean_p": 0.248169, "boundary_entropy": 0.289821, "bpb": 1.902787}, "ru": {"chunks": 6316.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.333298, "cps": 49.34375, "mean_p": 0.186783, "boundary_entropy": 0.085459, "bpb": 1.062717}, "el": {"chunks": 6807.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.331677, "cps": 53.179688, "mean_p": 0.19157, "boundary_entropy": 0.09122, "bpb": 1.055085}, "hi": {"chunks": 6903.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.410282, "cps": 53.929688, "mean_p": 0.278392, "boundary_entropy": 0.402036, "bpb": 0.903647}, "ta": {"chunks": 8420.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.450388, "cps": 65.78125, "mean_p": 0.259336, "boundary_entropy": 0.397766, "bpb": 0.78221}, "th": {"chunks": 7372.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.458315, "cps": 57.59375, "mean_p": 0.310206, "boundary_entropy": 0.420633, "bpb": 0.860442}, "zh": {"chunks": 2804.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.474771, "cps": 21.90625, "mean_p": 0.194252, "boundary_entropy": 0.118552, "bpb": 2.571999}, "my": {"chunks": 9442.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.498311, "cps": 73.765625, "mean_p": 0.306505, "boundary_entropy": 0.481227, "bpb": 0.879005}}, "bpb": {"en": 1.67785, "de": 1.81624, "vi": 1.45153, "tr": 1.76084, "fi": 1.90279, "ru": 1.06272, "el": 1.05509, "hi": 0.90365, "ta": 0.78221, "th": 0.86044, "zh": 2.572, "my": 0.87901}, "gini_cpc": 0.224591, "gini_cps": 0.235038, "bpic": 5.906504, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9727606253160889, "vi": 1.3954366297808685, "tr": 1.0950800956084856, "fi": 0.9664657133505485, "ru": 1.9925756637773744, "el": 1.9828817708573168, "hi": 2.4528145584994765, "ta": 2.6925795111440602, "th": 2.739972323687914, "zh": 2.838353487821281, "my": 2.979082530525062}, "t": 2113.5} |
| {"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.67733, "de": 1.8164, "vi": 1.46663, "tr": 1.75407, "fi": 1.91872, "ru": 1.06, "el": 1.05389, "hi": 0.91054, "ta": 0.78737, "th": 0.8587, "zh": 2.57994, "my": 0.88859}, "bytes_per_chunk": {"en": 5.607, "de": 6.119, "vi": 5.619, "tr": 5.882, "fi": 6.186, "ru": 5.624, "el": 5.577, "hi": 6.151, "ta": 6.048, "th": 6.159, "zh": 5.667, "my": 5.752}, "bpb_mean": 1.39768, "t": 2119.5} |
| {"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.06657, "bpb": 1.53873, "chunk_loss": 0.0301, "F": 0.1757, "G": 0.20955, "chunks": 179.92, "bytes_per_chunk": 5.691, "boundary_entropy": 0.15267, "router_noise": 0.0, "grad_norm": 0.4297, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.284, "alpha": 0.03, "t": 2119.7} |
| {"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 1.00156, "bpb": 1.44495, "chunk_loss": 0.030094, "F": 0.17502, "G": 0.22746, "chunks": 179.22, "bytes_per_chunk": 5.714, "boundary_entropy": 0.20178, "router_noise": 0.0, "grad_norm": 0.4093, "lr": 3.49e-05, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 2139.0} |
| {"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.09489, "bpb": 1.57959, "chunk_loss": 0.030073, "F": 0.17714, "G": 0.20843, "chunks": 181.39, "bytes_per_chunk": 5.645, "boundary_entropy": 0.13724, "router_noise": 0.0, "grad_norm": 0.4368, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 2158.1} |
| {"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 1.06155, "bpb": 1.53149, "chunk_loss": 0.030326, "F": 0.17648, "G": 0.21708, "chunks": 180.72, "bytes_per_chunk": 5.666, "boundary_entropy": 0.17314, "router_noise": 0.0, "grad_norm": 0.4507, "lr": 3.3e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 2177.2} |
| {"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.02222, "bpb": 1.47475, "chunk_loss": 0.030055, "F": 0.17453, "G": 0.21229, "chunks": 178.72, "bytes_per_chunk": 5.73, "boundary_entropy": 0.15785, "router_noise": 0.0, "grad_norm": 0.4106, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 2196.3} |
| {"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.04895, "bpb": 1.51331, "chunk_loss": 0.030094, "F": 0.17642, "G": 0.21995, "chunks": 180.66, "bytes_per_chunk": 5.668, "boundary_entropy": 0.1749, "router_noise": 0.0, "grad_norm": 0.4188, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 2215.3} |
| {"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 1.07223, "bpb": 1.54691, "chunk_loss": 0.030068, "F": 0.17484, "G": 0.21564, "chunks": 179.03, "bytes_per_chunk": 5.72, "boundary_entropy": 0.16396, "router_noise": 0.0, "grad_norm": 0.4153, "lr": 3.1e-05, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 2234.4} |
| {"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.04076, "bpb": 1.5015, "chunk_loss": 0.03008, "F": 0.17595, "G": 0.21643, "chunks": 180.17, "bytes_per_chunk": 5.683, "boundary_entropy": 0.16559, "router_noise": 0.0, "grad_norm": 0.4102, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 2253.3} |
| {"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.06115, "bpb": 1.53092, "chunk_loss": 0.030066, "F": 0.17587, "G": 0.21577, "chunks": 180.09, "bytes_per_chunk": 5.686, "boundary_entropy": 0.16651, "router_noise": 0.0, "grad_norm": 0.4242, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 2272.4} |
| {"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.04454, "bpb": 1.50695, "chunk_loss": 0.03009, "F": 0.17621, "G": 0.20873, "chunks": 180.44, "bytes_per_chunk": 5.675, "boundary_entropy": 0.14943, "router_noise": 0.0, "grad_norm": 0.4311, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 2291.4} |
| {"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2912.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.167327, "cps": 22.75, "mean_p": 0.178312, "boundary_entropy": 0.073475, "bpb": 1.668945}, "de": {"chunks": 3342.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.162762, "cps": 26.109375, "mean_p": 0.251475, "boundary_entropy": 0.27513, "bpb": 1.806194}, "vi": {"chunks": 4319.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.234283, "cps": 33.742188, "mean_p": 0.188065, "boundary_entropy": 0.078647, "bpb": 1.439914}, "tr": {"chunks": 3314.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183064, "cps": 25.890625, "mean_p": 0.180203, "boundary_entropy": 0.111231, "bpb": 1.748992}, "fi": {"chunks": 3084.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.163947, "cps": 24.09375, "mean_p": 0.249989, "boundary_entropy": 0.290589, "bpb": 1.890023}, "ru": {"chunks": 6304.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.332665, "cps": 49.25, "mean_p": 0.185929, "boundary_entropy": 0.085006, "bpb": 1.054373}, "el": {"chunks": 6813.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.331969, "cps": 53.226562, "mean_p": 0.190482, "boundary_entropy": 0.089846, "bpb": 1.048655}, "hi": {"chunks": 7030.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.417831, "cps": 54.921875, "mean_p": 0.282468, "boundary_entropy": 0.405095, "bpb": 0.898735}, "ta": {"chunks": 8482.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.453704, "cps": 66.265625, "mean_p": 0.261105, "boundary_entropy": 0.397701, "bpb": 0.778604}, "th": {"chunks": 7451.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.463227, "cps": 58.210938, "mean_p": 0.314455, "boundary_entropy": 0.424019, "bpb": 0.853434}, "zh": {"chunks": 2797.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.473586, "cps": 21.851562, "mean_p": 0.194123, "boundary_entropy": 0.116047, "bpb": 2.563438}, "my": {"chunks": 8849.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.467015, "cps": 69.132812, "mean_p": 0.303736, "boundary_entropy": 0.48178, "bpb": 0.872854}}, "bpb": {"en": 1.66895, "de": 1.80619, "vi": 1.43991, "tr": 1.74899, "fi": 1.89002, "ru": 1.05437, "el": 1.04865, "hi": 0.89874, "ta": 0.7786, "th": 0.85343, "zh": 2.56344, "my": 0.87285}, "gini_cpc": 0.219024, "gini_cps": 0.229595, "bpic": 5.931975, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9727176314407845, "vi": 1.40014434812543, "tr": 1.0940437821307014, "fi": 0.9797950375072803, "ru": 1.988106932645191, "el": 1.9839480383306212, "hi": 2.497083136031873, "ta": 2.711474647978392, "th": 2.7683834719057754, "zh": 2.830295450612713, "my": 2.7910239845256073}, "t": 2313.6} |
| {"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.66777, "de": 1.8065, "vi": 1.45538, "tr": 1.74207, "fi": 1.90593, "ru": 1.05193, "el": 1.04705, "hi": 0.90567, "ta": 0.7836, "th": 0.85169, "zh": 2.57111, "my": 0.88255}, "bytes_per_chunk": {"en": 5.606, "de": 6.129, "vi": 5.603, "tr": 5.901, "fi": 6.13, "ru": 5.632, "el": 5.571, "hi": 5.953, "ta": 6.01, "th": 6.049, "zh": 5.651, "my": 6.076}, "bpb_mean": 1.38927, "t": 2319.6} |
| {"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2912.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.167327, "cps": 22.75, "mean_p": 0.178312, "boundary_entropy": 0.073475, "bpb": 1.668945}, "de": {"chunks": 3342.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.162762, "cps": 26.109375, "mean_p": 0.251475, "boundary_entropy": 0.27513, "bpb": 1.806194}, "vi": {"chunks": 4319.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.234283, "cps": 33.742188, "mean_p": 0.188065, "boundary_entropy": 0.078647, "bpb": 1.439914}, "tr": {"chunks": 3314.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183064, "cps": 25.890625, "mean_p": 0.180203, "boundary_entropy": 0.111231, "bpb": 1.748992}, "fi": {"chunks": 3084.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.163947, "cps": 24.09375, "mean_p": 0.249989, "boundary_entropy": 0.290589, "bpb": 1.890023}, "ru": {"chunks": 6304.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.332665, "cps": 49.25, "mean_p": 0.185929, "boundary_entropy": 0.085006, "bpb": 1.054373}, "el": {"chunks": 6813.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.331969, "cps": 53.226562, "mean_p": 0.190482, "boundary_entropy": 0.089846, "bpb": 1.048655}, "hi": {"chunks": 7030.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.417831, "cps": 54.921875, "mean_p": 0.282468, "boundary_entropy": 0.405095, "bpb": 0.898735}, "ta": {"chunks": 8482.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.453704, "cps": 66.265625, "mean_p": 0.261105, "boundary_entropy": 0.397701, "bpb": 0.778604}, "th": {"chunks": 7451.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.463227, "cps": 58.210938, "mean_p": 0.314455, "boundary_entropy": 0.424019, "bpb": 0.853434}, "zh": {"chunks": 2797.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.473586, "cps": 21.851562, "mean_p": 0.194123, "boundary_entropy": 0.116047, "bpb": 2.563438}, "my": {"chunks": 8849.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.467015, "cps": 69.132812, "mean_p": 0.303736, "boundary_entropy": 0.48178, "bpb": 0.872854}}, "bpb": {"en": 1.66895, "de": 1.80619, "vi": 1.43991, "tr": 1.74899, "fi": 1.89002, "ru": 1.05437, "el": 1.04865, "hi": 0.89874, "ta": 0.7786, "th": 0.85343, "zh": 2.56344, "my": 0.87285}, "gini_cpc": 0.219024, "gini_cps": 0.229595, "bpic": 5.931975, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9727176314407845, "vi": 1.40014434812543, "tr": 1.0940437821307014, "fi": 0.9797950375072803, "ru": 1.988106932645191, "el": 1.9839480383306212, "hi": 2.497083136031873, "ta": 2.711474647978392, "th": 2.7683834719057754, "zh": 2.830295450612713, "my": 2.7910239845256073}, "t": 2323.0} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s0/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s0/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s0/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s0/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s0/th0.70"], "t": 2327.2} |
| {"event": "final", "run": "exp32_local_s0", "condition": "A", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 0.16732747227489514, "cps": 22.75, "bpb": 1.66235250794733, "bytes_per_chunk": 5.6043613041154465}, "de": {"cpc": 0.16276238250620953, "cps": 26.109375, "bpb": 1.8061266396580307, "bytes_per_chunk": 6.0965847645848115}, "vi": {"cpc": 0.23428261459180905, "cps": 33.7421875, "bpb": 1.4564274685993943, "bytes_per_chunk": 5.605799456835532}, "tr": {"cpc": 0.18306358062199635, "cps": 25.890625, "bpb": 1.7411195193976756, "bytes_per_chunk": 5.892928997740786}, "fi": {"cpc": 0.16394662697357928, "cps": 24.09375, "bpb": 1.9122971517185423, "bytes_per_chunk": 6.148635494728448}, "ru": {"cpc": 0.332664907651715, "cps": 49.25, "bpb": 1.0517109185427573, "bytes_per_chunk": 5.6303614768358425}, "el": {"cpc": 0.3319690103785996, "cps": 53.2265625, "bpb": 1.0439139453299882, "bytes_per_chunk": 5.591570325498059}, "hi": {"cpc": 0.41783060921248144, "cps": 54.921875, "bpb": 0.9071088086062249, "bytes_per_chunk": 5.944274692320338}, "ta": {"cpc": 0.45370419898368547, "cps": 66.265625, "bpb": 0.7817135276375686, "bytes_per_chunk": 6.0055555237369775}, "th": {"cpc": 0.46322660864159154, "cps": 58.2109375, "bpb": 0.8549487601185747, "bytes_per_chunk": 6.058051395821779}, "zh": {"cpc": 0.47358618354216053, "cps": 21.8515625, "bpb": 2.5747523843051807, "bytes_per_chunk": 5.67159593686784}, "my": {"cpc": 0.4670149883892759, "cps": 69.1328125, "bpb": 0.8805449111741973, "bytes_per_chunk": 6.099173457576445}}, "gini_cpc": 0.21902357373495107, "gini_cps": 0.2295946231406918, "bpic": 5.931975207505758, "seed": 0, "mean_boundary_entropy": 0.1855776458978653, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}], "t": 2340.3} |
|
|
