| {"event": "start", "run": "jitter_s0", "condition": "A", "params": 22516992, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.184, "vi": 1.39, "tr": 1.12, "fi": 1.104, "ru": 1.971, "el": 2.174, "hi": 2.552, "ta": 3.192, "th": 2.762, "zh": 0.915, "my": 3.524}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "global", "p3_objective": "global", "freeze_in_p3": false, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 1.7} |
| {"event": "probe_schedule", "n": 13, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 2000, 4000, 6000], "t": 1.7} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6384.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.366833, "cps": 49.875, "mean_p": 0.496725, "boundary_entropy": 0.686995, "bpb": 8.160238}, "de": {"chunks": 8628.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.420202, "cps": 67.40625, "mean_p": 0.498889, "boundary_entropy": 0.687821, "bpb": 8.156329}, "vi": {"chunks": 13115.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.711418, "cps": 102.460938, "mean_p": 0.505745, "boundary_entropy": 0.688385, "bpb": 8.087291}, "tr": {"chunks": 5290.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.292217, "cps": 41.328125, "mean_p": 0.489567, "boundary_entropy": 0.687288, "bpb": 8.171434}, "fi": {"chunks": 8895.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.472862, "cps": 69.492188, "mean_p": 0.500688, "boundary_entropy": 0.687512, "bpb": 8.214758}, "ru": {"chunks": 10744.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.566966, "cps": 83.9375, "mean_p": 0.493655, "boundary_entropy": 0.689893, "bpb": 8.099397}, "el": {"chunks": 15039.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.732788, "cps": 117.492188, "mean_p": 0.497482, "boundary_entropy": 0.690234, "bpb": 8.158044}, "hi": {"chunks": 2563.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.152333, "cps": 20.023438, "mean_p": 0.484638, "boundary_entropy": 0.690256, "bpb": 8.262162}, "ta": {"chunks": 38089.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.03739, "cps": 297.570312, "mean_p": 0.512976, "boundary_entropy": 0.690378, "bpb": 8.051803}, "th": {"chunks": 34006.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.114144, "cps": 265.671875, "mean_p": 0.511916, "boundary_entropy": 0.690552, "bpb": 8.248459}, "zh": {"chunks": 11821.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.001524, "cps": 92.351562, "mean_p": 0.52004, "boundary_entropy": 0.685674, "bpb": 8.123118}, "my": {"chunks": 48201.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.543857, "cps": 376.570312, "mean_p": 0.521563, "boundary_entropy": 0.690206, "bpb": 7.913994}}, "bpb": {"en": 8.16024, "de": 8.15633, "vi": 8.08729, "tr": 8.17143, "fi": 8.21476, "ru": 8.0994, "el": 8.15804, "hi": 8.26216, "ta": 8.0518, "th": 8.24846, "zh": 8.12312, "my": 7.91399}, "gini_cpc": 0.424496, "gini_cps": 0.425905, "bpic": 1.892645, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1454838515955792, "vi": 1.939350894526797, "tr": 0.7965927738475616, "fi": 1.2890367847879758, "ru": 1.5455676790921895, "el": 1.997603802482379, "hi": 0.41526447061189375, "ta": 5.553993192060118, "th": 5.7632270176529286, "zh": 5.456221801328583, "my": 6.934639901884223}, "t": 5.1} |
| {"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.64186, "bpb": 8.13949, "chunk_loss": 0.08991, "F": 1.0, "G": 0.4995, "chunks": 1024.0, "bytes_per_chunk": 1.0, "boundary_entropy": 0.69158, "router_noise": 0.0, "grad_norm": 7.8228, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.002, "alpha": 0.03, "t": 5.5} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6292.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.361547, "cps": 49.15625, "mean_p": 0.496424, "boundary_entropy": 0.686985, "bpb": 8.137942}, "de": {"chunks": 8544.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.416111, "cps": 66.75, "mean_p": 0.498658, "boundary_entropy": 0.687817, "bpb": 8.13676}, "vi": {"chunks": 12993.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.704801, "cps": 101.507812, "mean_p": 0.505487, "boundary_entropy": 0.688387, "bpb": 8.072181}, "tr": {"chunks": 5217.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.288184, "cps": 40.757812, "mean_p": 0.489318, "boundary_entropy": 0.687274, "bpb": 8.156829}, "fi": {"chunks": 8811.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.468396, "cps": 68.835938, "mean_p": 0.500401, "boundary_entropy": 0.687508, "bpb": 8.196416}, "ru": {"chunks": 10431.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.550449, "cps": 81.492188, "mean_p": 0.493206, "boundary_entropy": 0.689879, "bpb": 8.060083}, "el": {"chunks": 15001.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.730936, "cps": 117.195312, "mean_p": 0.497429, "boundary_entropy": 0.690233, "bpb": 8.140897}, "hi": {"chunks": 2503.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.148767, "cps": 19.554688, "mean_p": 0.484452, "boundary_entropy": 0.690243, "bpb": 8.249504}, "ta": {"chunks": 37861.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.025194, "cps": 295.789062, "mean_p": 0.512705, "boundary_entropy": 0.690391, "bpb": 8.028624}, "th": {"chunks": 33861.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.105129, "cps": 264.539062, "mean_p": 0.511768, "boundary_entropy": 0.690558, "bpb": 8.229615}, "zh": {"chunks": 11822.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.001693, "cps": 92.359375, "mean_p": 0.520019, "boundary_entropy": 0.685675, "bpb": 8.120583}, "my": {"chunks": 48181.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.542801, "cps": 376.414062, "mean_p": 0.521547, "boundary_entropy": 0.690207, "bpb": 7.893555}}, "bpb": {"en": 8.13794, "de": 8.13676, "vi": 8.07218, "tr": 8.15683, "fi": 8.19642, "ru": 8.06008, "el": 8.1409, "hi": 8.2495, "ta": 8.02862, "th": 8.22962, "zh": 8.12058, "my": 7.89355}, "gini_cpc": 0.426673, "gini_cps": 0.428193, "bpic": 1.90446, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.150917619502558, "vi": 1.9494033261656607, "tr": 0.7970869365552818, "fi": 1.2955337284054524, "ru": 1.52248189685105, "el": 2.021690973974, "hi": 0.41147284837275383, "ta": 5.601470037900537, "th": 5.822561987423376, "zh": 5.5364695874270105, "my": 7.033116960044964}, "t": 8.4} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6140.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.352813, "cps": 47.96875, "mean_p": 0.495813, "boundary_entropy": 0.686963, "bpb": 8.088626}, "de": {"chunks": 8384.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.408318, "cps": 65.5, "mean_p": 0.498195, "boundary_entropy": 0.687808, "bpb": 8.092933}, "vi": {"chunks": 12767.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.692541, "cps": 99.742188, "mean_p": 0.504896, "boundary_entropy": 0.688391, "bpb": 8.038432}, "tr": {"chunks": 5083.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.280782, "cps": 39.710938, "mean_p": 0.488774, "boundary_entropy": 0.687241, "bpb": 8.122514}, "fi": {"chunks": 8653.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.459997, "cps": 67.601562, "mean_p": 0.499815, "boundary_entropy": 0.687498, "bpb": 8.154597}, "ru": {"chunks": 9871.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.520897, "cps": 77.117188, "mean_p": 0.492489, "boundary_entropy": 0.689854, "bpb": 7.994827}, "el": {"chunks": 15018.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.731764, "cps": 117.328125, "mean_p": 0.497432, "boundary_entropy": 0.690233, "bpb": 8.120639}, "hi": {"chunks": 2362.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.140386, "cps": 18.453125, "mean_p": 0.484026, "boundary_entropy": 0.690214, "bpb": 8.216665}, "ta": {"chunks": 37462.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.003851, "cps": 292.671875, "mean_p": 0.512226, "boundary_entropy": 0.690413, "bpb": 7.987519}, "th": {"chunks": 33572.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.087162, "cps": 262.28125, "mean_p": 0.51154, "boundary_entropy": 0.690567, "bpb": 8.202526}, "zh": {"chunks": 11810.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.999661, "cps": 92.265625, "mean_p": 0.519969, "boundary_entropy": 0.685678, "bpb": 8.114145}, "my": {"chunks": 48128.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.540004, "cps": 376.0, "mean_p": 0.521493, "boundary_entropy": 0.69021, "bpb": 7.860478}}, "bpb": {"en": 8.08863, "de": 8.09293, "vi": 8.03843, "tr": 8.12251, "fi": 8.1546, "ru": 7.99483, "el": 8.12064, "hi": 8.21666, "ta": 7.98752, "th": 8.20253, "zh": 8.11415, "my": 7.86048}, "gini_cpc": 0.430739, "gini_cps": 0.432586, "bpic": 1.926128, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.15732307300348, "vi": 1.9629148721319472, "tr": 0.7958391639156395, "fi": 1.3037987770340904, "ru": 1.4764124087904908, "el": 2.0740871646191503, "hi": 0.39790607470076617, "ta": 5.679645622817159, "th": 5.915778412525478, "zh": 5.667769816153395, "my": 7.1992986118703755}, "t": 11.5} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5580.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.320634, "cps": 43.59375, "mean_p": 0.493699, "boundary_entropy": 0.686875, "bpb": 7.919777}, "de": {"chunks": 7847.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.382165, "cps": 61.304688, "mean_p": 0.496587, "boundary_entropy": 0.68777, "bpb": 7.9397}, "vi": {"chunks": 12023.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.652183, "cps": 93.929688, "mean_p": 0.50296, "boundary_entropy": 0.688395, "bpb": 7.926445}, "tr": {"chunks": 4607.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.254488, "cps": 35.992188, "mean_p": 0.486967, "boundary_entropy": 0.687126, "bpb": 8.005026}, "fi": {"chunks": 7997.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.425124, "cps": 62.476562, "mean_p": 0.497685, "boundary_entropy": 0.687452, "bpb": 8.007853}, "ru": {"chunks": 8139.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.429499, "cps": 63.585938, "mean_p": 0.490058, "boundary_entropy": 0.689757, "bpb": 7.777456}, "el": {"chunks": 14842.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.723189, "cps": 115.953125, "mean_p": 0.49725, "boundary_entropy": 0.69023, "bpb": 8.031972}, "hi": {"chunks": 1961.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.116553, "cps": 15.320312, "mean_p": 0.482238, "boundary_entropy": 0.690083, "bpb": 8.094255}, "ta": {"chunks": 35822.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.916127, "cps": 279.859375, "mean_p": 0.510362, "boundary_entropy": 0.690494, "bpb": 7.840834}, "th": {"chunks": 32414.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.015169, "cps": 253.234375, "mean_p": 0.510574, "boundary_entropy": 0.690602, "bpb": 8.088093}, "zh": {"chunks": 11765.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.992042, "cps": 91.914062, "mean_p": 0.519704, "boundary_entropy": 0.685693, "bpb": 8.087489}, "my": {"chunks": 47965.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.531402, "cps": 374.726562, "mean_p": 0.521282, "boundary_entropy": 0.690224, "bpb": 7.784123}}, "bpb": {"en": 7.91978, "de": 7.9397, "vi": 7.92645, "tr": 8.00503, "fi": 8.00785, "ru": 7.77746, "el": 8.03197, "hi": 8.09426, "ta": 7.84083, "th": 8.08809, "zh": 8.08749, "my": 7.78412}, "gini_cpc": 0.445746, "gini_cps": 0.447513, "bpic": 2.009724, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.191903696593315, "vi": 2.034040642653205, "tr": 0.7937021963329313, "fi": 1.3258827910688267, "ru": 1.3395278747127415, "el": 2.255493102676829, "hi": 0.36350671843295146, "ta": 5.976050809974492, "th": 6.284945033274171, "zh": 6.212814833830368, "my": 7.894979269806896}, "t": 14.6} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4774.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.274321, "cps": 37.296875, "mean_p": 0.490456, "boundary_entropy": 0.686711, "bpb": 7.491948}, "de": {"chunks": 7263.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.353723, "cps": 56.742188, "mean_p": 0.494684, "boundary_entropy": 0.687717, "bpb": 7.547367}, "vi": {"chunks": 10680.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.579333, "cps": 83.4375, "mean_p": 0.499673, "boundary_entropy": 0.688367, "bpb": 7.609612}, "tr": {"chunks": 4020.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.222063, "cps": 31.40625, "mean_p": 0.484883, "boundary_entropy": 0.686993, "bpb": 7.68169}, "fi": {"chunks": 6981.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.371113, "cps": 54.539062, "mean_p": 0.494274, "boundary_entropy": 0.687346, "bpb": 7.627208}, "ru": {"chunks": 5054.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.266702, "cps": 39.484375, "mean_p": 0.484986, "boundary_entropy": 0.689479, "bpb": 7.287633}, "el": {"chunks": 13847.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.674706, "cps": 108.179688, "mean_p": 0.496346, "boundary_entropy": 0.690209, "bpb": 7.792458}, "hi": {"chunks": 1194.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.070966, "cps": 9.328125, "mean_p": 0.477649, "boundary_entropy": 0.689689, "bpb": 7.723284}, "ta": {"chunks": 30803.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.64766, "cps": 240.648438, "mean_p": 0.505519, "boundary_entropy": 0.690645, "bpb": 7.428373}, "th": {"chunks": 28695.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.78396, "cps": 224.179688, "mean_p": 0.507303, "boundary_entropy": 0.6907, "bpb": 7.645596}, "zh": {"chunks": 11622.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.967829, "cps": 90.796875, "mean_p": 0.519029, "boundary_entropy": 0.685731, "bpb": 8.011482}, "my": {"chunks": 48036.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.535149, "cps": 375.28125, "mean_p": 0.521759, "boundary_entropy": 0.690171, "bpb": 7.563326}}, "bpb": {"en": 7.49195, "de": 7.54737, "vi": 7.60961, "tr": 7.68169, "fi": 7.62721, "ru": 7.28763, "el": 7.79246, "hi": 7.72328, "ta": 7.42837, "th": 7.6456, "zh": 8.01148, "my": 7.56333}, "gini_cpc": 0.476672, "gini_cps": 0.474585, "bpic": 2.218785, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2894524822161373, "vi": 2.1118828152083617, "tr": 0.8095006600155452, "fi": 1.3528431908729297, "ru": 0.9722271141064229, "el": 2.4595550791706797, "hi": 0.2586967420730542, "ta": 6.006330862108239, "th": 6.503196388478208, "zh": 7.1734674826719, "my": 9.241557406821252}, "t": 18.3} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2918.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.167672, "cps": 22.796875, "mean_p": 0.483036, "boundary_entropy": 0.686288, "bpb": 6.89556}, "de": {"chunks": 5637.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.274534, "cps": 44.039062, "mean_p": 0.489999, "boundary_entropy": 0.687637, "bpb": 6.982104}, "vi": {"chunks": 7859.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.426309, "cps": 61.398438, "mean_p": 0.49252, "boundary_entropy": 0.688228, "bpb": 7.188954}, "tr": {"chunks": 3001.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.165774, "cps": 23.445312, "mean_p": 0.481764, "boundary_entropy": 0.686833, "bpb": 7.198879}, "fi": {"chunks": 4624.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.245814, "cps": 36.125, "mean_p": 0.486731, "boundary_entropy": 0.687058, "bpb": 7.083572}, "ru": {"chunks": 919.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.048496, "cps": 7.179688, "mean_p": 0.470625, "boundary_entropy": 0.68813, "bpb": 6.56532}, "el": {"chunks": 6453.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.314428, "cps": 50.414062, "mean_p": 0.487918, "boundary_entropy": 0.689892, "bpb": 7.258176}, "hi": {"chunks": 485.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.028826, "cps": 3.789062, "mean_p": 0.470466, "boundary_entropy": 0.68891, "bpb": 7.066105}, "ta": {"chunks": 20147.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.077668, "cps": 157.398438, "mean_p": 0.496763, "boundary_entropy": 0.690722, "bpb": 6.869588}, "th": {"chunks": 20155.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.253031, "cps": 157.460938, "mean_p": 0.498736, "boundary_entropy": 0.690775, "bpb": 6.616165}, "zh": {"chunks": 11166.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.89062, "cps": 87.234375, "mean_p": 0.516898, "boundary_entropy": 0.685838, "bpb": 7.835642}, "my": {"chunks": 47523.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.508075, "cps": 371.273438, "mean_p": 0.522019, "boundary_entropy": 0.690101, "bpb": 6.934056}}, "bpb": {"en": 6.89556, "de": 6.9821, "vi": 7.18895, "tr": 7.19888, "fi": 7.08357, "ru": 6.56532, "el": 7.25818, "hi": 7.0661, "ta": 6.86959, "th": 6.61617, "zh": 7.83564, "my": 6.93406}, "gini_cpc": 0.561355, "gini_cps": 0.549132, "bpic": 2.932155, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.6373233685542794, "vi": 2.5425118132675557, "tr": 0.9886766483631835, "fi": 1.466036471376157, "ru": 0.28923119351997695, "el": 1.8752520659463727, "hi": 0.17191964208422997, "ta": 6.427228639238041, "th": 7.473096148097741, "zh": 11.275687043090205, "my": 14.958198951627361}, "t": 22.6} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 246.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.014135, "cps": 1.921875, "mean_p": 0.378113, "boundary_entropy": 0.654411, "bpb": 6.46997}, "de": {"chunks": 216.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.01052, "cps": 1.6875, "mean_p": 0.386177, "boundary_entropy": 0.659998, "bpb": 6.580452}, "vi": {"chunks": 456.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.024736, "cps": 3.5625, "mean_p": 0.413983, "boundary_entropy": 0.6719, "bpb": 7.054647}, "tr": {"chunks": 313.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.01729, "cps": 2.445312, "mean_p": 0.393962, "boundary_entropy": 0.662865, "bpb": 6.858326}, "fi": {"chunks": 297.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.015789, "cps": 2.320312, "mean_p": 0.389414, "boundary_entropy": 0.660704, "bpb": 6.63062}, "ru": {"chunks": 132.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.006966, "cps": 1.03125, "mean_p": 0.415013, "boundary_entropy": 0.674981, "bpb": 6.347138}, "el": {"chunks": 217.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.010574, "cps": 1.695312, "mean_p": 0.449091, "boundary_entropy": 0.684849, "bpb": 6.832869}, "hi": {"chunks": 179.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.010639, "cps": 1.398438, "mean_p": 0.454408, "boundary_entropy": 0.686448, "bpb": 6.423777}, "ta": {"chunks": 474.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.025354, "cps": 3.703125, "mean_p": 0.468511, "boundary_entropy": 0.688738, "bpb": 6.31379}, "th": {"chunks": 184.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.011439, "cps": 1.4375, "mean_p": 0.461773, "boundary_entropy": 0.68786, "bpb": 6.031388}, "zh": {"chunks": 9561.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.618862, "cps": 74.695312, "mean_p": 0.50882, "boundary_entropy": 0.686081, "bpb": 7.402198}, "my": {"chunks": 45213.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.386162, "cps": 353.226562, "mean_p": 0.520571, "boundary_entropy": 0.69015, "bpb": 5.868623}}, "bpb": {"en": 6.46997, "de": 6.58045, "vi": 7.05465, "tr": 6.85833, "fi": 6.63062, "ru": 6.34714, "el": 6.83287, "hi": 6.42378, "ta": 6.31379, "th": 6.03139, "zh": 7.4022, "my": 5.86862}, "gini_cpc": 0.819581, "gini_cps": 0.846406, "bpic": 6.675845, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7442011847674119, "vi": 1.749889856913215, "tr": 1.2231586733367195, "fi": 1.1169496052517272, "ru": 0.49278074522170023, "el": 0.7480108575387757, "hi": 0.7526394375385062, "ta": 1.7936672776730442, "th": 0.8092557071047862, "zh": 114.5246277039472, "my": 168.8064207819089}, "t": 27.9} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 165.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.009481, "cps": 1.289062, "mean_p": 0.394724, "boundary_entropy": 0.663453, "bpb": 5.934945}, "de": {"chunks": 167.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.008133, "cps": 1.304688, "mean_p": 0.401337, "boundary_entropy": 0.667516, "bpb": 6.064312}, "vi": {"chunks": 270.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.014646, "cps": 2.109375, "mean_p": 0.411652, "boundary_entropy": 0.671783, "bpb": 6.61001}, "tr": {"chunks": 190.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.010495, "cps": 1.484375, "mean_p": 0.405835, "boundary_entropy": 0.668896, "bpb": 6.359059}, "fi": {"chunks": 180.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.009569, "cps": 1.40625, "mean_p": 0.399347, "boundary_entropy": 0.666169, "bpb": 6.069432}, "ru": {"chunks": 129.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.006807, "cps": 1.007812, "mean_p": 0.372291, "boundary_entropy": 0.656344, "bpb": 5.646633}, "el": {"chunks": 128.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.006237, "cps": 1.0, "mean_p": 0.409407, "boundary_entropy": 0.673342, "bpb": 6.38104}, "hi": {"chunks": 144.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.008559, "cps": 1.125, "mean_p": 0.431837, "boundary_entropy": 0.681208, "bpb": 5.742968}, "ta": {"chunks": 151.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.008077, "cps": 1.179688, "mean_p": 0.436318, "boundary_entropy": 0.682596, "bpb": 5.752051}, "th": {"chunks": 133.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.008269, "cps": 1.039062, "mean_p": 0.418422, "boundary_entropy": 0.67722, "bpb": 5.415547}, "zh": {"chunks": 308.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.05215, "cps": 2.40625, "mean_p": 0.446985, "boundary_entropy": 0.680147, "bpb": 7.324651}, "my": {"chunks": 3850.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.203188, "cps": 30.078125, "mean_p": 0.485369, "boundary_entropy": 0.690631, "bpb": 5.158866}}, "bpb": {"en": 5.93495, "de": 6.06431, "vi": 6.61001, "tr": 6.35906, "fi": 6.06943, "ru": 5.64663, "el": 6.38104, "hi": 5.74297, "ta": 5.75205, "th": 5.41555, "zh": 7.32465, "my": 5.15887}, "gini_cpc": 0.635869, "gini_cps": 0.628131, "bpic": 65.998452, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8578359447983954, "vi": 1.5447592277535318, "tr": 1.106988796432535, "fi": 1.0092547397315883, "ru": 0.7179937634924441, "el": 0.6578233733328017, "hi": 0.9027086316358232, "ta": 0.8519059544360427, "th": 0.8721089665696441, "zh": 5.500440230274297, "my": 21.430757863626766}, "t": 32.8} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 151.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.008677, "cps": 1.179688, "mean_p": 0.418889, "boundary_entropy": 0.67239, "bpb": 4.62767}, "de": {"chunks": 145.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.007062, "cps": 1.132812, "mean_p": 0.423243, "boundary_entropy": 0.674893, "bpb": 4.837106}, "vi": {"chunks": 254.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.013778, "cps": 1.984375, "mean_p": 0.43657, "boundary_entropy": 0.679045, "bpb": 5.258264}, "tr": {"chunks": 172.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.009501, "cps": 1.34375, "mean_p": 0.421685, "boundary_entropy": 0.673879, "bpb": 5.256801}, "fi": {"chunks": 172.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.009144, "cps": 1.34375, "mean_p": 0.425847, "boundary_entropy": 0.675215, "bpb": 4.865459}, "ru": {"chunks": 135.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.007124, "cps": 1.054688, "mean_p": 0.423319, "boundary_entropy": 0.677312, "bpb": 3.877527}, "el": {"chunks": 128.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.006237, "cps": 1.0, "mean_p": 0.420729, "boundary_entropy": 0.677266, "bpb": 4.830958}, "hi": {"chunks": 219.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.013016, "cps": 1.710938, "mean_p": 0.466692, "boundary_entropy": 0.688226, "bpb": 3.740243}, "ta": {"chunks": 167.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.008933, "cps": 1.304688, "mean_p": 0.450738, "boundary_entropy": 0.68599, "bpb": 3.589666}, "th": {"chunks": 2380.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.147964, "cps": 18.59375, "mean_p": 0.468385, "boundary_entropy": 0.687722, "bpb": 3.306342}, "zh": {"chunks": 215.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.036404, "cps": 1.679688, "mean_p": 0.426009, "boundary_entropy": 0.674391, "bpb": 6.405209}, "my": {"chunks": 140.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.007389, "cps": 1.09375, "mean_p": 0.404089, "boundary_entropy": 0.672115, "bpb": 4.152467}}, "bpb": {"en": 4.62767, "de": 4.83711, "vi": 5.25826, "tr": 5.2568, "fi": 4.86546, "ru": 3.87753, "el": 4.83096, "hi": 3.74024, "ta": 3.58967, "th": 3.30634, "zh": 6.40521, "my": 4.15247}, "gini_cpc": 0.575425, "gini_cps": 0.522324, "bpic": 89.710379, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8138844818694377, "vi": 1.5879533783456101, "tr": 1.0950276069276872, "fi": 1.0538134478875083, "ru": 0.8210540110782995, "el": 0.7188136198669688, "hi": 1.5001552798087046, "ta": 1.0295280283533685, "th": 17.0530892382562, "zh": 4.1955817745114965, "my": 0.851553292594441}, "t": 41.2} |
| {"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.53738, "bpb": 3.66066, "chunk_loss": 0.02315, "F": 0.00571, "G": 0.36386, "chunks": 5.84, "bytes_per_chunk": 175.23, "boundary_entropy": 0.65001, "router_noise": 0.0, "grad_norm": 1.9851, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.216, "alpha": 0.03, "t": 48.7} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 177.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.010171, "cps": 1.382812, "mean_p": 0.34504, "boundary_entropy": 0.63286, "bpb": 3.720853}, "de": {"chunks": 302.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.014708, "cps": 2.359375, "mean_p": 0.348575, "boundary_entropy": 0.635562, "bpb": 3.978181}, "vi": {"chunks": 192.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.010415, "cps": 1.5, "mean_p": 0.378811, "boundary_entropy": 0.653899, "bpb": 3.321689}, "tr": {"chunks": 423.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.023366, "cps": 3.304688, "mean_p": 0.351562, "boundary_entropy": 0.633833, "bpb": 3.935694}, "fi": {"chunks": 204.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.010845, "cps": 1.59375, "mean_p": 0.317402, "boundary_entropy": 0.611621, "bpb": 3.969358}, "ru": {"chunks": 150.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.007916, "cps": 1.171875, "mean_p": 0.323817, "boundary_entropy": 0.615306, "bpb": 2.665593}, "el": {"chunks": 134.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.006529, "cps": 1.046875, "mean_p": 0.354725, "boundary_entropy": 0.643435, "bpb": 2.806589}, "hi": {"chunks": 177.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.01052, "cps": 1.382812, "mean_p": 0.343351, "boundary_entropy": 0.636062, "bpb": 2.165491}, "ta": {"chunks": 184.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.009842, "cps": 1.4375, "mean_p": 0.354411, "boundary_entropy": 0.640548, "bpb": 2.026701}, "th": {"chunks": 223.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.013864, "cps": 1.742188, "mean_p": 0.316373, "boundary_entropy": 0.616371, "bpb": 2.104417}, "zh": {"chunks": 198.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.033525, "cps": 1.546875, "mean_p": 0.340826, "boundary_entropy": 0.626518, "bpb": 5.003183}, "my": {"chunks": 130.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.006861, "cps": 1.015625, "mean_p": 0.405019, "boundary_entropy": 0.672024, "bpb": 2.221902}}, "bpb": {"en": 3.72085, "de": 3.97818, "vi": 3.32169, "tr": 3.93569, "fi": 3.96936, "ru": 2.66559, "el": 2.80659, "hi": 2.16549, "ta": 2.0267, "th": 2.10442, "zh": 5.00318, "my": 2.2219}, "gini_cpc": 0.270813, "gini_cps": 0.182171, "bpic": 153.881716, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.4461235200549427, "vi": 1.024021182992925, "tr": 2.297421440589021, "fi": 1.0662747815243336, "ru": 0.7782746746567686, "el": 0.6419701087741988, "hi": 1.0343536404160476, "ta": 0.967705495441997, "th": 1.3631217630912051, "zh": 3.2962686609997305, "my": 0.674575913382925}, "t": 54.5} |
| {"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.06785, "bpb": 2.98328, "chunk_loss": 0.022737, "F": 0.00482, "G": 0.37455, "chunks": 4.94, "bytes_per_chunk": 207.392, "boundary_entropy": 0.65379, "router_noise": 0.0, "grad_norm": 1.2115, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 68.0} |
| {"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.97127, "bpb": 2.84395, "chunk_loss": 0.022497, "F": 0.00098, "G": 0.37632, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.653, "router_noise": 0.0, "grad_norm": 0.793, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 85.3} |
| {"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.84488, "bpb": 2.66159, "chunk_loss": 0.022703, "F": 0.02719, "G": 0.4086, "chunks": 27.84, "bytes_per_chunk": 36.777, "boundary_entropy": 0.66821, "router_noise": 0.0, "grad_norm": 0.8627, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 102.2} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.6727, "de": 2.98896, "vi": 2.36676, "tr": 2.9987, "fi": 3.19708, "ru": 1.95199, "el": 2.06414, "hi": 1.53959, "ta": 1.26424, "th": 1.58473, "zh": 3.67557, "my": 1.46615}, "bytes_per_chunk": {"en": 11.696, "de": 14.228, "vi": 26.386, "tr": 14.915, "fi": 19.834, "ru": 14.979, "el": 52.418, "hi": 41.354, "ta": 82.931, "th": 12.679, "zh": 50.961, "my": 14.757}, "bpb_mean": 2.31422, "t": 120.4} |
| {"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.77073, "bpb": 2.55463, "chunk_loss": 0.036174, "F": 0.28519, "G": 0.40666, "chunks": 292.03, "bytes_per_chunk": 3.506, "boundary_entropy": 0.6685, "router_noise": 0.0, "grad_norm": 0.8882, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.091, "alpha": 0.03, "t": 120.4} |
| {"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.7461, "bpb": 2.51909, "chunk_loss": 0.023676, "F": 0.00098, "G": 0.34336, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.63283, "router_noise": 0.0, "grad_norm": 0.8669, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 137.5} |
| {"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.59531, "bpb": 2.30154, "chunk_loss": 0.04745, "F": 0.56873, "G": 0.3657, "chunks": 582.38, "bytes_per_chunk": 1.758, "boundary_entropy": 0.64586, "router_noise": 0.0, "grad_norm": 0.7713, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 154.6} |
| {"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.63156, "bpb": 2.35384, "chunk_loss": 0.021869, "F": 0.00412, "G": 0.39823, "chunks": 4.22, "bytes_per_chunk": 242.726, "boundary_entropy": 0.6633, "router_noise": 0.0, "grad_norm": 0.8277, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 172.1} |
| {"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.54875, "bpb": 2.23437, "chunk_loss": 0.022334, "F": 0.00446, "G": 0.38555, "chunks": 4.56, "bytes_per_chunk": 224.438, "boundary_entropy": 0.65782, "router_noise": 0.0, "grad_norm": 0.8032, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 189.8} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 865.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.049704, "cps": 6.757812, "mean_p": 0.387742, "boundary_entropy": 0.650626, "bpb": 2.342435}, "de": {"chunks": 1047.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.050991, "cps": 8.179688, "mean_p": 0.394094, "boundary_entropy": 0.656408, "bpb": 2.60724}, "vi": {"chunks": 1126.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.061079, "cps": 8.796875, "mean_p": 0.402093, "boundary_entropy": 0.662149, "bpb": 2.122694}, "tr": {"chunks": 887.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.048997, "cps": 6.929688, "mean_p": 0.391411, "boundary_entropy": 0.655639, "bpb": 2.600893}, "fi": {"chunks": 538.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.0286, "cps": 4.203125, "mean_p": 0.381657, "boundary_entropy": 0.650929, "bpb": 2.835173}, "ru": {"chunks": 1387.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.073193, "cps": 10.835938, "mean_p": 0.397072, "boundary_entropy": 0.663244, "bpb": 1.697863}, "el": {"chunks": 910.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.04434, "cps": 7.109375, "mean_p": 0.373454, "boundary_entropy": 0.650931, "bpb": 1.75602}, "hi": {"chunks": 658.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.039108, "cps": 5.140625, "mean_p": 0.375392, "boundary_entropy": 0.64351, "bpb": 1.391615}, "ta": {"chunks": 1316.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.070393, "cps": 10.28125, "mean_p": 0.410274, "boundary_entropy": 0.670053, "bpb": 1.138114}, "th": {"chunks": 1614.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.100342, "cps": 12.609375, "mean_p": 0.370276, "boundary_entropy": 0.642571, "bpb": 1.332886}, "zh": {"chunks": 316.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.053505, "cps": 2.46875, "mean_p": 0.374219, "boundary_entropy": 0.647369, "bpb": 3.454632}, "my": {"chunks": 2281.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.120382, "cps": 17.820312, "mean_p": 0.413421, "boundary_entropy": 0.671582, "bpb": 1.241147}}, "bpb": {"en": 2.34244, "de": 2.60724, "vi": 2.12269, "tr": 2.60089, "fi": 2.83517, "ru": 1.69786, "el": 1.75602, "hi": 1.39162, "ta": 1.13811, "th": 1.33289, "zh": 3.45463, "my": 1.24115}, "gini_cpc": 0.214487, "gini_cps": 0.254352, "bpic": 29.647045, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.025893521467909, "vi": 1.2288624145764449, "tr": 0.9857824478362255, "fi": 0.5754113246369499, "ru": 1.472567663611268, "el": 0.8920897715491345, "hi": 0.7868262374494326, "ta": 1.4162451398862481, "th": 2.01878689979714, "zh": 1.076469310136258, "my": 2.421976483250173}, "t": 209.0} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.34998, "de": 2.61291, "vi": 2.12962, "tr": 2.60764, "fi": 2.84089, "ru": 1.67632, "el": 1.74849, "hi": 1.38933, "ta": 1.14487, "th": 1.34978, "zh": 3.45794, "my": 1.25053}, "bytes_per_chunk": {"en": 21.62, "de": 20.52, "vi": 23.7, "tr": 28.276, "fi": 56.074, "ru": 28.798, "el": 50.931, "hi": 73.204, "ta": 45.733, "th": 25.158, "zh": 74.664, "my": 25.799}, "bpb_mean": 2.04652, "t": 210.2} |
| {"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.6244, "bpb": 2.34351, "chunk_loss": 0.022285, "F": 0.00098, "G": 0.38224, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.65507, "router_noise": 0.0, "grad_norm": 0.9058, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 210.3} |
| {"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.48308, "bpb": 2.13963, "chunk_loss": 0.025308, "F": 0.04755, "G": 0.35082, "chunks": 48.69, "bytes_per_chunk": 21.032, "boundary_entropy": 0.63459, "router_noise": 0.0, "grad_norm": 0.7813, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 227.7} |
| {"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 245.5} |
| {"event": "train", "step": 2400, "phase": 2, "objective": "global", "lm_loss": 1.40639, "bpb": 2.02899, "chunk_loss": 0.028925, "F": 0.14429, "G": 0.39408, "chunks": 147.75, "bytes_per_chunk": 6.931, "boundary_entropy": 0.66056, "router_noise": 0.0, "grad_norm": 0.859, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 245.6} |
| {"event": "train", "step": 2600, "phase": 2, "objective": "global", "lm_loss": 1.62783, "bpb": 2.34847, "chunk_loss": 0.022151, "F": 0.00308, "G": 0.3888, "chunks": 3.16, "bytes_per_chunk": 324.436, "boundary_entropy": 0.65898, "router_noise": 0.0, "grad_norm": 0.8836, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.207, "alpha": 0.03, "t": 286.9} |
| {"event": "train", "step": 2800, "phase": 2, "objective": "global", "lm_loss": 1.49748, "bpb": 2.1604, "chunk_loss": 0.026187, "F": 0.08603, "G": 0.38583, "chunks": 88.09, "bytes_per_chunk": 11.624, "boundary_entropy": 0.65703, "router_noise": 0.0, "grad_norm": 1.0019, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 308.8} |
| {"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.22275, "de": 2.45175, "vi": 2.04372, "tr": 2.44134, "fi": 2.6592, "ru": 1.5545, "el": 1.62173, "hi": 1.29278, "ta": 1.06456, "th": 1.24943, "zh": 3.32823, "my": 1.19027}, "bytes_per_chunk": {"en": 13.113, "de": 17.167, "vi": 32.56, "tr": 13.862, "fi": 10.431, "ru": 15.814, "el": 40.896, "hi": 19.984, "ta": 15.489, "th": 21.722, "zh": 51.07, "my": 24.631}, "bpb_mean": 1.92669, "t": 327.2} |
| {"event": "train", "step": 3000, "phase": 2, "objective": "global", "lm_loss": 1.46088, "bpb": 2.1076, "chunk_loss": 0.021531, "F": 0.00098, "G": 0.40331, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.66279, "router_noise": 0.0, "grad_norm": 0.8257, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 327.3} |
| {"event": "train", "step": 3200, "phase": 2, "objective": "global", "lm_loss": 1.40393, "bpb": 2.02544, "chunk_loss": 0.023008, "F": 0.00098, "G": 0.36203, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.64372, "router_noise": 0.0, "grad_norm": 0.7719, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 344.6} |
| {"event": "train", "step": 3400, "phase": 2, "objective": "global", "lm_loss": 1.42269, "bpb": 2.05251, "chunk_loss": 0.022703, "F": 0.00104, "G": 0.37063, "chunks": 1.06, "bytes_per_chunk": 963.765, "boundary_entropy": 0.64803, "router_noise": 0.0, "grad_norm": 0.8124, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 362.4} |
| {"event": "train", "step": 3600, "phase": 2, "objective": "global", "lm_loss": 1.67857, "bpb": 2.42166, "chunk_loss": 0.025092, "F": 0.05627, "G": 0.37274, "chunks": 57.62, "bytes_per_chunk": 17.77, "boundary_entropy": 0.6487, "router_noise": 0.0, "grad_norm": 0.9422, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 379.4} |
| {"event": "train", "step": 3800, "phase": 2, "objective": "global", "lm_loss": 1.5016, "bpb": 2.16636, "chunk_loss": 0.027607, "F": 0.11011, "G": 0.36684, "chunks": 112.75, "bytes_per_chunk": 9.082, "boundary_entropy": 0.64616, "router_noise": 0.0, "grad_norm": 0.8228, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 396.7} |
| {"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 743.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.042694, "cps": 5.804688, "mean_p": 0.386578, "boundary_entropy": 0.648773, "bpb": 2.142141}, "de": {"chunks": 991.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.048264, "cps": 7.742188, "mean_p": 0.393949, "boundary_entropy": 0.655005, "bpb": 2.352732}, "vi": {"chunks": 963.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.052238, "cps": 7.523438, "mean_p": 0.397769, "boundary_entropy": 0.659274, "bpb": 1.99228}, "tr": {"chunks": 996.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.055019, "cps": 7.78125, "mean_p": 0.39571, "boundary_entropy": 0.656209, "bpb": 2.340493}, "fi": {"chunks": 653.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.034714, "cps": 5.101562, "mean_p": 0.381919, "boundary_entropy": 0.648935, "bpb": 2.542253}, "ru": {"chunks": 890.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.046966, "cps": 6.953125, "mean_p": 0.389771, "boundary_entropy": 0.660209, "bpb": 1.499817}, "el": {"chunks": 1227.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.059787, "cps": 9.585938, "mean_p": 0.402776, "boundary_entropy": 0.66601, "bpb": 1.551699}, "hi": {"chunks": 505.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.030015, "cps": 3.945312, "mean_p": 0.35568, "boundary_entropy": 0.621076, "bpb": 1.245318}, "ta": {"chunks": 440.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.023536, "cps": 3.4375, "mean_p": 0.371573, "boundary_entropy": 0.648746, "bpb": 1.025044}, "th": {"chunks": 1650.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.10258, "cps": 12.890625, "mean_p": 0.373504, "boundary_entropy": 0.643095, "bpb": 1.171179}, "zh": {"chunks": 361.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.061124, "cps": 2.820312, "mean_p": 0.393553, "boundary_entropy": 0.65732, "bpb": 3.222654}, "my": {"chunks": 1762.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.092991, "cps": 13.765625, "mean_p": 0.388011, "boundary_entropy": 0.658824, "bpb": 1.140095}}, "bpb": {"en": 2.14214, "de": 2.35273, "vi": 1.99228, "tr": 2.34049, "fi": 2.54225, "ru": 1.49982, "el": 1.5517, "hi": 1.24532, "ta": 1.02504, "th": 1.17118, "zh": 3.22265, "my": 1.14009}, "gini_cpc": 0.222429, "gini_cps": 0.252564, "bpic": 34.32439, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1304635239376668, "vi": 1.2235407880658864, "tr": 1.2886770475718836, "fi": 0.8130862265020188, "ru": 1.100059304609069, "el": 1.40035782982937, "hi": 0.7030263639436125, "ta": 0.5512676574479397, "th": 2.4026924594317456, "zh": 1.4316902445171757, "my": 2.178100097426093}, "t": 415.4} |
| {"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 2.15046, "de": 2.35924, "vi": 1.99391, "tr": 2.34763, "fi": 2.54761, "ru": 1.47731, "el": 1.54451, "hi": 1.24194, "ta": 1.0329, "th": 1.18717, "zh": 3.22415, "my": 1.14907}, "bytes_per_chunk": {"en": 26.309, "de": 21.906, "vi": 27.344, "tr": 24.971, "fi": 41.005, "ru": 46.438, "el": 36.813, "hi": 105.576, "ta": 149.968, "th": 25.608, "zh": 51.899, "my": 29.418}, "bpb_mean": 1.85466, "t": 416.6} |
| {"event": "train", "step": 4000, "phase": 2, "objective": "global", "lm_loss": 1.37428, "bpb": 1.98267, "chunk_loss": 0.037107, "F": 0.31747, "G": 0.38364, "chunks": 325.09, "bytes_per_chunk": 3.15, "boundary_entropy": 0.6547, "router_noise": 0.0, "grad_norm": 0.8236, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.1, "alpha": 0.03, "t": 416.7} |
| {"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 434.4} |
| {"event": "train", "step": 4200, "phase": 3, "objective": "global", "lm_loss": 1.22543, "bpb": 1.76792, "chunk_loss": 0.053313, "F": 0.70139, "G": 0.36807, "chunks": 718.22, "bytes_per_chunk": 1.426, "boundary_entropy": 0.64404, "router_noise": 0.0, "grad_norm": 0.7423, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 434.5} |
| {"event": "train", "step": 4400, "phase": 3, "objective": "global", "lm_loss": 1.51993, "bpb": 2.19279, "chunk_loss": 0.032473, "F": 0.2225, "G": 0.37015, "chunks": 227.84, "bytes_per_chunk": 4.494, "boundary_entropy": 0.64718, "router_noise": 0.0, "grad_norm": 0.9441, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 451.9} |
| {"event": "train", "step": 4600, "phase": 3, "objective": "global", "lm_loss": 1.33993, "bpb": 1.93311, "chunk_loss": 0.023319, "F": 0.00107, "G": 0.35345, "chunks": 1.09, "bytes_per_chunk": 936.229, "boundary_entropy": 0.63719, "router_noise": 0.0, "grad_norm": 0.8465, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 469.5} |
| {"event": "train", "step": 4800, "phase": 3, "objective": "global", "lm_loss": 1.28915, "bpb": 1.85985, "chunk_loss": 0.0234, "F": 0.00729, "G": 0.35837, "chunks": 7.47, "bytes_per_chunk": 137.105, "boundary_entropy": 0.6404, "router_noise": 0.0, "grad_norm": 0.8537, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 486.9} |
| {"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 2.10988, "de": 2.301, "vi": 1.9648, "tr": 2.28913, "fi": 2.48488, "ru": 1.4289, "el": 1.49963, "hi": 1.21131, "ta": 1.00044, "th": 1.15444, "zh": 3.18178, "my": 1.12292}, "bytes_per_chunk": {"en": 35.92, "de": 33.225, "vi": 44.311, "tr": 39.167, "fi": 58.79, "ru": 55.539, "el": 57.013, "hi": 110.237, "ta": 188.865, "th": 47.715, "zh": 78.205, "my": 36.859}, "bpb_mean": 1.81243, "t": 505.3} |
| {"event": "train", "step": 5000, "phase": 3, "objective": "global", "lm_loss": 1.33088, "bpb": 1.92006, "chunk_loss": 0.022679, "F": 0.00098, "G": 0.37124, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.6485, "router_noise": 0.0, "grad_norm": 0.8763, "lr": 5e-05, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 505.5} |
| {"event": "train", "step": 5200, "phase": 3, "objective": "global", "lm_loss": 1.3245, "bpb": 1.91085, "chunk_loss": 0.061605, "F": 0.85696, "G": 0.37798, "chunks": 877.53, "bytes_per_chunk": 1.167, "boundary_entropy": 0.65228, "router_noise": 0.0, "grad_norm": 0.9709, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 523.0} |
| {"event": "train", "step": 5400, "phase": 3, "objective": "global", "lm_loss": 1.17281, "bpb": 1.69201, "chunk_loss": 0.023089, "F": 0.01028, "G": 0.37136, "chunks": 10.53, "bytes_per_chunk": 97.234, "boundary_entropy": 0.64689, "router_noise": 0.0, "grad_norm": 0.8181, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 540.1} |
| {"event": "train", "step": 5600, "phase": 3, "objective": "global", "lm_loss": 1.51532, "bpb": 2.18615, "chunk_loss": 0.022786, "F": 0.00098, "G": 0.36824, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.64788, "router_noise": 0.0, "grad_norm": 0.9137, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.123, "alpha": 0.03, "t": 564.7} |
| {"event": "train", "step": 5800, "phase": 3, "objective": "global", "lm_loss": 1.30505, "bpb": 1.88278, "chunk_loss": 0.023186, "F": 0.00098, "G": 0.35707, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.63876, "router_noise": 0.0, "grad_norm": 0.9114, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 602.1} |
| {"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 625.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.035913, "cps": 4.882812, "mean_p": 0.376954, "boundary_entropy": 0.64401, "bpb": 2.081873}, "de": {"chunks": 747.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.03638, "cps": 5.835938, "mean_p": 0.379785, "boundary_entropy": 0.647951, "bpb": 2.270786}, "vi": {"chunks": 738.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.040033, "cps": 5.765625, "mean_p": 0.380939, "boundary_entropy": 0.650053, "bpb": 1.947748}, "tr": {"chunks": 856.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.047285, "cps": 6.6875, "mean_p": 0.384487, "boundary_entropy": 0.650487, "bpb": 2.253192}, "fi": {"chunks": 514.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.027324, "cps": 4.015625, "mean_p": 0.379234, "boundary_entropy": 0.647883, "bpb": 2.443051}, "ru": {"chunks": 707.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.037309, "cps": 5.523438, "mean_p": 0.382472, "boundary_entropy": 0.656953, "bpb": 1.429963}, "el": {"chunks": 829.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.040394, "cps": 6.476562, "mean_p": 0.383706, "boundary_entropy": 0.657316, "bpb": 1.482602}, "hi": {"chunks": 449.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.026686, "cps": 3.507812, "mean_p": 0.347368, "boundary_entropy": 0.61441, "bpb": 1.195372}, "ta": {"chunks": 596.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.03188, "cps": 4.65625, "mean_p": 0.390185, "boundary_entropy": 0.66019, "bpb": 0.981009}, "th": {"chunks": 849.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.052782, "cps": 6.632812, "mean_p": 0.35477, "boundary_entropy": 0.628983, "bpb": 1.118028}, "zh": {"chunks": 277.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.046901, "cps": 2.164062, "mean_p": 0.374839, "boundary_entropy": 0.647492, "bpb": 3.147655}, "my": {"chunks": 1620.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.085497, "cps": 12.65625, "mean_p": 0.370474, "boundary_entropy": 0.648133, "bpb": 1.103244}}, "bpb": {"en": 2.08187, "de": 2.27079, "vi": 1.94775, "tr": 2.25319, "fi": 2.44305, "ru": 1.42996, "el": 1.4826, "hi": 1.19537, "ta": 0.98101, "th": 1.11803, "zh": 3.14766, "my": 1.10324}, "gini_cpc": 0.172699, "gini_cps": 0.211413, "bpic": 43.576814, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.013006652705401, "vi": 1.1146982587469487, "tr": 1.3166408219632104, "fi": 0.7608435064589868, "ru": 1.0388534881266491, "el": 1.1247546265165909, "hi": 0.7430796552748885, "ta": 0.8876972880449316, "th": 1.469706882188374, "zh": 1.3059616661022686, "my": 2.380651044965168}, "t": 620.9} |
| {"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 2.09086, "de": 2.27843, "vi": 1.94879, "tr": 2.25811, "fi": 2.45103, "ru": 1.40813, "el": 1.47652, "hi": 1.1921, "ta": 0.98822, "th": 1.13533, "zh": 3.14726, "my": 1.11249}, "bytes_per_chunk": {"en": 33.851, "de": 32.074, "vi": 41.205, "tr": 29.722, "fi": 49.667, "ru": 57.665, "el": 55.422, "hi": 112.46, "ta": 112.847, "th": 47.524, "zh": 92.272, "my": 33.149}, "bpb_mean": 1.79061, "t": 622.1} |
| {"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 625.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.035913, "cps": 4.882812, "mean_p": 0.376954, "boundary_entropy": 0.64401, "bpb": 2.081873}, "de": {"chunks": 747.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.03638, "cps": 5.835938, "mean_p": 0.379785, "boundary_entropy": 0.647951, "bpb": 2.270786}, "vi": {"chunks": 738.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.040033, "cps": 5.765625, "mean_p": 0.380939, "boundary_entropy": 0.650053, "bpb": 1.947748}, "tr": {"chunks": 856.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.047285, "cps": 6.6875, "mean_p": 0.384487, "boundary_entropy": 0.650487, "bpb": 2.253192}, "fi": {"chunks": 514.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.027324, "cps": 4.015625, "mean_p": 0.379234, "boundary_entropy": 0.647883, "bpb": 2.443051}, "ru": {"chunks": 707.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.037309, "cps": 5.523438, "mean_p": 0.382472, "boundary_entropy": 0.656953, "bpb": 1.429963}, "el": {"chunks": 829.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.040394, "cps": 6.476562, "mean_p": 0.383706, "boundary_entropy": 0.657316, "bpb": 1.482602}, "hi": {"chunks": 449.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.026686, "cps": 3.507812, "mean_p": 0.347368, "boundary_entropy": 0.61441, "bpb": 1.195372}, "ta": {"chunks": 596.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.03188, "cps": 4.65625, "mean_p": 0.390185, "boundary_entropy": 0.66019, "bpb": 0.981009}, "th": {"chunks": 849.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.052782, "cps": 6.632812, "mean_p": 0.35477, "boundary_entropy": 0.628983, "bpb": 1.118028}, "zh": {"chunks": 277.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.046901, "cps": 2.164062, "mean_p": 0.374839, "boundary_entropy": 0.647492, "bpb": 3.147655}, "my": {"chunks": 1620.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.085497, "cps": 12.65625, "mean_p": 0.370474, "boundary_entropy": 0.648133, "bpb": 1.103244}}, "bpb": {"en": 2.08187, "de": 2.27079, "vi": 1.94775, "tr": 2.25319, "fi": 2.44305, "ru": 1.42996, "el": 1.4826, "hi": 1.19537, "ta": 0.98101, "th": 1.11803, "zh": 3.14766, "my": 1.10324}, "gini_cpc": 0.172699, "gini_cps": 0.211413, "bpic": 43.576814, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.013006652705401, "vi": 1.1146982587469487, "tr": 1.3166408219632104, "fi": 0.7608435064589868, "ru": 1.0388534881266491, "el": 1.1247546265165909, "hi": 0.7430796552748885, "ta": 0.8876972880449316, "th": 1.469706882188374, "zh": 1.3059616661022686, "my": 2.380651044965168}, "t": 623.1} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/jitter_s0/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/jitter_s0/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/jitter_s0/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/jitter_s0/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/jitter_s0/th0.70"], "t": 626.6} |
| {"event": "final", "run": "jitter_s0", "condition": "A", "step": 6000, "params": 22516992, "per_lang": {"en": {"cpc": 0.03591334827328622, "cps": 4.8828125, "bpb": 2.102690803330191, "bytes_per_chunk": 33.77491464278812}, "de": {"cpc": 0.036380460721764964, "cps": 5.8359375, "bpb": 2.271415199876741, "bytes_per_chunk": 32.133366021083596}, "vi": {"cpc": 0.04003254678600488, "cps": 5.765625, "bpb": 1.951459073921452, "bytes_per_chunk": 40.595276809910956}, "tr": {"cpc": 0.04728498038999061, "cps": 6.6875, "bpb": 2.250277080982085, "bytes_per_chunk": 30.134958041154157}, "fi": {"cpc": 0.027324437828929883, "cps": 4.015625, "bpb": 2.4655344794072835, "bytes_per_chunk": 49.16890180999719}, "ru": {"cpc": 0.037308707124010555, "cps": 5.5234375, "bpb": 1.4088588409418283, "bytes_per_chunk": 58.11861212725862}, "el": {"cpc": 0.0403937046240803, "cps": 6.4765625, "bpb": 1.471620706197338, "bytes_per_chunk": 56.95068433630241}, "hi": {"cpc": 0.026686478454680536, "cps": 3.5078125, "bpb": 1.2017601651121157, "bytes_per_chunk": 111.29017193801741}, "ta": {"cpc": 0.031880181866809305, "cps": 4.65625, "bpb": 0.9874921775028731, "bytes_per_chunk": 111.97949594190517}, "th": {"cpc": 0.052782095119676714, "cps": 6.6328125, "bpb": 1.1332099012743477, "bytes_per_chunk": 47.20338525254344}, "zh": {"cpc": 0.0469014561462919, "cps": 2.1640625, "bpb": 3.1557357805467756, "bytes_per_chunk": 94.7394289844597}, "my": {"cpc": 0.08549715009499684, "cps": 12.65625, "bpb": 1.1230400189341303, "bytes_per_chunk": 33.42607586866433}}, "gini_cpc": 0.17269931081856482, "gini_cps": 0.21141326974754948, "bpic": 43.57681389803565, "seed": 0, "mean_boundary_entropy": 0.6520750522613525, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}], "t": 628.8} |
|
|
