Blame for log.jsonl

{"event": "start", "run": "exp32_local_s2", "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": 10285.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.59099, "cps": 80.351562, "mean_p": 0.506761, "boundary_entropy": 0.687331, "bpb": 8.019721}, "de": {"chunks": 11284.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.549554, "cps": 88.15625, "mean_p": 0.504462, "boundary_entropy": 0.688263, "bpb": 8.113016}, "vi": {"chunks": 5953.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.322918, "cps": 46.507812, "mean_p": 0.490341, "boundary_entropy": 0.688514, "bpb": 7.990959}, "tr": {"chunks": 6577.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.36331, "cps": 51.382812, "mean_p": 0.49627, "boundary_entropy": 0.687893, "bpb": 8.105784}, "fi": {"chunks": 2159.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.114773, "cps": 16.867188, "mean_p": 0.479659, "boundary_entropy": 0.68676, "bpb": 8.104721}, "ru": {"chunks": 23819.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.256939, "cps": 186.085938, "mean_p": 0.507388, "boundary_entropy": 0.690184, "bpb": 8.278473}, "el": {"chunks": 1446.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.070458, "cps": 11.296875, "mean_p": 0.479438, "boundary_entropy": 0.689435, "bpb": 7.88022}, "hi": {"chunks": 7945.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.472214, "cps": 62.070312, "mean_p": 0.494138, "boundary_entropy": 0.690834, "bpb": 7.805442}, "ta": {"chunks": 358.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.01915, "cps": 2.796875, "mean_p": 0.475419, "boundary_entropy": 0.689909, "bpb": 7.609112}, "th": {"chunks": 27847.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.73124, "cps": 217.554688, "mean_p": 0.503256, "boundary_entropy": 0.691078, "bpb": 7.878903}, "zh": {"chunks": 14701.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.489164, "cps": 114.851562, "mean_p": 0.531837, "boundary_entropy": 0.685229, "bpb": 8.024398}, "my": {"chunks": 785.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.041429, "cps": 6.132812, "mean_p": 0.475077, "boundary_entropy": 0.689998, "bpb": 8.249889}}, "bpb": {"en": 8.01972, "de": 8.11302, "vi": 7.99096, "tr": 8.10578, "fi": 8.10472, "ru": 8.27847, "el": 7.88022, "hi": 7.80544, "ta": 7.60911, "th": 7.8789, "zh": 8.0244, "my": 8.24989}, "gini_cpc": 0.559428, "gini_cps": 0.487853, "bpic": 3.39152, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9298876814280087, "vi": 0.5464023578619593, "tr": 0.6147479859952251, "fi": 0.19420507870592685, "ru": 2.126836643778949, "el": 0.11921949337891437, "hi": 0.7990218447355856, "ta": 0.032402415095948926, "th": 2.9293898587187748, "zh": 4.211853522393696, "my": 0.07010130532514902}, "t": 7.9}
{"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.58957, "bpb": 8.06405, "chunk_loss": 0.060675, "F": 0.57669, "G": 0.50291, "chunks": 590.53, "bytes_per_chunk": 1.734, "boundary_entropy": 0.69167, "router_noise": 0.0, "grad_norm": 9.534, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.006, "alpha": 0.03, "t": 8.7}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10213.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.586853, "cps": 79.789062, "mean_p": 0.506576, "boundary_entropy": 0.687333, "bpb": 7.983603}, "de": {"chunks": 11228.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.546827, "cps": 87.71875, "mean_p": 0.504323, "boundary_entropy": 0.688263, "bpb": 8.080536}, "vi": {"chunks": 5912.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.320694, "cps": 46.1875, "mean_p": 0.490221, "boundary_entropy": 0.688508, "bpb": 7.969592}, "tr": {"chunks": 6528.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.360603, "cps": 51.0, "mean_p": 0.496092, "boundary_entropy": 0.687887, "bpb": 8.0809}, "fi": {"chunks": 2147.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.114135, "cps": 16.773438, "mean_p": 0.479522, "boundary_entropy": 0.686745, "bpb": 8.081389}, "ru": {"chunks": 23823.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.25715, "cps": 186.117188, "mean_p": 0.507379, "boundary_entropy": 0.690185, "bpb": 8.225934}, "el": {"chunks": 1437.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.070019, "cps": 11.226562, "mean_p": 0.479409, "boundary_entropy": 0.689433, "bpb": 7.87039}, "hi": {"chunks": 7915.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.470431, "cps": 61.835938, "mean_p": 0.49411, "boundary_entropy": 0.690833, "bpb": 7.782157}, "ta": {"chunks": 357.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.019096, "cps": 2.789062, "mean_p": 0.475364, "boundary_entropy": 0.689904, "bpb": 7.594914}, "th": {"chunks": 27579.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.714579, "cps": 215.460938, "mean_p": 0.503152, "boundary_entropy": 0.691079, "bpb": 7.841983}, "zh": {"chunks": 14691.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.48747, "cps": 114.773438, "mean_p": 0.53175, "boundary_entropy": 0.685239, "bpb": 8.018842}, "my": {"chunks": 780.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.041165, "cps": 6.09375, "mean_p": 0.475043, "boundary_entropy": 0.689995, "bpb": 8.249016}}, "bpb": {"en": 7.9836, "de": 8.08054, "vi": 7.96959, "tr": 8.0809, "fi": 8.08139, "ru": 8.22593, "el": 7.87039, "hi": 7.78216, "ta": 7.59491, "th": 7.84198, "zh": 8.01884, "my": 8.24902}, "gini_cpc": 0.559878, "gini_cps": 0.487895, "bpic": 3.408054, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9317958783612076, "vi": 0.5464646479936666, "tr": 0.6144695730484945, "fi": 0.19448716658703713, "ru": 2.14219018287779, "el": 0.11931271031258718, "hi": 0.8016164754619618, "ta": 0.03253969927047601, "th": 2.9216503337475643, "zh": 4.238661199816745, "my": 0.07014585528372332}, "t": 14.1}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10073.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.578808, "cps": 78.695312, "mean_p": 0.506171, "boundary_entropy": 0.687337, "bpb": 7.913924}, "de": {"chunks": 11064.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.53884, "cps": 86.4375, "mean_p": 0.504011, "boundary_entropy": 0.688263, "bpb": 8.01818}, "vi": {"chunks": 5782.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.313643, "cps": 45.171875, "mean_p": 0.489936, "boundary_entropy": 0.688493, "bpb": 7.930122}, "tr": {"chunks": 6382.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.352538, "cps": 49.859375, "mean_p": 0.495715, "boundary_entropy": 0.687874, "bpb": 8.033022}, "fi": {"chunks": 2112.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.112275, "cps": 16.5, "mean_p": 0.479214, "boundary_entropy": 0.686711, "bpb": 8.037954}, "ru": {"chunks": 23819.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.256939, "cps": 186.085938, "mean_p": 0.507359, "boundary_entropy": 0.690187, "bpb": 8.118044}, "el": {"chunks": 1425.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.069434, "cps": 11.132812, "mean_p": 0.479368, "boundary_entropy": 0.689429, "bpb": 7.845137}, "hi": {"chunks": 7901.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.469599, "cps": 61.726562, "mean_p": 0.494081, "boundary_entropy": 0.690832, "bpb": 7.739976}, "ta": {"chunks": 355.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.018989, "cps": 2.773438, "mean_p": 0.475344, "boundary_entropy": 0.689902, "bpb": 7.564502}, "th": {"chunks": 27313.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.698042, "cps": 213.382812, "mean_p": 0.503044, "boundary_entropy": 0.69108, "bpb": 7.779997}, "zh": {"chunks": 14679.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.485439, "cps": 114.679688, "mean_p": 0.531536, "boundary_entropy": 0.685265, "bpb": 8.005385}, "my": {"chunks": 772.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.040743, "cps": 6.03125, "mean_p": 0.474901, "boundary_entropy": 0.68998, "bpb": 8.249489}}, "bpb": {"en": 7.91392, "de": 8.01818, "vi": 7.93012, "tr": 8.03302, "fi": 8.03795, "ru": 8.11804, "el": 7.84514, "hi": 7.73998, "ta": 7.5645, "th": 7.78, "zh": 8.00539, "my": 8.24949}, "gini_cpc": 0.561673, "gini_cps": 0.489426, "bpic": 3.436527, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9309471917197029, "vi": 0.5418763934587519, "tr": 0.6090760669780875, "fi": 0.19397569503825624, "ru": 2.17159881676427, "el": 0.11996078676821316, "hi": 0.8113201740223559, "ta": 0.03280712473190162, "th": 2.9336859822888806, "zh": 4.294062033239303, "my": 0.07039133640447531}, "t": 20.6}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 9512.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.546572, "cps": 74.3125, "mean_p": 0.504695, "boundary_entropy": 0.687348, "bpb": 7.669139}, "de": {"chunks": 10496.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.511177, "cps": 82.0, "mean_p": 0.502867, "boundary_entropy": 0.68826, "bpb": 7.795164}, "vi": {"chunks": 5419.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.293952, "cps": 42.335938, "mean_p": 0.489006, "boundary_entropy": 0.688444, "bpb": 7.791564}, "tr": {"chunks": 5879.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.324753, "cps": 45.929688, "mean_p": 0.494391, "boundary_entropy": 0.687823, "bpb": 7.862314}, "fi": {"chunks": 2027.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.107756, "cps": 15.835938, "mean_p": 0.478195, "boundary_entropy": 0.686597, "bpb": 7.88521}, "ru": {"chunks": 23699.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.250607, "cps": 185.148438, "mean_p": 0.507245, "boundary_entropy": 0.690196, "bpb": 7.797928}, "el": {"chunks": 1376.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.067047, "cps": 10.75, "mean_p": 0.479045, "boundary_entropy": 0.689399, "bpb": 7.764294}, "hi": {"chunks": 7744.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.460267, "cps": 60.5, "mean_p": 0.493896, "boundary_entropy": 0.690825, "bpb": 7.587137}, "ta": {"chunks": 348.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.018615, "cps": 2.71875, "mean_p": 0.475264, "boundary_entropy": 0.689896, "bpb": 7.467584}, "th": {"chunks": 26049.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.619459, "cps": 203.507812, "mean_p": 0.502592, "boundary_entropy": 0.691082, "bpb": 7.565227}, "zh": {"chunks": 14604.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.47274, "cps": 114.09375, "mean_p": 0.530854, "boundary_entropy": 0.685346, "bpb": 7.965746}, "my": {"chunks": 645.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.034041, "cps": 5.039062, "mean_p": 0.474582, "boundary_entropy": 0.689946, "bpb": 8.220808}}, "bpb": {"en": 7.66914, "de": 7.79516, "vi": 7.79156, "tr": 7.86231, "fi": 7.88521, "ru": 7.79793, "el": 7.76429, "hi": 7.58714, "ta": 7.46758, "th": 7.56523, "zh": 7.96575, "my": 8.22081}, "gini_cpc": 0.567995, "gini_cps": 0.494197, "bpic": 3.560187, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9352413356464027, "vi": 0.5378092748759479, "tr": 0.5941624303635485, "fi": 0.1971488026685172, "ru": 2.2880899061538154, "el": 0.12266759971079237, "hi": 0.8420978334085232, "ta": 0.03405697362670337, "th": 2.9629359887142868, "zh": 4.524084002327464, "my": 0.0622800019015346}, "t": 26.8}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 7440.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.427512, "cps": 58.125, "mean_p": 0.499528, "boundary_entropy": 0.687324, "bpb": 7.037112}, "de": {"chunks": 8524.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.415137, "cps": 66.59375, "mean_p": 0.498931, "boundary_entropy": 0.688216, "bpb": 7.186311}, "vi": {"chunks": 4270.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.231625, "cps": 33.359375, "mean_p": 0.48586, "boundary_entropy": 0.688258, "bpb": 7.425362}, "tr": {"chunks": 4458.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.246258, "cps": 34.828125, "mean_p": 0.49003, "boundary_entropy": 0.687608, "bpb": 7.367816}, "fi": {"chunks": 1738.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.092393, "cps": 13.578125, "mean_p": 0.475014, "boundary_entropy": 0.686215, "bpb": 7.452875}, "ru": {"chunks": 22391.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.181583, "cps": 174.929688, "mean_p": 0.506089, "boundary_entropy": 0.690241, "bpb": 6.807875}, "el": {"chunks": 1216.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.059251, "cps": 9.5, "mean_p": 0.477895, "boundary_entropy": 0.68929, "bpb": 7.489898}, "hi": {"chunks": 7525.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.447251, "cps": 58.789062, "mean_p": 0.493508, "boundary_entropy": 0.690807, "bpb": 7.022621}, "ta": {"chunks": 335.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.017919, "cps": 2.617188, "mean_p": 0.475403, "boundary_entropy": 0.689918, "bpb": 7.125947}, "th": {"chunks": 21329.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.326018, "cps": 166.632812, "mean_p": 0.500898, "boundary_entropy": 0.691084, "bpb": 6.855539}, "zh": {"chunks": 14321.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.424822, "cps": 111.882812, "mean_p": 0.528201, "boundary_entropy": 0.68564, "bpb": 7.840047}, "my": {"chunks": 591.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.031191, "cps": 4.617188, "mean_p": 0.473701, "boundary_entropy": 0.689851, "bpb": 8.106164}}, "bpb": {"en": 7.03711, "de": 7.18631, "vi": 7.42536, "tr": 7.36782, "fi": 7.45287, "ru": 6.80787, "el": 7.4899, "hi": 7.02262, "ta": 7.12595, "th": 6.85554, "zh": 7.84005, "my": 8.10616}, "gini_cpc": 0.585867, "gini_cps": 0.502516, "bpic": 4.076792, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9710513995566539, "vi": 0.5417961538798043, "tr": 0.5760241574645443, "fi": 0.21611707117146622, "ru": 2.7638563069197373, "el": 0.1385938357122536, "hi": 1.046170852705747, "ta": 0.04191510150204039, "th": 3.1017058820580186, "zh": 5.671932930426138, "my": 0.0729583980265174}, "t": 33.7}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3848.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.221111, "cps": 30.0625, "mean_p": 0.490513, "boundary_entropy": 0.687105, "bpb": 6.321003}, "de": {"chunks": 5776.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.281303, "cps": 45.125, "mean_p": 0.493885, "boundary_entropy": 0.688131, "bpb": 6.41345}, "vi": {"chunks": 2267.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.122973, "cps": 17.710938, "mean_p": 0.479382, "boundary_entropy": 0.687803, "bpb": 6.968585}, "tr": {"chunks": 2906.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.160526, "cps": 22.703125, "mean_p": 0.484539, "boundary_entropy": 0.687284, "bpb": 6.702424}, "fi": {"chunks": 1369.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.072777, "cps": 10.695312, "mean_p": 0.472708, "boundary_entropy": 0.685982, "bpb": 6.687504}, "ru": {"chunks": 11416.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.602427, "cps": 89.1875, "mean_p": 0.497533, "boundary_entropy": 0.690285, "bpb": 5.66852}, "el": {"chunks": 1213.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.059104, "cps": 9.476562, "mean_p": 0.476646, "boundary_entropy": 0.689159, "bpb": 6.692856}, "hi": {"chunks": 7967.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.473522, "cps": 62.242188, "mean_p": 0.493397, "boundary_entropy": 0.69079, "bpb": 5.938501}, "ta": {"chunks": 360.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.019256, "cps": 2.8125, "mean_p": 0.478758, "boundary_entropy": 0.690268, "bpb": 6.468143}, "th": {"chunks": 3815.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.237177, "cps": 29.804688, "mean_p": 0.491264, "boundary_entropy": 0.690891, "bpb": 5.565755}, "zh": {"chunks": 12318.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.085676, "cps": 96.234375, "mean_p": 0.517574, "boundary_entropy": 0.686517, "bpb": 7.462072}, "my": {"chunks": 629.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.033196, "cps": 4.914062, "mean_p": 0.472056, "boundary_entropy": 0.689659, "bpb": 7.398572}}, "bpb": {"en": 6.321, "de": 6.41345, "vi": 6.96859, "tr": 6.70242, "fi": 6.6875, "ru": 5.66852, "el": 6.69286, "hi": 5.9385, "ta": 6.46814, "th": 5.56575, "zh": 7.46207, "my": 7.39857}, "gini_cpc": 0.614177, "gini_cps": 0.469505, "bpic": 7.122355, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2722247327029872, "vi": 0.5561570348592318, "tr": 0.7259958119069868, "fi": 0.3291399672041318, "ru": 2.7245438537786826, "el": 0.2673061879866441, "hi": 2.1415528724518325, "ta": 0.0870895583385554, "th": 1.0726611065047496, "zh": 9.432695475829577, "my": 0.1501330564622205}, "t": 41.4}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3550.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203988, "cps": 27.734375, "mean_p": 0.493357, "boundary_entropy": 0.687493, "bpb": 5.817561}, "de": {"chunks": 6548.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.318901, "cps": 51.15625, "mean_p": 0.497251, "boundary_entropy": 0.688436, "bpb": 5.942417}, "vi": {"chunks": 1815.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.098454, "cps": 14.179688, "mean_p": 0.482189, "boundary_entropy": 0.688261, "bpb": 6.531708}, "tr": {"chunks": 3723.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205657, "cps": 29.085938, "mean_p": 0.490749, "boundary_entropy": 0.687893, "bpb": 6.251646}, "fi": {"chunks": 2503.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.13306, "cps": 19.554688, "mean_p": 0.488089, "boundary_entropy": 0.687747, "bpb": 5.963457}, "ru": {"chunks": 6789.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.358259, "cps": 53.039062, "mean_p": 0.494056, "boundary_entropy": 0.690264, "bpb": 5.140845}, "el": {"chunks": 11168.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.54417, "cps": 87.25, "mean_p": 0.495591, "boundary_entropy": 0.69041, "bpb": 5.476175}, "hi": {"chunks": 11403.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.677741, "cps": 89.085938, "mean_p": 0.495516, "boundary_entropy": 0.690837, "bpb": 5.205663}, "ta": {"chunks": 28834.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.542338, "cps": 225.265625, "mean_p": 0.50228, "boundary_entropy": 0.691289, "bpb": 4.931532}, "th": {"chunks": 7392.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.459559, "cps": 57.75, "mean_p": 0.495561, "boundary_entropy": 0.691059, "bpb": 4.804751}, "zh": {"chunks": 3127.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.529462, "cps": 24.429688, "mean_p": 0.488848, "boundary_entropy": 0.686579, "bpb": 6.984142}, "my": {"chunks": 11121.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.586922, "cps": 86.882812, "mean_p": 0.495612, "boundary_entropy": 0.691304, "bpb": 4.869809}}, "bpb": {"en": 5.81756, "de": 5.94242, "vi": 6.53171, "tr": 6.25165, "fi": 5.96346, "ru": 5.14085, "el": 5.47618, "hi": 5.20566, "ta": 4.93153, "th": 4.80475, "zh": 6.98414, "my": 4.86981}, "gini_cpc": 0.381298, "gini_cps": 0.410571, "bpic": 3.917212, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.563334927012031, "vi": 0.48264660378872093, "tr": 1.0081804043061886, "fi": 0.6522960269725311, "ru": 1.7562743617377086, "el": 2.6676590812231966, "hi": 3.322460439905406, "ta": 7.56092955486328, "th": 2.252872740327573, "zh": 2.5955548189237017, "my": 2.8772409440812066}, "t": 51.5}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4098.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.235477, "cps": 32.015625, "mean_p": 0.495763, "boundary_entropy": 0.687662, "bpb": 5.407331}, "de": {"chunks": 4332.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.210977, "cps": 33.84375, "mean_p": 0.494086, "boundary_entropy": 0.688412, "bpb": 5.555726}, "vi": {"chunks": 2535.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.13751, "cps": 19.804688, "mean_p": 0.487946, "boundary_entropy": 0.688774, "bpb": 5.975092}, "tr": {"chunks": 2153.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.118931, "cps": 16.820312, "mean_p": 0.48877, "boundary_entropy": 0.687904, "bpb": 5.863021}, "fi": {"chunks": 2762.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.146829, "cps": 21.578125, "mean_p": 0.490638, "boundary_entropy": 0.687982, "bpb": 5.551113}, "ru": {"chunks": 2515.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.132718, "cps": 19.648438, "mean_p": 0.491833, "boundary_entropy": 0.69028, "bpb": 4.844863}, "el": {"chunks": 3383.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.164839, "cps": 26.429688, "mean_p": 0.488371, "boundary_entropy": 0.690214, "bpb": 5.070181}, "hi": {"chunks": 6476.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.384903, "cps": 50.59375, "mean_p": 0.489527, "boundary_entropy": 0.690591, "bpb": 4.569975}, "ta": {"chunks": 9342.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.499706, "cps": 72.984375, "mean_p": 0.496082, "boundary_entropy": 0.691273, "bpb": 4.376409}, "th": {"chunks": 10699.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.665154, "cps": 83.585938, "mean_p": 0.492436, "boundary_entropy": 0.690817, "bpb": 4.317648}, "zh": {"chunks": 4785.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.810193, "cps": 37.382812, "mean_p": 0.497253, "boundary_entropy": 0.68721, "bpb": 6.672481}, "my": {"chunks": 15193.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.801826, "cps": 118.695312, "mean_p": 0.497252, "boundary_entropy": 0.691364, "bpb": 4.236122}}, "bpb": {"en": 5.40733, "de": 5.55573, "vi": 5.97509, "tr": 5.86302, "fi": 5.55111, "ru": 4.84486, "el": 5.07018, "hi": 4.56997, "ta": 4.37641, "th": 4.31765, "zh": 6.67248, "my": 4.23612}, "gini_cpc": 0.382304, "gini_cps": 0.351231, "bpic": 5.62127, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8959591455785312, "vi": 0.5839652278334586, "tr": 0.5050631052229276, "fi": 0.6235394806923117, "ru": 0.5636129214199346, "el": 0.7000246268711553, "hi": 1.6345715410771897, "ta": 2.122103489905186, "th": 2.824712738102986, "zh": 3.4406513414900792, "my": 3.405119266074895}, "t": 67.0}
{"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.16287, "bpb": 4.56305, "chunk_loss": 0.030228, "F": 0.17084, "G": 0.47339, "chunks": 174.94, "bytes_per_chunk": 5.854, "boundary_entropy": 0.68876, "router_noise": 0.0, "grad_norm": 1.6064, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 0.684, "alpha": 0.03, "t": 77.1}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3937.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.226225, "cps": 30.757812, "mean_p": 0.427314, "boundary_entropy": 0.661054, "bpb": 4.319407}, "de": {"chunks": 5527.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.269176, "cps": 43.179688, "mean_p": 0.431812, "boundary_entropy": 0.662952, "bpb": 4.521388}, "vi": {"chunks": 5384.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.292053, "cps": 42.0625, "mean_p": 0.457002, "boundary_entropy": 0.678336, "bpb": 4.61802}, "tr": {"chunks": 3834.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.211788, "cps": 29.953125, "mean_p": 0.440042, "boundary_entropy": 0.670801, "bpb": 4.723993}, "fi": {"chunks": 4075.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.216629, "cps": 31.835938, "mean_p": 0.4435, "boundary_entropy": 0.672042, "bpb": 4.535553}, "ru": {"chunks": 5283.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.278786, "cps": 41.273438, "mean_p": 0.483286, "boundary_entropy": 0.68916, "bpb": 3.178681}, "el": {"chunks": 12034.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.586367, "cps": 94.015625, "mean_p": 0.487278, "boundary_entropy": 0.687514, "bpb": 3.677759}, "hi": {"chunks": 7262.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.43162, "cps": 56.734375, "mean_p": 0.478514, "boundary_entropy": 0.687243, "bpb": 2.864003}, "ta": {"chunks": 7912.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.423215, "cps": 61.8125, "mean_p": 0.482517, "boundary_entropy": 0.689189, "bpb": 2.798385}, "th": {"chunks": 13107.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.814859, "cps": 102.398438, "mean_p": 0.468026, "boundary_entropy": 0.686401, "bpb": 2.696553}, "zh": {"chunks": 5057.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.856248, "cps": 39.507812, "mean_p": 0.497872, "boundary_entropy": 0.687209, "bpb": 6.198651}, "my": {"chunks": 19493.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.028763, "cps": 152.289062, "mean_p": 0.493676, "boundary_entropy": 0.690949, "bpb": 3.014542}}, "bpb": {"en": 4.31941, "de": 4.52139, "vi": 4.61802, "tr": 4.72399, "fi": 4.53555, "ru": 3.17868, "el": 3.67776, "hi": 2.864, "ta": 2.79838, "th": 2.69655, "zh": 6.19865, "my": 3.01454}, "gini_cpc": 0.311101, "gini_cps": 0.296761, "bpic": 4.130897, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1898597202931511, "vi": 1.2909832713074703, "tr": 0.9361819479314814, "fi": 0.9575786069864629, "ru": 1.232338741511256, "el": 2.591957447619032, "hi": 1.9079187545596488, "ta": 1.870766199087894, "th": 3.6019770351011013, "zh": 3.7849331767068546, "my": 4.547513658358539}, "t": 87.8}
{"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.52933, "bpb": 3.64905, "chunk_loss": 0.030857, "F": 0.19221, "G": 0.38466, "chunks": 196.83, "bytes_per_chunk": 5.203, "boundary_entropy": 0.62454, "router_noise": 0.0, "grad_norm": 2.0691, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.284, "alpha": 0.03, "t": 105.5}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2587.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.148653, "cps": 20.210938, "mean_p": 0.320718, "boundary_entropy": 0.557828, "bpb": 3.45617}, "de": {"chunks": 3573.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174013, "cps": 27.914062, "mean_p": 0.314637, "boundary_entropy": 0.551462, "bpb": 3.742538}, "vi": {"chunks": 4763.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.258367, "cps": 37.210938, "mean_p": 0.378733, "boundary_entropy": 0.604538, "bpb": 3.029825}, "tr": {"chunks": 3880.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.214329, "cps": 30.3125, "mean_p": 0.344302, "boundary_entropy": 0.566336, "bpb": 3.702488}, "fi": {"chunks": 3181.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.169103, "cps": 24.851562, "mean_p": 0.330859, "boundary_entropy": 0.570682, "bpb": 3.794671}, "ru": {"chunks": 5668.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299103, "cps": 44.28125, "mean_p": 0.275306, "boundary_entropy": 0.493586, "bpb": 2.605027}, "el": {"chunks": 7119.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.346879, "cps": 55.617188, "mean_p": 0.310738, "boundary_entropy": 0.560261, "bpb": 2.655338}, "hi": {"chunks": 8741.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.519525, "cps": 68.289062, "mean_p": 0.438578, "boundary_entropy": 0.645584, "bpb": 2.021656}, "ta": {"chunks": 9979.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.533779, "cps": 77.960938, "mean_p": 0.417434, "boundary_entropy": 0.633712, "bpb": 1.842463}, "th": {"chunks": 7313.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.454647, "cps": 57.132812, "mean_p": 0.393417, "boundary_entropy": 0.646264, "bpb": 2.053021}, "zh": {"chunks": 3131.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.530139, "cps": 24.460938, "mean_p": 0.438149, "boundary_entropy": 0.666391, "bpb": 4.876077}, "my": {"chunks": 12901.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.680863, "cps": 100.789062, "mean_p": 0.482708, "boundary_entropy": 0.67821, "bpb": 2.07712}}, "bpb": {"en": 3.45617, "de": 3.74254, "vi": 3.02982, "tr": 3.70249, "fi": 3.79467, "ru": 2.60503, "el": 2.65534, "hi": 2.02166, "ta": 1.84246, "th": 2.05302, "zh": 4.87608, "my": 2.07712}, "gini_cpc": 0.265469, "gini_cps": 0.278397, "bpic": 5.269111, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1705994090122889, "vi": 1.7380614658697506, "tr": 1.441812763160151, "fi": 1.1375735278297903, "ru": 2.0120942443086145, "el": 2.333489436771525, "hi": 3.4948918325769895, "ta": 3.5907836960844715, "th": 3.058455737235711, "zh": 3.5662954251315973, "my": 4.5802342569769}, "t": 121.4}
{"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.12363, "bpb": 3.06376, "chunk_loss": 0.031162, "F": 0.19568, "G": 0.34281, "chunks": 200.38, "bytes_per_chunk": 5.11, "boundary_entropy": 0.56114, "router_noise": 0.0, "grad_norm": 1.8201, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.261, "alpha": 0.03, "t": 131.6}
{"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.02422, "bpb": 2.92034, "chunk_loss": 0.031045, "F": 0.19971, "G": 0.30083, "chunks": 204.5, "bytes_per_chunk": 5.007, "boundary_entropy": 0.48828, "router_noise": 0.0, "grad_norm": 1.131, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 155.5}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.78182, "bpb": 2.57062, "chunk_loss": 0.030832, "F": 0.19073, "G": 0.2997, "chunks": 195.31, "bytes_per_chunk": 5.243, "boundary_entropy": 0.46324, "router_noise": 0.0, "grad_norm": 0.9029, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 178.3}
{"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.76157, "bpb": 2.54141, "chunk_loss": 0.030718, "F": 0.1891, "G": 0.28075, "chunks": 193.64, "bytes_per_chunk": 5.288, "boundary_entropy": 0.41366, "router_noise": 0.0, "grad_norm": 0.8891, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 202.2}
{"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.65944, "bpb": 2.39406, "chunk_loss": 0.030554, "F": 0.1879, "G": 0.27541, "chunks": 192.41, "bytes_per_chunk": 5.322, "boundary_entropy": 0.38535, "router_noise": 0.0, "grad_norm": 0.8136, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 225.6}
{"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.7172, "bpb": 2.4774, "chunk_loss": 0.031079, "F": 0.20461, "G": 0.27358, "chunks": 209.52, "bytes_per_chunk": 4.887, "boundary_entropy": 0.35272, "router_noise": 0.0, "grad_norm": 0.9253, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.23, "alpha": 0.03, "t": 248.7}
{"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.60356, "bpb": 2.31344, "chunk_loss": 0.030908, "F": 0.19975, "G": 0.27406, "chunks": 204.55, "bytes_per_chunk": 5.006, "boundary_entropy": 0.35943, "router_noise": 0.0, "grad_norm": 0.6479, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 272.6}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2897.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.166466, "cps": 22.632812, "mean_p": 0.216414, "boundary_entropy": 0.269659, "bpb": 2.302406}, "de": {"chunks": 3495.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.170214, "cps": 27.304688, "mean_p": 0.213595, "boundary_entropy": 0.264779, "bpb": 2.573204}, "vi": {"chunks": 4618.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.250502, "cps": 36.078125, "mean_p": 0.235551, "boundary_entropy": 0.326909, "bpb": 2.107698}, "tr": {"chunks": 4078.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.225267, "cps": 31.859375, "mean_p": 0.2495, "boundary_entropy": 0.311278, "bpb": 2.580318}, "fi": {"chunks": 3857.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.20504, "cps": 30.132812, "mean_p": 0.251127, "boundary_entropy": 0.305259, "bpb": 2.828595}, "ru": {"chunks": 5834.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.307863, "cps": 45.578125, "mean_p": 0.28362, "boundary_entropy": 0.456359, "bpb": 1.657855}, "el": {"chunks": 5942.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.289529, "cps": 46.421875, "mean_p": 0.287183, "boundary_entropy": 0.471406, "bpb": 1.735339}, "hi": {"chunks": 7943.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.472095, "cps": 62.054688, "mean_p": 0.334013, "boundary_entropy": 0.538744, "bpb": 1.353938}, "ta": {"chunks": 11160.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.596951, "cps": 87.1875, "mean_p": 0.303316, "boundary_entropy": 0.483207, "bpb": 1.116459}, "th": {"chunks": 8222.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.511159, "cps": 64.234375, "mean_p": 0.303808, "boundary_entropy": 0.494078, "bpb": 1.311037}, "zh": {"chunks": 4459.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.754995, "cps": 34.835938, "mean_p": 0.295377, "boundary_entropy": 0.194007, "bpb": 3.389499}, "my": {"chunks": 7487.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.395134, "cps": 58.492188, "mean_p": 0.238776, "boundary_entropy": 0.42601, "bpb": 1.25451}}, "bpb": {"en": 2.30241, "de": 2.5732, "vi": 2.1077, "tr": 2.58032, "fi": 2.82859, "ru": 1.65786, "el": 1.73534, "hi": 1.35394, "ta": 1.11646, "th": 1.31104, "zh": 3.3895, "my": 1.25451}, "gini_cpc": 0.27125, "gini_cps": 0.219963, "bpic": 5.483212, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0225166721440406, "vi": 1.5048264344613953, "tr": 1.3532321120861142, "fi": 1.2317239340292283, "ru": 1.8494084409037463, "el": 1.7392716870811367, "hi": 2.8359927393250484, "ta": 3.586033564062684, "th": 3.07066212472165, "zh": 4.5354423887013215, "my": 2.3736685851122874}, "t": 298.4}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.31298, "de": 2.56537, "vi": 2.1117, "tr": 2.57363, "fi": 2.83487, "ru": 1.65762, "el": 1.73058, "hi": 1.36019, "ta": 1.11701, "th": 1.30248, "zh": 3.39554, "my": 1.26648}, "bytes_per_chunk": {"en": 5.456, "de": 5.561, "vi": 5.315, "tr": 4.716, "fi": 4.835, "ru": 5.916, "el": 6.384, "hi": 5.349, "ta": 4.552, "th": 5.432, "zh": 3.628, "my": 6.952}, "bpb_mean": 2.01904, "t": 305.1}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.54132, "bpb": 2.22366, "chunk_loss": 0.030765, "F": 0.19746, "G": 0.25992, "chunks": 202.2, "bytes_per_chunk": 5.064, "boundary_entropy": 0.32816, "router_noise": 0.0, "grad_norm": 0.7536, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.328, "alpha": 0.03, "t": 305.4}
{"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.50567, "bpb": 2.17222, "chunk_loss": 0.030591, "F": 0.1908, "G": 0.25205, "chunks": 195.38, "bytes_per_chunk": 5.241, "boundary_entropy": 0.32513, "router_noise": 0.0, "grad_norm": 0.6509, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 329.1}
{"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.4968, "bpb": 2.15942, "chunk_loss": 0.030597, "F": 0.19183, "G": 0.25047, "chunks": 196.44, "bytes_per_chunk": 5.213, "boundary_entropy": 0.31155, "router_noise": 0.0, "grad_norm": 0.6069, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 353.1}
{"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.54827, "bpb": 2.23369, "chunk_loss": 0.030674, "F": 0.19373, "G": 0.24048, "chunks": 198.38, "bytes_per_chunk": 5.162, "boundary_entropy": 0.27507, "router_noise": 0.0, "grad_norm": 0.6397, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.221, "alpha": 0.03, "t": 375.2}
{"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.37605, "bpb": 1.98522, "chunk_loss": 0.030486, "F": 0.18828, "G": 0.24381, "chunks": 192.8, "bytes_per_chunk": 5.311, "boundary_entropy": 0.30529, "router_noise": 0.0, "grad_norm": 0.5329, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 398.8}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.49104, "bpb": 2.15112, "chunk_loss": 0.030508, "F": 0.18672, "G": 0.23231, "chunks": 191.2, "bytes_per_chunk": 5.356, "boundary_entropy": 0.26897, "router_noise": 0.0, "grad_norm": 0.5542, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.222, "alpha": 0.03, "t": 421.0}
{"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.52071, "bpb": 2.19392, "chunk_loss": 0.03048, "F": 0.18777, "G": 0.23607, "chunks": 192.28, "bytes_per_chunk": 5.326, "boundary_entropy": 0.26389, "router_noise": 0.0, "grad_norm": 0.6635, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 444.9}
{"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.38482, "bpb": 1.99788, "chunk_loss": 0.030282, "F": 0.17892, "G": 0.23892, "chunks": 183.22, "bytes_per_chunk": 5.589, "boundary_entropy": 0.30017, "router_noise": 0.0, "grad_norm": 0.4956, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.223, "alpha": 0.03, "t": 467.2}
{"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.39463, "bpb": 2.01202, "chunk_loss": 0.030858, "F": 0.19951, "G": 0.24641, "chunks": 204.3, "bytes_per_chunk": 5.012, "boundary_entropy": 0.26879, "router_noise": 0.0, "grad_norm": 0.4963, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 490.7}
{"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.38868, "bpb": 2.00344, "chunk_loss": 0.030973, "F": 0.19862, "G": 0.24595, "chunks": 203.39, "bytes_per_chunk": 5.035, "boundary_entropy": 0.26831, "router_noise": 0.0, "grad_norm": 0.4935, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.221, "alpha": 0.03, "t": 512.8}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3021.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.173591, "cps": 23.601562, "mean_p": 0.198435, "boundary_entropy": 0.193027, "bpb": 2.08145}, "de": {"chunks": 3418.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.166464, "cps": 26.703125, "mean_p": 0.192849, "boundary_entropy": 0.204573, "bpb": 2.285441}, "vi": {"chunks": 3773.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.204665, "cps": 29.476562, "mean_p": 0.193751, "boundary_entropy": 0.247224, "bpb": 1.940006}, "tr": {"chunks": 3575.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197481, "cps": 27.929688, "mean_p": 0.213271, "boundary_entropy": 0.231356, "bpb": 2.25387}, "fi": {"chunks": 3528.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18755, "cps": 27.5625, "mean_p": 0.207041, "boundary_entropy": 0.200081, "bpb": 2.432771}, "ru": {"chunks": 6268.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.330765, "cps": 48.96875, "mean_p": 0.268366, "boundary_entropy": 0.388908, "bpb": 1.418167}, "el": {"chunks": 5882.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.286605, "cps": 45.953125, "mean_p": 0.26904, "boundary_entropy": 0.404401, "bpb": 1.479971}, "hi": {"chunks": 7997.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.475305, "cps": 62.476562, "mean_p": 0.266626, "boundary_entropy": 0.423175, "bpb": 1.193529}, "ta": {"chunks": 8196.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.438406, "cps": 64.03125, "mean_p": 0.279609, "boundary_entropy": 0.455577, "bpb": 1.001598}, "th": {"chunks": 7352.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.457072, "cps": 57.4375, "mean_p": 0.235922, "boundary_entropy": 0.383653, "bpb": 1.105343}, "zh": {"chunks": 4390.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.743312, "cps": 34.296875, "mean_p": 0.283417, "boundary_entropy": 0.14567, "bpb": 3.093401}, "my": {"chunks": 7555.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.398723, "cps": 59.023438, "mean_p": 0.265884, "boundary_entropy": 0.465972, "bpb": 1.130864}}, "bpb": {"en": 2.08145, "de": 2.28544, "vi": 1.94001, "tr": 2.25387, "fi": 2.43277, "ru": 1.41817, "el": 1.47997, "hi": 1.19353, "ta": 1.0016, "th": 1.10534, "zh": 3.0934, "my": 1.13086}, "gini_cpc": 0.262286, "gini_cps": 0.197813, "bpic": 5.908414, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9589435583410163, "vi": 1.179008831318821, "tr": 1.1376243772463661, "fi": 1.080413713436596, "ru": 1.9054307446816874, "el": 1.6510399043173931, "hi": 2.7380754923558865, "ta": 2.5255145513898034, "th": 2.6330422510255382, "zh": 4.281978403125295, "my": 2.296912692008799}, "t": 540.1}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.08968, "de": 2.27758, "vi": 1.94376, "tr": 2.24553, "fi": 2.44819, "ru": 1.41852, "el": 1.47571, "hi": 1.20128, "ta": 1.00219, "th": 1.10265, "zh": 3.09955, "my": 1.14025}, "bytes_per_chunk": {"en": 5.3, "de": 5.759, "vi": 6.375, "tr": 5.388, "fi": 5.221, "ru": 5.432, "el": 6.237, "hi": 5.451, "ta": 6.147, "th": 6.031, "zh": 3.693, "my": 7.16}, "bpb_mean": 1.78707, "t": 546.6}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.39653, "bpb": 2.01477, "chunk_loss": 0.03074, "F": 0.19519, "G": 0.23625, "chunks": 199.88, "bytes_per_chunk": 5.123, "boundary_entropy": 0.2506, "router_noise": 0.0, "grad_norm": 0.5464, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.34, "alpha": 0.03, "t": 546.8}
{"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.44152, "bpb": 2.07968, "chunk_loss": 0.030727, "F": 0.19885, "G": 0.24112, "chunks": 203.62, "bytes_per_chunk": 5.029, "boundary_entropy": 0.2502, "router_noise": 0.0, "grad_norm": 0.4711, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 570.1}
{"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.40131, "bpb": 2.02167, "chunk_loss": 0.030564, "F": 0.19131, "G": 0.23035, "chunks": 195.91, "bytes_per_chunk": 5.227, "boundary_entropy": 0.23783, "router_noise": 0.0, "grad_norm": 0.4404, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.223, "alpha": 0.03, "t": 592.4}
{"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.33833, "bpb": 1.93079, "chunk_loss": 0.030507, "F": 0.19098, "G": 0.22536, "chunks": 195.56, "bytes_per_chunk": 5.236, "boundary_entropy": 0.22964, "router_noise": 0.0, "grad_norm": 0.4382, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 616.2}
{"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.36607, "bpb": 1.97082, "chunk_loss": 0.030545, "F": 0.19208, "G": 0.23269, "chunks": 196.69, "bytes_per_chunk": 5.206, "boundary_entropy": 0.24139, "router_noise": 0.0, "grad_norm": 0.4534, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.216, "alpha": 0.03, "t": 637.7}
{"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.22162, "bpb": 1.76243, "chunk_loss": 0.03038, "F": 0.18433, "G": 0.22529, "chunks": 188.75, "bytes_per_chunk": 5.425, "boundary_entropy": 0.24605, "router_noise": 0.0, "grad_norm": 0.4118, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 661.5}
{"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.21786, "bpb": 1.757, "chunk_loss": 0.030391, "F": 0.18475, "G": 0.22699, "chunks": 189.19, "bytes_per_chunk": 5.413, "boundary_entropy": 0.24639, "router_noise": 0.0, "grad_norm": 0.4014, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.218, "alpha": 0.03, "t": 683.3}
{"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.36582, "bpb": 1.97047, "chunk_loss": 0.030523, "F": 0.19258, "G": 0.22136, "chunks": 197.2, "bytes_per_chunk": 5.193, "boundary_entropy": 0.20466, "router_noise": 0.0, "grad_norm": 0.4521, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 706.7}
{"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.1902, "bpb": 1.7171, "chunk_loss": 0.030444, "F": 0.18704, "G": 0.23329, "chunks": 191.53, "bytes_per_chunk": 5.346, "boundary_entropy": 0.24938, "router_noise": 0.0, "grad_norm": 0.4229, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.217, "alpha": 0.03, "t": 728.4}
{"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.24503, "bpb": 1.7962, "chunk_loss": 0.030405, "F": 0.18579, "G": 0.22582, "chunks": 190.25, "bytes_per_chunk": 5.382, "boundary_entropy": 0.23171, "router_noise": 0.0, "grad_norm": 0.4081, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 751.9}
{"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3027.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.173936, "cps": 23.648438, "mean_p": 0.184619, "boundary_entropy": 0.127875, "bpb": 1.972089}, "de": {"chunks": 3498.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.17036, "cps": 27.328125, "mean_p": 0.185302, "boundary_entropy": 0.14858, "bpb": 2.159564}, "vi": {"chunks": 4040.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.219148, "cps": 31.5625, "mean_p": 0.202808, "boundary_entropy": 0.231036, "bpb": 1.833548}, "tr": {"chunks": 3473.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191847, "cps": 27.132812, "mean_p": 0.196366, "boundary_entropy": 0.160734, "bpb": 2.121212}, "fi": {"chunks": 3573.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189942, "cps": 27.914062, "mean_p": 0.196995, "boundary_entropy": 0.135888, "bpb": 2.288775}, "ru": {"chunks": 5468.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.288549, "cps": 42.71875, "mean_p": 0.229294, "boundary_entropy": 0.3168, "bpb": 1.317192}, "el": {"chunks": 6007.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.292696, "cps": 46.929688, "mean_p": 0.252051, "boundary_entropy": 0.35605, "bpb": 1.345805}, "hi": {"chunks": 7040.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.418425, "cps": 55.0, "mean_p": 0.206913, "boundary_entropy": 0.313177, "bpb": 1.126782}, "ta": {"chunks": 7955.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.425515, "cps": 62.148438, "mean_p": 0.281738, "boundary_entropy": 0.452397, "bpb": 0.941686}, "th": {"chunks": 7656.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.475971, "cps": 59.8125, "mean_p": 0.213346, "boundary_entropy": 0.325502, "bpb": 1.04045}, "zh": {"chunks": 4242.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.718253, "cps": 33.140625, "mean_p": 0.27622, "boundary_entropy": 0.122555, "bpb": 2.908659}, "my": {"chunks": 9467.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.499631, "cps": 73.960938, "mean_p": 0.294482, "boundary_entropy": 0.482653, "bpb": 1.049956}}, "bpb": {"en": 1.97209, "de": 2.15956, "vi": 1.83355, "tr": 2.12121, "fi": 2.28878, "ru": 1.31719, "el": 1.34581, "hi": 1.12678, "ta": 0.94169, "th": 1.04045, "zh": 2.90866, "my": 1.04996}, "gini_cpc": 0.261027, "gini_cps": 0.209773, "bpic": 5.864086, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9794428432714154, "vi": 1.259940169609936, "tr": 1.1029756661131955, "fi": 1.092025631419189, "ru": 1.6589411915452223, "el": 1.6827844338342242, "hi": 2.405632516857937, "ta": 2.4463940608341455, "th": 2.7364817666799244, "zh": 4.129418706078078, "my": 2.8725043847508798}, "t": 779.2}
{"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 1.97695, "de": 2.15307, "vi": 1.84124, "tr": 2.11456, "fi": 2.3043, "ru": 1.31621, "el": 1.34179, "hi": 1.13354, "ta": 0.94236, "th": 1.0389, "zh": 2.91619, "my": 1.05835}, "bytes_per_chunk": {"en": 5.323, "de": 5.706, "vi": 6.055, "tr": 5.442, "fi": 5.255, "ru": 6.082, "el": 5.969, "hi": 6.176, "ta": 6.34, "th": 6.043, "zh": 3.733, "my": 5.626}, "bpb_mean": 1.67812, "t": 785.7}
{"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.27383, "bpb": 1.83775, "chunk_loss": 0.030374, "F": 0.18544, "G": 0.21796, "chunks": 189.89, "bytes_per_chunk": 5.393, "boundary_entropy": 0.20439, "router_noise": 0.0, "grad_norm": 0.4139, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.34, "alpha": 0.03, "t": 785.9}
{"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.21217, "bpb": 1.74879, "chunk_loss": 0.030214, "F": 0.18059, "G": 0.21597, "chunks": 184.92, "bytes_per_chunk": 5.537, "boundary_entropy": 0.21727, "router_noise": 0.0, "grad_norm": 0.4161, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.22, "alpha": 0.03, "t": 807.9}
{"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.28285, "bpb": 1.85076, "chunk_loss": 0.030558, "F": 0.19162, "G": 0.22909, "chunks": 196.22, "bytes_per_chunk": 5.219, "boundary_entropy": 0.22648, "router_noise": 0.0, "grad_norm": 0.4504, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 831.6}
{"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.27349, "bpb": 1.83726, "chunk_loss": 0.030334, "F": 0.18272, "G": 0.21667, "chunks": 187.11, "bytes_per_chunk": 5.473, "boundary_entropy": 0.20328, "router_noise": 0.0, "grad_norm": 0.4121, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.223, "alpha": 0.03, "t": 853.9}
{"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.33838, "bpb": 1.93088, "chunk_loss": 0.03063, "F": 0.19629, "G": 0.22384, "chunks": 201.0, "bytes_per_chunk": 5.095, "boundary_entropy": 0.19043, "router_noise": 0.0, "grad_norm": 0.4102, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 876.3}
{"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.1863, "bpb": 1.71147, "chunk_loss": 0.030434, "F": 0.18846, "G": 0.22456, "chunks": 192.98, "bytes_per_chunk": 5.306, "boundary_entropy": 0.21103, "router_noise": 0.0, "grad_norm": 0.3931, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 899.9}
{"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.24252, "bpb": 1.79258, "chunk_loss": 0.030306, "F": 0.18391, "G": 0.21186, "chunks": 188.33, "bytes_per_chunk": 5.437, "boundary_entropy": 0.19311, "router_noise": 0.0, "grad_norm": 0.4265, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 924.0}
{"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.21838, "bpb": 1.75775, "chunk_loss": 0.030511, "F": 0.19168, "G": 0.23113, "chunks": 196.28, "bytes_per_chunk": 5.217, "boundary_entropy": 0.22143, "router_noise": 0.0, "grad_norm": 0.423, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 948.4}
{"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.24498, "bpb": 1.79613, "chunk_loss": 0.030552, "F": 0.19504, "G": 0.22344, "chunks": 199.72, "bytes_per_chunk": 5.127, "boundary_entropy": 0.19056, "router_noise": 0.0, "grad_norm": 0.3928, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 972.7}
{"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.23522, "bpb": 1.78205, "chunk_loss": 0.030422, "F": 0.18643, "G": 0.21749, "chunks": 190.91, "bytes_per_chunk": 5.364, "boundary_entropy": 0.19057, "router_noise": 0.0, "grad_norm": 0.4125, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 997.0}
{"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2992.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.171924, "cps": 23.375, "mean_p": 0.179806, "boundary_entropy": 0.105708, "bpb": 1.890059}, "de": {"chunks": 3513.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.17109, "cps": 27.445312, "mean_p": 0.181392, "boundary_entropy": 0.127575, "bpb": 2.066453}, "vi": {"chunks": 3981.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.215948, "cps": 31.101562, "mean_p": 0.207723, "boundary_entropy": 0.235759, "bpb": 1.750616}, "tr": {"chunks": 3372.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.186267, "cps": 26.34375, "mean_p": 0.184499, "boundary_entropy": 0.132754, "bpb": 2.002063}, "fi": {"chunks": 3551.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188773, "cps": 27.742188, "mean_p": 0.19363, "boundary_entropy": 0.116012, "bpb": 2.163546}, "ru": {"chunks": 5693.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.300422, "cps": 44.476562, "mean_p": 0.223777, "boundary_entropy": 0.280462, "bpb": 1.264432}, "el": {"chunks": 5883.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.286654, "cps": 45.960938, "mean_p": 0.236275, "boundary_entropy": 0.314009, "bpb": 1.277671}, "hi": {"chunks": 6766.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.40214, "cps": 52.859375, "mean_p": 0.203389, "boundary_entropy": 0.319931, "bpb": 1.071085}, "ta": {"chunks": 8267.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.442204, "cps": 64.585938, "mean_p": 0.309488, "boundary_entropy": 0.505931, "bpb": 0.930521}, "th": {"chunks": 8056.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.500839, "cps": 62.9375, "mean_p": 0.20802, "boundary_entropy": 0.299153, "bpb": 1.005487}, "zh": {"chunks": 4226.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.715544, "cps": 33.015625, "mean_p": 0.271671, "boundary_entropy": 0.106492, "bpb": 2.802499}, "my": {"chunks": 9587.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.505964, "cps": 74.898438, "mean_p": 0.364325, "boundary_entropy": 0.564837, "bpb": 1.011839}}, "bpb": {"en": 1.89006, "de": 2.06645, "vi": 1.75062, "tr": 2.00206, "fi": 2.16355, "ru": 1.26443, "el": 1.27767, "hi": 1.07108, "ta": 0.93052, "th": 1.00549, "zh": 2.8025, "my": 1.01184}, "gini_cpc": 0.265769, "gini_cps": 0.2171, "bpic": 5.824836, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9951493728238695, "vi": 1.2560634161942017, "tr": 1.083426728397013, "fi": 1.097997420584213, "ru": 1.7474087275831418, "el": 1.6673260449147835, "hi": 2.3390497095771985, "ta": 2.5720831193962184, "th": 2.913137094213728, "zh": 4.161966508273106, "my": 2.9429432148384715}, "t": 1026.9}
{"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.89378, "de": 2.06258, "vi": 1.76009, "tr": 1.99786, "fi": 2.17991, "ru": 1.26365, "el": 1.27521, "hi": 1.0775, "ta": 0.93218, "th": 1.00286, "zh": 2.81206, "my": 1.02044}, "bytes_per_chunk": {"en": 5.347, "de": 5.656, "vi": 6.098, "tr": 5.621, "fi": 5.314, "ru": 5.826, "el": 6.049, "hi": 6.638, "ta": 6.133, "th": 5.785, "zh": 3.742, "my": 5.76}, "bpb_mean": 1.60651, "t": 1033.3}
{"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.16332, "bpb": 1.67832, "chunk_loss": 0.030251, "F": 0.17912, "G": 0.20755, "chunks": 183.42, "bytes_per_chunk": 5.583, "boundary_entropy": 0.19147, "router_noise": 0.0, "grad_norm": 0.3834, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.365, "alpha": 0.03, "t": 1033.6}
{"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.23365, "bpb": 1.77978, "chunk_loss": 0.030395, "F": 0.18484, "G": 0.21526, "chunks": 189.28, "bytes_per_chunk": 5.41, "boundary_entropy": 0.18569, "router_noise": 0.0, "grad_norm": 0.4163, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.254, "alpha": 0.03, "t": 1058.9}
{"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.21217, "bpb": 1.74879, "chunk_loss": 0.030668, "F": 0.1931, "G": 0.23099, "chunks": 197.73, "bytes_per_chunk": 5.179, "boundary_entropy": 0.20894, "router_noise": 0.0, "grad_norm": 0.3973, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.256, "alpha": 0.03, "t": 1084.6}
{"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.14594, "bpb": 1.65324, "chunk_loss": 0.03045, "F": 0.18619, "G": 0.22352, "chunks": 190.66, "bytes_per_chunk": 5.371, "boundary_entropy": 0.20688, "router_noise": 0.0, "grad_norm": 0.3854, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.252, "alpha": 0.03, "t": 1109.8}
{"event": "phase_transition", "step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1134.6}
{"event": "train", "step": 4400, "phase": 2, "objective": "global", "lm_loss": 1.19202, "bpb": 1.71972, "chunk_loss": 0.030621, "F": 0.19357, "G": 0.22149, "chunks": 198.22, "bytes_per_chunk": 5.166, "boundary_entropy": 0.18347, "router_noise": 0.0, "grad_norm": 0.3926, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.251, "alpha": 0.03, "t": 1134.9}
{"event": "train", "step": 4500, "phase": 2, "objective": "global", "lm_loss": 1.28544, "bpb": 1.8545, "chunk_loss": 0.030383, "F": 0.18813, "G": 0.2077, "chunks": 192.64, "bytes_per_chunk": 5.316, "boundary_entropy": 0.14957, "router_noise": 0.0, "grad_norm": 0.407, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.252, "alpha": 0.03, "t": 1160.1}
{"event": "train", "step": 4600, "phase": 2, "objective": "global", "lm_loss": 1.13604, "bpb": 1.63897, "chunk_loss": 0.030304, "F": 0.18092, "G": 0.21579, "chunks": 185.27, "bytes_per_chunk": 5.527, "boundary_entropy": 0.19549, "router_noise": 0.0, "grad_norm": 0.3872, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.247, "alpha": 0.03, "t": 1184.8}
{"event": "train", "step": 4700, "phase": 2, "objective": "global", "lm_loss": 1.19376, "bpb": 1.72223, "chunk_loss": 0.030444, "F": 0.19281, "G": 0.22331, "chunks": 197.44, "bytes_per_chunk": 5.186, "boundary_entropy": 0.18834, "router_noise": 0.0, "grad_norm": 0.3722, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 1209.0}
{"event": "train", "step": 4800, "phase": 2, "objective": "global", "lm_loss": 1.18128, "bpb": 1.70422, "chunk_loss": 0.030347, "F": 0.18529, "G": 0.21314, "chunks": 189.73, "bytes_per_chunk": 5.397, "boundary_entropy": 0.17198, "router_noise": 0.0, "grad_norm": 0.3846, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.246, "alpha": 0.03, "t": 1233.6}
{"event": "train", "step": 4900, "phase": 2, "objective": "global", "lm_loss": 1.24627, "bpb": 1.79798, "chunk_loss": 0.030602, "F": 0.19437, "G": 0.21782, "chunks": 199.03, "bytes_per_chunk": 5.145, "boundary_entropy": 0.16056, "router_noise": 0.0, "grad_norm": 0.3991, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1257.6}
{"event": "probe", "step": 5000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2975.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.170948, "cps": 23.242188, "mean_p": 0.176294, "boundary_entropy": 0.095342, "bpb": 1.836424}, "de": {"chunks": 3427.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.166902, "cps": 26.773438, "mean_p": 0.175656, "boundary_entropy": 0.112065, "bpb": 2.003216}, "vi": {"chunks": 3726.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.202116, "cps": 29.109375, "mean_p": 0.208207, "boundary_entropy": 0.246234, "bpb": 1.692401}, "tr": {"chunks": 3370.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.186157, "cps": 26.328125, "mean_p": 0.182266, "boundary_entropy": 0.124678, "bpb": 1.933129}, "fi": {"chunks": 3549.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188666, "cps": 27.726562, "mean_p": 0.191526, "boundary_entropy": 0.103397, "bpb": 2.091487}, "ru": {"chunks": 5488.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.289604, "cps": 42.875, "mean_p": 0.216058, "boundary_entropy": 0.263828, "bpb": 1.221504}, "el": {"chunks": 5578.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.271793, "cps": 43.578125, "mean_p": 0.237305, "boundary_entropy": 0.327846, "bpb": 1.226843}, "hi": {"chunks": 7517.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.446776, "cps": 58.726562, "mean_p": 0.204787, "boundary_entropy": 0.282459, "bpb": 1.021844}, "ta": {"chunks": 8114.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.43402, "cps": 63.390625, "mean_p": 0.311574, "boundary_entropy": 0.49603, "bpb": 0.869521}, "th": {"chunks": 7080.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.440162, "cps": 55.3125, "mean_p": 0.189882, "boundary_entropy": 0.286413, "bpb": 0.971862}, "zh": {"chunks": 4229.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.716051, "cps": 33.039062, "mean_p": 0.270277, "boundary_entropy": 0.097812, "bpb": 2.717876}, "my": {"chunks": 8024.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.423475, "cps": 62.6875, "mean_p": 0.31907, "boundary_entropy": 0.520207, "bpb": 0.974293}}, "bpb": {"en": 1.83642, "de": 2.00322, "vi": 1.6924, "tr": 1.93313, "fi": 2.09149, "ru": 1.2215, "el": 1.22684, "hi": 1.02184, "ta": 0.86952, "th": 0.97186, "zh": 2.71788, "my": 0.97429}, "gini_cpc": 0.258463, "gini_cps": 0.200172, "bpic": 6.084326, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.976334975425908, "vi": 1.1823249618806024, "tr": 1.0889714644074502, "fi": 1.1036497418594253, "ru": 1.6941116871022817, "el": 1.5899183248800177, "hi": 2.613524811767203, "ta": 2.538906362866088, "th": 2.574834636372423, "zh": 4.188720600329533, "my": 2.4772206640731023}, "t": 1286.1}
{"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.83791, "de": 2.00119, "vi": 1.70343, "tr": 1.92812, "fi": 2.10573, "ru": 1.21975, "el": 1.22632, "hi": 1.02787, "ta": 0.87322, "th": 0.96806, "zh": 2.72646, "my": 0.9827}, "bytes_per_chunk": {"en": 5.368, "de": 5.768, "vi": 6.387, "tr": 5.66, "fi": 5.328, "ru": 5.997, "el": 6.408, "hi": 5.776, "ta": 6.263, "th": 6.417, "zh": 3.74, "my": 6.807}, "bpb_mean": 1.55006, "t": 1292.7}
{"event": "train", "step": 5000, "phase": 2, "objective": "global", "lm_loss": 1.19266, "bpb": 1.72065, "chunk_loss": 0.030347, "F": 0.18677, "G": 0.22167, "chunks": 191.25, "bytes_per_chunk": 5.354, "boundary_entropy": 0.1891, "router_noise": 0.0, "grad_norm": 0.3805, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.353, "alpha": 0.03, "t": 1292.9}
{"event": "train", "step": 5100, "phase": 2, "objective": "global", "lm_loss": 1.19278, "bpb": 1.72082, "chunk_loss": 0.030494, "F": 0.19197, "G": 0.21318, "chunks": 196.58, "bytes_per_chunk": 5.209, "boundary_entropy": 0.15701, "router_noise": 0.0, "grad_norm": 0.3971, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.245, "alpha": 0.03, "t": 1317.4}
{"event": "train", "step": 5200, "phase": 2, "objective": "global", "lm_loss": 1.19402, "bpb": 1.72261, "chunk_loss": 0.030514, "F": 0.1915, "G": 0.22078, "chunks": 196.09, "bytes_per_chunk": 5.222, "boundary_entropy": 0.17654, "router_noise": 0.0, "grad_norm": 0.3757, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 1341.8}
{"event": "train", "step": 5300, "phase": 2, "objective": "global", "lm_loss": 1.15334, "bpb": 1.66392, "chunk_loss": 0.030488, "F": 0.19153, "G": 0.22144, "chunks": 196.12, "bytes_per_chunk": 5.221, "boundary_entropy": 0.18215, "router_noise": 0.0, "grad_norm": 0.3695, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.248, "alpha": 0.03, "t": 1366.6}
{"event": "train", "step": 5400, "phase": 2, "objective": "global", "lm_loss": 1.12856, "bpb": 1.62817, "chunk_loss": 0.030318, "F": 0.18257, "G": 0.20784, "chunks": 186.95, "bytes_per_chunk": 5.477, "boundary_entropy": 0.16823, "router_noise": 0.0, "grad_norm": 0.3853, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 1390.4}
{"event": "train", "step": 5500, "phase": 2, "objective": "global", "lm_loss": 1.12423, "bpb": 1.62192, "chunk_loss": 0.030321, "F": 0.18501, "G": 0.21832, "chunks": 189.45, "bytes_per_chunk": 5.405, "boundary_entropy": 0.19101, "router_noise": 0.0, "grad_norm": 0.3656, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 1413.8}
{"event": "train", "step": 5600, "phase": 2, "objective": "global", "lm_loss": 1.13309, "bpb": 1.6347, "chunk_loss": 0.030647, "F": 0.19337, "G": 0.22376, "chunks": 198.02, "bytes_per_chunk": 5.171, "boundary_entropy": 0.18249, "router_noise": 0.0, "grad_norm": 0.3827, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 1438.2}
{"event": "train", "step": 5700, "phase": 2, "objective": "global", "lm_loss": 1.17683, "bpb": 1.69781, "chunk_loss": 0.030459, "F": 0.19351, "G": 0.22085, "chunks": 198.16, "bytes_per_chunk": 5.168, "boundary_entropy": 0.16397, "router_noise": 0.0, "grad_norm": 0.3723, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1462.1}
{"event": "train", "step": 5800, "phase": 2, "objective": "global", "lm_loss": 1.09784, "bpb": 1.58385, "chunk_loss": 0.030273, "F": 0.18074, "G": 0.21919, "chunks": 185.08, "bytes_per_chunk": 5.533, "boundary_entropy": 0.20035, "router_noise": 0.0, "grad_norm": 0.3551, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.247, "alpha": 0.03, "t": 1486.8}
{"event": "train", "step": 5900, "phase": 2, "objective": "global", "lm_loss": 1.13852, "bpb": 1.64254, "chunk_loss": 0.030537, "F": 0.18753, "G": 0.21844, "chunks": 192.03, "bytes_per_chunk": 5.332, "boundary_entropy": 0.1801, "router_noise": 0.0, "grad_norm": 0.3793, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.249, "alpha": 0.03, "t": 1511.7}
{"event": "probe", "step": 6000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2973.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.170833, "cps": 23.226562, "mean_p": 0.174741, "boundary_entropy": 0.085658, "bpb": 1.799042}, "de": {"chunks": 3476.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.169288, "cps": 27.15625, "mean_p": 0.176727, "boundary_entropy": 0.106628, "bpb": 1.951651}, "vi": {"chunks": 3875.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.210198, "cps": 30.273438, "mean_p": 0.220469, "boundary_entropy": 0.264904, "bpb": 1.633117}, "tr": {"chunks": 3373.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.186323, "cps": 26.351562, "mean_p": 0.180521, "boundary_entropy": 0.108954, "bpb": 1.872285}, "fi": {"chunks": 3521.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187178, "cps": 27.507812, "mean_p": 0.189863, "boundary_entropy": 0.094946, "bpb": 2.044992}, "ru": {"chunks": 5538.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.292243, "cps": 43.265625, "mean_p": 0.215519, "boundary_entropy": 0.253506, "bpb": 1.18655}, "el": {"chunks": 5717.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.278566, "cps": 44.664062, "mean_p": 0.238062, "boundary_entropy": 0.316383, "bpb": 1.18421}, "hi": {"chunks": 7162.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.425676, "cps": 55.953125, "mean_p": 0.195203, "boundary_entropy": 0.276843, "bpb": 0.992444}, "ta": {"chunks": 7832.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.418936, "cps": 61.1875, "mean_p": 0.319265, "boundary_entropy": 0.509511, "bpb": 0.846328}, "th": {"chunks": 8020.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.498601, "cps": 62.65625, "mean_p": 0.199828, "boundary_entropy": 0.26677, "bpb": 0.931999}, "zh": {"chunks": 4224.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.715205, "cps": 33.0, "mean_p": 0.268781, "boundary_entropy": 0.090036, "bpb": 2.644951}, "my": {"chunks": 9021.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.476092, "cps": 70.476562, "mean_p": 0.340022, "boundary_entropy": 0.530193, "bpb": 0.933397}}, "bpb": {"en": 1.79904, "de": 1.95165, "vi": 1.63312, "tr": 1.87229, "fi": 2.04499, "ru": 1.18655, "el": 1.18421, "hi": 0.99244, "ta": 0.84633, "th": 0.932, "zh": 2.64495, "my": 0.9334}, "gini_cpc": 0.265064, "gini_cps": 0.210983, "bpic": 5.928768, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9909610199976872, "vi": 1.2304324491346992, "tr": 1.0906741014933439, "fi": 1.095679037014916, "ru": 1.7106964261466653, "el": 1.6306342425099292, "hi": 2.4917728801411814, "ta": 2.4523159356918875, "th": 2.9186533322752144, "zh": 4.186582732571471, "my": 2.7868944311341655}, "t": 1539.5}
{"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.79893, "de": 1.94952, "vi": 1.64716, "tr": 1.86733, "fi": 2.06189, "ru": 1.18543, "el": 1.18405, "hi": 0.99697, "ta": 0.85068, "th": 0.92831, "zh": 2.65298, "my": 0.94235}, "bytes_per_chunk": {"en": 5.383, "de": 5.74, "vi": 6.2, "tr": 5.619, "fi": 5.369, "ru": 5.952, "el": 6.233, "hi": 6.137, "ta": 6.579, "th": 5.659, "zh": 3.741, "my": 6.051}, "bpb_mean": 1.50547, "t": 1546.0}
{"event": "train", "step": 6000, "phase": 2, "objective": "global", "lm_loss": 1.1258, "bpb": 1.62418, "chunk_loss": 0.030472, "F": 0.1886, "G": 0.22553, "chunks": 193.12, "bytes_per_chunk": 5.302, "boundary_entropy": 0.19381, "router_noise": 0.0, "grad_norm": 0.393, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.346, "alpha": 0.03, "t": 1546.3}
{"event": "train", "step": 6100, "phase": 2, "objective": "global", "lm_loss": 1.19631, "bpb": 1.72591, "chunk_loss": 0.030713, "F": 0.20038, "G": 0.23162, "chunks": 205.19, "bytes_per_chunk": 4.991, "boundary_entropy": 0.17455, "router_noise": 0.0, "grad_norm": 0.3948, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.246, "alpha": 0.03, "t": 1570.8}
{"event": "train", "step": 6200, "phase": 2, "objective": "global", "lm_loss": 1.17561, "bpb": 1.69605, "chunk_loss": 0.03055, "F": 0.19052, "G": 0.21632, "chunks": 195.09, "bytes_per_chunk": 5.249, "boundary_entropy": 0.16278, "router_noise": 0.0, "grad_norm": 0.3905, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.251, "alpha": 0.03, "t": 1595.9}
{"event": "train", "step": 6300, "phase": 2, "objective": "global", "lm_loss": 1.20346, "bpb": 1.73622, "chunk_loss": 0.030525, "F": 0.19376, "G": 0.2121, "chunks": 198.41, "bytes_per_chunk": 5.161, "boundary_entropy": 0.1416, "router_noise": 0.0, "grad_norm": 0.3863, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1620.0}
{"event": "train", "step": 6400, "phase": 2, "objective": "global", "lm_loss": 1.11875, "bpb": 1.61401, "chunk_loss": 0.030438, "F": 0.18929, "G": 0.21559, "chunks": 193.83, "bytes_per_chunk": 5.283, "boundary_entropy": 0.16626, "router_noise": 0.0, "grad_norm": 0.3733, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 1644.3}
{"event": "train", "step": 6500, "phase": 2, "objective": "global", "lm_loss": 1.15254, "bpb": 1.66276, "chunk_loss": 0.030397, "F": 0.18559, "G": 0.21282, "chunks": 190.05, "bytes_per_chunk": 5.388, "boundary_entropy": 0.16574, "router_noise": 0.0, "grad_norm": 0.3905, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.247, "alpha": 0.03, "t": 1669.0}
{"event": "train", "step": 6600, "phase": 2, "objective": "global", "lm_loss": 1.09662, "bpb": 1.58208, "chunk_loss": 0.030247, "F": 0.18224, "G": 0.21755, "chunks": 186.61, "bytes_per_chunk": 5.487, "boundary_entropy": 0.18599, "router_noise": 0.0, "grad_norm": 0.3749, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 1692.9}
{"event": "train", "step": 6700, "phase": 2, "objective": "global", "lm_loss": 1.11294, "bpb": 1.60564, "chunk_loss": 0.030522, "F": 0.19441, "G": 0.22163, "chunks": 199.08, "bytes_per_chunk": 5.144, "boundary_entropy": 0.17074, "router_noise": 0.0, "grad_norm": 0.3787, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 1717.1}
{"event": "train", "step": 6800, "phase": 2, "objective": "global", "lm_loss": 1.1767, "bpb": 1.69762, "chunk_loss": 0.03059, "F": 0.19525, "G": 0.21386, "chunks": 199.94, "bytes_per_chunk": 5.122, "boundary_entropy": 0.13928, "router_noise": 0.0, "grad_norm": 0.3777, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1741.1}
{"event": "train", "step": 6900, "phase": 2, "objective": "global", "lm_loss": 1.14055, "bpb": 1.64547, "chunk_loss": 0.030451, "F": 0.19211, "G": 0.2132, "chunks": 196.72, "bytes_per_chunk": 5.205, "boundary_entropy": 0.14456, "router_noise": 0.0, "grad_norm": 0.3784, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.245, "alpha": 0.03, "t": 1765.7}
{"event": "probe", "step": 7000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2979.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.171177, "cps": 23.273438, "mean_p": 0.175419, "boundary_entropy": 0.076704, "bpb": 1.764356}, "de": {"chunks": 3452.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.16812, "cps": 26.96875, "mean_p": 0.174383, "boundary_entropy": 0.091889, "bpb": 1.912944}, "vi": {"chunks": 3961.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.214863, "cps": 30.945312, "mean_p": 0.22175, "boundary_entropy": 0.253143, "bpb": 1.590504}, "tr": {"chunks": 3393.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.187427, "cps": 26.507812, "mean_p": 0.181364, "boundary_entropy": 0.098137, "bpb": 1.829388}, "fi": {"chunks": 3585.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.19058, "cps": 28.007812, "mean_p": 0.192307, "boundary_entropy": 0.086192, "bpb": 1.982273}, "ru": {"chunks": 5526.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.291609, "cps": 43.171875, "mean_p": 0.213961, "boundary_entropy": 0.246046, "bpb": 1.160208}, "el": {"chunks": 5830.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.284072, "cps": 45.546875, "mean_p": 0.236081, "boundary_entropy": 0.307934, "bpb": 1.151816}, "hi": {"chunks": 7437.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.442021, "cps": 58.101562, "mean_p": 0.200949, "boundary_entropy": 0.276665, "bpb": 0.963136}, "ta": {"chunks": 7816.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.41808, "cps": 61.0625, "mean_p": 0.331159, "boundary_entropy": 0.520657, "bpb": 0.825833}, "th": {"chunks": 7914.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.492011, "cps": 61.828125, "mean_p": 0.192813, "boundary_entropy": 0.257023, "bpb": 0.911884}, "zh": {"chunks": 4227.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.715713, "cps": 33.023438, "mean_p": 0.26697, "boundary_entropy": 0.085052, "bpb": 2.592714}, "my": {"chunks": 8538.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.450602, "cps": 66.703125, "mean_p": 0.343921, "boundary_entropy": 0.53668, "bpb": 0.914713}}, "bpb": {"en": 1.76436, "de": 1.91294, "vi": 1.5905, "tr": 1.82939, "fi": 1.98227, "ru": 1.16021, "el": 1.15182, "hi": 0.96314, "ta": 0.82583, "th": 0.91188, "zh": 2.59271, "my": 0.91471}, "gini_cpc": 0.261359, "gini_cps": 0.204347, "bpic": 5.935553, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9821368289368294, "vi": 1.255206898520181, "tr": 1.0949314371455852, "fi": 1.1133478981010179, "ru": 1.7035515627864708, "el": 1.659515551510282, "hi": 2.5822383429923277, "ta": 2.4423769812059652, "th": 2.874276854362258, "zh": 4.181117978235048, "my": 2.63236671902813}, "t": 1793.4}
{"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.76397, "de": 1.911, "vi": 1.6038, "tr": 1.82583, "fi": 1.99723, "ru": 1.15989, "el": 1.15114, "hi": 0.96898, "ta": 0.83013, "th": 0.91013, "zh": 2.60047, "my": 0.92213}, "bytes_per_chunk": {"en": 5.346, "de": 5.761, "vi": 6.083, "tr": 5.552, "fi": 5.262, "ru": 5.947, "el": 6.114, "hi": 5.87, "ta": 6.552, "th": 5.79, "zh": 3.743, "my": 6.256}, "bpb_mean": 1.47039, "t": 1799.9}
{"event": "train", "step": 7000, "phase": 2, "objective": "global", "lm_loss": 1.25017, "bpb": 1.80362, "chunk_loss": 0.030676, "F": 0.19511, "G": 0.2166, "chunks": 199.8, "bytes_per_chunk": 5.125, "boundary_entropy": 0.13996, "router_noise": 0.0, "grad_norm": 0.4087, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.345, "alpha": 0.03, "t": 1800.2}
{"event": "train", "step": 7100, "phase": 2, "objective": "global", "lm_loss": 1.15692, "bpb": 1.66908, "chunk_loss": 0.030572, "F": 0.19511, "G": 0.22292, "chunks": 199.8, "bytes_per_chunk": 5.125, "boundary_entropy": 0.16196, "router_noise": 0.0, "grad_norm": 0.3906, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.246, "alpha": 0.03, "t": 1824.7}
{"event": "train", "step": 7200, "phase": 2, "objective": "global", "lm_loss": 1.06627, "bpb": 1.5383, "chunk_loss": 0.03038, "F": 0.18541, "G": 0.21363, "chunks": 189.86, "bytes_per_chunk": 5.393, "boundary_entropy": 0.16708, "router_noise": 0.0, "grad_norm": 0.3897, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.248, "alpha": 0.03, "t": 1849.6}
{"event": "train", "step": 7300, "phase": 2, "objective": "global", "lm_loss": 1.14833, "bpb": 1.65669, "chunk_loss": 0.030522, "F": 0.1908, "G": 0.22023, "chunks": 195.38, "bytes_per_chunk": 5.241, "boundary_entropy": 0.16188, "router_noise": 0.0, "grad_norm": 0.3893, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 1873.8}
{"event": "train", "step": 7400, "phase": 2, "objective": "global", "lm_loss": 1.06949, "bpb": 1.54295, "chunk_loss": 0.030302, "F": 0.18153, "G": 0.21873, "chunks": 185.89, "bytes_per_chunk": 5.509, "boundary_entropy": 0.18671, "router_noise": 0.0, "grad_norm": 0.3752, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 1898.3}
{"event": "train", "step": 7500, "phase": 2, "objective": "global", "lm_loss": 1.08822, "bpb": 1.56997, "chunk_loss": 0.030416, "F": 0.18701, "G": 0.21653, "chunks": 191.5, "bytes_per_chunk": 5.347, "boundary_entropy": 0.16645, "router_noise": 0.0, "grad_norm": 0.3913, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 1922.5}
{"event": "train", "step": 7600, "phase": 2, "objective": "global", "lm_loss": 1.0747, "bpb": 1.55046, "chunk_loss": 0.030329, "F": 0.18451, "G": 0.2198, "chunks": 188.94, "bytes_per_chunk": 5.42, "boundary_entropy": 0.18168, "router_noise": 0.0, "grad_norm": 0.3784, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 1946.7}
{"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1971.0}
{"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.06168, "bpb": 1.53168, "chunk_loss": 0.03023, "F": 0.18289, "G": 0.20629, "chunks": 187.28, "bytes_per_chunk": 5.468, "boundary_entropy": 0.15445, "router_noise": 0.0, "grad_norm": 0.3804, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.245, "alpha": 0.03, "t": 1971.2}
{"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.12625, "bpb": 1.62484, "chunk_loss": 0.030393, "F": 0.18697, "G": 0.21079, "chunks": 191.45, "bytes_per_chunk": 5.349, "boundary_entropy": 0.14462, "router_noise": 0.0, "grad_norm": 0.3955, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 1995.3}
{"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.10669, "bpb": 1.59661, "chunk_loss": 0.030397, "F": 0.19002, "G": 0.21721, "chunks": 194.58, "bytes_per_chunk": 5.263, "boundary_entropy": 0.15916, "router_noise": 0.0, "grad_norm": 0.3885, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.249, "alpha": 0.03, "t": 2020.2}
{"event": "probe", "step": 8000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2974.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17089, "cps": 23.234375, "mean_p": 0.174588, "boundary_entropy": 0.072975, "bpb": 1.738376}, "de": {"chunks": 3494.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.170165, "cps": 27.296875, "mean_p": 0.174515, "boundary_entropy": 0.090066, "bpb": 1.881086}, "vi": {"chunks": 4039.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.219094, "cps": 31.554688, "mean_p": 0.227558, "boundary_entropy": 0.263486, "bpb": 1.558125}, "tr": {"chunks": 3328.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183837, "cps": 26.0, "mean_p": 0.176655, "boundary_entropy": 0.095707, "bpb": 1.792659}, "fi": {"chunks": 3548.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188613, "cps": 27.71875, "mean_p": 0.189172, "boundary_entropy": 0.080435, "bpb": 1.941745}, "ru": {"chunks": 5634.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.297309, "cps": 44.015625, "mean_p": 0.212901, "boundary_entropy": 0.233235, "bpb": 1.138443}, "el": {"chunks": 6217.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.302928, "cps": 48.570312, "mean_p": 0.243684, "boundary_entropy": 0.304355, "bpb": 1.124741}, "hi": {"chunks": 7388.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.439108, "cps": 57.71875, "mean_p": 0.197234, "boundary_entropy": 0.267417, "bpb": 0.937419}, "ta": {"chunks": 8770.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.469109, "cps": 68.515625, "mean_p": 0.343139, "boundary_entropy": 0.52038, "bpb": 0.803142}, "th": {"chunks": 8643.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.537333, "cps": 67.523438, "mean_p": 0.202306, "boundary_entropy": 0.246779, "bpb": 0.89025}, "zh": {"chunks": 4238.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.717575, "cps": 33.109375, "mean_p": 0.267637, "boundary_entropy": 0.077781, "bpb": 2.544831}, "my": {"chunks": 8384.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.442474, "cps": 65.5, "mean_p": 0.354361, "boundary_entropy": 0.547823, "bpb": 0.890914}}, "bpb": {"en": 1.73838, "de": 1.88109, "vi": 1.55813, "tr": 1.79266, "fi": 1.94175, "ru": 1.13844, "el": 1.12474, "hi": 0.93742, "ta": 0.80314, "th": 0.89025, "zh": 2.54483, "my": 0.89091}, "gini_cpc": 0.266835, "gini_cps": 0.21622, "bpic": 5.75755, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9957576451717741, "vi": 1.2820762857704116, "tr": 1.075761313660556, "fi": 1.1037097618875933, "ru": 1.739765780120765, "el": 1.7726507478397537, "hi": 2.569537557294472, "ta": 2.745094375017649, "th": 3.1443190272591455, "zh": 4.199046323356435, "my": 2.5892324996922875}, "t": 2048.6}
{"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.73855, "de": 1.87996, "vi": 1.57292, "tr": 1.78995, "fi": 1.958, "ru": 1.13759, "el": 1.124, "hi": 0.94344, "ta": 0.80825, "th": 0.88699, "zh": 2.55269, "my": 0.90066}, "bytes_per_chunk": {"en": 5.346, "de": 5.733, "vi": 5.983, "tr": 5.636, "fi": 5.33, "ru": 5.799, "el": 5.769, "hi": 5.867, "ta": 5.855, "th": 5.386, "zh": 3.738, "my": 6.421}, "bpb_mean": 1.44108, "t": 2055.9}
{"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.10253, "bpb": 1.59061, "chunk_loss": 0.030393, "F": 0.19051, "G": 0.21687, "chunks": 195.08, "bytes_per_chunk": 5.249, "boundary_entropy": 0.16182, "router_noise": 0.0, "grad_norm": 0.4123, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.359, "alpha": 0.03, "t": 2056.2}
{"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.09598, "bpb": 1.58117, "chunk_loss": 0.030296, "F": 0.18362, "G": 0.20949, "chunks": 188.03, "bytes_per_chunk": 5.446, "boundary_entropy": 0.16117, "router_noise": 0.0, "grad_norm": 0.4031, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 2080.4}
{"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.14611, "bpb": 1.65349, "chunk_loss": 0.030363, "F": 0.18895, "G": 0.21808, "chunks": 193.48, "bytes_per_chunk": 5.292, "boundary_entropy": 0.15516, "router_noise": 0.0, "grad_norm": 0.3978, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 2104.7}
{"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.07953, "bpb": 1.55744, "chunk_loss": 0.030269, "F": 0.18086, "G": 0.20754, "chunks": 185.2, "bytes_per_chunk": 5.529, "boundary_entropy": 0.15565, "router_noise": 0.0, "grad_norm": 0.4066, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.245, "alpha": 0.03, "t": 2129.2}
{"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.10136, "bpb": 1.58892, "chunk_loss": 0.030512, "F": 0.18813, "G": 0.21468, "chunks": 192.64, "bytes_per_chunk": 5.316, "boundary_entropy": 0.16086, "router_noise": 0.0, "grad_norm": 0.4119, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.247, "alpha": 0.03, "t": 2153.8}
{"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.08566, "bpb": 1.56627, "chunk_loss": 0.030526, "F": 0.19576, "G": 0.21973, "chunks": 200.45, "bytes_per_chunk": 5.108, "boundary_entropy": 0.15818, "router_noise": 0.0, "grad_norm": 0.4071, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 2177.8}
{"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.0555, "bpb": 1.52277, "chunk_loss": 0.030233, "F": 0.18147, "G": 0.21043, "chunks": 185.83, "bytes_per_chunk": 5.51, "boundary_entropy": 0.1668, "router_noise": 0.0, "grad_norm": 0.3794, "lr": 6.33e-05, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 2202.1}
{"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 1.07388, "bpb": 1.54928, "chunk_loss": 0.030552, "F": 0.19186, "G": 0.22098, "chunks": 196.47, "bytes_per_chunk": 5.212, "boundary_entropy": 0.17077, "router_noise": 0.0, "grad_norm": 0.4058, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.245, "alpha": 0.03, "t": 2226.6}
{"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.06911, "bpb": 1.5424, "chunk_loss": 0.030282, "F": 0.18462, "G": 0.22183, "chunks": 189.05, "bytes_per_chunk": 5.417, "boundary_entropy": 0.17565, "router_noise": 0.0, "grad_norm": 0.3796, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 2250.4}
{"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.05706, "bpb": 1.52501, "chunk_loss": 0.030307, "F": 0.18336, "G": 0.20841, "chunks": 187.77, "bytes_per_chunk": 5.454, "boundary_entropy": 0.15291, "router_noise": 0.0, "grad_norm": 0.3917, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 2274.8}
{"event": "probe", "step": 9000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2963.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.170258, "cps": 23.148438, "mean_p": 0.173023, "boundary_entropy": 0.070164, "bpb": 1.720493}, "de": {"chunks": 3484.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.169678, "cps": 27.21875, "mean_p": 0.17239, "boundary_entropy": 0.084829, "bpb": 1.856185}, "vi": {"chunks": 3939.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.21367, "cps": 30.773438, "mean_p": 0.217208, "boundary_entropy": 0.242572, "bpb": 1.531658}, "tr": {"chunks": 3333.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184113, "cps": 26.039062, "mean_p": 0.177821, "boundary_entropy": 0.091438, "bpb": 1.769063}, "fi": {"chunks": 3500.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186061, "cps": 27.34375, "mean_p": 0.18586, "boundary_entropy": 0.075586, "bpb": 1.909569}, "ru": {"chunks": 5567.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.293773, "cps": 43.492188, "mean_p": 0.210342, "boundary_entropy": 0.228604, "bpb": 1.119724}, "el": {"chunks": 6078.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.296156, "cps": 47.484375, "mean_p": 0.246253, "boundary_entropy": 0.311576, "bpb": 1.104011}, "hi": {"chunks": 7331.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.435721, "cps": 57.273438, "mean_p": 0.198387, "boundary_entropy": 0.260923, "bpb": 0.92014}, "ta": {"chunks": 8472.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.453169, "cps": 66.1875, "mean_p": 0.349693, "boundary_entropy": 0.528848, "bpb": 0.788139}, "th": {"chunks": 7981.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.496177, "cps": 62.351562, "mean_p": 0.192918, "boundary_entropy": 0.242753, "bpb": 0.874814}, "zh": {"chunks": 4242.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.718253, "cps": 33.140625, "mean_p": 0.267078, "boundary_entropy": 0.075819, "bpb": 2.51035}, "my": {"chunks": 8943.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.471976, "cps": 69.867188, "mean_p": 0.366892, "boundary_entropy": 0.548121, "bpb": 0.86297}}, "bpb": {"en": 1.72049, "de": 1.85618, "vi": 1.53166, "tr": 1.76906, "fi": 1.90957, "ru": 1.11972, "el": 1.10401, "hi": 0.92014, "ta": 0.78814, "th": 0.87481, "zh": 2.51035, "my": 0.86297}, "gini_cpc": 0.264667, "gini_cps": 0.215706, "bpic": 5.829614, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9965938616728184, "vi": 1.254975673666569, "tr": 1.0813772557224905, "fi": 1.0928199876157654, "ru": 1.7254583308473816, "el": 1.739451480800358, "hi": 2.5591787168039297, "ta": 2.6616622574378828, "th": 2.9142628109128745, "zh": 4.21861303520025, "my": 2.772121897320489}, "t": 2302.7}
{"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.71924, "de": 1.85534, "vi": 1.54694, "tr": 1.76593, "fi": 1.92697, "ru": 1.11944, "el": 1.10416, "hi": 0.92629, "ta": 0.79275, "th": 0.87143, "zh": 2.51819, "my": 0.87244}, "bytes_per_chunk": {"en": 5.346, "de": 5.744, "vi": 6.08, "tr": 5.62, "fi": 5.38, "ru": 5.895, "el": 5.908, "hi": 5.914, "ta": 6.091, "th": 5.744, "zh": 3.736, "my": 6.019}, "bpb_mean": 1.41826, "t": 2309.7}
{"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.10011, "bpb": 1.58713, "chunk_loss": 0.030479, "F": 0.19147, "G": 0.21576, "chunks": 196.06, "bytes_per_chunk": 5.223, "boundary_entropy": 0.14493, "router_noise": 0.0, "grad_norm": 0.4108, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.352, "alpha": 0.03, "t": 2310.0}
{"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.03793, "bpb": 1.49742, "chunk_loss": 0.03022, "F": 0.1814, "G": 0.20425, "chunks": 185.75, "bytes_per_chunk": 5.513, "boundary_entropy": 0.15359, "router_noise": 0.0, "grad_norm": 0.3858, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 2333.5}
{"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 1.12033, "bpb": 1.6163, "chunk_loss": 0.030427, "F": 0.18831, "G": 0.21146, "chunks": 192.83, "bytes_per_chunk": 5.31, "boundary_entropy": 0.14109, "router_noise": 0.0, "grad_norm": 0.435, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 2357.7}
{"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 0.98102, "bpb": 1.41531, "chunk_loss": 0.030339, "F": 0.18553, "G": 0.21822, "chunks": 189.98, "bytes_per_chunk": 5.39, "boundary_entropy": 0.18049, "router_noise": 0.0, "grad_norm": 0.3909, "lr": 4.71e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 2381.7}
{"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.10329, "bpb": 1.59171, "chunk_loss": 0.030772, "F": 0.19974, "G": 0.21386, "chunks": 204.53, "bytes_per_chunk": 5.007, "boundary_entropy": 0.13013, "router_noise": 0.0, "grad_norm": 0.395, "lr": 4.52e-05, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 2405.2}
{"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.03146, "bpb": 1.48808, "chunk_loss": 0.030355, "F": 0.18423, "G": 0.21381, "chunks": 188.66, "bytes_per_chunk": 5.428, "boundary_entropy": 0.16853, "router_noise": 0.0, "grad_norm": 0.3926, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 2429.1}
{"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 1.11083, "bpb": 1.60259, "chunk_loss": 0.030416, "F": 0.18742, "G": 0.20723, "chunks": 191.92, "bytes_per_chunk": 5.336, "boundary_entropy": 0.13355, "router_noise": 0.0, "grad_norm": 0.4184, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 2453.0}
{"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 1.08794, "bpb": 1.56957, "chunk_loss": 0.030398, "F": 0.18657, "G": 0.21373, "chunks": 191.05, "bytes_per_chunk": 5.36, "boundary_entropy": 0.15443, "router_noise": 0.0, "grad_norm": 0.4046, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 2476.5}
{"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 1.04804, "bpb": 1.51201, "chunk_loss": 0.030339, "F": 0.18509, "G": 0.21271, "chunks": 189.53, "bytes_per_chunk": 5.403, "boundary_entropy": 0.1617, "router_noise": 0.0, "grad_norm": 0.4067, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 2500.1}
{"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.05694, "bpb": 1.52484, "chunk_loss": 0.030398, "F": 0.18913, "G": 0.20983, "chunks": 193.67, "bytes_per_chunk": 5.287, "boundary_entropy": 0.14452, "router_noise": 0.0, "grad_norm": 0.3977, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 2523.2}
{"event": "probe", "step": 10000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2962.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.170201, "cps": 23.140625, "mean_p": 0.172065, "boundary_entropy": 0.067245, "bpb": 1.706849}, "de": {"chunks": 3446.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.167827, "cps": 26.921875, "mean_p": 0.171566, "boundary_entropy": 0.080251, "bpb": 1.840923}, "vi": {"chunks": 3914.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.212314, "cps": 30.578125, "mean_p": 0.213841, "boundary_entropy": 0.234274, "bpb": 1.514394}, "tr": {"chunks": 3342.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18461, "cps": 26.109375, "mean_p": 0.177088, "boundary_entropy": 0.087814, "bpb": 1.750858}, "fi": {"chunks": 3508.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186487, "cps": 27.40625, "mean_p": 0.185927, "boundary_entropy": 0.072418, "bpb": 1.891634}, "ru": {"chunks": 5429.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.286491, "cps": 42.414062, "mean_p": 0.204528, "boundary_entropy": 0.219013, "bpb": 1.107897}, "el": {"chunks": 5981.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.291429, "cps": 46.726562, "mean_p": 0.237278, "boundary_entropy": 0.296492, "bpb": 1.089762}, "hi": {"chunks": 7272.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.432214, "cps": 56.8125, "mean_p": 0.197115, "boundary_entropy": 0.255504, "bpb": 0.906316}, "ta": {"chunks": 8115.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.434073, "cps": 63.398438, "mean_p": 0.349687, "boundary_entropy": 0.530632, "bpb": 0.778247}, "th": {"chunks": 7992.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.49686, "cps": 62.4375, "mean_p": 0.190597, "boundary_entropy": 0.240735, "bpb": 0.862846}, "zh": {"chunks": 4241.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.718083, "cps": 33.132812, "mean_p": 0.266145, "boundary_entropy": 0.072899, "bpb": 2.492343}, "my": {"chunks": 8774.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.463057, "cps": 68.546875, "mean_p": 0.361002, "boundary_entropy": 0.549133, "bpb": 0.849126}}, "bpb": {"en": 1.70685, "de": 1.84092, "vi": 1.51439, "tr": 1.75086, "fi": 1.89163, "ru": 1.1079, "el": 1.08976, "hi": 0.90632, "ta": 0.77825, "th": 0.86285, "zh": 2.49234, "my": 0.84913}, "gini_cpc": 0.264704, "gini_cps": 0.211514, "bpic": 5.906504, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9860567990007554, "vi": 1.2474316113680803, "tr": 1.084663335613904, "fi": 1.095687651833292, "ru": 1.6832541479674825, "el": 1.7122691153677714, "hi": 2.539439457496792, "ta": 2.550363713753908, "th": 2.9192647013450026, "zh": 4.2190424574237895, "my": 2.7206540391935166}, "t": 2550.3}
{"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.70595, "de": 1.84126, "vi": 1.52951, "tr": 1.748, "fi": 1.90901, "ru": 1.10721, "el": 1.08924, "hi": 0.91231, "ta": 0.78304, "th": 0.85966, "zh": 2.50026, "my": 0.85814}, "bytes_per_chunk": {"en": 5.348, "de": 5.745, "vi": 6.115, "tr": 5.611, "fi": 5.363, "ru": 6.01, "el": 5.976, "hi": 5.898, "ta": 6.356, "th": 5.76, "zh": 3.74, "my": 6.117}, "bpb_mean": 1.40363, "t": 2557.2}
{"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.07244, "bpb": 1.54721, "chunk_loss": 0.030334, "F": 0.18571, "G": 0.21263, "chunks": 190.17, "bytes_per_chunk": 5.385, "boundary_entropy": 0.15272, "router_noise": 0.0, "grad_norm": 0.3917, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.342, "alpha": 0.03, "t": 2557.4}
{"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 1.06149, "bpb": 1.53141, "chunk_loss": 0.030472, "F": 0.19511, "G": 0.22419, "chunks": 199.8, "bytes_per_chunk": 5.125, "boundary_entropy": 0.15832, "router_noise": 0.0, "grad_norm": 0.4284, "lr": 3.49e-05, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 2580.8}
{"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.03903, "bpb": 1.49901, "chunk_loss": 0.030353, "F": 0.18475, "G": 0.21723, "chunks": 189.19, "bytes_per_chunk": 5.413, "boundary_entropy": 0.1665, "router_noise": 0.0, "grad_norm": 0.4369, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 2604.9}
{"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 0.98749, "bpb": 1.42464, "chunk_loss": 0.030347, "F": 0.18468, "G": 0.21957, "chunks": 189.11, "bytes_per_chunk": 5.415, "boundary_entropy": 0.17785, "router_noise": 0.0, "grad_norm": 0.3998, "lr": 3.3e-05, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 2629.3}
{"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.08237, "bpb": 1.56153, "chunk_loss": 0.0304, "F": 0.18953, "G": 0.22106, "chunks": 194.08, "bytes_per_chunk": 5.276, "boundary_entropy": 0.16073, "router_noise": 0.0, "grad_norm": 0.4104, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 2653.3}
{"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.0145, "bpb": 1.46361, "chunk_loss": 0.030538, "F": 0.19093, "G": 0.22018, "chunks": 195.52, "bytes_per_chunk": 5.237, "boundary_entropy": 0.16967, "router_noise": 0.0, "grad_norm": 0.4079, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.245, "alpha": 0.03, "t": 2677.8}
{"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 0.97775, "bpb": 1.41059, "chunk_loss": 0.030485, "F": 0.19182, "G": 0.22026, "chunks": 196.42, "bytes_per_chunk": 5.213, "boundary_entropy": 0.1716, "router_noise": 0.0, "grad_norm": 0.4041, "lr": 3.1e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 2701.9}
{"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.05147, "bpb": 1.51695, "chunk_loss": 0.030451, "F": 0.19046, "G": 0.2206, "chunks": 195.03, "bytes_per_chunk": 5.25, "boundary_entropy": 0.16605, "router_noise": 0.0, "grad_norm": 0.4076, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.248, "alpha": 0.03, "t": 2726.7}
{"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.02158, "bpb": 1.47383, "chunk_loss": 0.030229, "F": 0.18394, "G": 0.20439, "chunks": 188.36, "bytes_per_chunk": 5.436, "boundary_entropy": 0.13311, "router_noise": 0.0, "grad_norm": 0.4135, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 2750.6}
{"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.03733, "bpb": 1.49655, "chunk_loss": 0.030331, "F": 0.18652, "G": 0.22417, "chunks": 191.0, "bytes_per_chunk": 5.361, "boundary_entropy": 0.17818, "router_noise": 0.0, "grad_norm": 0.4239, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 2774.0}
{"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2967.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.170488, "cps": 23.179688, "mean_p": 0.172415, "boundary_entropy": 0.065801, "bpb": 1.698245}, "de": {"chunks": 3461.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.168558, "cps": 27.039062, "mean_p": 0.171921, "boundary_entropy": 0.079009, "bpb": 1.832111}, "vi": {"chunks": 3891.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211066, "cps": 30.398438, "mean_p": 0.211948, "boundary_entropy": 0.227373, "bpb": 1.505704}, "tr": {"chunks": 3367.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185991, "cps": 26.304688, "mean_p": 0.178805, "boundary_entropy": 0.085726, "bpb": 1.740017}, "fi": {"chunks": 3522.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187231, "cps": 27.515625, "mean_p": 0.186393, "boundary_entropy": 0.07051, "bpb": 1.87719}, "ru": {"chunks": 5534.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.292032, "cps": 43.234375, "mean_p": 0.207003, "boundary_entropy": 0.217838, "bpb": 1.100804}, "el": {"chunks": 5907.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.287823, "cps": 46.148438, "mean_p": 0.234161, "boundary_entropy": 0.291897, "bpb": 1.082956}, "hi": {"chunks": 7200.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.427935, "cps": 56.25, "mean_p": 0.194745, "boundary_entropy": 0.257202, "bpb": 0.902268}, "ta": {"chunks": 8621.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.461139, "cps": 67.351562, "mean_p": 0.35492, "boundary_entropy": 0.533629, "bpb": 0.771619}, "th": {"chunks": 7946.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.494001, "cps": 62.078125, "mean_p": 0.191691, "boundary_entropy": 0.243071, "bpb": 0.85675}, "zh": {"chunks": 4233.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.716729, "cps": 33.070312, "mean_p": 0.265604, "boundary_entropy": 0.071549, "bpb": 2.476444}, "my": {"chunks": 8538.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.450602, "cps": 66.703125, "mean_p": 0.350283, "boundary_entropy": 0.53808, "bpb": 0.840514}}, "bpb": {"en": 1.69825, "de": 1.83211, "vi": 1.5057, "tr": 1.74002, "fi": 1.87719, "ru": 1.1008, "el": 1.08296, "hi": 0.90227, "ta": 0.77162, "th": 0.85675, "zh": 2.47644, "my": 0.84051}, "gini_cpc": 0.263947, "gini_cps": 0.21204, "bpic": 5.887386, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9886800389439436, "vi": 1.2380114539017886, "tr": 1.0909356621114308, "fi": 1.098206581470655, "ru": 1.7129177682742356, "el": 1.6882342221090376, "hi": 2.510059390291723, "ta": 2.704822369412021, "th": 2.8975708862140053, "zh": 4.203987341967192, "my": 2.643013298275969}, "t": 2802.1}
{"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.69764, "de": 1.83233, "vi": 1.52067, "tr": 1.73729, "fi": 1.89431, "ru": 1.10067, "el": 1.08256, "hi": 0.90806, "ta": 0.77664, "th": 0.85389, "zh": 2.48307, "my": 0.84866}, "bytes_per_chunk": {"en": 5.344, "de": 5.726, "vi": 6.167, "tr": 5.579, "fi": 5.331, "ru": 5.928, "el": 6.062, "hi": 6.022, "ta": 5.951, "th": 5.825, "zh": 3.742, "my": 6.263}, "bpb_mean": 1.39465, "t": 2808.7}
{"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2967.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.170488, "cps": 23.179688, "mean_p": 0.172415, "boundary_entropy": 0.065801, "bpb": 1.698245}, "de": {"chunks": 3461.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.168558, "cps": 27.039062, "mean_p": 0.171921, "boundary_entropy": 0.079009, "bpb": 1.832111}, "vi": {"chunks": 3891.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211066, "cps": 30.398438, "mean_p": 0.211948, "boundary_entropy": 0.227373, "bpb": 1.505704}, "tr": {"chunks": 3367.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185991, "cps": 26.304688, "mean_p": 0.178805, "boundary_entropy": 0.085726, "bpb": 1.740017}, "fi": {"chunks": 3522.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187231, "cps": 27.515625, "mean_p": 0.186393, "boundary_entropy": 0.07051, "bpb": 1.87719}, "ru": {"chunks": 5534.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.292032, "cps": 43.234375, "mean_p": 0.207003, "boundary_entropy": 0.217838, "bpb": 1.100804}, "el": {"chunks": 5907.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.287823, "cps": 46.148438, "mean_p": 0.234161, "boundary_entropy": 0.291897, "bpb": 1.082956}, "hi": {"chunks": 7200.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.427935, "cps": 56.25, "mean_p": 0.194745, "boundary_entropy": 0.257202, "bpb": 0.902268}, "ta": {"chunks": 8621.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.461139, "cps": 67.351562, "mean_p": 0.35492, "boundary_entropy": 0.533629, "bpb": 0.771619}, "th": {"chunks": 7946.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.494001, "cps": 62.078125, "mean_p": 0.191691, "boundary_entropy": 0.243071, "bpb": 0.85675}, "zh": {"chunks": 4233.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.716729, "cps": 33.070312, "mean_p": 0.265604, "boundary_entropy": 0.071549, "bpb": 2.476444}, "my": {"chunks": 8538.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.450602, "cps": 66.703125, "mean_p": 0.350283, "boundary_entropy": 0.53808, "bpb": 0.840514}}, "bpb": {"en": 1.69825, "de": 1.83211, "vi": 1.5057, "tr": 1.74002, "fi": 1.87719, "ru": 1.1008, "el": 1.08296, "hi": 0.90227, "ta": 0.77162, "th": 0.85675, "zh": 2.47644, "my": 0.84051}, "gini_cpc": 0.263947, "gini_cps": 0.21204, "bpic": 5.887386, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9886800389439436, "vi": 1.2380114539017886, "tr": 1.0909356621114308, "fi": 1.098206581470655, "ru": 1.7129177682742356, "el": 1.6882342221090376, "hi": 2.510059390291723, "ta": 2.704822369412021, "th": 2.8975708862140053, "zh": 4.203987341967192, "my": 2.643013298275969}, "t": 2812.5}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s2/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s2/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s2/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s2/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s2/th0.70"], "t": 2816.8}
{"event": "final", "run": "exp32_local_s2", "condition": "A", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 0.17048784692294433, "cps": 23.1796875, "bpb": 1.6931716484971258, "bytes_per_chunk": 5.338519578244246}, "de": {"cpc": 0.1685579311352457, "cps": 27.0390625, "bpb": 1.8315910466286345, "bytes_per_chunk": 5.687531188301404}, "vi": {"cpc": 0.21106590724165988, "cps": 30.3984375, "bpb": 1.5223879424380602, "bytes_per_chunk": 6.167332270719499}, "tr": {"cpc": 0.18599127216483455, "cps": 26.3046875, "bpb": 1.7367852878730408, "bytes_per_chunk": 5.575693122480884}, "fi": {"cpc": 0.187230875551539, "cps": 27.515625, "bpb": 1.899413071429155, "bytes_per_chunk": 5.327128537825714}, "ru": {"cpc": 0.2920316622691293, "cps": 43.234375, "bpb": 1.101538844663503, "bytes_per_chunk": 5.922518624787489}, "el": {"cpc": 0.2878234176290016, "cps": 46.1484375, "bpb": 1.079825282567976, "bytes_per_chunk": 6.0586114449477675}, "hi": {"cpc": 0.4279346210995542, "cps": 56.25, "bpb": 0.9105429205574486, "bytes_per_chunk": 6.009307016940605}, "ta": {"cpc": 0.46113934207007223, "cps": 67.3515625, "bpb": 0.7750441210539766, "bytes_per_chunk": 5.947342748567864}, "th": {"cpc": 0.49400062169723347, "cps": 62.078125, "bpb": 0.8569683953833194, "bytes_per_chunk": 5.815310044200917}, "zh": {"cpc": 0.7167287504232983, "cps": 33.0703125, "bpb": 2.4890658228800207, "bytes_per_chunk": 3.7616266555697457}, "my": {"cpc": 0.4506016466117796, "cps": 66.703125, "bpb": 0.8467113339624879, "bytes_per_chunk": 6.260865411598927}}, "gini_cpc": 0.2639468220288741, "gini_cps": 0.2120399772960868, "bpic": 5.887385521653091, "seed": 2, "mean_boundary_entropy": 0.14645597338676453, "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": 2830.4}