| {"event": "start", "run": "exp32_local_onechunk_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": 3.1} |
| {"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": 3.1} |
| {"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": 8.082365}, "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": 8.12759}, "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": 8.063676}, "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": 8.093682}, "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": 8.104954}, "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": 8.241678}, "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": 7.978207}, "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": 7.864654}, "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": 7.649439}, "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": 7.880325}, "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": 8.098555}, "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": 8.237587}}, "bpb": {"en": 8.08237, "de": 8.12759, "vi": 8.06368, "tr": 8.09368, "fi": 8.10495, "ru": 8.24168, "el": 7.97821, "hi": 7.86465, "ta": 7.64944, "th": 7.88033, "zh": 8.09856, "my": 8.23759}, "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": 6.7} |
| {"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.60972, "bpb": 8.09311, "chunk_loss": 0.017966, "F": 0.00098, "G": 0.50291, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.69167, "router_noise": 0.0, "grad_norm": 5.2915, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.007, "alpha": 0.03, "t": 7.7} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10331.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.593633, "cps": 80.710938, "mean_p": 0.50688, "boundary_entropy": 0.687329, "bpb": 8.06406}, "de": {"chunks": 11370.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.553743, "cps": 88.828125, "mean_p": 0.504619, "boundary_entropy": 0.688262, "bpb": 8.110691}, "vi": {"chunks": 6012.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.326119, "cps": 46.96875, "mean_p": 0.490468, "boundary_entropy": 0.68852, "bpb": 8.051765}, "tr": {"chunks": 6608.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.365022, "cps": 51.625, "mean_p": 0.496346, "boundary_entropy": 0.687895, "bpb": 8.079979}, "fi": {"chunks": 2169.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.115305, "cps": 16.945312, "mean_p": 0.479777, "boundary_entropy": 0.686772, "bpb": 8.087438}, "ru": {"chunks": 24063.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.269815, "cps": 187.992188, "mean_p": 0.507583, "boundary_entropy": 0.69018, "bpb": 8.212877}, "el": {"chunks": 1453.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.070799, "cps": 11.351562, "mean_p": 0.479482, "boundary_entropy": 0.689439, "bpb": 7.967357}, "hi": {"chunks": 8027.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.477088, "cps": 62.710938, "mean_p": 0.494201, "boundary_entropy": 0.690836, "bpb": 7.855505}, "ta": {"chunks": 358.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.01915, "cps": 2.796875, "mean_p": 0.475443, "boundary_entropy": 0.689912, "bpb": 7.641634}, "th": {"chunks": 28157.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.750513, "cps": 219.976562, "mean_p": 0.503355, "boundary_entropy": 0.691078, "bpb": 7.862766}, "zh": {"chunks": 14705.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.489841, "cps": 114.882812, "mean_p": 0.531851, "boundary_entropy": 0.685228, "bpb": 8.096862}, "my": {"chunks": 786.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.041482, "cps": 6.140625, "mean_p": 0.475106, "boundary_entropy": 0.690001, "bpb": 8.241419}}, "bpb": {"en": 8.06406, "de": 8.11069, "vi": 8.05177, "tr": 8.07998, "fi": 8.08744, "ru": 8.21288, "el": 7.96736, "hi": 7.8555, "ta": 7.64163, "th": 7.86277, "zh": 8.09686, "my": 8.24142}, "gini_cpc": 0.558878, "gini_cps": 0.488449, "bpic": 3.365349, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9328027465960712, "vi": 0.5493607010711913, "tr": 0.6148953963422921, "fi": 0.19423586644771376, "ru": 2.139056792720324, "el": 0.11926321919440926, "hi": 0.8036740559112975, "ta": 0.03225813950845365, "th": 2.9488119255141605, "zh": 4.194240648093647, "my": 0.06987807437038583}, "t": 10.4} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10402.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.597713, "cps": 81.265625, "mean_p": 0.507078, "boundary_entropy": 0.687327, "bpb": 8.029064}, "de": {"chunks": 11495.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.559831, "cps": 89.804688, "mean_p": 0.504888, "boundary_entropy": 0.688261, "bpb": 8.079148}, "vi": {"chunks": 6107.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.331272, "cps": 47.710938, "mean_p": 0.490668, "boundary_entropy": 0.688531, "bpb": 8.028907}, "tr": {"chunks": 6677.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.368834, "cps": 52.164062, "mean_p": 0.496487, "boundary_entropy": 0.6879, "bpb": 8.055864}, "fi": {"chunks": 2191.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.116474, "cps": 17.117188, "mean_p": 0.479995, "boundary_entropy": 0.686794, "bpb": 8.057345}, "ru": {"chunks": 24530.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.294459, "cps": 191.640625, "mean_p": 0.508027, "boundary_entropy": 0.69017, "bpb": 8.150727}, "el": {"chunks": 1462.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.071237, "cps": 11.421875, "mean_p": 0.479555, "boundary_entropy": 0.689445, "bpb": 7.946318}, "hi": {"chunks": 8155.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.484695, "cps": 63.710938, "mean_p": 0.494313, "boundary_entropy": 0.690839, "bpb": 7.839139}, "ta": {"chunks": 361.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.01931, "cps": 2.820312, "mean_p": 0.47556, "boundary_entropy": 0.689923, "bpb": 7.616892}, "th": {"chunks": 28596.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.777805, "cps": 223.40625, "mean_p": 0.503506, "boundary_entropy": 0.691076, "bpb": 7.835476}, "zh": {"chunks": 14716.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.491703, "cps": 114.96875, "mean_p": 0.531905, "boundary_entropy": 0.685222, "bpb": 8.092958}, "my": {"chunks": 787.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.041535, "cps": 6.148438, "mean_p": 0.475114, "boundary_entropy": 0.690002, "bpb": 8.240585}}, "bpb": {"en": 8.02906, "de": 8.07915, "vi": 8.02891, "tr": 8.05586, "fi": 8.05734, "ru": 8.15073, "el": 7.94632, "hi": 7.83914, "ta": 7.61689, "th": 7.83548, "zh": 8.09296, "my": 8.24058}, "gini_cpc": 0.557878, "gini_cps": 0.489218, "bpic": 3.323383, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.93662088854429, "vi": 0.5542325762289251, "tr": 0.6170752041478389, "fi": 0.1948667602566094, "ru": 2.1656865764093465, "el": 0.11918285872689129, "hi": 0.8109165485092162, "ta": 0.03230643252251724, "th": 2.9743460419874537, "zh": 4.168728460291799, "my": 0.06948941043758453}, "t": 13.9} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10688.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.614147, "cps": 83.5, "mean_p": 0.507876, "boundary_entropy": 0.687315, "bpb": 7.900827}, "de": {"chunks": 12059.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.587298, "cps": 94.210938, "mean_p": 0.505943, "boundary_entropy": 0.688252, "bpb": 7.962293}, "vi": {"chunks": 6464.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.350637, "cps": 50.5, "mean_p": 0.491462, "boundary_entropy": 0.688573, "bpb": 7.946924}, "tr": {"chunks": 6945.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.383638, "cps": 54.257812, "mean_p": 0.49706, "boundary_entropy": 0.687916, "bpb": 7.96671}, "fi": {"chunks": 2321.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.123385, "cps": 18.132812, "mean_p": 0.480849, "boundary_entropy": 0.686878, "bpb": 7.94604}, "ru": {"chunks": 25630.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.352507, "cps": 200.234375, "mean_p": 0.50917, "boundary_entropy": 0.690141, "bpb": 7.972678}, "el": {"chunks": 1512.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.073673, "cps": 11.8125, "mean_p": 0.479774, "boundary_entropy": 0.689464, "bpb": 7.879969}, "hi": {"chunks": 8806.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.523388, "cps": 68.796875, "mean_p": 0.494799, "boundary_entropy": 0.690851, "bpb": 7.767137}, "ta": {"chunks": 375.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.020059, "cps": 2.929688, "mean_p": 0.475989, "boundary_entropy": 0.689967, "bpb": 7.542837}, "th": {"chunks": 30185.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.876593, "cps": 235.820312, "mean_p": 0.504076, "boundary_entropy": 0.691071, "bpb": 7.733191}, "zh": {"chunks": 14750.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.49746, "cps": 115.234375, "mean_p": 0.532116, "boundary_entropy": 0.685201, "bpb": 8.081916}, "my": {"chunks": 800.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.042221, "cps": 6.25, "mean_p": 0.475352, "boundary_entropy": 0.690026, "bpb": 8.218595}}, "bpb": {"en": 7.90083, "de": 7.96229, "vi": 7.94692, "tr": 7.96671, "fi": 7.94604, "ru": 7.97268, "el": 7.87997, "hi": 7.76714, "ta": 7.54284, "th": 7.73319, "zh": 8.08192, "my": 8.2186}, "gini_cpc": 0.552754, "gini_cps": 0.488979, "bpic": 3.18398, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9562832812799376, "vi": 0.5709339660792329, "tr": 0.6246681652576193, "fi": 0.2009050863904003, "ru": 2.2022522731581686, "el": 0.11996059773451197, "hi": 0.8522191389879973, "ta": 0.032661300493423424, "th": 3.0556090666816815, "zh": 4.06655127562349, "my": 0.06874708467801571}, "t": 17.7} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 11642.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.668965, "cps": 90.953125, "mean_p": 0.510565, "boundary_entropy": 0.687264, "bpb": 7.474033}, "de": {"chunks": 13743.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.669313, "cps": 107.367188, "mean_p": 0.50952, "boundary_entropy": 0.688192, "bpb": 7.571513}, "vi": {"chunks": 7716.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.418552, "cps": 60.28125, "mean_p": 0.494313, "boundary_entropy": 0.688706, "bpb": 7.675218}, "tr": {"chunks": 7868.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.434624, "cps": 61.46875, "mean_p": 0.499055, "boundary_entropy": 0.687967, "bpb": 7.666572}, "fi": {"chunks": 2891.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.153687, "cps": 22.585938, "mean_p": 0.48406, "boundary_entropy": 0.687169, "bpb": 7.573027}, "ru": {"chunks": 29090.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.535092, "cps": 227.265625, "mean_p": 0.51311, "boundary_entropy": 0.690001, "bpb": 7.397121}, "el": {"chunks": 1804.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.087901, "cps": 14.09375, "mean_p": 0.480797, "boundary_entropy": 0.689551, "bpb": 7.669829}, "hi": {"chunks": 12133.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.721129, "cps": 94.789062, "mean_p": 0.497074, "boundary_entropy": 0.690898, "bpb": 7.497598}, "ta": {"chunks": 452.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.024178, "cps": 3.53125, "mean_p": 0.47812, "boundary_entropy": 0.690175, "bpb": 7.249286}, "th": {"chunks": 35239.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.190799, "cps": 275.304688, "mean_p": 0.506232, "boundary_entropy": 0.691038, "bpb": 7.397427}, "zh": {"chunks": 14851.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.514561, "cps": 116.023438, "mean_p": 0.532759, "boundary_entropy": 0.685135, "bpb": 8.05105}, "my": {"chunks": 880.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.046443, "cps": 6.875, "mean_p": 0.476395, "boundary_entropy": 0.690131, "bpb": 8.129238}}, "bpb": {"en": 7.47403, "de": 7.57151, "vi": 7.67522, "tr": 7.66657, "fi": 7.57303, "ru": 7.39712, "el": 7.66983, "hi": 7.4976, "ta": 7.24929, "th": 7.39743, "zh": 8.05105, "my": 8.12924}, "gini_cpc": 0.53395, "gini_cps": 0.481507, "bpic": 2.774809, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0005197469234959, "vi": 0.6256703898471473, "tr": 0.6496961976621153, "fi": 0.22973794578683676, "ru": 2.2947270346334965, "el": 0.13139904635844474, "hi": 1.0779773852575076, "ta": 0.03614177605509173, "th": 3.2749074836871226, "zh": 3.758882764059482, "my": 0.06942498927525298}, "t": 22.3} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14600.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.838936, "cps": 114.0625, "mean_p": 0.520349, "boundary_entropy": 0.686881, "bpb": 6.608184}, "de": {"chunks": 18178.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.885307, "cps": 142.015625, "mean_p": 0.52204, "boundary_entropy": 0.687602, "bpb": 6.751516}, "vi": {"chunks": 13415.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.727692, "cps": 104.804688, "mean_p": 0.50482, "boundary_entropy": 0.688938, "bpb": 7.101362}, "tr": {"chunks": 11946.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.659891, "cps": 93.328125, "mean_p": 0.507463, "boundary_entropy": 0.688019, "bpb": 6.995438}, "fi": {"chunks": 6698.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.356068, "cps": 52.328125, "mean_p": 0.49683, "boundary_entropy": 0.687923, "bpb": 6.789606}, "ru": {"chunks": 33384.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.761689, "cps": 260.8125, "mean_p": 0.524624, "boundary_entropy": 0.689243, "bpb": 6.487485}, "el": {"chunks": 4096.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.199581, "cps": 32.0, "mean_p": 0.486108, "boundary_entropy": 0.689932, "bpb": 7.111682}, "hi": {"chunks": 33888.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.014146, "cps": 264.75, "mean_p": 0.508104, "boundary_entropy": 0.690834, "bpb": 6.841785}, "ta": {"chunks": 3035.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.162343, "cps": 23.710938, "mean_p": 0.488218, "boundary_entropy": 0.690917, "bpb": 6.708469}, "th": {"chunks": 43043.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.675971, "cps": 336.273438, "mean_p": 0.513623, "boundary_entropy": 0.690791, "bpb": 6.567366}, "zh": {"chunks": 15115.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.559262, "cps": 118.085938, "mean_p": 0.534864, "boundary_entropy": 0.684915, "bpb": 8.007759}, "my": {"chunks": 1520.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.08022, "cps": 11.875, "mean_p": 0.481411, "boundary_entropy": 0.690572, "bpb": 7.70341}}, "bpb": {"en": 6.60818, "de": 6.75152, "vi": 7.10136, "tr": 6.99544, "fi": 6.78961, "ru": 6.48749, "el": 7.11168, "hi": 6.84179, "ta": 6.70847, "th": 6.56737, "zh": 8.00776, "my": 7.70341}, "gini_cpc": 0.455605, "gini_cps": 0.422091, "bpic": 1.929343, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0552733154581098, "vi": 0.8673987650055174, "tr": 0.7865805864312204, "fi": 0.42442848581018244, "ru": 2.0999087432681534, "el": 0.23789776789021871, "hi": 2.4008339840013027, "ta": 0.19351048005656776, "th": 3.1897212539548034, "zh": 3.050604968710714, "my": 0.09562060260091786}, "t": 28.2} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17313.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.994828, "cps": 135.257812, "mean_p": 0.571187, "boundary_entropy": 0.678293, "bpb": 5.948925}, "de": {"chunks": 20762.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.011153, "cps": 162.203125, "mean_p": 0.573868, "boundary_entropy": 0.678249, "bpb": 6.115675}, "vi": {"chunks": 23947.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.298996, "cps": 187.085938, "mean_p": 0.553782, "boundary_entropy": 0.683636, "bpb": 6.852539}, "tr": {"chunks": 19511.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.077777, "cps": 152.429688, "mean_p": 0.558574, "boundary_entropy": 0.681855, "bpb": 6.425496}, "fi": {"chunks": 19361.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.029238, "cps": 151.257812, "mean_p": 0.557586, "boundary_entropy": 0.682083, "bpb": 6.109033}, "ru": {"chunks": 34552.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823325, "cps": 269.9375, "mean_p": 0.56965, "boundary_entropy": 0.681092, "bpb": 5.735624}, "el": {"chunks": 23418.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.141061, "cps": 182.953125, "mean_p": 0.505995, "boundary_entropy": 0.690414, "bpb": 6.423537}, "hi": {"chunks": 43121.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.562912, "cps": 336.882812, "mean_p": 0.544128, "boundary_entropy": 0.687282, "bpb": 5.994162}, "ta": {"chunks": 50715.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.712757, "cps": 396.210938, "mean_p": 0.525126, "boundary_entropy": 0.690157, "bpb": 5.760057}, "th": {"chunks": 46042.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.862418, "cps": 359.703125, "mean_p": 0.531217, "boundary_entropy": 0.68936, "bpb": 5.525597}, "zh": {"chunks": 15750.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.66678, "cps": 123.046875, "mean_p": 0.546313, "boundary_entropy": 0.683256, "bpb": 7.660289}, "my": {"chunks": 25328.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.336711, "cps": 197.875, "mean_p": 0.501827, "boundary_entropy": 0.691326, "bpb": 6.48999}}, "bpb": {"en": 5.94893, "de": 6.11568, "vi": 6.85254, "tr": 6.4255, "fi": 6.10903, "ru": 5.73562, "el": 6.42354, "hi": 5.99416, "ta": 5.76006, "th": 5.5256, "zh": 7.66029, "my": 6.48999}, "gini_cpc": 0.230818, "gini_cps": 0.220442, "bpic": 1.129366, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0164091609448602, "vi": 1.305749184932171, "tr": 1.0833798870903844, "fi": 1.0345886095967196, "ru": 1.8328029191540451, "el": 1.146992947792458, "hi": 2.5762353912308895, "ta": 2.7268594357461144, "th": 2.877298414725918, "zh": 2.680642548543833, "my": 1.3436597647596078}, "t": 37.4} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17415.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.00069, "cps": 136.054688, "mean_p": 0.789112, "boundary_entropy": 0.511313, "bpb": 5.493525}, "de": {"chunks": 20857.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.015779, "cps": 162.945312, "mean_p": 0.787427, "boundary_entropy": 0.513464, "bpb": 5.642568}, "vi": {"chunks": 24249.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.315378, "cps": 189.445312, "mean_p": 0.756747, "boundary_entropy": 0.549325, "bpb": 6.384471}, "tr": {"chunks": 19720.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.089322, "cps": 154.0625, "mean_p": 0.777315, "boundary_entropy": 0.525601, "bpb": 6.022082}, "fi": {"chunks": 19564.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.04003, "cps": 152.84375, "mean_p": 0.778319, "boundary_entropy": 0.524292, "bpb": 5.652083}, "ru": {"chunks": 34556.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823536, "cps": 269.96875, "mean_p": 0.668856, "boundary_entropy": 0.633338, "bpb": 4.97095}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.573427, "boundary_entropy": 0.680229, "bpb": 5.326794}, "hi": {"chunks": 43161.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.56529, "cps": 337.195312, "mean_p": 0.656315, "boundary_entropy": 0.641897, "bpb": 5.053241}, "ta": {"chunks": 50905.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.722921, "cps": 397.695312, "mean_p": 0.64735, "boundary_entropy": 0.647802, "bpb": 4.840288}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.644251, "boundary_entropy": 0.649495, "bpb": 4.644002}, "zh": {"chunks": 16168.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.737555, "cps": 126.3125, "mean_p": 0.650786, "boundary_entropy": 0.64168, "bpb": 6.889569}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.564947, "boundary_entropy": 0.683115, "bpb": 4.814937}}, "bpb": {"en": 5.49353, "de": 5.64257, "vi": 6.38447, "tr": 6.02208, "fi": 5.65208, "ru": 4.97095, "el": 5.32679, "hi": 5.05324, "ta": 4.84029, "th": 4.644, "zh": 6.88957, "my": 4.81494}, "gini_cpc": 0.221594, "gini_cps": 0.23394, "bpic": 1.000245, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.015079542760791, "vi": 1.3144719802588745, "tr": 1.0885716021675755, "fi": 1.0393131257019654, "ru": 1.822279092493733, "el": 1.8144678080763403, "hi": 2.56352210588556, "ta": 2.7210443082078952, "th": 2.8631174769480716, "zh": 2.735668685956529, "my": 2.834387153598852}, "t": 52.7} |
| {"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.22996, "bpb": 4.65985, "chunk_loss": 0.008736, "F": 0.00098, "G": 0.76083, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.52761, "router_noise": 0.0, "grad_norm": 2.2966, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 0.597, "alpha": 0.03, "t": 67.4} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.874422, "boundary_entropy": 0.372272, "bpb": 4.372731}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.866228, "boundary_entropy": 0.385043, "bpb": 4.569438}, "vi": {"chunks": 24256.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.315758, "cps": 189.5, "mean_p": 0.74091, "boundary_entropy": 0.558975, "bpb": 4.743176}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.816928, "boundary_entropy": 0.460395, "bpb": 4.828585}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.855091, "boundary_entropy": 0.401896, "bpb": 4.574809}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.725397, "boundary_entropy": 0.578817, "bpb": 3.225809}, "el": {"chunks": 37256.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815329, "cps": 291.0625, "mean_p": 0.67721, "boundary_entropy": 0.622398, "bpb": 3.637341}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.693742, "boundary_entropy": 0.611402, "bpb": 2.973934}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.721022, "boundary_entropy": 0.585941, "bpb": 2.933417}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.695045, "boundary_entropy": 0.61312, "bpb": 2.771121}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.87285, "boundary_entropy": 0.37563, "bpb": 6.228336}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.732912, "boundary_entropy": 0.577487, "bpb": 3.033359}}, "bpb": {"en": 4.37273, "de": 4.56944, "vi": 4.74318, "tr": 4.82858, "fi": 4.57481, "ru": 3.22581, "el": 3.63734, "hi": 2.97393, "ta": 2.93342, "th": 2.77112, "zh": 6.22834, "my": 3.03336}, "gini_cpc": 0.221541, "gini_cps": 0.233868, "bpic": 1.000068, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3146249668678838, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8137658212258567, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 80.6} |
| {"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.54909, "bpb": 3.67756, "chunk_loss": 0.003741, "F": 0.00098, "G": 0.90039, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.30308, "router_noise": 0.0, "grad_norm": 2.3719, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.415, "alpha": 0.03, "t": 108.9} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.960886, "boundary_entropy": 0.160693, "bpb": 3.566096}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.958238, "boundary_entropy": 0.162531, "bpb": 3.835583}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.935162, "boundary_entropy": 0.228944, "bpb": 3.122082}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.938256, "boundary_entropy": 0.203085, "bpb": 3.814288}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.960837, "boundary_entropy": 0.153146, "bpb": 3.863931}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.977107, "boundary_entropy": 0.10256, "bpb": 2.661046}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.943675, "boundary_entropy": 0.209799, "bpb": 2.740596}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.962379, "boundary_entropy": 0.159288, "bpb": 2.107798}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.95717, "boundary_entropy": 0.175554, "bpb": 1.960569}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.971044, "boundary_entropy": 0.129638, "bpb": 2.101646}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.943799, "boundary_entropy": 0.21011, "bpb": 4.853319}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.942012, "boundary_entropy": 0.219047, "bpb": 2.18629}}, "bpb": {"en": 3.5661, "de": 3.83558, "vi": 3.12208, "tr": 3.81429, "fi": 3.86393, "ru": 2.66105, "el": 2.7406, "hi": 2.1078, "ta": 1.96057, "th": 2.10165, "zh": 4.85332, "my": 2.18629}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 132.3} |
| {"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.15399, "bpb": 3.10755, "chunk_loss": 0.001012, "F": 0.00098, "G": 0.97663, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.10613, "router_noise": 0.0, "grad_norm": 1.5651, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.4, "alpha": 0.03, "t": 148.8} |
| {"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.06981, "bpb": 2.98611, "chunk_loss": 0.000427, "F": 0.00098, "G": 0.99299, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.04069, "router_noise": 0.0, "grad_norm": 1.8449, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.341, "alpha": 0.03, "t": 182.9} |
| {"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.826, "bpb": 2.63437, "chunk_loss": 0.000275, "F": 0.00098, "G": 0.99724, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.01871, "router_noise": 0.0, "grad_norm": 1.027, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.322, "alpha": 0.03, "t": 215.1} |
| {"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.80724, "bpb": 2.60729, "chunk_loss": 0.00023, "F": 0.00098, "G": 0.99848, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.0112, "router_noise": 0.0, "grad_norm": 1.0211, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.303, "alpha": 0.03, "t": 245.4} |
| {"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.6979, "bpb": 2.44955, "chunk_loss": 0.00021, "F": 0.00098, "G": 0.99905, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00746, "router_noise": 0.0, "grad_norm": 0.8661, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.286, "alpha": 0.03, "t": 274.0} |
| {"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.75228, "bpb": 2.52801, "chunk_loss": 0.0002, "F": 0.00098, "G": 0.99933, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00551, "router_noise": 0.0, "grad_norm": 0.985, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.272, "alpha": 0.03, "t": 301.2} |
| {"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.64603, "bpb": 2.37472, "chunk_loss": 0.000191, "F": 0.00098, "G": 0.99956, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00379, "router_noise": 0.0, "grad_norm": 0.8198, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.259, "alpha": 0.03, "t": 327.0} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999649, "boundary_entropy": 0.003111, "bpb": 2.346938}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999625, "boundary_entropy": 0.003306, "bpb": 2.630814}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.999717, "boundary_entropy": 0.002579, "bpb": 2.127817}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999617, "boundary_entropy": 0.003364, "bpb": 2.631123}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999757, "boundary_entropy": 0.00225, "bpb": 2.916959}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999663, "boundary_entropy": 0.003001, "bpb": 1.681474}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999701, "boundary_entropy": 0.002709, "bpb": 1.764139}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.999788, "boundary_entropy": 0.00199, "bpb": 1.370406}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999767, "boundary_entropy": 0.00216, "bpb": 1.132609}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.999788, "boundary_entropy": 0.001984, "bpb": 1.341722}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999709, "boundary_entropy": 0.002643, "bpb": 3.484131}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999805, "boundary_entropy": 0.001844, "bpb": 1.27939}}, "bpb": {"en": 2.34694, "de": 2.63081, "vi": 2.12782, "tr": 2.63112, "fi": 2.91696, "ru": 1.68147, "el": 1.76414, "hi": 1.37041, "ta": 1.13261, "th": 1.34172, "zh": 3.48413, "my": 1.27939}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 354.6} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.35885, "de": 2.62299, "vi": 2.12955, "tr": 2.62495, "fi": 2.92294, "ru": 1.68131, "el": 1.75867, "hi": 1.3783, "ta": 1.13197, "th": 1.33844, "zh": 3.48927, "my": 1.29053}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 2.06065, "t": 359.1} |
| {"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.57266, "bpb": 2.26887, "chunk_loss": 0.000187, "F": 0.00098, "G": 0.99968, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00287, "router_noise": 0.0, "grad_norm": 0.7288, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.323, "alpha": 0.03, "t": 359.3} |
| {"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.54054, "bpb": 2.22253, "chunk_loss": 0.000184, "F": 0.00098, "G": 0.99976, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00221, "router_noise": 0.0, "grad_norm": 0.7757, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.248, "alpha": 0.03, "t": 384.1} |
| {"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.52912, "bpb": 2.20605, "chunk_loss": 0.000183, "F": 0.00098, "G": 0.99979, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00195, "router_noise": 0.0, "grad_norm": 0.647, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 408.4} |
| {"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.58191, "bpb": 2.28222, "chunk_loss": 0.000181, "F": 0.00098, "G": 0.99985, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00149, "router_noise": 0.0, "grad_norm": 0.7142, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 432.5} |
| {"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.4059, "bpb": 2.02828, "chunk_loss": 0.000181, "F": 0.00098, "G": 0.99986, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00133, "router_noise": 0.0, "grad_norm": 0.6465, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 456.7} |
| {"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.52008, "bpb": 2.19301, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00115, "router_noise": 0.0, "grad_norm": 0.6275, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 480.8} |
| {"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.55188, "bpb": 2.23888, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00113, "router_noise": 0.0, "grad_norm": 0.8056, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 505.1} |
| {"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.41026, "bpb": 2.03457, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00107, "router_noise": 0.0, "grad_norm": 0.5625, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 529.0} |
| {"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.42972, "bpb": 2.06265, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00105, "router_noise": 0.0, "grad_norm": 0.5685, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 552.8} |
| {"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.42208, "bpb": 2.05162, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00106, "router_noise": 0.0, "grad_norm": 0.6514, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 576.9} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999899, "boundary_entropy": 0.001026, "bpb": 2.131579}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999898, "boundary_entropy": 0.001036, "bpb": 2.333208}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.981089}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999898, "boundary_entropy": 0.001039, "bpb": 2.306539}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 2.497241}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999895, "boundary_entropy": 0.001064, "bpb": 1.428493}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999898, "boundary_entropy": 0.001044, "bpb": 1.49207}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.999898, "boundary_entropy": 0.001044, "bpb": 1.201507}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999897, "boundary_entropy": 0.001047, "bpb": 1.000922}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.999899, "boundary_entropy": 0.001027, "bpb": 1.13907}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999899, "boundary_entropy": 0.00103, "bpb": 3.20064}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999899, "boundary_entropy": 0.001029, "bpb": 1.13713}}, "bpb": {"en": 2.13158, "de": 2.33321, "vi": 1.98109, "tr": 2.30654, "fi": 2.49724, "ru": 1.42849, "el": 1.49207, "hi": 1.20151, "ta": 1.00092, "th": 1.13907, "zh": 3.20064, "my": 1.13713}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 603.2} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.14038, "de": 2.32438, "vi": 1.98399, "tr": 2.29838, "fi": 2.50969, "ru": 1.42955, "el": 1.48768, "hi": 1.21097, "ta": 1.00136, "th": 1.13744, "zh": 3.20866, "my": 1.14703}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.82329, "t": 607.7} |
| {"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.42939, "bpb": 2.06217, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00104, "router_noise": 0.0, "grad_norm": 0.6971, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.311, "alpha": 0.03, "t": 607.9} |
| {"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.47724, "bpb": 2.1312, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00104, "router_noise": 0.0, "grad_norm": 0.5388, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 632.0} |
| {"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.43374, "bpb": 2.06845, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.5634, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 655.8} |
| {"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.37039, "bpb": 1.97706, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.5166, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 679.6} |
| {"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.40063, "bpb": 2.02068, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00104, "router_noise": 0.0, "grad_norm": 0.5403, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 703.3} |
| {"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.25858, "bpb": 1.81575, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.5054, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 727.1} |
| {"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.25261, "bpb": 1.80713, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.4929, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 750.7} |
| {"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.40633, "bpb": 2.0289, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.5312, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 774.6} |
| {"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.22716, "bpb": 1.77042, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.5215, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 798.5} |
| {"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.277, "bpb": 1.84232, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4745, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 822.1} |
| {"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 2.025413}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.9999, "boundary_entropy": 0.001017, "bpb": 2.204017}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.877283}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 2.168087}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.358537}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.9999, "boundary_entropy": 0.001022, "bpb": 1.333921}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001025, "bpb": 1.372279}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 1.127186}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 0.94357}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.066379}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 3.006985}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 1.068172}}, "bpb": {"en": 2.02541, "de": 2.20402, "vi": 1.87728, "tr": 2.16809, "fi": 2.35854, "ru": 1.33392, "el": 1.37228, "hi": 1.12719, "ta": 0.94357, "th": 1.06638, "zh": 3.00699, "my": 1.06817}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 847.7} |
| {"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 2.03327, "de": 2.1969, "vi": 1.88385, "tr": 2.16055, "fi": 2.37196, "ru": 1.33398, "el": 1.36817, "hi": 1.13456, "ta": 0.94426, "th": 1.06445, "zh": 3.01527, "my": 1.07808}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.71544, "t": 852.2} |
| {"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.30572, "bpb": 1.88376, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4925, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.304, "alpha": 0.03, "t": 852.4} |
| {"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.24404, "bpb": 1.79477, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5457, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 876.0} |
| {"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.32116, "bpb": 1.90603, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.533, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 899.5} |
| {"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.31204, "bpb": 1.89288, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5013, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 923.0} |
| {"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.38045, "bpb": 1.99157, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4867, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 946.2} |
| {"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.22463, "bpb": 1.76677, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4619, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 969.4} |
| {"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.27766, "bpb": 1.84327, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4882, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 992.5} |
| {"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.25836, "bpb": 1.81543, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4711, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.23, "alpha": 0.03, "t": 1015.5} |
| {"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.2935, "bpb": 1.86613, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4724, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.23, "alpha": 0.03, "t": 1038.5} |
| {"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.27291, "bpb": 1.83643, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5092, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 1061.3} |
| {"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.968007}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001016, "bpb": 2.127875}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.813491}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 2.077894}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.243945}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 1.286449}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 1.311803}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.07594}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.914392}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.029996}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.902738}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 1.028367}}, "bpb": {"en": 1.96801, "de": 2.12788, "vi": 1.81349, "tr": 2.07789, "fi": 2.24395, "ru": 1.28645, "el": 1.3118, "hi": 1.07594, "ta": 0.91439, "th": 1.03, "zh": 2.90274, "my": 1.02837}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1085.9} |
| {"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.97485, "de": 2.12029, "vi": 1.82047, "tr": 2.06998, "fi": 2.25771, "ru": 1.28718, "el": 1.30926, "hi": 1.0838, "ta": 0.91616, "th": 1.02815, "zh": 2.9123, "my": 1.03775}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.65149, "t": 1090.4} |
| {"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.20292, "bpb": 1.73545, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4665, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.293, "alpha": 0.03, "t": 1090.6} |
| {"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.2793, "bpb": 1.84564, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4728, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1113.0} |
| {"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.26536, "bpb": 1.82553, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5145, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1135.5} |
| {"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.18282, "bpb": 1.70645, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4661, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1157.9} |
| {"event": "phase_transition", "step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1180.1} |
| {"event": "train", "step": 4400, "phase": 2, "objective": "global", "lm_loss": 1.25353, "bpb": 1.80847, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4843, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1180.3} |
| {"event": "train", "step": 4500, "phase": 2, "objective": "global", "lm_loss": 1.3382, "bpb": 1.93061, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5037, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.223, "alpha": 0.03, "t": 1202.7} |
| {"event": "train", "step": 4600, "phase": 2, "objective": "global", "lm_loss": 1.17938, "bpb": 1.70148, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4647, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1225.1} |
| {"event": "train", "step": 4700, "phase": 2, "objective": "global", "lm_loss": 1.25244, "bpb": 1.80689, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.493, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1247.6} |
| {"event": "train", "step": 4800, "phase": 2, "objective": "global", "lm_loss": 1.23779, "bpb": 1.78575, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.473, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.223, "alpha": 0.03, "t": 1269.9} |
| {"event": "train", "step": 4900, "phase": 2, "objective": "global", "lm_loss": 1.30928, "bpb": 1.88889, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4769, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1292.4} |
| {"event": "probe", "step": 5000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.927852}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001016, "bpb": 2.080217}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.754173}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 2.034617}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.184518}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.248158}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.264894}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.039333}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.877681}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.992665}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.833386}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.000339}}, "bpb": {"en": 1.92785, "de": 2.08022, "vi": 1.75417, "tr": 2.03462, "fi": 2.18452, "ru": 1.24816, "el": 1.26489, "hi": 1.03933, "ta": 0.87768, "th": 0.99267, "zh": 2.83339, "my": 1.00034}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1317.0} |
| {"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.93332, "de": 2.07175, "vi": 1.76276, "tr": 2.02738, "fi": 2.19903, "ru": 1.24728, "el": 1.2622, "hi": 1.04615, "ta": 0.8798, "th": 0.9903, "zh": 2.84403, "my": 1.00979}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.60615, "t": 1321.5} |
| {"event": "train", "step": 5000, "phase": 2, "objective": "global", "lm_loss": 1.24691, "bpb": 1.7989, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.443, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.293, "alpha": 0.03, "t": 1321.7} |
| {"event": "train", "step": 5100, "phase": 2, "objective": "global", "lm_loss": 1.24452, "bpb": 1.79547, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4931, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1344.1} |
| {"event": "train", "step": 5200, "phase": 2, "objective": "global", "lm_loss": 1.25239, "bpb": 1.80682, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4619, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.222, "alpha": 0.03, "t": 1366.3} |
| {"event": "train", "step": 5300, "phase": 2, "objective": "global", "lm_loss": 1.21336, "bpb": 1.75051, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4887, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1388.7} |
| {"event": "train", "step": 5400, "phase": 2, "objective": "global", "lm_loss": 1.1901, "bpb": 1.71696, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4702, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1411.1} |
| {"event": "train", "step": 5500, "phase": 2, "objective": "global", "lm_loss": 1.17463, "bpb": 1.69463, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4579, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.223, "alpha": 0.03, "t": 1433.5} |
| {"event": "train", "step": 5600, "phase": 2, "objective": "global", "lm_loss": 1.19882, "bpb": 1.72954, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.463, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1456.0} |
| {"event": "train", "step": 5700, "phase": 2, "objective": "global", "lm_loss": 1.24095, "bpb": 1.79031, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4422, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1478.4} |
| {"event": "train", "step": 5800, "phase": 2, "objective": "global", "lm_loss": 1.14753, "bpb": 1.65554, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4446, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1500.7} |
| {"event": "train", "step": 5900, "phase": 2, "objective": "global", "lm_loss": 1.20076, "bpb": 1.73234, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4731, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 1523.3} |
| {"event": "probe", "step": 6000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.897457}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.037341}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.702881}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.985746}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.135568}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.217959}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.22782}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 1.011656}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.858498}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.967799}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.783787}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.966406}}, "bpb": {"en": 1.89746, "de": 2.03734, "vi": 1.70288, "tr": 1.98575, "fi": 2.13557, "ru": 1.21796, "el": 1.22782, "hi": 1.01166, "ta": 0.8585, "th": 0.9678, "zh": 2.78379, "my": 0.96641}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1548.0} |
| {"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.90131, "de": 2.03109, "vi": 1.71425, "tr": 1.97892, "fi": 2.15234, "ru": 1.21729, "el": 1.22444, "hi": 1.01655, "ta": 0.86068, "th": 0.96528, "zh": 2.79363, "my": 0.97661}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.56936, "t": 1552.5} |
| {"event": "train", "step": 6000, "phase": 2, "objective": "global", "lm_loss": 1.18845, "bpb": 1.71457, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4555, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.294, "alpha": 0.03, "t": 1552.8} |
| {"event": "train", "step": 6100, "phase": 2, "objective": "global", "lm_loss": 1.26196, "bpb": 1.82062, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4779, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 1575.4} |
| {"event": "train", "step": 6200, "phase": 2, "objective": "global", "lm_loss": 1.23298, "bpb": 1.77881, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4699, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1597.8} |
| {"event": "train", "step": 6300, "phase": 2, "objective": "global", "lm_loss": 1.2722, "bpb": 1.8354, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4664, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 1620.5} |
| {"event": "train", "step": 6400, "phase": 2, "objective": "global", "lm_loss": 1.1893, "bpb": 1.7158, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4424, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1643.0} |
| {"event": "train", "step": 6500, "phase": 2, "objective": "global", "lm_loss": 1.20595, "bpb": 1.73981, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4566, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1665.5} |
| {"event": "train", "step": 6600, "phase": 2, "objective": "global", "lm_loss": 1.14554, "bpb": 1.65267, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4354, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 1688.1} |
| {"event": "train", "step": 6700, "phase": 2, "objective": "global", "lm_loss": 1.18438, "bpb": 1.7087, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4552, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1710.6} |
| {"event": "train", "step": 6800, "phase": 2, "objective": "global", "lm_loss": 1.24887, "bpb": 1.80174, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4852, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 1733.1} |
| {"event": "train", "step": 6900, "phase": 2, "objective": "global", "lm_loss": 1.21978, "bpb": 1.75977, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4588, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1755.5} |
| {"event": "probe", "step": 7000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.866877}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.003762}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.668069}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 1.949305}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.084413}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.19662}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.19724}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.986278}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.84231}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.952583}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.745974}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.952827}}, "bpb": {"en": 1.86688, "de": 2.00376, "vi": 1.66807, "tr": 1.9493, "fi": 2.08441, "ru": 1.19662, "el": 1.19724, "hi": 0.98628, "ta": 0.84231, "th": 0.95258, "zh": 2.74597, "my": 0.95283}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1780.2} |
| {"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.87039, "de": 1.99592, "vi": 1.6789, "tr": 1.94291, "fi": 2.1003, "ru": 1.19543, "el": 1.19495, "hi": 0.99297, "ta": 0.84506, "th": 0.94998, "zh": 2.75525, "my": 0.96072}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.54023, "t": 1784.7} |
| {"event": "train", "step": 7000, "phase": 2, "objective": "global", "lm_loss": 1.32256, "bpb": 1.90805, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4959, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.294, "alpha": 0.03, "t": 1785.0} |
| {"event": "train", "step": 7100, "phase": 2, "objective": "global", "lm_loss": 1.22434, "bpb": 1.76635, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4838, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1807.5} |
| {"event": "train", "step": 7200, "phase": 2, "objective": "global", "lm_loss": 1.13297, "bpb": 1.63454, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4521, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 1830.1} |
| {"event": "train", "step": 7300, "phase": 2, "objective": "global", "lm_loss": 1.20998, "bpb": 1.74563, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4807, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 1852.9} |
| {"event": "train", "step": 7400, "phase": 2, "objective": "global", "lm_loss": 1.1267, "bpb": 1.62548, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4759, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.223, "alpha": 0.03, "t": 1875.2} |
| {"event": "train", "step": 7500, "phase": 2, "objective": "global", "lm_loss": 1.15766, "bpb": 1.67015, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4853, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1897.7} |
| {"event": "train", "step": 7600, "phase": 2, "objective": "global", "lm_loss": 1.12735, "bpb": 1.62642, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4544, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1920.3} |
| {"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1942.6} |
| {"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.12771, "bpb": 1.62694, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4551, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1942.8} |
| {"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.20078, "bpb": 1.73236, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4877, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1965.3} |
| {"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.17887, "bpb": 1.70074, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4661, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1987.9} |
| {"event": "probe", "step": 8000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.84596}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.973663}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.643251}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 1.911833}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.052031}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.177226}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.174636}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.970608}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.823772}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.935156}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.714747}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.93486}}, "bpb": {"en": 1.84596, "de": 1.97366, "vi": 1.64325, "tr": 1.91183, "fi": 2.05203, "ru": 1.17723, "el": 1.17464, "hi": 0.97061, "ta": 0.82377, "th": 0.93516, "zh": 2.71475, "my": 0.93486}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2012.5} |
| {"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.84872, "de": 1.9672, "vi": 1.65581, "tr": 1.90731, "fi": 2.06912, "ru": 1.17594, "el": 1.17199, "hi": 0.97682, "ta": 0.82703, "th": 0.93241, "zh": 2.72573, "my": 0.94374}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.51682, "t": 2017.0} |
| {"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.17171, "bpb": 1.69042, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5199, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.294, "alpha": 0.03, "t": 2017.3} |
| {"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.15179, "bpb": 1.66169, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4893, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 2039.9} |
| {"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.2136, "bpb": 1.75086, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4948, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 2062.3} |
| {"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.14591, "bpb": 1.6532, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4902, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 2084.9} |
| {"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.17655, "bpb": 1.6974, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5118, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 2107.3} |
| {"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.16527, "bpb": 1.68114, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4822, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 2130.9} |
| {"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.12257, "bpb": 1.61953, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4694, "lr": 6.33e-05, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 2155.3} |
| {"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 1.13846, "bpb": 1.64245, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5093, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 2179.3} |
| {"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.12749, "bpb": 1.62663, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4787, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 2203.6} |
| {"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.12726, "bpb": 1.6263, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4871, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 2228.0} |
| {"event": "probe", "step": 9000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.827725}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.95175}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.617235}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.887294}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.019707}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.162819}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.158033}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.955633}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.812495}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.922505}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.687899}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.912098}}, "bpb": {"en": 1.82773, "de": 1.95175, "vi": 1.61724, "tr": 1.88729, "fi": 2.01971, "ru": 1.16282, "el": 1.15803, "hi": 0.95563, "ta": 0.81249, "th": 0.9225, "zh": 2.6879, "my": 0.9121}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2254.2} |
| {"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.82907, "de": 1.94543, "vi": 1.6309, "tr": 1.88355, "fi": 2.03794, "ru": 1.16253, "el": 1.15604, "hi": 0.96179, "ta": 0.81548, "th": 0.91997, "zh": 2.69724, "my": 0.92146}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.49678, "t": 2258.8} |
| {"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.17039, "bpb": 1.68851, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4872, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.31, "alpha": 0.03, "t": 2259.0} |
| {"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.09725, "bpb": 1.58299, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4789, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 2282.5} |
| {"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 1.21792, "bpb": 1.75709, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.526, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 2306.5} |
| {"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 1.05422, "bpb": 1.52092, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4794, "lr": 4.71e-05, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 2330.4} |
| {"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.17884, "bpb": 1.7007, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5196, "lr": 4.52e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 2354.5} |
| {"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.10666, "bpb": 1.59657, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4932, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 2378.5} |
| {"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 1.18167, "bpb": 1.70479, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5192, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 2402.5} |
| {"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 1.15407, "bpb": 1.66498, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4905, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 2426.5} |
| {"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 1.11854, "bpb": 1.61372, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4876, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 2450.1} |
| {"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.1291, "bpb": 1.62895, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4848, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 2473.8} |
| {"event": "probe", "step": 10000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.81331}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.936091}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.60292}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.866757}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.004824}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.15327}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.146434}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.945283}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.803691}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.914692}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.669738}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.901103}}, "bpb": {"en": 1.81331, "de": 1.93609, "vi": 1.60292, "tr": 1.86676, "fi": 2.00482, "ru": 1.15327, "el": 1.14643, "hi": 0.94528, "ta": 0.80369, "th": 0.91469, "zh": 2.66974, "my": 0.9011}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2499.9} |
| {"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.81538, "de": 1.93085, "vi": 1.6165, "tr": 1.86261, "fi": 2.02274, "ru": 1.1527, "el": 1.14405, "hi": 0.95148, "ta": 0.80715, "th": 0.91247, "zh": 2.68053, "my": 0.91046}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.48391, "t": 2504.7} |
| {"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.1463, "bpb": 1.65377, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4781, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.311, "alpha": 0.03, "t": 2504.9} |
| {"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 1.16776, "bpb": 1.68473, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.507, "lr": 3.49e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 2528.6} |
| {"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.10516, "bpb": 1.59442, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5166, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 2552.6} |
| {"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 1.05476, "bpb": 1.5217, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5081, "lr": 3.3e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 2576.2} |
| {"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.1507, "bpb": 1.6601, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.512, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 2600.3} |
| {"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.0939, "bpb": 1.57816, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5111, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 2623.7} |
| {"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 1.05241, "bpb": 1.51831, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5081, "lr": 3.1e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 2647.2} |
| {"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.12058, "bpb": 1.61665, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5172, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 2670.4} |
| {"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.10309, "bpb": 1.59142, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.514, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 2694.5} |
| {"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.11008, "bpb": 1.6015, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5047, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 2717.7} |
| {"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.80443}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.927924}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.595555}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 1.855408}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.989966}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.148936}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.140229}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.940118}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.800195}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.908552}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.66096}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.894089}}, "bpb": {"en": 1.80443, "de": 1.92792, "vi": 1.59555, "tr": 1.85541, "fi": 1.98997, "ru": 1.14894, "el": 1.14023, "hi": 0.94012, "ta": 0.8002, "th": 0.90855, "zh": 2.66096, "my": 0.89409}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2743.6} |
| {"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.80646, "de": 1.92271, "vi": 1.60885, "tr": 1.85201, "fi": 2.00776, "ru": 1.14807, "el": 1.13812, "hi": 0.94598, "ta": 0.8037, "th": 0.90656, "zh": 2.67129, "my": 0.90281}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.47619, "t": 2748.1} |
| {"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.80443}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.927924}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.595555}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 1.855408}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.989966}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.148936}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.140229}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.940118}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.800195}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.908552}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.66096}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.894089}}, "bpb": {"en": 1.80443, "de": 1.92792, "vi": 1.59555, "tr": 1.85541, "fi": 1.98997, "ru": 1.14894, "el": 1.14023, "hi": 0.94012, "ta": 0.8002, "th": 0.90855, "zh": 2.66096, "my": 0.89409}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2750.8} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_onechunk_s2/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_onechunk_s2/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_onechunk_s2/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_onechunk_s2/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_onechunk_s2/th0.70"], "t": 2765.9} |
| {"event": "final", "run": "exp32_local_onechunk_s2", "condition": "A", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 1.000861920358559, "cps": 136.078125, "bpb": 1.80217505878265, "bytes_per_chunk": 1024.0}, "de": {"cpc": 1.0160229873861588, "cps": 162.984375, "bpb": 1.9249649423266384, "bytes_per_chunk": 1024.0}, "vi": {"cpc": 1.3167344724708434, "cps": 189.640625, "bpb": 1.612284391931281, "bytes_per_chunk": 1024.0}, "tr": {"cpc": 1.0905374799756946, "cps": 154.234375, "bpb": 1.8497320131177883, "bytes_per_chunk": 1024.0}, "fi": {"cpc": 1.040455052894583, "cps": 152.90625, "bpb": 2.0144766451320475, "bytes_per_chunk": 1024.0}, "ru": {"cpc": 1.8237467018469657, "cps": 270.0, "bpb": 1.1482993527169538, "bytes_per_chunk": 1024.0}, "el": {"cpc": 1.8157189494713248, "cps": 291.125, "bpb": 1.1346589733459393, "bytes_per_chunk": 1024.0}, "hi": {"cpc": 2.566002971768202, "cps": 337.2890625, "bpb": 0.9477836568143168, "bytes_per_chunk": 1024.0}, "ta": {"cpc": 2.7234554693768387, "cps": 397.7734375, "bpb": 0.802089134577335, "bytes_per_chunk": 1024.0}, "th": {"cpc": 2.8650917003419334, "cps": 360.0390625, "bpb": 0.9082885988422281, "bytes_per_chunk": 1024.0}, "zh": {"cpc": 2.7384016254656283, "cps": 126.3515625, "bpb": 2.673498039864812, "bytes_per_chunk": 1024.0}, "my": {"cpc": 2.8363415663922313, "cps": 419.8671875, "bpb": 0.9015376187375486, "bytes_per_chunk": 1024.0}}, "gini_cpc": 0.22151585453568945, "gini_cps": 0.2338532044751218, "bpic": 1.0, "seed": 2, "mean_boundary_entropy": 0.001020480995066464, "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": 2774.6} |
|
|
