Download log.jsonl from AdaptiveChunking/exp32_onechunk_s2: direct link, hf CLI and curl.
- Browser
- Download file 107 kB
-
https://huggingface.co/AdaptiveChunking/exp32_onechunk_s2/resolve/main/log.jsonl
- Command line
-
hf download hf://AdaptiveChunking/exp32_onechunk_s2/log.jsonl
-
curl -L -o log.jsonl https://huggingface.co/AdaptiveChunking/exp32_onechunk_s2/resolve/main/log.jsonl
107 kB
| {"event": "start", "run": "exp32_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": 2.0} | |
| {"event": "probe_schedule", "n": 21, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 11000], "t": 2.0} | |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10185.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.585244, "cps": 79.570312, "mean_p": 0.50663, "boundary_entropy": 0.687329, "bpb": 8.08194}, "de": {"chunks": 10966.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.534067, "cps": 85.671875, "mean_p": 0.503879, "boundary_entropy": 0.688251, "bpb": 8.128434}, "vi": {"chunks": 5741.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.311418, "cps": 44.851562, "mean_p": 0.489692, "boundary_entropy": 0.688475, "bpb": 8.061751}, "tr": {"chunks": 6370.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.351875, "cps": 49.765625, "mean_p": 0.495802, "boundary_entropy": 0.687878, "bpb": 8.093617}, "fi": {"chunks": 2115.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.112434, "cps": 16.523438, "mean_p": 0.479404, "boundary_entropy": 0.686744, "bpb": 8.107398}, "ru": {"chunks": 23602.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.245488, "cps": 184.390625, "mean_p": 0.507156, "boundary_entropy": 0.690197, "bpb": 8.243086}, "el": {"chunks": 1455.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.070896, "cps": 11.367188, "mean_p": 0.480429, "boundary_entropy": 0.689544, "bpb": 7.978709}, "hi": {"chunks": 7663.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.455453, "cps": 59.867188, "mean_p": 0.494285, "boundary_entropy": 0.690846, "bpb": 7.867437}, "ta": {"chunks": 357.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.019096, "cps": 2.789062, "mean_p": 0.475379, "boundary_entropy": 0.689908, "bpb": 7.649677}, "th": {"chunks": 27234.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.69313, "cps": 212.765625, "mean_p": 0.50318, "boundary_entropy": 0.691076, "bpb": 7.878103}, "zh": {"chunks": 14702.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.489333, "cps": 114.859375, "mean_p": 0.531852, "boundary_entropy": 0.685234, "bpb": 8.098421}, "my": {"chunks": 789.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.04164, "cps": 6.164062, "mean_p": 0.475663, "boundary_entropy": 0.690063, "bpb": 8.232494}}, "bpb": {"en": 8.08194, "de": 8.12843, "vi": 8.06175, "tr": 8.09362, "fi": 8.1074, "ru": 8.24309, "el": 7.97871, "hi": 7.86744, "ta": 7.64968, "th": 7.8781, "zh": 8.09842, "my": 8.23249}, "gini_cpc": 0.562508, "gini_cps": 0.489131, "bpic": 3.45192, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9125547315729478, "vi": 0.5321174384544991, "tr": 0.6012456783662331, "fi": 0.19211513271032077, "ru": 2.1281521716321925, "el": 0.1211393488838307, "hi": 0.7782279770749311, "ta": 0.032629155488401715, "th": 2.8930334475858417, "zh": 4.253496332079356, "my": 0.07115029646287195}, "t": 6.3} | |
| {"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.61035, "bpb": 8.09403, "chunk_loss": 0.01799, "F": 0.00098, "G": 0.50224, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.6917, "router_noise": 0.0, "grad_norm": 5.3146, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.011, "alpha": 0.03, "t": 7.7} | |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10234.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.58806, "cps": 79.953125, "mean_p": 0.50675, "boundary_entropy": 0.687327, "bpb": 8.063676}, "de": {"chunks": 11046.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.537963, "cps": 86.296875, "mean_p": 0.504034, "boundary_entropy": 0.688251, "bpb": 8.111841}, "vi": {"chunks": 5792.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.314185, "cps": 45.25, "mean_p": 0.489818, "boundary_entropy": 0.688482, "bpb": 8.04984}, "tr": {"chunks": 6395.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.353256, "cps": 49.960938, "mean_p": 0.495875, "boundary_entropy": 0.68788, "bpb": 8.080471}, "fi": {"chunks": 2125.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.112966, "cps": 16.601562, "mean_p": 0.479518, "boundary_entropy": 0.686755, "bpb": 8.090909}, "ru": {"chunks": 23865.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.259367, "cps": 186.445312, "mean_p": 0.507346, "boundary_entropy": 0.690193, "bpb": 8.214535}, "el": {"chunks": 1462.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.071237, "cps": 11.421875, "mean_p": 0.480475, "boundary_entropy": 0.689547, "bpb": 7.96803}, "hi": {"chunks": 7739.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.45997, "cps": 60.460938, "mean_p": 0.494348, "boundary_entropy": 0.690848, "bpb": 7.860318}, "ta": {"chunks": 358.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.01915, "cps": 2.796875, "mean_p": 0.475404, "boundary_entropy": 0.689911, "bpb": 7.644308}, "th": {"chunks": 27534.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.711781, "cps": 215.109375, "mean_p": 0.503276, "boundary_entropy": 0.691075, "bpb": 7.863334}, "zh": {"chunks": 14708.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.490349, "cps": 114.90625, "mean_p": 0.531866, "boundary_entropy": 0.685233, "bpb": 8.09682}, "my": {"chunks": 791.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.041746, "cps": 6.179688, "mean_p": 0.475692, "boundary_entropy": 0.690066, "bpb": 8.238831}}, "bpb": {"en": 8.06368, "de": 8.11184, "vi": 8.04984, "tr": 8.08047, "fi": 8.09091, "ru": 8.21454, "el": 7.96803, "hi": 7.86032, "ta": 7.64431, "th": 7.86333, "zh": 8.09682, "my": 8.23883}, "gini_cpc": 0.562016, "gini_cps": 0.489815, "bpic": 3.425118, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9148109183022118, "vi": 0.5342740967940908, "tr": 0.6007153159621181, "fi": 0.1920992894699521, "ru": 2.1415633799694023, "el": 0.1211393488838307, "hi": 0.7821831955422897, "ta": 0.032563888925330735, "th": 2.9108977498800375, "zh": 4.234858327993228, "my": 0.07098912366395813}, "t": 12.0} | |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10303.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.592024, "cps": 80.492188, "mean_p": 0.506938, "boundary_entropy": 0.687325, "bpb": 8.028293}, "de": {"chunks": 11173.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.544148, "cps": 87.289062, "mean_p": 0.504289, "boundary_entropy": 0.68825, "bpb": 8.079776}, "vi": {"chunks": 5879.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.318904, "cps": 45.929688, "mean_p": 0.49001, "boundary_entropy": 0.688493, "bpb": 8.026676}, "tr": {"chunks": 6464.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.357068, "cps": 50.5, "mean_p": 0.496005, "boundary_entropy": 0.687885, "bpb": 8.055764}, "fi": {"chunks": 2145.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.114029, "cps": 16.757812, "mean_p": 0.479722, "boundary_entropy": 0.686776, "bpb": 8.059483}, "ru": {"chunks": 24338.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.284327, "cps": 190.140625, "mean_p": 0.507786, "boundary_entropy": 0.690183, "bpb": 8.153155}, "el": {"chunks": 1471.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.071676, "cps": 11.492188, "mean_p": 0.48055, "boundary_entropy": 0.689554, "bpb": 7.946765}, "hi": {"chunks": 7881.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.46841, "cps": 61.570312, "mean_p": 0.494468, "boundary_entropy": 0.690851, "bpb": 7.841852}, "ta": {"chunks": 359.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.019203, "cps": 2.804688, "mean_p": 0.475525, "boundary_entropy": 0.689923, "bpb": 7.618729}, "th": {"chunks": 27977.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.739322, "cps": 218.570312, "mean_p": 0.503433, "boundary_entropy": 0.691074, "bpb": 7.831688}, "zh": {"chunks": 14719.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.492211, "cps": 114.992188, "mean_p": 0.531918, "boundary_entropy": 0.685227, "bpb": 8.092531}, "my": {"chunks": 692.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.036521, "cps": 5.40625, "mean_p": 0.475705, "boundary_entropy": 0.690067, "bpb": 8.236741}}, "bpb": {"en": 8.02829, "de": 8.07978, "vi": 8.02668, "tr": 8.05576, "fi": 8.05948, "ru": 8.15315, "el": 7.94677, "hi": 7.84185, "ta": 7.61873, "th": 7.83169, "zh": 8.09253, "my": 8.23674}, "gini_cpc": 0.561862, "gini_cps": 0.491748, "bpic": 3.384282, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9191318422187119, "vi": 0.5386674566177767, "tr": 0.6031303967281295, "fi": 0.19260867048405664, "ru": 2.1693823020013383, "el": 0.12106880282022757, "hi": 0.7912007221313084, "ta": 0.03243615741767426, "th": 2.9379236006328724, "zh": 4.209643156088763, "my": 0.06168834781116342}, "t": 17.2} | |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10591.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.608573, "cps": 82.742188, "mean_p": 0.507708, "boundary_entropy": 0.687314, "bpb": 7.898943}, "de": {"chunks": 11674.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.568548, "cps": 91.203125, "mean_p": 0.505303, "boundary_entropy": 0.688245, "bpb": 7.961827}, "vi": {"chunks": 6185.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.335503, "cps": 48.320312, "mean_p": 0.490779, "boundary_entropy": 0.688536, "bpb": 7.944}, "tr": {"chunks": 6720.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.371209, "cps": 52.5, "mean_p": 0.496545, "boundary_entropy": 0.687902, "bpb": 7.965875}, "fi": {"chunks": 2251.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.119664, "cps": 17.585938, "mean_p": 0.480528, "boundary_entropy": 0.686855, "bpb": 7.947241}, "ru": {"chunks": 25470.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.344063, "cps": 198.984375, "mean_p": 0.508925, "boundary_entropy": 0.690154, "bpb": 7.97318}, "el": {"chunks": 1518.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.073966, "cps": 11.859375, "mean_p": 0.48077, "boundary_entropy": 0.689572, "bpb": 7.880757}, "hi": {"chunks": 8485.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.504309, "cps": 66.289062, "mean_p": 0.494953, "boundary_entropy": 0.690863, "bpb": 7.771769}, "ta": {"chunks": 373.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.019952, "cps": 2.914062, "mean_p": 0.47596, "boundary_entropy": 0.689968, "bpb": 7.542348}, "th": {"chunks": 29617.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.841281, "cps": 231.382812, "mean_p": 0.503993, "boundary_entropy": 0.691068, "bpb": 7.730084}, "zh": {"chunks": 14752.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.497799, "cps": 115.25, "mean_p": 0.53213, "boundary_entropy": 0.685205, "bpb": 8.081804}, "my": {"chunks": 809.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.042696, "cps": 6.320312, "mean_p": 0.475958, "boundary_entropy": 0.690093, "bpb": 8.212375}}, "bpb": {"en": 7.89894, "de": 7.96183, "vi": 7.944, "tr": 7.96587, "fi": 7.94724, "ru": 7.97318, "el": 7.88076, "hi": 7.77177, "ta": 7.54235, "th": 7.73008, "zh": 8.0818, "my": 8.21237}, "gini_cpc": 0.556496, "gini_cps": 0.491345, "bpic": 3.240162, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9342313438820317, "vi": 0.5512945690796776, "tr": 0.6099663455488001, "fi": 0.19663044574397964, "ru": 2.2085482047907954, "el": 0.12153967720135822, "hi": 0.8286744064706038, "ta": 0.03278464719561733, "th": 3.025569630603776, "zh": 4.104352125642717, "my": 0.07015720805403802}, "t": 25.1} | |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 11510.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.66138, "cps": 89.921875, "mean_p": 0.510305, "boundary_entropy": 0.687266, "bpb": 7.473458}, "de": {"chunks": 13311.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.648274, "cps": 103.992188, "mean_p": 0.508754, "boundary_entropy": 0.6882, "bpb": 7.572299}, "vi": {"chunks": 7317.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.396908, "cps": 57.164062, "mean_p": 0.493527, "boundary_entropy": 0.688676, "bpb": 7.672785}, "tr": {"chunks": 7564.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.417831, "cps": 59.09375, "mean_p": 0.498417, "boundary_entropy": 0.687957, "bpb": 7.665916}, "fi": {"chunks": 2747.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.146032, "cps": 21.460938, "mean_p": 0.483562, "boundary_entropy": 0.687135, "bpb": 7.574939}, "ru": {"chunks": 29018.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.531293, "cps": 226.703125, "mean_p": 0.512805, "boundary_entropy": 0.69002, "bpb": 7.399488}, "el": {"chunks": 1820.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.088681, "cps": 14.21875, "mean_p": 0.481791, "boundary_entropy": 0.689653, "bpb": 7.670049}, "hi": {"chunks": 12005.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.713522, "cps": 93.789062, "mean_p": 0.497262, "boundary_entropy": 0.690909, "bpb": 7.498777}, "ta": {"chunks": 451.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.024124, "cps": 3.523438, "mean_p": 0.478095, "boundary_entropy": 0.690177, "bpb": 7.2501}, "th": {"chunks": 34623.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.152502, "cps": 270.492188, "mean_p": 0.506143, "boundary_entropy": 0.691036, "bpb": 7.392383}, "zh": {"chunks": 14855.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.515239, "cps": 116.054688, "mean_p": 0.532779, "boundary_entropy": 0.685139, "bpb": 8.050415}, "my": {"chunks": 881.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.046496, "cps": 6.882812, "mean_p": 0.477014, "boundary_entropy": 0.690196, "bpb": 8.123215}}, "bpb": {"en": 7.47346, "de": 7.5723, "vi": 7.67278, "tr": 7.66592, "fi": 7.57494, "ru": 7.39949, "el": 7.67005, "hi": 7.49878, "ta": 7.2501, "th": 7.39238, "zh": 8.05042, "my": 8.12321}, "gini_cpc": 0.538019, "gini_cps": 0.484441, "bpic": 2.819804, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9801827898156189, "vi": 0.600120841607291, "tr": 0.6317565674130458, "fi": 0.2207982223462888, "ru": 2.3152988636702285, "el": 0.13408473542832344, "hi": 1.0788371375494918, "ta": 0.03647538368557035, "th": 3.2545610836436643, "zh": 3.803014752112667, "my": 0.0703009718674017}, "t": 32.6} | |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14522.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.834454, "cps": 113.453125, "mean_p": 0.520042, "boundary_entropy": 0.6869, "bpb": 6.607662}, "de": {"chunks": 17982.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.875761, "cps": 140.484375, "mean_p": 0.521174, "boundary_entropy": 0.68767, "bpb": 6.751942}, "vi": {"chunks": 12838.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.696393, "cps": 100.296875, "mean_p": 0.503993, "boundary_entropy": 0.68895, "bpb": 7.099751}, "tr": {"chunks": 11549.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.637961, "cps": 90.226562, "mean_p": 0.506671, "boundary_entropy": 0.688047, "bpb": 6.995113}, "fi": {"chunks": 6337.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.336877, "cps": 49.507812, "mean_p": 0.496016, "boundary_entropy": 0.687908, "bpb": 6.789918}, "ru": {"chunks": 33354.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.760106, "cps": 260.578125, "mean_p": 0.524095, "boundary_entropy": 0.689295, "bpb": 6.489069}, "el": {"chunks": 4382.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.213517, "cps": 34.234375, "mean_p": 0.487228, "boundary_entropy": 0.690017, "bpb": 7.11175}, "hi": {"chunks": 35152.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.089272, "cps": 274.625, "mean_p": 0.508494, "boundary_entropy": 0.690832, "bpb": 6.841411}, "ta": {"chunks": 2677.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.143193, "cps": 20.914062, "mean_p": 0.488218, "boundary_entropy": 0.690922, "bpb": 6.70765}, "th": {"chunks": 43002.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.673422, "cps": 335.953125, "mean_p": 0.513592, "boundary_entropy": 0.690787, "bpb": 6.562241}, "zh": {"chunks": 15118.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.55977, "cps": 118.109375, "mean_p": 0.53494, "boundary_entropy": 0.684909, "bpb": 8.006043}, "my": {"chunks": 1561.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.082383, "cps": 12.195312, "mean_p": 0.482097, "boundary_entropy": 0.69063, "bpb": 7.693454}}, "bpb": {"en": 6.60766, "de": 6.75194, "vi": 7.09975, "tr": 6.99511, "fi": 6.78992, "ru": 6.48907, "el": 7.11175, "hi": 6.84141, "ta": 6.70765, "th": 6.56224, "zh": 8.00604, "my": 7.69345}, "gini_cpc": 0.460797, "gini_cps": 0.428917, "bpic": 1.933659, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0495020082250526, "vi": 0.8345491462183748, "tr": 0.7645246941663438, "fi": 0.4037100040310669, "ru": 2.109290506006899, "el": 0.25587580138153304, "hi": 2.503759755399043, "ta": 0.17160130626387926, "th": 3.2037991172173275, "zh": 3.067598990249529, "my": 0.09872728970366554}, "t": 44.7} | |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17314.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.994886, "cps": 135.265625, "mean_p": 0.571433, "boundary_entropy": 0.678226, "bpb": 5.949339}, "de": {"chunks": 20763.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.011201, "cps": 162.210938, "mean_p": 0.573834, "boundary_entropy": 0.678269, "bpb": 6.116813}, "vi": {"chunks": 23946.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.298942, "cps": 187.078125, "mean_p": 0.553272, "boundary_entropy": 0.683761, "bpb": 6.851446}, "tr": {"chunks": 19512.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.077832, "cps": 152.4375, "mean_p": 0.558478, "boundary_entropy": 0.681896, "bpb": 6.424486}, "fi": {"chunks": 19361.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.029238, "cps": 151.257812, "mean_p": 0.557359, "boundary_entropy": 0.682157, "bpb": 6.109016}, "ru": {"chunks": 34548.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823113, "cps": 269.90625, "mean_p": 0.568895, "boundary_entropy": 0.681306, "bpb": 5.7374}, "el": {"chunks": 25717.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.253082, "cps": 200.914062, "mean_p": 0.507631, "boundary_entropy": 0.69039, "bpb": 6.420839}, "hi": {"chunks": 43122.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.562972, "cps": 336.890625, "mean_p": 0.544333, "boundary_entropy": 0.687254, "bpb": 6.000188}, "ta": {"chunks": 50712.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.712597, "cps": 396.1875, "mean_p": 0.524963, "boundary_entropy": 0.690176, "bpb": 5.765682}, "th": {"chunks": 46043.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.862481, "cps": 359.710938, "mean_p": 0.531595, "boundary_entropy": 0.689314, "bpb": 5.529632}, "zh": {"chunks": 15761.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.668642, "cps": 123.132812, "mean_p": 0.54674, "boundary_entropy": 0.683179, "bpb": 7.657582}, "my": {"chunks": 27381.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.44506, "cps": 213.914062, "mean_p": 0.502668, "boundary_entropy": 0.691329, "bpb": 6.487391}}, "bpb": {"en": 5.94934, "de": 6.11681, "vi": 6.85145, "tr": 6.42449, "fi": 6.10902, "ru": 5.7374, "el": 6.42084, "hi": 6.00019, "ta": 5.76568, "th": 5.52963, "zh": 7.65758, "my": 6.48739}, "gini_cpc": 0.227474, "gini_cps": 0.219399, "bpic": 1.115059, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0163994089117936, "vi": 1.3056192456510687, "tr": 1.083372838025394, "fi": 1.0345288551431215, "ru": 1.832484895624905, "el": 1.259523186249181, "hi": 2.5761463371849835, "ta": 2.7265406456760477, "th": 2.877194720679519, "zh": 2.6823598100634873, "my": 1.4524882780177013}, "t": 67.4} | |
| {"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": 0.788908, "boundary_entropy": 0.51154, "bpb": 5.494141}, "de": {"chunks": 20857.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.015779, "cps": 162.945312, "mean_p": 0.787145, "boundary_entropy": 0.513812, "bpb": 5.644748}, "vi": {"chunks": 24225.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.314076, "cps": 189.257812, "mean_p": 0.755808, "boundary_entropy": 0.550424, "bpb": 6.383703}, "tr": {"chunks": 19724.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.089543, "cps": 154.09375, "mean_p": 0.776988, "boundary_entropy": 0.525966, "bpb": 6.022794}, "fi": {"chunks": 19566.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040136, "cps": 152.859375, "mean_p": 0.77793, "boundary_entropy": 0.524743, "bpb": 5.652631}, "ru": {"chunks": 34556.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823536, "cps": 269.96875, "mean_p": 0.668701, "boundary_entropy": 0.633456, "bpb": 4.979994}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.574141, "boundary_entropy": 0.68003, "bpb": 5.33477}, "hi": {"chunks": 43161.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.56529, "cps": 337.195312, "mean_p": 0.659574, "boundary_entropy": 0.639817, "bpb": 5.101633}, "ta": {"chunks": 50905.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.722921, "cps": 397.695312, "mean_p": 0.649167, "boundary_entropy": 0.646703, "bpb": 4.871878}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.646573, "boundary_entropy": 0.648124, "bpb": 4.696074}, "zh": {"chunks": 16168.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.737555, "cps": 126.3125, "mean_p": 0.651736, "boundary_entropy": 0.641112, "bpb": 6.895481}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.562871, "boundary_entropy": 0.683648, "bpb": 4.886777}}, "bpb": {"en": 5.49414, "de": 5.64475, "vi": 6.3837, "tr": 6.02279, "fi": 5.65263, "ru": 4.97999, "el": 5.33477, "hi": 5.10163, "ta": 4.87188, "th": 4.69607, "zh": 6.89548, "my": 4.88678}, "gini_cpc": 0.221611, "gini_cps": 0.233948, "bpic": 1.000292, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.015079542760791, "vi": 1.3131710058877164, "tr": 1.0887924077663924, "fi": 1.0394193732102155, "ru": 1.822279092493733, "el": 1.8144678080763403, "hi": 2.56352210588556, "ta": 2.7210443082078952, "th": 2.8631174769480716, "zh": 2.735668685956529, "my": 2.834387153598852}, "t": 105.9} | |
| {"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.26282, "bpb": 4.70726, "chunk_loss": 0.008566, "F": 0.00098, "G": 0.76556, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.52308, "router_noise": 0.0, "grad_norm": 2.5676, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 1.321, "alpha": 0.03, "t": 139.8} | |
| {"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": 0.846942, "boundary_entropy": 0.422023, "bpb": 4.385084}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.841951, "boundary_entropy": 0.428268, "bpb": 4.606185}, "vi": {"chunks": 24176.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.311418, "cps": 188.875, "mean_p": 0.715413, "boundary_entropy": 0.585405, "bpb": 4.792718}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.792251, "boundary_entropy": 0.497017, "bpb": 4.866796}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.827244, "boundary_entropy": 0.449211, "bpb": 4.613761}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.724318, "boundary_entropy": 0.58279, "bpb": 3.251563}, "el": {"chunks": 37242.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.814647, "cps": 290.953125, "mean_p": 0.672792, "boundary_entropy": 0.624651, "bpb": 3.759496}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.695857, "boundary_entropy": 0.609233, "bpb": 3.079709}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.72005, "boundary_entropy": 0.587741, "bpb": 3.028434}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.690565, "boundary_entropy": 0.616347, "bpb": 2.839549}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.870258, "boundary_entropy": 0.380469, "bpb": 6.243047}, "my": {"chunks": 53742.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836289, "cps": 419.859375, "mean_p": 0.725209, "boundary_entropy": 0.585995, "bpb": 3.168297}}, "bpb": {"en": 4.38508, "de": 4.60618, "vi": 4.79272, "tr": 4.8668, "fi": 4.61376, "ru": 3.25156, "el": 3.7595, "hi": 3.07971, "ta": 3.02843, "th": 2.83955, "zh": 6.24305, "my": 3.1683}, "gini_cpc": 0.221639, "gini_cps": 0.233932, "bpic": 1.000315, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3102891325444408, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.813084247211009, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.83384624060585}, "t": 172.0} | |
| {"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.54391, "bpb": 3.67008, "chunk_loss": 0.003668, "F": 0.00098, "G": 0.90244, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.29648, "router_noise": 0.0, "grad_norm": 1.768, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 1.119, "alpha": 0.03, "t": 251.7} | |
| {"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": 0.961794, "boundary_entropy": 0.15746, "bpb": 3.536706}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.957262, "boundary_entropy": 0.163906, "bpb": 3.790089}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.928139, "boundary_entropy": 0.241345, "bpb": 3.096836}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.933336, "boundary_entropy": 0.212362, "bpb": 3.766403}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.955763, "boundary_entropy": 0.162938, "bpb": 3.8437}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.977916, "boundary_entropy": 0.099033, "bpb": 2.648285}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.944648, "boundary_entropy": 0.206981, "bpb": 2.69852}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.964263, "boundary_entropy": 0.153044, "bpb": 2.110731}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.960081, "boundary_entropy": 0.166211, "bpb": 1.982711}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.973261, "boundary_entropy": 0.12174, "bpb": 2.106003}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.951885, "boundary_entropy": 0.188009, "bpb": 4.933666}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.935354, "boundary_entropy": 0.236254, "bpb": 2.209696}}, "bpb": {"en": 3.53671, "de": 3.79009, "vi": 3.09684, "tr": 3.7664, "fi": 3.8437, "ru": 2.64829, "el": 2.69852, "hi": 2.11073, "ta": 1.98271, "th": 2.106, "zh": 4.93367, "my": 2.2097}, "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": 308.6} | |
| {"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.16187, "bpb": 3.11892, "chunk_loss": 0.001051, "F": 0.00098, "G": 0.97554, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.10936, "router_noise": 0.0, "grad_norm": 1.5437, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 1.014, "alpha": 0.03, "t": 353.1} | |
| {"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.08238, "bpb": 3.00424, "chunk_loss": 0.000441, "F": 0.00098, "G": 0.99259, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.04267, "router_noise": 0.0, "grad_norm": 1.5806, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.947, "alpha": 0.03, "t": 447.9} | |
| {"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.84816, "bpb": 2.66633, "chunk_loss": 0.000282, "F": 0.00098, "G": 0.99705, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.01977, "router_noise": 0.0, "grad_norm": 1.1541, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.807, "alpha": 0.03, "t": 528.6} | |
| {"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.82812, "bpb": 2.63742, "chunk_loss": 0.000235, "F": 0.00098, "G": 0.99835, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.012, "router_noise": 0.0, "grad_norm": 0.9102, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.818, "alpha": 0.03, "t": 610.4} | |
| {"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.72063, "bpb": 2.48234, "chunk_loss": 0.000213, "F": 0.00098, "G": 0.99896, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00805, "router_noise": 0.0, "grad_norm": 0.8826, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.699, "alpha": 0.03, "t": 680.3} | |
| {"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.79245, "bpb": 2.58596, "chunk_loss": 0.000201, "F": 0.00098, "G": 0.99929, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00571, "router_noise": 0.0, "grad_norm": 1.0748, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.707, "alpha": 0.03, "t": 751.0} | |
| {"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.67981, "bpb": 2.42345, "chunk_loss": 0.000193, "F": 0.00098, "G": 0.99953, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00403, "router_noise": 0.0, "grad_norm": 0.8223, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.574, "alpha": 0.03, "t": 808.4} | |
| {"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": 0.999548, "boundary_entropy": 0.003875, "bpb": 2.374913}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999477, "boundary_entropy": 0.004422, "bpb": 2.661748}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.999677, "boundary_entropy": 0.002894, "bpb": 2.148025}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999587, "boundary_entropy": 0.003592, "bpb": 2.660552}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999649, "boundary_entropy": 0.003117, "bpb": 2.914249}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999653, "boundary_entropy": 0.003086, "bpb": 1.783184}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999696, "boundary_entropy": 0.002749, "bpb": 1.872246}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.999807, "boundary_entropy": 0.001823, "bpb": 1.456636}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999761, "boundary_entropy": 0.002215, "bpb": 1.144009}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.999719, "boundary_entropy": 0.002555, "bpb": 1.387097}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999624, "boundary_entropy": 0.003307, "bpb": 3.521792}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999736, "boundary_entropy": 0.002418, "bpb": 1.308173}}, "bpb": {"en": 2.37491, "de": 2.66175, "vi": 2.14803, "tr": 2.66055, "fi": 2.91425, "ru": 1.78318, "el": 1.87225, "hi": 1.45664, "ta": 1.14401, "th": 1.3871, "zh": 3.52179, "my": 1.30817}, "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": 874.7} | |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.3878, "de": 2.65312, "vi": 2.15003, "tr": 2.65415, "fi": 2.92218, "ru": 1.78089, "el": 1.86613, "hi": 1.46334, "ta": 1.14286, "th": 1.37869, "zh": 3.52574, "my": 1.32205}, "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": 2.10391, "t": 882.0} | |
| {"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.60122, "bpb": 2.31008, "chunk_loss": 0.000189, "F": 0.00098, "G": 0.99962, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00333, "router_noise": 0.0, "grad_norm": 0.7638, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.744, "alpha": 0.03, "t": 882.7} | |
| {"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.57131, "bpb": 2.26693, "chunk_loss": 0.000186, "F": 0.00098, "G": 0.99972, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00253, "router_noise": 0.0, "grad_norm": 0.7363, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.587, "alpha": 0.03, "t": 941.4} | |
| {"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.55914, "bpb": 2.24936, "chunk_loss": 0.000184, "F": 0.00098, "G": 0.99978, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00202, "router_noise": 0.0, "grad_norm": 0.7608, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.362, "alpha": 0.03, "t": 977.6} | |
| {"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.61139, "bpb": 2.32475, "chunk_loss": 0.000182, "F": 0.00098, "G": 0.99982, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00175, "router_noise": 0.0, "grad_norm": 0.7693, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.505, "alpha": 0.03, "t": 1028.1} | |
| {"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.43099, "bpb": 2.06448, "chunk_loss": 0.000181, "F": 0.00098, "G": 0.99984, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00153, "router_noise": 0.0, "grad_norm": 0.6059, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.57, "alpha": 0.03, "t": 1085.1} | |
| {"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.54616, "bpb": 2.23063, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99987, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00128, "router_noise": 0.0, "grad_norm": 0.6467, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.553, "alpha": 0.03, "t": 1140.4} | |
| {"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.57095, "bpb": 2.2664, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99988, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00117, "router_noise": 0.0, "grad_norm": 0.6996, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.52, "alpha": 0.03, "t": 1192.4} | |
| {"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.43957, "bpb": 2.07686, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00111, "router_noise": 0.0, "grad_norm": 0.602, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.584, "alpha": 0.03, "t": 1250.8} | |
| {"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.44897, "bpb": 2.09042, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00108, "router_noise": 0.0, "grad_norm": 0.5861, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.536, "alpha": 0.03, "t": 1304.4} | |
| {"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.44018, "bpb": 2.07774, "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": 0.5891, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.545, "alpha": 0.03, "t": 1359.0} | |
| {"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": 0.999898, "boundary_entropy": 0.001037, "bpb": 2.132945}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999898, "boundary_entropy": 0.001034, "bpb": 2.359128}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.994021}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999896, "boundary_entropy": 0.001055, "bpb": 2.322806}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999898, "boundary_entropy": 0.001041, "bpb": 2.521403}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999897, "boundary_entropy": 0.001048, "bpb": 1.464432}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999894, "boundary_entropy": 0.001077, "bpb": 1.573144}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.999898, "boundary_entropy": 0.001037, "bpb": 1.254611}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001024, "bpb": 1.043897}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.999898, "boundary_entropy": 0.001037, "bpb": 1.172051}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999898, "boundary_entropy": 0.001035, "bpb": 3.243292}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001022, "bpb": 1.164388}}, "bpb": {"en": 2.13295, "de": 2.35913, "vi": 1.99402, "tr": 2.32281, "fi": 2.5214, "ru": 1.46443, "el": 1.57314, "hi": 1.25461, "ta": 1.0439, "th": 1.17205, "zh": 3.24329, "my": 1.16439}, "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": 1415.8} | |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.14324, "de": 2.34607, "vi": 1.99724, "tr": 2.31282, "fi": 2.53469, "ru": 1.46314, "el": 1.56739, "hi": 1.26263, "ta": 1.04394, "th": 1.16963, "zh": 3.24983, "my": 1.17394}, "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": 1.85538, "t": 1423.0} | |
| {"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.44445, "bpb": 2.08391, "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": 0.6828, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.645, "alpha": 0.03, "t": 1423.5} | |
| {"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.49431, "bpb": 2.15583, "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": 0.5779, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.506, "alpha": 0.03, "t": 1474.1} | |
| {"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.45041, "bpb": 2.0925, "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": 0.5285, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.531, "alpha": 0.03, "t": 1527.2} | |
| {"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.38695, "bpb": 2.00094, "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": 0.5171, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.511, "alpha": 0.03, "t": 1578.3} | |
| {"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.41638, "bpb": 2.04341, "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": 0.5376, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.487, "alpha": 0.03, "t": 1627.1} | |
| {"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.26892, "bpb": 1.83067, "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": 0.4846, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.519, "alpha": 0.03, "t": 1679.0} | |
| {"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.26733, "bpb": 1.82837, "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": 0.5106, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.555, "alpha": 0.03, "t": 1734.5} | |
| {"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.41532, "bpb": 2.04188, "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": 0.5158, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.485, "alpha": 0.03, "t": 1783.0} | |
| {"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.23513, "bpb": 1.78191, "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": 0.5291, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.511, "alpha": 0.03, "t": 1834.2} | |
| {"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.28986, "bpb": 1.86088, "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": 0.4864, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.562, "alpha": 0.03, "t": 1890.4} | |
| {"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": 0.001017, "bpb": 2.030042}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 2.216238}, "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": 1.893862}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 2.178957}, "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": 2.37774}, "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": 1.352365}, "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": 1.40471}, "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": 1.14832}, "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": 0.974776}, "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": 1.092496}, "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": 3.034474}, "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": 1.088953}}, "bpb": {"en": 2.03004, "de": 2.21624, "vi": 1.89386, "tr": 2.17896, "fi": 2.37774, "ru": 1.35236, "el": 1.40471, "hi": 1.14832, "ta": 0.97478, "th": 1.0925, "zh": 3.03447, "my": 1.08895}, "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": 1946.4} | |
| {"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 2.03771, "de": 2.20728, "vi": 1.89882, "tr": 2.17119, "fi": 2.38996, "ru": 1.35071, "el": 1.3997, "hi": 1.15594, "ta": 0.9739, "th": 1.08952, "zh": 3.041, "my": 1.09961}, "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": 1.73461, "t": 1950.1} | |
| {"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.31694, "bpb": 1.89995, "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": 0.000264, "router_trainable": true, "sec_per_step": 0.6, "alpha": 0.03, "t": 1950.4} | |
| {"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.25397, "bpb": 1.8091, "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": 0.5, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.374, "alpha": 0.03, "t": 1987.8} | |
| {"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.33226, "bpb": 1.92204, "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": 0.5381, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.539, "alpha": 0.03, "t": 2041.7} | |
| {"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.31999, "bpb": 1.90434, "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": 0.4919, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.549, "alpha": 0.03, "t": 2096.6} | |
| {"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.38927, "bpb": 2.00429, "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": 0.5016, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.476, "alpha": 0.03, "t": 2144.2} | |
| {"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.23335, "bpb": 1.77934, "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": 0.5039, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.516, "alpha": 0.03, "t": 2195.9} | |
| {"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.28927, "bpb": 1.86002, "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": 0.5592, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.487, "alpha": 0.03, "t": 2244.6} | |
| {"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.26529, "bpb": 1.82542, "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": 0.4756, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.412, "alpha": 0.03, "t": 2285.7} | |
| {"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.29903, "bpb": 1.87411, "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": 0.4811, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.443, "alpha": 0.03, "t": 2330.0} | |
| {"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.27725, "bpb": 1.84268, "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": 0.4863, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.46, "alpha": 0.03, "t": 2376.0} | |
| {"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": 0.9999, "boundary_entropy": 0.001015, "bpb": 1.954386}, "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": 2.126761}, "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": 1.801485}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001017, "bpb": 2.071279}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 2.25408}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.295664}, "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": 1.327228}, "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": 1.100784}, "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": 0.946147}, "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": 1.049092}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 2.936816}, "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": 1.047029}}, "bpb": {"en": 1.95439, "de": 2.12676, "vi": 1.80149, "tr": 2.07128, "fi": 2.25408, "ru": 1.29566, "el": 1.32723, "hi": 1.10078, "ta": 0.94615, "th": 1.04909, "zh": 2.93682, "my": 1.04703}, "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": 2422.9} | |
| {"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.95983, "de": 2.12085, "vi": 1.81013, "tr": 2.06344, "fi": 2.26775, "ru": 1.29425, "el": 1.32374, "hi": 1.10864, "ta": 0.94738, "th": 1.04669, "zh": 2.94699, "my": 1.05599}, "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": 1.66214, "t": 2429.0} | |
| {"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.20782, "bpb": 1.74252, "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": 0.4629, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.535, "alpha": 0.03, "t": 2429.4} | |
| {"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.28107, "bpb": 1.8482, "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": 0.4686, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.445, "alpha": 0.03, "t": 2474.0} | |
| {"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.26231, "bpb": 1.82113, "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": 0.518, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.478, "alpha": 0.03, "t": 2521.8} | |
| {"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.18299, "bpb": 1.7067, "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": 0.468, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.492, "alpha": 0.03, "t": 2571.0} | |
| {"event": "phase_transition", "step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 2617.6} | |
| {"event": "train", "step": 4400, "phase": 2, "objective": "global", "lm_loss": 1.22538, "bpb": 1.76785, "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": 0.4603, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.471, "alpha": 0.03, "t": 2618.1} | |
| {"event": "train", "step": 4500, "phase": 2, "objective": "global", "lm_loss": 1.32804, "bpb": 1.91596, "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": 0.5056, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.562, "alpha": 0.03, "t": 2674.3} | |
| {"event": "train", "step": 4600, "phase": 2, "objective": "global", "lm_loss": 1.16395, "bpb": 1.67923, "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": 0.4946, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.554, "alpha": 0.03, "t": 2729.8} | |
| {"event": "train", "step": 4700, "phase": 2, "objective": "global", "lm_loss": 1.23432, "bpb": 1.78075, "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": 0.5067, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.41, "alpha": 0.03, "t": 2770.8} | |
| {"event": "train", "step": 4800, "phase": 2, "objective": "global", "lm_loss": 1.21291, "bpb": 1.74986, "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": 0.4599, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.476, "alpha": 0.03, "t": 2818.4} | |
| {"event": "train", "step": 4900, "phase": 2, "objective": "global", "lm_loss": 1.28215, "bpb": 1.84975, "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": 0.5041, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.501, "alpha": 0.03, "t": 2868.5} | |
| {"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": 0.001014, "bpb": 1.855089}, "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": 2.008177}, "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": 1.715965}, "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": 1.932892}, "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": 2.135034}, "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": 1.250074}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.273283}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.062805}, "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": 0.892829}, "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": 1.00816}, "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": 2.859351}, "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": 1.013315}}, "bpb": {"en": 1.85509, "de": 2.00818, "vi": 1.71597, "tr": 1.93289, "fi": 2.13503, "ru": 1.25007, "el": 1.27328, "hi": 1.0628, "ta": 0.89283, "th": 1.00816, "zh": 2.85935, "my": 1.01331}, "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": 2916.2} | |
| {"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.85686, "de": 2.00949, "vi": 1.72788, "tr": 1.92652, "fi": 2.14414, "ru": 1.24889, "el": 1.27011, "hi": 1.06933, "ta": 0.89471, "th": 1.00583, "zh": 2.87101, "my": 1.02055}, "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": 1.58711, "t": 2919.9} | |
| {"event": "train", "step": 5000, "phase": 2, "objective": "global", "lm_loss": 1.21945, "bpb": 1.7593, "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": 0.4405, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.516, "alpha": 0.03, "t": 2920.1} | |
| {"event": "train", "step": 5100, "phase": 2, "objective": "global", "lm_loss": 1.22379, "bpb": 1.76556, "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": 0.4742, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.211, "alpha": 0.03, "t": 2941.2} | |
| {"event": "train", "step": 5200, "phase": 2, "objective": "global", "lm_loss": 1.22329, "bpb": 1.76484, "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": 0.447, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.207, "alpha": 0.03, "t": 2961.8} | |
| {"event": "train", "step": 5300, "phase": 2, "objective": "global", "lm_loss": 1.18323, "bpb": 1.70705, "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": 0.4796, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.208, "alpha": 0.03, "t": 2982.7} | |
| {"event": "train", "step": 5400, "phase": 2, "objective": "global", "lm_loss": 1.15528, "bpb": 1.66672, "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": 0.4736, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.208, "alpha": 0.03, "t": 3003.5} | |
| {"event": "train", "step": 5500, "phase": 2, "objective": "global", "lm_loss": 1.15309, "bpb": 1.66356, "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": 0.4554, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.215, "alpha": 0.03, "t": 3024.9} | |
| {"event": "train", "step": 5600, "phase": 2, "objective": "global", "lm_loss": 1.16728, "bpb": 1.68403, "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": 0.4516, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.203, "alpha": 0.03, "t": 3045.3} | |
| {"event": "train", "step": 5700, "phase": 2, "objective": "global", "lm_loss": 1.21211, "bpb": 1.7487, "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": 0.4515, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.217, "alpha": 0.03, "t": 3067.0} | |
| {"event": "train", "step": 5800, "phase": 2, "objective": "global", "lm_loss": 1.12307, "bpb": 1.62024, "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": 0.4329, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.219, "alpha": 0.03, "t": 3088.9} | |
| {"event": "train", "step": 5900, "phase": 2, "objective": "global", "lm_loss": 1.16167, "bpb": 1.67594, "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": 0.4369, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.201, "alpha": 0.03, "t": 3108.9} | |
| {"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": 1.812421}, "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": 1.953461}, "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": 1.635334}, "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": 1.852972}, "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": 2.070761}, "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": 1.221246}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.233697}, "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": 1.024445}, "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": 0.870796}, "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": 0.976417}, "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": 2.803311}, "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": 0.983772}}, "bpb": {"en": 1.81242, "de": 1.95346, "vi": 1.63533, "tr": 1.85297, "fi": 2.07076, "ru": 1.22125, "el": 1.2337, "hi": 1.02445, "ta": 0.8708, "th": 0.97642, "zh": 2.80331, "my": 0.98377}, "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": 3132.6} | |
| {"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.81293, "de": 1.95591, "vi": 1.65018, "tr": 1.84667, "fi": 2.08183, "ru": 1.21952, "el": 1.2315, "hi": 1.02978, "ta": 0.87316, "th": 0.97363, "zh": 2.81412, "my": 0.99166}, "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": 1.54007, "t": 3136.4} | |
| {"event": "train", "step": 6000, "phase": 2, "objective": "global", "lm_loss": 1.16858, "bpb": 1.68591, "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": 0.4524, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.277, "alpha": 0.03, "t": 3136.6} | |
| {"event": "train", "step": 6100, "phase": 2, "objective": "global", "lm_loss": 1.23357, "bpb": 1.77966, "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": 0.4751, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.22, "alpha": 0.03, "t": 3158.6} | |
| {"event": "train", "step": 6200, "phase": 2, "objective": "global", "lm_loss": 1.21354, "bpb": 1.75077, "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": 0.4494, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.219, "alpha": 0.03, "t": 3180.5} | |
| {"event": "train", "step": 6300, "phase": 2, "objective": "global", "lm_loss": 1.24666, "bpb": 1.79855, "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": 0.4628, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.203, "alpha": 0.03, "t": 3200.9} | |
| {"event": "train", "step": 6400, "phase": 2, "objective": "global", "lm_loss": 1.15375, "bpb": 1.66451, "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": 0.4343, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.22, "alpha": 0.03, "t": 3222.9} | |
| {"event": "train", "step": 6500, "phase": 2, "objective": "global", "lm_loss": 1.17349, "bpb": 1.69299, "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": 0.4616, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.219, "alpha": 0.03, "t": 3244.8} | |
| {"event": "train", "step": 6600, "phase": 2, "objective": "global", "lm_loss": 1.11994, "bpb": 1.61573, "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": 0.4343, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.217, "alpha": 0.03, "t": 3266.5} | |
| {"event": "train", "step": 6700, "phase": 2, "objective": "global", "lm_loss": 1.16017, "bpb": 1.67378, "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": 0.4476, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.222, "alpha": 0.03, "t": 3288.7} | |
| {"event": "train", "step": 6800, "phase": 2, "objective": "global", "lm_loss": 1.21299, "bpb": 1.74997, "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": 0.0001233, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 3308.9} | |
| {"event": "train", "step": 6900, "phase": 2, "objective": "global", "lm_loss": 1.18272, "bpb": 1.7063, "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": 0.4457, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 3331.9} | |
| {"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": 1.78148}, "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": 1.913783}, "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": 1.585512}, "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": 1.812516}, "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": 2.013005}, "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": 1.197633}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.199709}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.998981}, "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": 0.853005}, "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": 0.95778}, "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": 2.758098}, "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": 0.962938}}, "bpb": {"en": 1.78148, "de": 1.91378, "vi": 1.58551, "tr": 1.81252, "fi": 2.01301, "ru": 1.19763, "el": 1.19971, "hi": 0.99898, "ta": 0.853, "th": 0.95778, "zh": 2.7581, "my": 0.96294}, "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": 3354.0} | |
| {"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.78182, "de": 1.91689, "vi": 1.60052, "tr": 1.80642, "fi": 2.02283, "ru": 1.19548, "el": 1.19895, "hi": 1.00541, "ta": 0.85529, "th": 0.95537, "zh": 2.76864, "my": 0.96985}, "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": 1.50646, "t": 3359.5} | |
| {"event": "train", "step": 7000, "phase": 2, "objective": "global", "lm_loss": 1.27375, "bpb": 1.83763, "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": 0.0001157, "router_trainable": true, "sec_per_step": 0.278, "alpha": 0.03, "t": 3359.7} | |
| {"event": "train", "step": 7100, "phase": 2, "objective": "global", "lm_loss": 1.1895, "bpb": 1.71609, "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": 0.4708, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 3379.9} | |
| {"event": "train", "step": 7200, "phase": 2, "objective": "global", "lm_loss": 1.09231, "bpb": 1.57586, "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": 0.4476, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 3403.3} | |
| {"event": "train", "step": 7300, "phase": 2, "objective": "global", "lm_loss": 1.17203, "bpb": 1.69088, "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": 0.4629, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 3426.0} | |
| {"event": "train", "step": 7400, "phase": 2, "objective": "global", "lm_loss": 1.08828, "bpb": 1.57005, "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": 0.449, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 3446.2} | |
| {"event": "train", "step": 7500, "phase": 2, "objective": "global", "lm_loss": 1.11931, "bpb": 1.61482, "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": 0.4559, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 3468.9} | |
| {"event": "train", "step": 7600, "phase": 2, "objective": "global", "lm_loss": 1.10432, "bpb": 1.59319, "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": 0.4686, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.222, "alpha": 0.03, "t": 3491.1} | |
| {"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 3511.9} | |
| {"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.09316, "bpb": 1.57709, "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": 0.4473, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.21, "alpha": 0.03, "t": 3512.1} | |
| {"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.16574, "bpb": 1.68181, "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": 0.4705, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 3535.4} | |
| {"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.14771, "bpb": 1.65579, "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": 0.4516, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 3555.6} | |
| {"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": 1.759737}, "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": 1.885159}, "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": 1.550676}, "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": 1.780346}, "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": 1.979678}, "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": 1.175738}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.174093}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.981011}, "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": 0.831983}, "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": 0.938604}, "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": 2.712014}, "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": 0.937626}}, "bpb": {"en": 1.75974, "de": 1.88516, "vi": 1.55068, "tr": 1.78035, "fi": 1.97968, "ru": 1.17574, "el": 1.17409, "hi": 0.98101, "ta": 0.83198, "th": 0.9386, "zh": 2.71201, "my": 0.93763}, "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": 3580.3} | |
| {"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.75995, "de": 1.88908, "vi": 1.56738, "tr": 1.77429, "fi": 1.98994, "ru": 1.17349, "el": 1.17307, "hi": 0.98717, "ta": 0.83603, "th": 0.93637, "zh": 2.72444, "my": 0.94423}, "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": 1.47962, "t": 3584.2} | |
| {"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.11915, "bpb": 1.61459, "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": 0.5011, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.288, "alpha": 0.03, "t": 3584.4} | |
| {"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.12559, "bpb": 1.62389, "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": 0.4924, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 3607.5} | |
| {"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.18221, "bpb": 1.70557, "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": 0.4884, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.23, "alpha": 0.03, "t": 3630.5} | |
| {"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.1134, "bpb": 1.60629, "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": 0.4524, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 3650.7} | |
| {"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.13227, "bpb": 1.63351, "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": 0.4868, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.229, "alpha": 0.03, "t": 3673.6} | |
| {"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.12806, "bpb": 1.62744, "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": 0.4961, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 3696.8} | |
| {"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.08699, "bpb": 1.56819, "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": 6.33e-05, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 3717.0} | |
| {"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 1.11141, "bpb": 1.60343, "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": 0.478, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 3739.8} | |
| {"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.09764, "bpb": 1.58355, "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": 0.4809, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.205, "alpha": 0.03, "t": 3760.2} | |
| {"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.08993, "bpb": 1.57244, "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": 0.4758, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 3782.8} | |
| {"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": 1.742854}, "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": 1.865004}, "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": 1.524138}, "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": 1.75853}, "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": 1.952908}, "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": 1.157212}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.157728}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.964293}, "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": 0.817266}, "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": 0.923839}, "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": 2.658749}, "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": 0.916331}}, "bpb": {"en": 1.74285, "de": 1.865, "vi": 1.52414, "tr": 1.75853, "fi": 1.95291, "ru": 1.15721, "el": 1.15773, "hi": 0.96429, "ta": 0.81727, "th": 0.92384, "zh": 2.65875, "my": 0.91633}, "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": 3807.6} | |
| {"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.74293, "de": 1.86904, "vi": 1.5413, "tr": 1.75264, "fi": 1.96364, "ru": 1.1552, "el": 1.15637, "hi": 0.97063, "ta": 0.82131, "th": 0.92145, "zh": 2.67005, "my": 0.9243}, "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": 1.4574, "t": 3811.3} | |
| {"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.1415, "bpb": 1.64684, "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": 0.4666, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.287, "alpha": 0.03, "t": 3811.5} | |
| {"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.06572, "bpb": 1.53751, "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": 0.4621, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 3833.9} | |
| {"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 1.17172, "bpb": 1.69043, "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": 0.5078, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.209, "alpha": 0.03, "t": 3854.8} | |
| {"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 1.02546, "bpb": 1.47943, "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": 0.4575, "lr": 4.71e-05, "router_trainable": true, "sec_per_step": 0.23, "alpha": 0.03, "t": 3877.8} | |
| {"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.13989, "bpb": 1.64452, "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": 4.52e-05, "router_trainable": true, "sec_per_step": 0.204, "alpha": 0.03, "t": 3898.2} | |
| {"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.07073, "bpb": 1.54473, "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": 0.4605, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 3920.8} | |
| {"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 1.14274, "bpb": 1.64862, "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": 0.5038, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 3944.1} | |
| {"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 1.11718, "bpb": 1.61175, "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": 0.4695, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.201, "alpha": 0.03, "t": 3964.2} | |
| {"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 1.08038, "bpb": 1.55866, "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": 0.4594, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 3986.9} | |
| {"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.09688, "bpb": 1.58247, "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": 0.4638, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.204, "alpha": 0.03, "t": 4007.3} | |
| {"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": 1.734159}, "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": 1.853131}, "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": 1.506734}, "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": 1.745538}, "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": 1.936887}, "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": 1.141592}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.142875}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.949643}, "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": 0.80819}, "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": 0.909108}, "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": 2.622728}, "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": 0.90155}}, "bpb": {"en": 1.73416, "de": 1.85313, "vi": 1.50673, "tr": 1.74554, "fi": 1.93689, "ru": 1.14159, "el": 1.14288, "hi": 0.94964, "ta": 0.80819, "th": 0.90911, "zh": 2.62273, "my": 0.90155}, "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": 4031.8} | |
| {"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.73364, "de": 1.85751, "vi": 1.52451, "tr": 1.73919, "fi": 1.94816, "ru": 1.13913, "el": 1.14196, "hi": 0.95642, "ta": 0.81209, "th": 0.90764, "zh": 2.63309, "my": 0.90899}, "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": 1.44186, "t": 4035.5} | |
| {"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.10415, "bpb": 1.59296, "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": 0.4678, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.284, "alpha": 0.03, "t": 4035.7} | |
| {"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 1.14364, "bpb": 1.64992, "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": 0.5388, "lr": 3.49e-05, "router_trainable": true, "sec_per_step": 0.2, "alpha": 0.03, "t": 4055.7} | |
| {"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.0688, "bpb": 1.54196, "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": 0.4854, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.199, "alpha": 0.03, "t": 4075.6} | |
| {"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 1.01424, "bpb": 1.46324, "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": 0.4756, "lr": 3.3e-05, "router_trainable": true, "sec_per_step": 0.215, "alpha": 0.03, "t": 4097.1} | |
| {"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.11567, "bpb": 1.60956, "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": 0.4736, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 4121.2} | |
| {"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.06439, "bpb": 1.53559, "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": 0.4904, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.245, "alpha": 0.03, "t": 4145.7} | |
| {"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 1.02206, "bpb": 1.47453, "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": 3.1e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 4169.3} | |
| {"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.08472, "bpb": 1.56492, "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": 0.4916, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 4193.7} | |
| {"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.0691, "bpb": 1.54238, "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": 0.5035, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 4218.2} | |
| {"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.07976, "bpb": 1.55777, "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": 0.4857, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.216, "alpha": 0.03, "t": 4239.7} | |
| {"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": 1.726671}, "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": 1.845727}, "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": 1.498326}, "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": 1.7359}, "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": 1.925667}, "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": 1.131095}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.135689}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.93866}, "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": 0.800298}, "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": 0.899546}, "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": 2.597827}, "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": 0.889353}}, "bpb": {"en": 1.72667, "de": 1.84573, "vi": 1.49833, "tr": 1.7359, "fi": 1.92567, "ru": 1.1311, "el": 1.13569, "hi": 0.93866, "ta": 0.8003, "th": 0.89955, "zh": 2.59783, "my": 0.88935}, "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": 4265.1} | |
| {"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.72651, "de": 1.85008, "vi": 1.51601, "tr": 1.73008, "fi": 1.93667, "ru": 1.12912, "el": 1.13483, "hi": 0.94569, "ta": 0.8044, "th": 0.89879, "zh": 2.60751, "my": 0.89634}, "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": 1.43134, "t": 4268.8} | |
| {"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": 1.726671}, "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": 1.845727}, "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": 1.498326}, "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": 1.7359}, "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": 1.925667}, "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": 1.131095}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.135689}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.93866}, "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": 0.800298}, "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": 0.899546}, "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": 2.597827}, "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": 0.889353}}, "bpb": {"en": 1.72667, "de": 1.84573, "vi": 1.49833, "tr": 1.7359, "fi": 1.92567, "ru": 1.1311, "el": 1.13569, "hi": 0.93866, "ta": 0.8003, "th": 0.89955, "zh": 2.59783, "my": 0.88935}, "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": 4271.4} | |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s2/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s2/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s2/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s2/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s2/th0.70"], "t": 4286.8} | |
| {"event": "final", "run": "exp32_onechunk_s2", "condition": "A", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 1.000861920358559, "cps": 136.078125, "bpb": 1.7225185745598732, "bytes_per_chunk": 1024.0}, "de": {"cpc": 1.0160229873861588, "cps": 162.984375, "bpb": 1.8479500363787047, "bytes_per_chunk": 1024.0}, "vi": {"cpc": 1.3167344724708434, "cps": 189.640625, "bpb": 1.5163563388896506, "bytes_per_chunk": 1024.0}, "tr": {"cpc": 1.0905374799756946, "cps": 154.234375, "bpb": 1.7299184720785474, "bytes_per_chunk": 1024.0}, "fi": {"cpc": 1.040455052894583, "cps": 152.90625, "bpb": 1.9428830730448525, "bytes_per_chunk": 1024.0}, "ru": {"cpc": 1.8237467018469657, "cps": 270.0, "bpb": 1.1305190607189484, "bytes_per_chunk": 1024.0}, "el": {"cpc": 1.8157189494713248, "cps": 291.125, "bpb": 1.1316378938541363, "bytes_per_chunk": 1024.0}, "hi": {"cpc": 2.566002971768202, "cps": 337.2890625, "bpb": 0.9472820210429548, "bytes_per_chunk": 1024.0}, "ta": {"cpc": 2.7234554693768387, "cps": 397.7734375, "bpb": 0.8032655335304415, "bytes_per_chunk": 1024.0}, "th": {"cpc": 2.8650917003419334, "cps": 360.0390625, "bpb": 0.9013324756818163, "bytes_per_chunk": 1024.0}, "zh": {"cpc": 2.7384016254656283, "cps": 126.3515625, "bpb": 2.608464971722008, "bytes_per_chunk": 1024.0}, "my": {"cpc": 2.8363415663922313, "cps": 419.8671875, "bpb": 0.8951136743924004, "bytes_per_chunk": 1024.0}}, "gini_cpc": 0.22151585453568945, "gini_cps": 0.2338532044751218, "bpic": 1.0, "seed": 2, "mean_boundary_entropy": 0.001020369934849441, "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": 4294.2} | |