| {"event": "start", "run": "exp32_onechunk_s1", "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": 14124.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.811584, "cps": 110.34375, "mean_p": 0.52085, "boundary_entropy": 0.686717, "bpb": 8.07852}, "de": {"chunks": 18700.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.910729, "cps": 146.09375, "mean_p": 0.525595, "boundary_entropy": 0.687217, "bpb": 8.13096}, "vi": {"chunks": 21351.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.158177, "cps": 166.804688, "mean_p": 0.524018, "boundary_entropy": 0.687874, "bpb": 8.101211}, "tr": {"chunks": 17846.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.985803, "cps": 139.421875, "mean_p": 0.529939, "boundary_entropy": 0.686373, "bpb": 8.170882}, "fi": {"chunks": 17574.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.934241, "cps": 137.296875, "mean_p": 0.526553, "boundary_entropy": 0.686826, "bpb": 8.065992}, "ru": {"chunks": 34128.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.80095, "cps": 266.625, "mean_p": 0.540146, "boundary_entropy": 0.687211, "bpb": 8.223579}, "el": {"chunks": 36913.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.798616, "cps": 288.382812, "mean_p": 0.535776, "boundary_entropy": 0.688142, "bpb": 8.073042}, "hi": {"chunks": 34684.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.061456, "cps": 270.96875, "mean_p": 0.509835, "boundary_entropy": 0.69075, "bpb": 8.178762}, "ta": {"chunks": 50678.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.710778, "cps": 395.921875, "mean_p": 0.522012, "boundary_entropy": 0.690366, "bpb": 8.093916}, "th": {"chunks": 45409.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.823065, "cps": 354.757812, "mean_p": 0.531604, "boundary_entropy": 0.689071, "bpb": 7.971837}, "zh": {"chunks": 4820.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.816119, "cps": 37.65625, "mean_p": 0.492076, "boundary_entropy": 0.68644, "bpb": 8.091891}, "my": {"chunks": 28215.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.489075, "cps": 220.429688, "mean_p": 0.506229, "boundary_entropy": 0.690851, "bpb": 8.16653}}, "bpb": {"en": 8.07852, "de": 8.13096, "vi": 8.10121, "tr": 8.17088, "fi": 8.06599, "ru": 8.22358, "el": 8.07304, "hi": 8.17876, "ta": 8.09392, "th": 7.97184, "zh": 8.09189, "my": 8.16653}, "gini_cpc": 0.247425, "gini_cps": 0.271932, "bpic": 1.182896, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1221621360827407, "vi": 1.4270575576222595, "tr": 1.2146656456674263, "fi": 1.1511320610100555, "ru": 2.2190548395702145, "el": 2.2161793753498933, "hi": 2.540039766651812, "ta": 3.3401072258858213, "th": 3.478462165603346, "zh": 1.0055878258102513, "my": 1.8347761655608366}, "t": 6.0} |
| {"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.62517, "bpb": 8.11541, "chunk_loss": 0.017265, "F": 0.00098, "G": 0.52249, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.69066, "router_noise": 0.0, "grad_norm": 4.9749, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.01, "alpha": 0.03, "t": 7.4} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14169.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.81417, "cps": 110.695312, "mean_p": 0.521063, "boundary_entropy": 0.6867, "bpb": 8.06238}, "de": {"chunks": 18732.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.912288, "cps": 146.34375, "mean_p": 0.525786, "boundary_entropy": 0.687198, "bpb": 8.116294}, "vi": {"chunks": 21376.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.159533, "cps": 167.0, "mean_p": 0.524194, "boundary_entropy": 0.687857, "bpb": 8.092582}, "tr": {"chunks": 17865.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.986853, "cps": 139.570312, "mean_p": 0.530065, "boundary_entropy": 0.686358, "bpb": 8.159145}, "fi": {"chunks": 17587.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.934932, "cps": 137.398438, "mean_p": 0.52666, "boundary_entropy": 0.686814, "bpb": 8.051263}, "ru": {"chunks": 34131.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.801108, "cps": 266.648438, "mean_p": 0.540307, "boundary_entropy": 0.687185, "bpb": 8.20486}, "el": {"chunks": 36914.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.798665, "cps": 288.390625, "mean_p": 0.535775, "boundary_entropy": 0.688142, "bpb": 8.066066}, "hi": {"chunks": 34616.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.057415, "cps": 270.4375, "mean_p": 0.509782, "boundary_entropy": 0.690752, "bpb": 8.152452}, "ta": {"chunks": 50676.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.710671, "cps": 395.90625, "mean_p": 0.52199, "boundary_entropy": 0.690367, "bpb": 8.073703}, "th": {"chunks": 45415.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.823438, "cps": 354.804688, "mean_p": 0.531645, "boundary_entropy": 0.689066, "bpb": 7.946902}, "zh": {"chunks": 4825.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.816966, "cps": 37.695312, "mean_p": 0.492103, "boundary_entropy": 0.686441, "bpb": 8.089827}, "my": {"chunks": 28265.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.491714, "cps": 220.820312, "mean_p": 0.506265, "boundary_entropy": 0.69085, "bpb": 8.16501}}, "bpb": {"en": 8.06238, "de": 8.11629, "vi": 8.09258, "tr": 8.15915, "fi": 8.05126, "ru": 8.20486, "el": 8.06607, "hi": 8.15245, "ta": 8.0737, "th": 7.9469, "zh": 8.08983, "my": 8.16501}, "gini_cpc": 0.24698, "gini_cps": 0.271565, "bpic": 1.182425, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.120512386813916, "vi": 1.4241909402257453, "tr": 1.2120970353975486, "fi": 1.1483249475630715, "ru": 2.2122016830007762, "el": 2.2092007532559874, "hi": 2.527008653867027, "ta": 3.3293678225129804, "th": 3.467872910237411, "zh": 1.0034339596153128, "my": 1.8321901062140002}, "t": 10.9} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14270.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.819974, "cps": 111.484375, "mean_p": 0.52148, "boundary_entropy": 0.686668, "bpb": 8.027654}, "de": {"chunks": 18790.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.915112, "cps": 146.796875, "mean_p": 0.526157, "boundary_entropy": 0.687162, "bpb": 8.083743}, "vi": {"chunks": 21427.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.1623, "cps": 167.398438, "mean_p": 0.524537, "boundary_entropy": 0.687824, "bpb": 8.072316}, "tr": {"chunks": 17881.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.987737, "cps": 139.695312, "mean_p": 0.5303, "boundary_entropy": 0.686328, "bpb": 8.133799}, "fi": {"chunks": 17609.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.936101, "cps": 137.570312, "mean_p": 0.526884, "boundary_entropy": 0.68679, "bpb": 8.020699}, "ru": {"chunks": 34138.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.801478, "cps": 266.703125, "mean_p": 0.540611, "boundary_entropy": 0.687136, "bpb": 8.167214}, "el": {"chunks": 36914.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.798665, "cps": 288.390625, "mean_p": 0.535808, "boundary_entropy": 0.688138, "bpb": 8.052915}, "hi": {"chunks": 34559.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.054027, "cps": 269.992188, "mean_p": 0.509746, "boundary_entropy": 0.690753, "bpb": 8.109531}, "ta": {"chunks": 50675.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.710618, "cps": 395.898438, "mean_p": 0.521971, "boundary_entropy": 0.690369, "bpb": 8.038336}, "th": {"chunks": 45426.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.824122, "cps": 354.890625, "mean_p": 0.531725, "boundary_entropy": 0.689056, "bpb": 7.908335}, "zh": {"chunks": 4842.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.819844, "cps": 37.828125, "mean_p": 0.492166, "boundary_entropy": 0.686443, "bpb": 8.086839}, "my": {"chunks": 28349.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.496147, "cps": 221.476562, "mean_p": 0.506324, "boundary_entropy": 0.690848, "bpb": 8.153007}}, "bpb": {"en": 8.02765, "de": 8.08374, "vi": 8.07232, "tr": 8.1338, "fi": 8.0207, "ru": 8.16721, "el": 8.05292, "hi": 8.10953, "ta": 8.03834, "th": 7.90834, "zh": 8.08684, "my": 8.15301}, "gini_cpc": 0.246137, "gini_cps": 0.270884, "bpic": 1.181301, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.116026533389719, "vi": 1.4174846831107384, "tr": 1.204595950424769, "fi": 1.1416236500307693, "ru": 2.196994687824153, "el": 2.1935645040563476, "hi": 2.504991412693636, "ta": 3.305738036900767, "th": 3.4441620603926366, "zh": 0.999842261299485, "my": 1.8246287556590908}, "t": 15.5} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14575.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.837499, "cps": 113.867188, "mean_p": 0.523021, "boundary_entropy": 0.686541, "bpb": 7.90254}, "de": {"chunks": 18954.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.923099, "cps": 148.078125, "mean_p": 0.527504, "boundary_entropy": 0.687024, "bpb": 7.967272}, "vi": {"chunks": 21621.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.172823, "cps": 168.914062, "mean_p": 0.52574, "boundary_entropy": 0.687704, "bpb": 8.001472}, "tr": {"chunks": 17955.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.991825, "cps": 140.273438, "mean_p": 0.531156, "boundary_entropy": 0.686219, "bpb": 8.045495}, "fi": {"chunks": 17692.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.940514, "cps": 138.21875, "mean_p": 0.52772, "boundary_entropy": 0.686698, "bpb": 7.91367}, "ru": {"chunks": 34162.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.802744, "cps": 266.890625, "mean_p": 0.541788, "boundary_entropy": 0.686944, "bpb": 8.042532}, "el": {"chunks": 36895.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.797739, "cps": 288.242188, "mean_p": 0.535746, "boundary_entropy": 0.688146, "bpb": 8.010024}, "hi": {"chunks": 34477.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.049153, "cps": 269.351562, "mean_p": 0.509684, "boundary_entropy": 0.690755, "bpb": 7.994748}, "ta": {"chunks": 50676.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.710671, "cps": 395.90625, "mean_p": 0.521925, "boundary_entropy": 0.690372, "bpb": 7.936321}, "th": {"chunks": 45454.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.825863, "cps": 355.109375, "mean_p": 0.531938, "boundary_entropy": 0.68903, "bpb": 7.791962}, "zh": {"chunks": 4893.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.82848, "cps": 38.226562, "mean_p": 0.49238, "boundary_entropy": 0.68645, "bpb": 8.075175}, "my": {"chunks": 28654.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.512244, "cps": 223.859375, "mean_p": 0.506531, "boundary_entropy": 0.690843, "bpb": 8.116142}}, "bpb": {"en": 7.90254, "de": 7.96727, "vi": 8.00147, "tr": 8.04549, "fi": 7.91367, "ru": 8.04253, "el": 8.01002, "hi": 7.99475, "ta": 7.93632, "th": 7.79196, "zh": 8.07518, "my": 8.11614}, "gini_cpc": 0.243668, "gini_cps": 0.268708, "bpic": 1.177213, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1022091852474605, "vi": 1.4003873904930448, "tr": 1.1842691463721648, "fi": 1.1230021921591988, "ru": 2.1525320709459304, "el": 2.1465560164310484, "hi": 2.446752004159456, "ta": 3.2366252265651054, "th": 3.3741671981662593, "zh": 0.9892301168855233, "my": 1.805666069564405}, "t": 20.9} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 15446.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.887548, "cps": 120.671875, "mean_p": 0.528358, "boundary_entropy": 0.686026, "bpb": 7.492177}, "de": {"chunks": 19432.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.946379, "cps": 151.8125, "mean_p": 0.532106, "boundary_entropy": 0.686498, "bpb": 7.583274}, "vi": {"chunks": 22235.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.20613, "cps": 173.710938, "mean_p": 0.529993, "boundary_entropy": 0.687232, "bpb": 7.76567}, "tr": {"chunks": 18178.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.004143, "cps": 142.015625, "mean_p": 0.533998, "boundary_entropy": 0.685833, "bpb": 7.74741}, "fi": {"chunks": 17952.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.954335, "cps": 140.25, "mean_p": 0.530526, "boundary_entropy": 0.68637, "bpb": 7.562521}, "ru": {"chunks": 34240.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.80686, "cps": 267.5, "mean_p": 0.545548, "boundary_entropy": 0.686295, "bpb": 7.637631}, "el": {"chunks": 36889.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.797447, "cps": 288.195312, "mean_p": 0.535736, "boundary_entropy": 0.688145, "bpb": 7.794848}, "hi": {"chunks": 33989.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.020149, "cps": 265.539062, "mean_p": 0.509316, "boundary_entropy": 0.690767, "bpb": 7.58418}, "ta": {"chunks": 50635.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.708478, "cps": 395.585938, "mean_p": 0.521583, "boundary_entropy": 0.690399, "bpb": 7.550281}, "th": {"chunks": 45522.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.83009, "cps": 355.640625, "mean_p": 0.532465, "boundary_entropy": 0.688966, "bpb": 7.348368}, "zh": {"chunks": 5168.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.875042, "cps": 40.375, "mean_p": 0.493377, "boundary_entropy": 0.686485, "bpb": 8.029066}, "my": {"chunks": 29682.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.566498, "cps": 231.890625, "mean_p": 0.507222, "boundary_entropy": 0.690826, "bpb": 7.936474}}, "bpb": {"en": 7.49218, "de": 7.58327, "vi": 7.76567, "tr": 7.74741, "fi": 7.56252, "ru": 7.63763, "el": 7.79485, "hi": 7.58418, "ta": 7.55028, "th": 7.34837, "zh": 8.02907, "my": 7.93647}, "gini_cpc": 0.234787, "gini_cps": 0.261817, "bpic": 1.165204, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0662847165005218, "vi": 1.3589456303685157, "tr": 1.1313673396470736, "fi": 1.0752489963112144, "ru": 2.035788381140253, "el": 2.0251823181943163, "hi": 2.276100342101582, "ta": 3.051640953131906, "th": 3.188661065166275, "zh": 0.9859097284800121, "my": 1.7649722209397802}, "t": 28.0} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 16605.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.954146, "cps": 129.726562, "mean_p": 0.543932, "boundary_entropy": 0.68389, "bpb": 6.616142}, "de": {"chunks": 20125.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.98013, "cps": 157.226562, "mean_p": 0.545561, "boundary_entropy": 0.68449, "bpb": 6.744444}, "vi": {"chunks": 23130.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.254679, "cps": 180.703125, "mean_p": 0.542576, "boundary_entropy": 0.685395, "bpb": 7.292494}, "tr": {"chunks": 18646.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.029995, "cps": 145.671875, "mean_p": 0.542107, "boundary_entropy": 0.684591, "bpb": 7.054267}, "fi": {"chunks": 18496.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.983254, "cps": 144.5, "mean_p": 0.539696, "boundary_entropy": 0.685074, "bpb": 6.791509}, "ru": {"chunks": 34430.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.816887, "cps": 268.984375, "mean_p": 0.556191, "boundary_entropy": 0.684196, "bpb": 6.580367}, "el": {"chunks": 36896.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.797788, "cps": 288.25, "mean_p": 0.535654, "boundary_entropy": 0.688158, "bpb": 7.312595}, "hi": {"chunks": 33359.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.982704, "cps": 260.617188, "mean_p": 0.508716, "boundary_entropy": 0.690791, "bpb": 6.720734}, "ta": {"chunks": 50625.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.707943, "cps": 395.507812, "mean_p": 0.521319, "boundary_entropy": 0.690422, "bpb": 6.826574}, "th": {"chunks": 45741.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.843705, "cps": 357.351562, "mean_p": 0.533308, "boundary_entropy": 0.688892, "bpb": 6.569245}, "zh": {"chunks": 6596.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.11683, "cps": 51.53125, "mean_p": 0.498176, "boundary_entropy": 0.686607, "bpb": 7.934873}, "my": {"chunks": 36169.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.908856, "cps": 282.570312, "mean_p": 0.511078, "boundary_entropy": 0.690715, "bpb": 7.525181}}, "bpb": {"en": 6.61614, "de": 6.74444, "vi": 7.29249, "tr": 7.05427, "fi": 6.79151, "ru": 6.58037, "el": 7.31259, "hi": 6.72073, "ta": 6.82657, "th": 6.56925, "zh": 7.93487, "my": 7.52518}, "gini_cpc": 0.21662, "gini_cps": 0.248447, "bpic": 1.126058, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0272324309632486, "vi": 1.3149757111891704, "tr": 1.0794943378552349, "fi": 1.0305075395304129, "ru": 1.9042021389431134, "el": 1.8841856016214826, "hi": 2.07798874379036, "ta": 2.8380811355586144, "th": 2.980367576383255, "zh": 1.1705027667727856, "my": 2.00059125350243}, "t": 39.4} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17275.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.992645, "cps": 134.960938, "mean_p": 0.601673, "boundary_entropy": 0.667301, "bpb": 5.919932}, "de": {"chunks": 20723.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.009253, "cps": 161.898438, "mean_p": 0.603515, "boundary_entropy": 0.667439, "bpb": 6.09096}, "vi": {"chunks": 23930.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.298074, "cps": 186.953125, "mean_p": 0.581867, "boundary_entropy": 0.675488, "bpb": 6.917889}, "tr": {"chunks": 19510.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.077722, "cps": 152.421875, "mean_p": 0.593075, "boundary_entropy": 0.670841, "bpb": 6.480591}, "fi": {"chunks": 19330.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.02759, "cps": 151.015625, "mean_p": 0.595129, "boundary_entropy": 0.670025, "bpb": 6.137241}, "ru": {"chunks": 34545.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.822955, "cps": 269.882812, "mean_p": 0.563891, "boundary_entropy": 0.682613, "bpb": 5.670681}, "el": {"chunks": 37200.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.8126, "cps": 290.625, "mean_p": 0.547528, "boundary_entropy": 0.686328, "bpb": 6.632483}, "hi": {"chunks": 23104.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.373195, "cps": 180.5, "mean_p": 0.502344, "boundary_entropy": 0.690976, "bpb": 5.685825}, "ta": {"chunks": 50058.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.677614, "cps": 391.078125, "mean_p": 0.51702, "boundary_entropy": 0.690767, "bpb": 5.746647}, "th": {"chunks": 46034.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.861921, "cps": 359.640625, "mean_p": 0.532859, "boundary_entropy": 0.689093, "bpb": 5.499508}, "zh": {"chunks": 13152.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.226888, "cps": 102.75, "mean_p": 0.52394, "boundary_entropy": 0.685793, "bpb": 7.635509}, "my": {"chunks": 52775.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.785254, "cps": 412.304688, "mean_p": 0.527662, "boundary_entropy": 0.689673, "bpb": 6.658989}}, "bpb": {"en": 5.91993, "de": 6.09096, "vi": 6.91789, "tr": 6.48059, "fi": 6.13724, "ru": 5.67068, "el": 6.63248, "hi": 5.68583, "ta": 5.74665, "th": 5.49951, "zh": 7.63551, "my": 6.65899}, "gini_cpc": 0.222014, "gini_cps": 0.244329, "bpic": 1.073105, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0167315118657954, "vi": 1.3076924634878992, "tr": 1.085707360510563, "fi": 1.0352042224270508, "ru": 1.8364624249204817, "el": 1.8260310535079312, "hi": 1.3833694071300935, "ta": 2.6974542563637707, "th": 2.8831265954608902, "zh": 2.2433881509789404, "my": 2.8058918658331264}, "t": 56.1} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17393.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.999425, "cps": 135.882812, "mean_p": 0.791879, "boundary_entropy": 0.505486, "bpb": 5.491072}, "de": {"chunks": 20836.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.014757, "cps": 162.78125, "mean_p": 0.791695, "boundary_entropy": 0.506938, "bpb": 5.653666}, "vi": {"chunks": 24205.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.312992, "cps": 189.101562, "mean_p": 0.73336, "boundary_entropy": 0.572812, "bpb": 6.445454}, "tr": {"chunks": 19723.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.089488, "cps": 154.085938, "mean_p": 0.779881, "boundary_entropy": 0.520812, "bpb": 6.071185}, "fi": {"chunks": 19557.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.039658, "cps": 152.789062, "mean_p": 0.780447, "boundary_entropy": 0.519557, "bpb": 5.689945}, "ru": {"chunks": 34556.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823536, "cps": 269.96875, "mean_p": 0.646752, "boundary_entropy": 0.647725, "bpb": 4.932327}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.589214, "boundary_entropy": 0.675225, "bpb": 5.326845}, "hi": {"chunks": 43165.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.565527, "cps": 337.226562, "mean_p": 0.564824, "boundary_entropy": 0.682937, "bpb": 4.742515}, "ta": {"chunks": 50903.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.722814, "cps": 397.679688, "mean_p": 0.562865, "boundary_entropy": 0.683704, "bpb": 4.973289}, "th": {"chunks": 46084.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.86503, "cps": 360.03125, "mean_p": 0.618054, "boundary_entropy": 0.663439, "bpb": 4.83704}, "zh": {"chunks": 16167.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.737386, "cps": 126.304688, "mean_p": 0.641479, "boundary_entropy": 0.64723, "bpb": 6.846068}, "my": {"chunks": 53718.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.835022, "cps": 419.671875, "mean_p": 0.547031, "boundary_entropy": 0.687071, "bpb": 4.852362}}, "bpb": {"en": 5.49107, "de": 5.65367, "vi": 6.44545, "tr": 6.07118, "fi": 5.68995, "ru": 4.93233, "el": 5.32685, "hi": 4.74252, "ta": 4.97329, "th": 4.83704, "zh": 6.84607, "my": 4.85236}, "gini_cpc": 0.221723, "gini_cps": 0.234026, "bpic": 1.000547, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.015340161301107, "vi": 1.3137464886433134, "tr": 1.0901143246637082, "fi": 1.0402553919883355, "ru": 1.8245840508123017, "el": 1.816762886083451, "hi": 2.5670025233460985, "ta": 2.7243790516998545, "th": 2.8666767619936353, "zh": 2.7389595527812265, "my": 2.8366521447502757}, "t": 88.3} |
| {"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.37003, "bpb": 4.86192, "chunk_loss": 0.00795, "F": 0.00098, "G": 0.78279, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.50152, "router_noise": 0.0, "grad_norm": 1.7019, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 1.145, "alpha": 0.03, "t": 121.8} |
| {"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": 0.846257, "boundary_entropy": 0.421303, "bpb": 4.359591}, "de": {"chunks": 20858.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.015828, "cps": 162.953125, "mean_p": 0.844366, "boundary_entropy": 0.419764, "bpb": 4.60278}, "vi": {"chunks": 24234.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.314565, "cps": 189.328125, "mean_p": 0.711698, "boundary_entropy": 0.587707, "bpb": 4.802098}, "tr": {"chunks": 19699.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.088162, "cps": 153.898438, "mean_p": 0.788789, "boundary_entropy": 0.491176, "bpb": 4.793253}, "fi": {"chunks": 19571.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040402, "cps": 152.898438, "mean_p": 0.826996, "boundary_entropy": 0.444716, "bpb": 4.611134}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.749717, "boundary_entropy": 0.558621, "bpb": 3.263767}, "el": {"chunks": 37258.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815427, "cps": 291.078125, "mean_p": 0.690442, "boundary_entropy": 0.613706, "bpb": 3.771699}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.683425, "boundary_entropy": 0.620218, "bpb": 2.832803}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.706656, "boundary_entropy": 0.60184, "bpb": 2.984118}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.715636, "boundary_entropy": 0.59356, "bpb": 2.880086}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.847105, "boundary_entropy": 0.422392, "bpb": 6.198407}, "my": {"chunks": 53722.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.835233, "cps": 419.703125, "mean_p": 0.683237, "boundary_entropy": 0.613231, "bpb": 3.073309}}, "bpb": {"en": 4.35959, "de": 4.60278, "vi": 4.8021, "tr": 4.79325, "fi": 4.61113, "ru": 3.26377, "el": 3.7717, "hi": 2.8328, "ta": 2.98412, "th": 2.88009, "zh": 6.19841, "my": 3.07331}, "gini_cpc": 0.221615, "gini_cps": 0.233929, "bpic": 1.0003, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.014953370056241, "vi": 1.313432612428937, "tr": 1.0872250816313866, "fi": 1.0395059211933828, "ru": 1.8221761311426539, "el": 1.8138631889422634, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8327916292253263}, "t": 150.0} |
| {"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.38505, "bpb": 3.4409, "chunk_loss": 0.00374, "F": 0.00098, "G": 0.9004, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.2982, "router_noise": 0.0, "grad_norm": 1.9327, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.936, "alpha": 0.03, "t": 215.4} |
| {"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": 0.955399, "boundary_entropy": 0.175415, "bpb": 3.528553}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.948927, "boundary_entropy": 0.187039, "bpb": 3.766754}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.919336, "boundary_entropy": 0.263466, "bpb": 3.085781}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.938352, "boundary_entropy": 0.201428, "bpb": 3.789451}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.956616, "boundary_entropy": 0.165736, "bpb": 3.802806}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.981877, "boundary_entropy": 0.084762, "bpb": 2.634994}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.958026, "boundary_entropy": 0.171367, "bpb": 2.684278}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.960952, "boundary_entropy": 0.163689, "bpb": 2.083255}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.954407, "boundary_entropy": 0.183216, "bpb": 1.957156}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.962572, "boundary_entropy": 0.157608, "bpb": 2.091091}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.955147, "boundary_entropy": 0.177247, "bpb": 4.740311}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.928075, "boundary_entropy": 0.254876, "bpb": 2.186837}}, "bpb": {"en": 3.52855, "de": 3.76675, "vi": 3.08578, "tr": 3.78945, "fi": 3.80281, "ru": 2.63499, "el": 2.68428, "hi": 2.08325, "ta": 1.95716, "th": 2.09109, "zh": 4.74031, "my": 2.18684}, "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": 266.2} |
| {"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.13229, "bpb": 3.07624, "chunk_loss": 0.001048, "F": 0.00098, "G": 0.97563, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.10775, "router_noise": 0.0, "grad_norm": 1.5961, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.888, "alpha": 0.03, "t": 304.2} |
| {"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.01282, "bpb": 2.90388, "chunk_loss": 0.000433, "F": 0.00098, "G": 0.9928, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.04162, "router_noise": 0.0, "grad_norm": 1.1669, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.788, "alpha": 0.03, "t": 382.9} |
| {"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.81705, "bpb": 2.62145, "chunk_loss": 0.000284, "F": 0.00098, "G": 0.99699, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.02011, "router_noise": 0.0, "grad_norm": 1.1634, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.722, "alpha": 0.03, "t": 455.2} |
| {"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.82452, "bpb": 2.63222, "chunk_loss": 0.000234, "F": 0.00098, "G": 0.99837, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.01191, "router_noise": 0.0, "grad_norm": 0.8703, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.669, "alpha": 0.03, "t": 522.1} |
| {"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.80594, "bpb": 2.60543, "chunk_loss": 0.000213, "F": 0.00098, "G": 0.99895, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00811, "router_noise": 0.0, "grad_norm": 0.8274, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.646, "alpha": 0.03, "t": 586.7} |
| {"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.65519, "bpb": 2.38793, "chunk_loss": 0.000198, "F": 0.00098, "G": 0.99938, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00514, "router_noise": 0.0, "grad_norm": 0.7183, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.572, "alpha": 0.03, "t": 643.9} |
| {"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.59543, "bpb": 2.30171, "chunk_loss": 0.000195, "F": 0.00098, "G": 0.99947, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00448, "router_noise": 0.0, "grad_norm": 0.7539, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.379, "alpha": 0.03, "t": 681.7} |
| {"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": 0.999494, "boundary_entropy": 0.004288, "bpb": 2.3855}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999435, "boundary_entropy": 0.00474, "bpb": 2.676373}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.999649, "boundary_entropy": 0.003115, "bpb": 2.144179}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999492, "boundary_entropy": 0.004319, "bpb": 2.670115}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999651, "boundary_entropy": 0.003107, "bpb": 2.890802}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999685, "boundary_entropy": 0.002839, "bpb": 1.775984}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999654, "boundary_entropy": 0.003082, "bpb": 1.873201}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.999761, "boundary_entropy": 0.00222, "bpb": 1.412445}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999737, "boundary_entropy": 0.002419, "bpb": 1.15156}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.999623, "boundary_entropy": 0.003326, "bpb": 1.375854}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999584, "boundary_entropy": 0.003623, "bpb": 3.552695}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999739, "boundary_entropy": 0.002382, "bpb": 1.307019}}, "bpb": {"en": 2.3855, "de": 2.67637, "vi": 2.14418, "tr": 2.67011, "fi": 2.8908, "ru": 1.77598, "el": 1.8732, "hi": 1.41244, "ta": 1.15156, "th": 1.37585, "zh": 3.55269, "my": 1.30702}, "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": 706.9} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.39954, "de": 2.6714, "vi": 2.14799, "tr": 2.66339, "fi": 2.90004, "ru": 1.77388, "el": 1.86711, "hi": 1.42037, "ta": 1.15042, "th": 1.37434, "zh": 3.55675, "my": 1.32056}, "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": 2.10382, "t": 711.7} |
| {"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.57254, "bpb": 2.2687, "chunk_loss": 0.00019, "F": 0.00098, "G": 0.99961, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00342, "router_noise": 0.0, "grad_norm": 0.6196, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.302, "alpha": 0.03, "t": 711.9} |
| {"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.48837, "bpb": 2.14726, "chunk_loss": 0.000186, "F": 0.00098, "G": 0.99972, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00257, "router_noise": 0.0, "grad_norm": 0.6745, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.465, "alpha": 0.03, "t": 758.4} |
| {"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.54648, "bpb": 2.23111, "chunk_loss": 0.000184, "F": 0.00098, "G": 0.99977, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00214, "router_noise": 0.0, "grad_norm": 0.6827, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.496, "alpha": 0.03, "t": 808.0} |
| {"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.42472, "bpb": 2.05544, "chunk_loss": 0.000183, "F": 0.00098, "G": 0.99981, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00184, "router_noise": 0.0, "grad_norm": 0.8012, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.486, "alpha": 0.03, "t": 856.6} |
| {"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.37302, "bpb": 1.98085, "chunk_loss": 0.000181, "F": 0.00098, "G": 0.99985, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00145, "router_noise": 0.0, "grad_norm": 0.6203, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.483, "alpha": 0.03, "t": 904.9} |
| {"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.44943, "bpb": 2.09109, "chunk_loss": 0.000181, "F": 0.00098, "G": 0.99986, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00134, "router_noise": 0.0, "grad_norm": 0.6465, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.493, "alpha": 0.03, "t": 954.2} |
| {"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.49282, "bpb": 2.15368, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99988, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00118, "router_noise": 0.0, "grad_norm": 0.6489, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.478, "alpha": 0.03, "t": 1002.0} |
| {"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.51482, "bpb": 2.18542, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00114, "router_noise": 0.0, "grad_norm": 0.6234, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.449, "alpha": 0.03, "t": 1046.9} |
| {"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.32179, "bpb": 1.90694, "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": 0.5376, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.462, "alpha": 0.03, "t": 1093.1} |
| {"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.30726, "bpb": 1.88597, "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": 0.5566, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.45, "alpha": 0.03, "t": 1138.2} |
| {"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": 0.999897, "boundary_entropy": 0.001045, "bpb": 2.126673}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999896, "boundary_entropy": 0.001057, "bpb": 2.343981}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001022, "bpb": 1.985167}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999898, "boundary_entropy": 0.001042, "bpb": 2.317885}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999899, "boundary_entropy": 0.001033, "bpb": 2.515348}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 1.462356}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001024, "bpb": 1.528537}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 1.244786}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 1.031114}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001023, "bpb": 1.178296}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999899, "boundary_entropy": 0.001027, "bpb": 3.215881}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001023, "bpb": 1.166352}}, "bpb": {"en": 2.12667, "de": 2.34398, "vi": 1.98517, "tr": 2.31789, "fi": 2.51535, "ru": 1.46236, "el": 1.52854, "hi": 1.24479, "ta": 1.03111, "th": 1.1783, "zh": 3.21588, "my": 1.16635}, "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": 1190.2} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.13533, "de": 2.33829, "vi": 1.99041, "tr": 2.31163, "fi": 2.52932, "ru": 1.46169, "el": 1.52361, "hi": 1.25282, "ta": 1.03082, "th": 1.17243, "zh": 3.22286, "my": 1.178}, "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": 1.8456, "t": 1196.3} |
| {"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.40656, "bpb": 2.02923, "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": 0.5473, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.586, "alpha": 0.03, "t": 1196.7} |
| {"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.42757, "bpb": 2.05954, "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": 0.5662, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.442, "alpha": 0.03, "t": 1241.0} |
| {"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.28304, "bpb": 1.85104, "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": 0.5149, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.471, "alpha": 0.03, "t": 1288.0} |
| {"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.43442, "bpb": 2.06943, "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": 0.5567, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.453, "alpha": 0.03, "t": 1333.3} |
| {"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.3369, "bpb": 1.92873, "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": 0.5928, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.465, "alpha": 0.03, "t": 1379.8} |
| {"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.30192, "bpb": 1.87827, "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": 0.5226, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.464, "alpha": 0.03, "t": 1426.2} |
| {"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.40722, "bpb": 2.03019, "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": 0.5623, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.447, "alpha": 0.03, "t": 1470.9} |
| {"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.25885, "bpb": 1.81614, "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": 0.4791, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.474, "alpha": 0.03, "t": 1518.4} |
| {"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.33066, "bpb": 1.91974, "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": 0.4836, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.464, "alpha": 0.03, "t": 1564.8} |
| {"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.39619, "bpb": 2.01428, "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": 0.5627, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.474, "alpha": 0.03, "t": 1612.2} |
| {"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 2.021902}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 2.198301}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001017, "bpb": 1.883058}, "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": 2.177204}, "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": 2.376204}, "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": 1.344308}, "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": 1.396025}, "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": 1.153428}, "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": 0.967682}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 1.098362}, "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": 3.036136}, "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": 1.093608}}, "bpb": {"en": 2.0219, "de": 2.1983, "vi": 1.88306, "tr": 2.1772, "fi": 2.3762, "ru": 1.34431, "el": 1.39603, "hi": 1.15343, "ta": 0.96768, "th": 1.09836, "zh": 3.03614, "my": 1.09361}, "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": 1660.9} |
| {"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 2.03227, "de": 2.18889, "vi": 1.889, "tr": 2.17113, "fi": 2.39147, "ru": 1.34412, "el": 1.39222, "hi": 1.16151, "ta": 0.96805, "th": 1.09399, "zh": 3.04526, "my": 1.10343}, "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": 1.73178, "t": 1667.0} |
| {"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.43601, "bpb": 2.07172, "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": 0.5179, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.553, "alpha": 0.03, "t": 1667.5} |
| {"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.30938, "bpb": 1.88904, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5093, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.458, "alpha": 0.03, "t": 1713.3} |
| {"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.34333, "bpb": 1.93802, "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": 0.5068, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.459, "alpha": 0.03, "t": 1759.1} |
| {"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.21052, "bpb": 1.74641, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4759, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.463, "alpha": 0.03, "t": 1805.4} |
| {"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.26345, "bpb": 1.82278, "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": 0.4878, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.454, "alpha": 0.03, "t": 1850.8} |
| {"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.24719, "bpb": 1.79932, "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": 0.4578, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.459, "alpha": 0.03, "t": 1896.7} |
| {"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.30712, "bpb": 1.88577, "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": 0.4806, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.448, "alpha": 0.03, "t": 1941.5} |
| {"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.22288, "bpb": 1.76424, "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": 0.5135, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.455, "alpha": 0.03, "t": 1987.1} |
| {"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.37072, "bpb": 1.97754, "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": 0.0002394, "router_trainable": true, "sec_per_step": 0.441, "alpha": 0.03, "t": 2031.1} |
| {"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.22087, "bpb": 1.76134, "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": 0.4581, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.464, "alpha": 0.03, "t": 2077.6} |
| {"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.961905}, "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": 2.131719}, "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": 1.809539}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 2.086071}, "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": 2.255816}, "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": 1.288024}, "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": 1.316495}, "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": 1.094673}, "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": 0.921813}, "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": 1.042572}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.930919}, "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": 1.041019}}, "bpb": {"en": 1.9619, "de": 2.13172, "vi": 1.80954, "tr": 2.08607, "fi": 2.25582, "ru": 1.28802, "el": 1.31649, "hi": 1.09467, "ta": 0.92181, "th": 1.04257, "zh": 2.93092, "my": 1.04102}, "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": 2124.8} |
| {"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.96811, "de": 2.12235, "vi": 1.81627, "tr": 2.08164, "fi": 2.27158, "ru": 1.28545, "el": 1.31243, "hi": 1.10313, "ta": 0.92311, "th": 1.04053, "zh": 2.9372, "my": 1.04987}, "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": 1.65931, "t": 2131.2} |
| {"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.23414, "bpb": 1.78049, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4731, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.54, "alpha": 0.03, "t": 2131.6} |
| {"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.30675, "bpb": 1.88525, "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": 0.487, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.454, "alpha": 0.03, "t": 2177.0} |
| {"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.11919, "bpb": 1.61465, "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": 0.4523, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.46, "alpha": 0.03, "t": 2223.0} |
| {"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.28561, "bpb": 1.85474, "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": 0.4565, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.471, "alpha": 0.03, "t": 2270.1} |
| {"event": "phase_transition", "step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 2315.9} |
| {"event": "train", "step": 4400, "phase": 2, "objective": "global", "lm_loss": 1.30292, "bpb": 1.87971, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4931, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.462, "alpha": 0.03, "t": 2316.3} |
| {"event": "train", "step": 4500, "phase": 2, "objective": "global", "lm_loss": 1.29443, "bpb": 1.86747, "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": 0.5066, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.462, "alpha": 0.03, "t": 2362.5} |
| {"event": "train", "step": 4600, "phase": 2, "objective": "global", "lm_loss": 1.22138, "bpb": 1.76208, "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": 0.4715, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.439, "alpha": 0.03, "t": 2406.3} |
| {"event": "train", "step": 4700, "phase": 2, "objective": "global", "lm_loss": 1.24529, "bpb": 1.79658, "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": 0.4557, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.478, "alpha": 0.03, "t": 2454.2} |
| {"event": "train", "step": 4800, "phase": 2, "objective": "global", "lm_loss": 1.25721, "bpb": 1.81377, "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": 0.4797, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.474, "alpha": 0.03, "t": 2501.6} |
| {"event": "train", "step": 4900, "phase": 2, "objective": "global", "lm_loss": 1.24074, "bpb": 1.79001, "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": 0.4866, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.46, "alpha": 0.03, "t": 2547.6} |
| {"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": 1.919019}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 2.070714}, "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": 1.740401}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 2.023419}, "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": 2.185722}, "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": 1.248271}, "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": 1.260194}, "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": 1.053749}, "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": 0.895854}, "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": 1.01166}, "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": 2.845529}, "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": 1.008299}}, "bpb": {"en": 1.91902, "de": 2.07071, "vi": 1.7404, "tr": 2.02342, "fi": 2.18572, "ru": 1.24827, "el": 1.26019, "hi": 1.05375, "ta": 0.89585, "th": 1.01166, "zh": 2.84553, "my": 1.0083}, "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": 2596.9} |
| {"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.92511, "de": 2.06331, "vi": 1.75046, "tr": 2.01774, "fi": 2.20199, "ru": 1.24657, "el": 1.25605, "hi": 1.06223, "ta": 0.89838, "th": 1.0107, "zh": 2.85469, "my": 1.01792}, "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": 1.60876, "t": 2603.0} |
| {"event": "train", "step": 5000, "phase": 2, "objective": "global", "lm_loss": 1.18836, "bpb": 1.71445, "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": 0.4847, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.558, "alpha": 0.03, "t": 2603.4} |
| {"event": "train", "step": 5100, "phase": 2, "objective": "global", "lm_loss": 1.17404, "bpb": 1.69379, "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": 0.4639, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.471, "alpha": 0.03, "t": 2650.5} |
| {"event": "train", "step": 5200, "phase": 2, "objective": "global", "lm_loss": 1.29667, "bpb": 1.8707, "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": 0.4989, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.465, "alpha": 0.03, "t": 2697.1} |
| {"event": "train", "step": 5300, "phase": 2, "objective": "global", "lm_loss": 1.2043, "bpb": 1.73744, "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": 0.4679, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.486, "alpha": 0.03, "t": 2745.7} |
| {"event": "train", "step": 5400, "phase": 2, "objective": "global", "lm_loss": 1.14544, "bpb": 1.65252, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4787, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.486, "alpha": 0.03, "t": 2794.2} |
| {"event": "train", "step": 5500, "phase": 2, "objective": "global", "lm_loss": 1.15977, "bpb": 1.67319, "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": 0.4353, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.492, "alpha": 0.03, "t": 2843.4} |
| {"event": "train", "step": 5600, "phase": 2, "objective": "global", "lm_loss": 1.18819, "bpb": 1.7142, "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": 0.4875, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.464, "alpha": 0.03, "t": 2889.8} |
| {"event": "train", "step": 5700, "phase": 2, "objective": "global", "lm_loss": 1.27542, "bpb": 1.84004, "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": 0.4981, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.455, "alpha": 0.03, "t": 2935.3} |
| {"event": "train", "step": 5800, "phase": 2, "objective": "global", "lm_loss": 1.28163, "bpb": 1.84901, "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": 0.4993, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.478, "alpha": 0.03, "t": 2983.1} |
| {"event": "train", "step": 5900, "phase": 2, "objective": "global", "lm_loss": 1.2282, "bpb": 1.77191, "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": 0.5491, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.689, "alpha": 0.03, "t": 3051.9} |
| {"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": 1.876228}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.018606}, "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": 1.689079}, "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": 1.960993}, "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": 2.133036}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001017, "bpb": 1.216594}, "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": 1.226907}, "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": 1.031199}, "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": 0.876282}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.987844}, "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": 2.794262}, "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": 0.976122}}, "bpb": {"en": 1.87623, "de": 2.01861, "vi": 1.68908, "tr": 1.96099, "fi": 2.13304, "ru": 1.21659, "el": 1.22691, "hi": 1.0312, "ta": 0.87628, "th": 0.98784, "zh": 2.79426, "my": 0.97612}, "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": 3117.8} |
| {"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.87968, "de": 2.01293, "vi": 1.70129, "tr": 1.9559, "fi": 2.14921, "ru": 1.21399, "el": 1.22428, "hi": 1.0386, "ta": 0.87753, "th": 0.98615, "zh": 2.80199, "my": 0.98561}, "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": 1.56893, "t": 3126.7} |
| {"event": "train", "step": 6000, "phase": 2, "objective": "global", "lm_loss": 1.24126, "bpb": 1.79076, "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": 0.4488, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.753, "alpha": 0.03, "t": 3127.2} |
| {"event": "train", "step": 6100, "phase": 2, "objective": "global", "lm_loss": 1.1919, "bpb": 1.71955, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.493, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.632, "alpha": 0.03, "t": 3190.4} |
| {"event": "train", "step": 6200, "phase": 2, "objective": "global", "lm_loss": 1.12627, "bpb": 1.62486, "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": 0.0001469, "router_trainable": true, "sec_per_step": 0.652, "alpha": 0.03, "t": 3255.6} |
| {"event": "train", "step": 6300, "phase": 2, "objective": "global", "lm_loss": 1.20824, "bpb": 1.74313, "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": 0.0001429, "router_trainable": true, "sec_per_step": 0.619, "alpha": 0.03, "t": 3317.5} |
| {"event": "train", "step": 6400, "phase": 2, "objective": "global", "lm_loss": 1.15282, "bpb": 1.66317, "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": 0.4635, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.622, "alpha": 0.03, "t": 3379.7} |
| {"event": "train", "step": 6500, "phase": 2, "objective": "global", "lm_loss": 1.1383, "bpb": 1.64222, "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": 0.4637, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.618, "alpha": 0.03, "t": 3441.5} |
| {"event": "train", "step": 6600, "phase": 2, "objective": "global", "lm_loss": 1.19032, "bpb": 1.71727, "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": 0.4589, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.634, "alpha": 0.03, "t": 3504.9} |
| {"event": "train", "step": 6700, "phase": 2, "objective": "global", "lm_loss": 1.26556, "bpb": 1.82582, "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": 0.5202, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.638, "alpha": 0.03, "t": 3568.7} |
| {"event": "train", "step": 6800, "phase": 2, "objective": "global", "lm_loss": 1.12411, "bpb": 1.62174, "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": 0.4696, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.64, "alpha": 0.03, "t": 3632.7} |
| {"event": "train", "step": 6900, "phase": 2, "objective": "global", "lm_loss": 1.19449, "bpb": 1.72329, "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": 0.479, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.645, "alpha": 0.03, "t": 3697.1} |
| {"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": 1.826424}, "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": 1.966216}, "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": 1.63983}, "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": 1.896031}, "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": 2.07507}, "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": 1.193735}, "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": 1.191317}, "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": 1.000439}, "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": 0.8572}, "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": 0.961261}, "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": 2.751201}, "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": 0.946257}}, "bpb": {"en": 1.82642, "de": 1.96622, "vi": 1.63983, "tr": 1.89603, "fi": 2.07507, "ru": 1.19373, "el": 1.19132, "hi": 1.00044, "ta": 0.8572, "th": 0.96126, "zh": 2.7512, "my": 0.94626}, "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": 3764.7} |
| {"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.82828, "de": 1.96255, "vi": 1.65313, "tr": 1.88985, "fi": 2.09117, "ru": 1.19085, "el": 1.18865, "hi": 1.00817, "ta": 0.86075, "th": 0.9592, "zh": 2.7593, "my": 0.95713}, "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": 1.52908, "t": 3773.2} |
| {"event": "train", "step": 7000, "phase": 2, "objective": "global", "lm_loss": 1.19716, "bpb": 1.72714, "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": 0.5223, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.77, "alpha": 0.03, "t": 3774.1} |
| {"event": "train", "step": 7100, "phase": 2, "objective": "global", "lm_loss": 1.09011, "bpb": 1.57269, "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": 0.4658, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.645, "alpha": 0.03, "t": 3838.5} |
| {"event": "train", "step": 7200, "phase": 2, "objective": "global", "lm_loss": 1.14549, "bpb": 1.65259, "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": 0.4763, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.643, "alpha": 0.03, "t": 3902.8} |
| {"event": "train", "step": 7300, "phase": 2, "objective": "global", "lm_loss": 1.18524, "bpb": 1.70994, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4852, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.636, "alpha": 0.03, "t": 3966.5} |
| {"event": "train", "step": 7400, "phase": 2, "objective": "global", "lm_loss": 1.14428, "bpb": 1.65085, "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": 0.4828, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.627, "alpha": 0.03, "t": 4029.2} |
| {"event": "train", "step": 7500, "phase": 2, "objective": "global", "lm_loss": 1.19052, "bpb": 1.71756, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4619, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.634, "alpha": 0.03, "t": 4092.6} |
| {"event": "train", "step": 7600, "phase": 2, "objective": "global", "lm_loss": 1.13428, "bpb": 1.63642, "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": 0.4813, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.635, "alpha": 0.03, "t": 4156.1} |
| {"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 4216.8} |
| {"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.13905, "bpb": 1.64331, "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": 0.4979, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.612, "alpha": 0.03, "t": 4217.3} |
| {"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.1708, "bpb": 1.68911, "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": 0.508, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.62, "alpha": 0.03, "t": 4279.3} |
| {"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.07589, "bpb": 1.55218, "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": 0.464, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.613, "alpha": 0.03, "t": 4340.5} |
| {"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": 1.785432}, "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": 1.921114}, "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": 1.607149}, "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": 1.837592}, "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": 2.026169}, "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": 1.174902}, "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": 1.170175}, "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": 0.98272}, "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": 0.837744}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.944539}, "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": 2.715775}, "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": 0.929058}}, "bpb": {"en": 1.78543, "de": 1.92111, "vi": 1.60715, "tr": 1.83759, "fi": 2.02617, "ru": 1.1749, "el": 1.17017, "hi": 0.98272, "ta": 0.83774, "th": 0.94454, "zh": 2.71578, "my": 0.92906}, "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": 4405.4} |
| {"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.78605, "de": 1.92059, "vi": 1.62198, "tr": 1.83205, "fi": 2.03969, "ru": 1.17229, "el": 1.16798, "hi": 0.98946, "ta": 0.84111, "th": 0.94299, "zh": 2.72533, "my": 0.93988}, "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": 1.49828, "t": 4414.0} |
| {"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.09568, "bpb": 1.58073, "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": 0.4673, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.741, "alpha": 0.03, "t": 4414.6} |
| {"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.16252, "bpb": 1.67716, "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": 0.4993, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.612, "alpha": 0.03, "t": 4475.8} |
| {"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.15962, "bpb": 1.67298, "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": 0.4729, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.619, "alpha": 0.03, "t": 4537.7} |
| {"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.10488, "bpb": 1.594, "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": 0.4835, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.58, "alpha": 0.03, "t": 4595.6} |
| {"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.25739, "bpb": 1.81402, "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": 0.5012, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.583, "alpha": 0.03, "t": 4653.9} |
| {"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.142, "bpb": 1.64756, "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": 0.4609, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.594, "alpha": 0.03, "t": 4713.3} |
| {"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.16385, "bpb": 1.67908, "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": 0.4879, "lr": 6.33e-05, "router_trainable": true, "sec_per_step": 0.599, "alpha": 0.03, "t": 4773.2} |
| {"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 1.12217, "bpb": 1.61895, "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": 0.5002, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.594, "alpha": 0.03, "t": 4832.5} |
| {"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.18752, "bpb": 1.71322, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5092, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.599, "alpha": 0.03, "t": 4892.4} |
| {"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.13595, "bpb": 1.63882, "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": 0.4748, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.585, "alpha": 0.03, "t": 4951.0} |
| {"event": "probe", "step": 9000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.763681}, "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": 1.892021}, "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": 1.575312}, "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": 1.800481}, "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": 1.988581}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001017, "bpb": 1.160376}, "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": 1.151382}, "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": 0.968843}, "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": 0.823375}, "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": 0.932539}, "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": 2.690151}, "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": 0.912895}}, "bpb": {"en": 1.76368, "de": 1.89202, "vi": 1.57531, "tr": 1.80048, "fi": 1.98858, "ru": 1.16038, "el": 1.15138, "hi": 0.96884, "ta": 0.82337, "th": 0.93254, "zh": 2.69015, "my": 0.91289}, "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": 5013.6} |
| {"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.76338, "de": 1.89248, "vi": 1.59044, "tr": 1.79478, "fi": 2.00187, "ru": 1.15765, "el": 1.14904, "hi": 0.97599, "ta": 0.82742, "th": 0.9315, "zh": 2.69944, "my": 0.92093}, "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": 1.47541, "t": 5022.7} |
| {"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.17776, "bpb": 1.69915, "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": 0.5438, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.725, "alpha": 0.03, "t": 5023.4} |
| {"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.14828, "bpb": 1.65662, "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": 0.5028, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.496, "alpha": 0.03, "t": 5073.0} |
| {"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 1.07213, "bpb": 1.54676, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4848, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.331, "alpha": 0.03, "t": 5106.1} |
| {"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 1.11264, "bpb": 1.6052, "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": 4.71e-05, "router_trainable": true, "sec_per_step": 0.451, "alpha": 0.03, "t": 5151.2} |
| {"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.06726, "bpb": 1.53974, "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": 0.4763, "lr": 4.52e-05, "router_trainable": true, "sec_per_step": 0.541, "alpha": 0.03, "t": 5205.4} |
| {"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.16316, "bpb": 1.67808, "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": 0.5158, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.565, "alpha": 0.03, "t": 5261.9} |
| {"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 1.10293, "bpb": 1.59119, "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": 0.4842, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.49, "alpha": 0.03, "t": 5310.8} |
| {"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 1.13325, "bpb": 1.63493, "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": 0.525, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.564, "alpha": 0.03, "t": 5367.3} |
| {"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 1.04941, "bpb": 1.51398, "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": 0.481, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.574, "alpha": 0.03, "t": 5424.7} |
| {"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.11488, "bpb": 1.60844, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4702, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.496, "alpha": 0.03, "t": 5474.3} |
| {"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": 1.747221}, "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": 1.871453}, "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": 1.559703}, "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": 1.777536}, "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": 1.96494}, "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": 1.14876}, "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": 1.139473}, "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": 0.958662}, "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": 0.813799}, "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": 0.922694}, "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": 2.670765}, "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": 0.901915}}, "bpb": {"en": 1.74722, "de": 1.87145, "vi": 1.5597, "tr": 1.77754, "fi": 1.96494, "ru": 1.14876, "el": 1.13947, "hi": 0.95866, "ta": 0.8138, "th": 0.92269, "zh": 2.67077, "my": 0.90192}, "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": 5536.0} |
| {"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.7473, "de": 1.87276, "vi": 1.57519, "tr": 1.77193, "fi": 1.97785, "ru": 1.14723, "el": 1.13665, "hi": 0.9647, "ta": 0.81866, "th": 0.92125, "zh": 2.68084, "my": 0.91063}, "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": 1.46042, "t": 5543.5} |
| {"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.13798, "bpb": 1.64176, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4959, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.699, "alpha": 0.03, "t": 5544.1} |
| {"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 1.19317, "bpb": 1.72138, "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": 3.49e-05, "router_trainable": true, "sec_per_step": 0.554, "alpha": 0.03, "t": 5599.6} |
| {"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.10926, "bpb": 1.60032, "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": 0.5074, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.519, "alpha": 0.03, "t": 5651.4} |
| {"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 1.09967, "bpb": 1.58649, "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": 3.3e-05, "router_trainable": true, "sec_per_step": 0.546, "alpha": 0.03, "t": 5706.1} |
| {"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.10044, "bpb": 1.5876, "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": 0.4997, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.555, "alpha": 0.03, "t": 5761.6} |
| {"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.05744, "bpb": 1.52556, "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": 0.4972, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.511, "alpha": 0.03, "t": 5812.6} |
| {"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 1.15602, "bpb": 1.66778, "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": 0.5034, "lr": 3.1e-05, "router_trainable": true, "sec_per_step": 0.545, "alpha": 0.03, "t": 5867.2} |
| {"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.12326, "bpb": 1.62052, "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": 0.5183, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.544, "alpha": 0.03, "t": 5921.6} |
| {"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.10082, "bpb": 1.58815, "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": 0.5077, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.473, "alpha": 0.03, "t": 5968.9} |
| {"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.18211, "bpb": 1.70543, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5092, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.548, "alpha": 0.03, "t": 6023.7} |
| {"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": 1.739141}, "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": 1.861922}, "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": 1.545427}, "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": 1.765674}, "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": 1.951378}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001017, "bpb": 1.144666}, "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": 1.131391}, "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": 0.951759}, "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": 0.808829}, "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": 0.915927}, "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": 2.659544}, "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": 0.895348}}, "bpb": {"en": 1.73914, "de": 1.86192, "vi": 1.54543, "tr": 1.76567, "fi": 1.95138, "ru": 1.14467, "el": 1.13139, "hi": 0.95176, "ta": 0.80883, "th": 0.91593, "zh": 2.65954, "my": 0.89535}, "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": 6081.4} |
| {"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.73816, "de": 1.86345, "vi": 1.56098, "tr": 1.75988, "fi": 1.96484, "ru": 1.14265, "el": 1.1294, "hi": 0.95815, "ta": 0.81337, "th": 0.91438, "zh": 2.66835, "my": 0.90382}, "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": 1.45145, "t": 6088.6} |
| {"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": 1.739141}, "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": 1.861922}, "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": 1.545427}, "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": 1.765674}, "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": 1.951378}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001017, "bpb": 1.144666}, "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": 1.131391}, "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": 0.951759}, "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": 0.808829}, "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": 0.915927}, "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": 2.659544}, "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": 0.895348}}, "bpb": {"en": 1.73914, "de": 1.86192, "vi": 1.54543, "tr": 1.76567, "fi": 1.95138, "ru": 1.14467, "el": 1.13139, "hi": 0.95176, "ta": 0.80883, "th": 0.91593, "zh": 2.65954, "my": 0.89535}, "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": 6092.1} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s1/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s1/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s1/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s1/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s1/th0.70"], "t": 6108.2} |
| {"event": "final", "run": "exp32_onechunk_s1", "condition": "A", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 1.000861920358559, "cps": 136.078125, "bpb": 1.7334981189736058, "bytes_per_chunk": 1024.0}, "de": {"cpc": 1.0160229873861588, "cps": 162.984375, "bpb": 1.8639862576887685, "bytes_per_chunk": 1024.0}, "vi": {"cpc": 1.3167344724708434, "cps": 189.640625, "bpb": 1.5639013189327229, "bytes_per_chunk": 1024.0}, "tr": {"cpc": 1.0905374799756946, "cps": 154.234375, "bpb": 1.7597192603569218, "bytes_per_chunk": 1024.0}, "fi": {"cpc": 1.040455052894583, "cps": 152.90625, "bpb": 1.9707200989221814, "bytes_per_chunk": 1024.0}, "ru": {"cpc": 1.8237467018469657, "cps": 270.0, "bpb": 1.1441115267983237, "bytes_per_chunk": 1024.0}, "el": {"cpc": 1.8157189494713248, "cps": 291.125, "bpb": 1.1264634357142533, "bytes_per_chunk": 1024.0}, "hi": {"cpc": 2.566002971768202, "cps": 337.2890625, "bpb": 0.9587633839596158, "bytes_per_chunk": 1024.0}, "ta": {"cpc": 2.7234554693768387, "cps": 397.7734375, "bpb": 0.8122669872385285, "bytes_per_chunk": 1024.0}, "th": {"cpc": 2.8650917003419334, "cps": 360.0390625, "bpb": 0.916371718939286, "bytes_per_chunk": 1024.0}, "zh": {"cpc": 2.7384016254656283, "cps": 126.3515625, "bpb": 2.67161658191112, "bytes_per_chunk": 1024.0}, "my": {"cpc": 2.8363415663922313, "cps": 419.8671875, "bpb": 0.9015909226104004, "bytes_per_chunk": 1024.0}}, "gini_cpc": 0.22151585453568945, "gini_cps": 0.2338532044751218, "bpic": 1.0, "seed": 1, "mean_boundary_entropy": 0.0010204949649050832, "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": 6116.1} |
|
|
