| {"event": "start", "run": "specfull_B_s1", "condition": "B", "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": "parity_a", "p3_objective": "none", "freeze_in_p3": true, "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": 2.5} |
| {"event": "probe_schedule", "n": 44, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 500, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, 6000, 6500, 7000, 7500, 8000, 12000, 16000, 20000, 24000, 28000, 32000, 36000, 40000, 44000, 48000, 52000, 56000, 60000, 64000, 68000, 72000, 76000, 76300], "t": 2.5} |
| {"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": 8.06011}, "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": 8.1963}, "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": 8.061647}, "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": 8.199}, "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": 8.038283}, "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": 8.262097}, "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": 8.178552}, "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": 8.397747}, "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": 8.106024}, "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": 8.030889}, "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": 8.099598}, "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": 8.204861}}, "bpb": {"en": 8.06011, "de": 8.1963, "vi": 8.06165, "tr": 8.199, "fi": 8.03828, "ru": 8.2621, "el": 8.17855, "hi": 8.39775, "ta": 8.10602, "th": 8.03089, "zh": 8.0996, "my": 8.20486}, "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": 13.0} |
| {"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.63341, "bpb": 8.1273, "chunk_loss": 0.084693, "F": 0.86952, "G": 0.52249, "chunks": 890.39, "bytes_per_chunk": 1.15, "boundary_entropy": 0.69066, "router_noise": 0.0, "grad_norm": 9.4224, "lr": 1e-07, "router_trainable": true, "sec_per_step": 0.001, "alpha": 0.03, "t": 14.0} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14109.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.810722, "cps": 110.226562, "mean_p": 0.520777, "boundary_entropy": 0.686722, "bpb": 8.051355}, "de": {"chunks": 18692.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.910339, "cps": 146.03125, "mean_p": 0.525525, "boundary_entropy": 0.687223, "bpb": 8.188255}, "vi": {"chunks": 21344.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.157798, "cps": 166.75, "mean_p": 0.52399, "boundary_entropy": 0.687876, "bpb": 8.057962}, "tr": {"chunks": 17839.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.985417, "cps": 139.367188, "mean_p": 0.529884, "boundary_entropy": 0.686379, "bpb": 8.192882}, "fi": {"chunks": 17564.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.933709, "cps": 137.21875, "mean_p": 0.526475, "boundary_entropy": 0.686834, "bpb": 8.030674}, "ru": {"chunks": 34128.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.80095, "cps": 266.625, "mean_p": 0.540125, "boundary_entropy": 0.687214, "bpb": 8.253459}, "el": {"chunks": 36912.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.798567, "cps": 288.375, "mean_p": 0.535748, "boundary_entropy": 0.688146, "bpb": 8.175513}, "hi": {"chunks": 34656.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.059792, "cps": 270.75, "mean_p": 0.509807, "boundary_entropy": 0.690751, "bpb": 8.385792}, "ta": {"chunks": 50675.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.710618, "cps": 395.898438, "mean_p": 0.521967, "boundary_entropy": 0.690369, "bpb": 8.097378}, "th": {"chunks": 45408.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.823003, "cps": 354.75, "mean_p": 0.531572, "boundary_entropy": 0.689075, "bpb": 8.019155}, "zh": {"chunks": 4816.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.815442, "cps": 37.625, "mean_p": 0.492067, "boundary_entropy": 0.686439, "bpb": 8.098282}, "my": {"chunks": 28224.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.48955, "cps": 220.5, "mean_p": 0.506233, "boundary_entropy": 0.690851, "bpb": 8.20137}}, "bpb": {"en": 8.05136, "de": 8.18825, "vi": 8.05796, "tr": 8.19288, "fi": 8.03067, "ru": 8.25346, "el": 8.17551, "hi": 8.38579, "ta": 8.09738, "th": 8.01916, "zh": 8.09828, "my": 8.20137}, "gini_cpc": 0.247529, "gini_cps": 0.27203, "bpic": 1.183169, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1228745843326384, "vi": 1.4281063723269043, "tr": 1.2154800661121117, "fi": 1.1517001720834326, "ru": 2.2214140303416055, "el": 2.2184754073286346, "hi": 2.540687487579456, "ta": 3.3434603293340386, "th": 3.482083612395371, "zh": 1.0058215177622571, "my": 1.8373126877768047}, "t": 24.2} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14076.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.808826, "cps": 109.96875, "mean_p": 0.520627, "boundary_entropy": 0.686733, "bpb": 8.032883}, "de": {"chunks": 18676.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.90956, "cps": 145.90625, "mean_p": 0.525379, "boundary_entropy": 0.687237, "bpb": 8.171113}, "vi": {"chunks": 21332.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.157147, "cps": 166.65625, "mean_p": 0.523925, "boundary_entropy": 0.687882, "bpb": 8.049573}, "tr": {"chunks": 17820.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.984367, "cps": 139.21875, "mean_p": 0.529775, "boundary_entropy": 0.686391, "bpb": 8.180059}, "fi": {"chunks": 17541.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.932486, "cps": 137.039062, "mean_p": 0.52633, "boundary_entropy": 0.686848, "bpb": 8.015445}, "ru": {"chunks": 34127.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.800897, "cps": 266.617188, "mean_p": 0.540082, "boundary_entropy": 0.687221, "bpb": 8.236868}, "el": {"chunks": 36912.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.798567, "cps": 288.375, "mean_p": 0.535697, "boundary_entropy": 0.688153, "bpb": 8.167881}, "hi": {"chunks": 34554.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.05373, "cps": 269.953125, "mean_p": 0.509737, "boundary_entropy": 0.690753, "bpb": 8.364884}, "ta": {"chunks": 50665.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.710083, "cps": 395.820312, "mean_p": 0.521884, "boundary_entropy": 0.690376, "bpb": 8.081885}, "th": {"chunks": 45404.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.822754, "cps": 354.71875, "mean_p": 0.531512, "boundary_entropy": 0.689082, "bpb": 8.000356}, "zh": {"chunks": 4806.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.813749, "cps": 37.546875, "mean_p": 0.492041, "boundary_entropy": 0.686438, "bpb": 8.096176}, "my": {"chunks": 28194.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.487967, "cps": 220.265625, "mean_p": 0.506216, "boundary_entropy": 0.690851, "bpb": 8.193704}}, "bpb": {"en": 8.03288, "de": 8.17111, "vi": 8.04957, "tr": 8.18006, "fi": 8.01544, "ru": 8.23687, "el": 8.16788, "hi": 8.36488, "ta": 8.08188, "th": 8.00036, "zh": 8.09618, "my": 8.1937}, "gini_cpc": 0.247741, "gini_cps": 0.272257, "bpic": 1.184118, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1245436566483604, "vi": 1.43064965706333, "tr": 1.2170320359217566, "fi": 1.1528885531279585, "ru": 2.2265567057383926, "el": 2.2236764366297033, "hi": 2.539148599810749, "ta": 3.350637459537462, "th": 3.489939608447344, "zh": 1.0060861857073693, "my": 1.8396626087172485}, "t": 35.4} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 13949.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.801528, "cps": 108.976562, "mean_p": 0.520083, "boundary_entropy": 0.686773, "bpb": 7.967698}, "de": {"chunks": 18594.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.905567, "cps": 145.265625, "mean_p": 0.524856, "boundary_entropy": 0.687287, "bpb": 8.110588}, "vi": {"chunks": 21285.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.154597, "cps": 166.289062, "mean_p": 0.52371, "boundary_entropy": 0.687901, "bpb": 8.020783}, "tr": {"chunks": 17774.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.981826, "cps": 138.859375, "mean_p": 0.529387, "boundary_entropy": 0.686433, "bpb": 8.134868}, "fi": {"chunks": 17460.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.92818, "cps": 136.40625, "mean_p": 0.52583, "boundary_entropy": 0.686898, "bpb": 7.962407}, "ru": {"chunks": 34116.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.800317, "cps": 266.53125, "mean_p": 0.539896, "boundary_entropy": 0.687249, "bpb": 8.175695}, "el": {"chunks": 36886.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.797301, "cps": 288.171875, "mean_p": 0.535536, "boundary_entropy": 0.688175, "bpb": 8.144738}, "hi": {"chunks": 34332.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.040535, "cps": 268.21875, "mean_p": 0.509586, "boundary_entropy": 0.690758, "bpb": 8.310307}, "ta": {"chunks": 50647.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.70912, "cps": 395.679688, "mean_p": 0.52164, "boundary_entropy": 0.690397, "bpb": 8.036709}, "th": {"chunks": 45393.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.82207, "cps": 354.632812, "mean_p": 0.531363, "boundary_entropy": 0.6891, "bpb": 7.942384}, "zh": {"chunks": 4781.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.809516, "cps": 37.351562, "mean_p": 0.491975, "boundary_entropy": 0.686435, "bpb": 8.088393}, "my": {"chunks": 28181.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.487281, "cps": 220.164062, "mean_p": 0.50621, "boundary_entropy": 0.690851, "bpb": 8.173304}}, "bpb": {"en": 7.9677, "de": 8.11059, "vi": 8.02078, "tr": 8.13487, "fi": 7.96241, "ru": 8.17569, "el": 8.14474, "hi": 8.31031, "ta": 8.03671, "th": 7.94238, "zh": 8.08839, "my": 8.1733}, "gini_cpc": 0.248599, "gini_cps": 0.273078, "bpic": 1.186714, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1297997268067315, "vi": 1.440494347623777, "tr": 1.22494240793932, "fi": 1.1580129119939833, "ru": 2.2461043934831575, "el": 2.2423415413651764, "hi": 2.5458046585965834, "ta": 3.3799424223546475, "th": 3.5208608926701386, "zh": 1.0099650541106922, "my": 1.8555560173945882}, "t": 46.8} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 13443.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.772453, "cps": 105.023438, "mean_p": 0.518097, "boundary_entropy": 0.686908, "bpb": 7.741691}, "de": {"chunks": 18219.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.887303, "cps": 142.335938, "mean_p": 0.52294, "boundary_entropy": 0.687458, "bpb": 7.895808}, "vi": {"chunks": 21067.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.142772, "cps": 164.585938, "mean_p": 0.522824, "boundary_entropy": 0.687978, "bpb": 7.918221}, "tr": {"chunks": 17587.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.971496, "cps": 137.398438, "mean_p": 0.528002, "boundary_entropy": 0.686578, "bpb": 7.973464}, "fi": {"chunks": 17206.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.914678, "cps": 134.421875, "mean_p": 0.524062, "boundary_entropy": 0.687064, "bpb": 7.780619}, "ru": {"chunks": 34092.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.79905, "cps": 266.34375, "mean_p": 0.539225, "boundary_entropy": 0.687352, "bpb": 7.965768}, "el": {"chunks": 36845.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.795303, "cps": 287.851562, "mean_p": 0.534822, "boundary_entropy": 0.68827, "bpb": 8.029879}, "hi": {"chunks": 33448.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.987994, "cps": 261.3125, "mean_p": 0.509017, "boundary_entropy": 0.690776, "bpb": 8.102732}, "ta": {"chunks": 50544.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.703611, "cps": 394.875, "mean_p": 0.520595, "boundary_entropy": 0.690481, "bpb": 7.839659}, "th": {"chunks": 45311.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.816972, "cps": 353.992188, "mean_p": 0.530629, "boundary_entropy": 0.689188, "bpb": 7.702121}, "zh": {"chunks": 4694.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.794785, "cps": 36.671875, "mean_p": 0.491654, "boundary_entropy": 0.686419, "bpb": 8.054144}, "my": {"chunks": 28067.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.481265, "cps": 219.273438, "mean_p": 0.506145, "boundary_entropy": 0.690853, "bpb": 8.07013}}, "bpb": {"en": 7.74169, "de": 7.89581, "vi": 7.91822, "tr": 7.97346, "fi": 7.78062, "ru": 7.96577, "el": 8.02988, "hi": 8.10273, "ta": 7.83966, "th": 7.70212, "zh": 8.05414, "my": 8.07013}, "gini_cpc": 0.251857, "gini_cps": 0.276591, "bpic": 1.197359, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1486826204393377, "vi": 1.4794063377136395, "tr": 1.2576770433728899, "fi": 1.1841206549259764, "ru": 2.329009108525648, "el": 2.3241579385087547, "hi": 2.5736115870442577, "ta": 3.5000323675028837, "th": 3.6467878849092883, "zh": 1.0289104170349233, "my": 1.9176111230223245}, "t": 59.6} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 11476.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.659427, "cps": 89.65625, "mean_p": 0.511412, "boundary_entropy": 0.687247, "bpb": 7.121111}, "de": {"chunks": 16525.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.804802, "cps": 129.101562, "mean_p": 0.516229, "boundary_entropy": 0.687944, "bpb": 7.253621}, "vi": {"chunks": 20282.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.10019, "cps": 158.453125, "mean_p": 0.520056, "boundary_entropy": 0.688199, "bpb": 7.60873}, "tr": {"chunks": 16743.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.924874, "cps": 130.804688, "mean_p": 0.523182, "boundary_entropy": 0.687028, "bpb": 7.460795}, "fi": {"chunks": 15909.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.845729, "cps": 124.289062, "mean_p": 0.518119, "boundary_entropy": 0.68753, "bpb": 7.258129}, "ru": {"chunks": 33907.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.789288, "cps": 264.898438, "mean_p": 0.535323, "boundary_entropy": 0.68792, "bpb": 7.115043}, "el": {"chunks": 36670.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.786776, "cps": 286.484375, "mean_p": 0.532019, "boundary_entropy": 0.688624, "bpb": 7.618716}, "hi": {"chunks": 30529.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.814502, "cps": 238.507812, "mean_p": 0.507138, "boundary_entropy": 0.690827, "bpb": 7.350337}, "ta": {"chunks": 50062.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.677828, "cps": 391.109375, "mean_p": 0.516952, "boundary_entropy": 0.69074, "bpb": 7.212583}, "th": {"chunks": 44864.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.789182, "cps": 350.5, "mean_p": 0.527802, "boundary_entropy": 0.689508, "bpb": 6.884494}, "zh": {"chunks": 4354.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.737216, "cps": 34.015625, "mean_p": 0.490402, "boundary_entropy": 0.686354, "bpb": 7.924509}, "my": {"chunks": 27481.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.450338, "cps": 214.695312, "mean_p": 0.505849, "boundary_entropy": 0.690857, "bpb": 7.684707}}, "bpb": {"en": 7.12111, "de": 7.25362, "vi": 7.60873, "tr": 7.4608, "fi": 7.25813, "ru": 7.11504, "el": 7.61872, "hi": 7.35034, "ta": 7.21258, "th": 6.88449, "zh": 7.92451, "my": 7.68471}, "gini_cpc": 0.266657, "gini_cps": 0.292548, "bpic": 1.242806, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2204574467234537, "vi": 1.6684039794656893, "tr": 1.4025433921963177, "fi": 1.2825212777266197, "ru": 2.713399449671725, "el": 2.7095904030141598, "hi": 2.751636658091802, "ta": 4.060844008464098, "th": 4.229709175061512, "zh": 1.1179659147015335, "my": 2.1993924843785564}, "t": 74.6} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 7653.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.439752, "cps": 59.789062, "mean_p": 0.501474, "boundary_entropy": 0.687521, "bpb": 6.207224}, "de": {"chunks": 12032.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.585984, "cps": 94.0, "mean_p": 0.505485, "boundary_entropy": 0.688407, "bpb": 6.345212}, "vi": {"chunks": 17306.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.938758, "cps": 135.203125, "mean_p": 0.511665, "boundary_entropy": 0.688702, "bpb": 6.958296}, "tr": {"chunks": 14309.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.790421, "cps": 111.789062, "mean_p": 0.513463, "boundary_entropy": 0.68776, "bpb": 6.662018}, "fi": {"chunks": 11977.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.636702, "cps": 93.570312, "mean_p": 0.507173, "boundary_entropy": 0.688053, "bpb": 6.428425}, "ru": {"chunks": 29098.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.535515, "cps": 227.328125, "mean_p": 0.513772, "boundary_entropy": 0.689966, "bpb": 5.717651}, "el": {"chunks": 35383.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.724066, "cps": 276.429688, "mean_p": 0.521995, "boundary_entropy": 0.689664, "bpb": 6.490437}, "hi": {"chunks": 19356.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.150431, "cps": 151.21875, "mean_p": 0.499868, "boundary_entropy": 0.690892, "bpb": 5.758734}, "ta": {"chunks": 25990.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.390211, "cps": 203.046875, "mean_p": 0.503214, "boundary_entropy": 0.691243, "bpb": 5.847493}, "th": {"chunks": 39717.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.469195, "cps": 310.289062, "mean_p": 0.514351, "boundary_entropy": 0.690615, "bpb": 5.469761}, "zh": {"chunks": 3379.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.57213, "cps": 26.398438, "mean_p": 0.486235, "boundary_entropy": 0.68609, "bpb": 7.513632}, "my": {"chunks": 22608.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.19316, "cps": 176.625, "mean_p": 0.502581, "boundary_entropy": 0.690883, "bpb": 6.296939}}, "bpb": {"en": 6.20722, "de": 6.34521, "vi": 6.9583, "tr": 6.66202, "fi": 6.42843, "ru": 5.71765, "el": 6.49044, "hi": 5.75873, "ta": 5.84749, "th": 5.46976, "zh": 7.51363, "my": 6.29694}, "gini_cpc": 0.276357, "gini_cps": 0.302151, "bpic": 1.607069, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.332532539381729, "vi": 2.1347447997918674, "tr": 1.7974264949729164, "fi": 1.4478666775057445, "ru": 3.4917756500891053, "el": 3.9205429336072215, "hi": 2.616091606414872, "ta": 3.1613546345118553, "th": 5.614974373571799, "zh": 1.3010295359030668, "my": 2.713258519241048}, "t": 94.0} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3726.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.214101, "cps": 29.109375, "mean_p": 0.493681, "boundary_entropy": 0.687529, "bpb": 5.764897}, "de": {"chunks": 5599.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.272683, "cps": 43.742188, "mean_p": 0.495628, "boundary_entropy": 0.688443, "bpb": 5.904925}, "vi": {"chunks": 6755.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.366423, "cps": 52.773438, "mean_p": 0.493275, "boundary_entropy": 0.688797, "bpb": 6.44843}, "tr": {"chunks": 8612.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.475722, "cps": 67.28125, "mean_p": 0.500731, "boundary_entropy": 0.688189, "bpb": 6.285104}, "fi": {"chunks": 6623.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.352081, "cps": 51.742188, "mean_p": 0.497763, "boundary_entropy": 0.688191, "bpb": 5.963693}, "ru": {"chunks": 4643.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.245013, "cps": 36.273438, "mean_p": 0.490318, "boundary_entropy": 0.690106, "bpb": 5.087271}, "el": {"chunks": 1808.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.088096, "cps": 14.125, "mean_p": 0.487851, "boundary_entropy": 0.690233, "bpb": 5.58798}, "hi": {"chunks": 445.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.026449, "cps": 3.476562, "mean_p": 0.476845, "boundary_entropy": 0.689714, "bpb": 5.095663}, "ta": {"chunks": 2246.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.120139, "cps": 17.546875, "mean_p": 0.485592, "boundary_entropy": 0.690772, "bpb": 5.173248}, "th": {"chunks": 4189.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.260429, "cps": 32.726562, "mean_p": 0.489977, "boundary_entropy": 0.690794, "bpb": 4.841222}, "zh": {"chunks": 3194.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.540806, "cps": 24.953125, "mean_p": 0.486716, "boundary_entropy": 0.686225, "bpb": 6.947229}, "my": {"chunks": 7677.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.405161, "cps": 59.976562, "mean_p": 0.481198, "boundary_entropy": 0.690226, "bpb": 4.673367}}, "bpb": {"en": 5.7649, "de": 5.90492, "vi": 6.44843, "tr": 6.2851, "fi": 5.96369, "ru": 5.08727, "el": 5.58798, "hi": 5.09566, "ta": 5.17325, "th": 4.84122, "zh": 6.94723, "my": 4.67337}, "gini_cpc": 0.302851, "gini_cps": 0.296364, "bpic": 6.912856, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2736184154441932, "vi": 1.7114471047030804, "tr": 2.2219523405298522, "fi": 1.644463131903413, "ru": 1.144381264366351, "el": 0.41147063861346245, "hi": 0.12353391572333364, "ta": 0.5611326665572743, "th": 1.2163836242345791, "zh": 2.525938304505421, "my": 1.8923847264672495}, "t": 119.2} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3789.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.217721, "cps": 29.601562, "mean_p": 0.492928, "boundary_entropy": 0.687464, "bpb": 5.317709}, "de": {"chunks": 4273.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.208104, "cps": 33.382812, "mean_p": 0.492, "boundary_entropy": 0.68826, "bpb": 5.495295}, "vi": {"chunks": 4863.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.263792, "cps": 37.992188, "mean_p": 0.489884, "boundary_entropy": 0.688714, "bpb": 5.949344}, "tr": {"chunks": 5674.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.313429, "cps": 44.328125, "mean_p": 0.494339, "boundary_entropy": 0.687975, "bpb": 5.778634}, "fi": {"chunks": 4702.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.24996, "cps": 36.734375, "mean_p": 0.493106, "boundary_entropy": 0.688004, "bpb": 5.487152}, "ru": {"chunks": 9766.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.515356, "cps": 76.296875, "mean_p": 0.492177, "boundary_entropy": 0.69002, "bpb": 4.70706}, "el": {"chunks": 4341.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.211519, "cps": 33.914062, "mean_p": 0.494208, "boundary_entropy": 0.690568, "bpb": 5.053378}, "hi": {"chunks": 13574.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.806776, "cps": 106.046875, "mean_p": 0.497699, "boundary_entropy": 0.690893, "bpb": 4.159872}, "ta": {"chunks": 19853.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.061942, "cps": 155.101562, "mean_p": 0.499986, "boundary_entropy": 0.691242, "bpb": 4.278638}, "th": {"chunks": 15282.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.950078, "cps": 119.390625, "mean_p": 0.496986, "boundary_entropy": 0.69089, "bpb": 4.216549}, "zh": {"chunks": 2829.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.479004, "cps": 22.101562, "mean_p": 0.489165, "boundary_entropy": 0.686712, "bpb": 6.669987}, "my": {"chunks": 5751.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.303515, "cps": 44.929688, "mean_p": 0.481547, "boundary_entropy": 0.690312, "bpb": 4.208355}}, "bpb": {"en": 5.31771, "de": 5.4953, "vi": 5.94934, "tr": 5.77863, "fi": 5.48715, "ru": 4.70706, "el": 5.05338, "hi": 4.15987, "ta": 4.27864, "th": 4.21655, "zh": 6.66999, "my": 4.20836}, "gini_cpc": 0.330651, "gini_cps": 0.340464, "bpic": 4.052726, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9558285546095217, "vi": 1.2116038440253907, "tr": 1.4395882556265416, "fi": 1.1480749900676788, "ru": 2.3670477587183005, "el": 0.9715126404151394, "hi": 3.705546665349018, "ta": 4.87753268124371, "th": 4.363737773718451, "zh": 2.2000827694047604, "my": 1.3940537096902788}, "t": 151.6} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2820.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.162041, "cps": 22.03125, "mean_p": 0.410863, "boundary_entropy": 0.655452, "bpb": 4.25529}, "de": {"chunks": 3211.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.156382, "cps": 25.085938, "mean_p": 0.417509, "boundary_entropy": 0.661081, "bpb": 4.452408}, "vi": {"chunks": 4107.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.222783, "cps": 32.085938, "mean_p": 0.435227, "boundary_entropy": 0.668687, "bpb": 4.2637}, "tr": {"chunks": 3255.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179804, "cps": 25.429688, "mean_p": 0.437894, "boundary_entropy": 0.671786, "bpb": 4.580906}, "fi": {"chunks": 2025.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.10765, "cps": 15.820312, "mean_p": 0.434912, "boundary_entropy": 0.672197, "bpb": 4.456461}, "ru": {"chunks": 6202.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.327282, "cps": 48.453125, "mean_p": 0.449026, "boundary_entropy": 0.677518, "bpb": 3.124918}, "el": {"chunks": 5430.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.264581, "cps": 42.421875, "mean_p": 0.467941, "boundary_entropy": 0.68661, "bpb": 3.361675}, "hi": {"chunks": 11235.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.667756, "cps": 87.773438, "mean_p": 0.481149, "boundary_entropy": 0.689015, "bpb": 2.607884}, "ta": {"chunks": 10911.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.583632, "cps": 85.242188, "mean_p": 0.485844, "boundary_entropy": 0.689495, "bpb": 2.590836}, "th": {"chunks": 10295.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.640037, "cps": 80.429688, "mean_p": 0.494868, "boundary_entropy": 0.690928, "bpb": 2.615768}, "zh": {"chunks": 2988.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.505926, "cps": 23.34375, "mean_p": 0.48519, "boundary_entropy": 0.686072, "bpb": 5.936137}, "my": {"chunks": 9747.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.514408, "cps": 76.148438, "mean_p": 0.447472, "boundary_entropy": 0.682399, "bpb": 2.730379}}, "bpb": {"en": 4.25529, "de": 4.45241, "vi": 4.2637, "tr": 4.58091, "fi": 4.45646, "ru": 3.12492, "el": 3.36167, "hi": 2.60788, "ta": 2.59084, "th": 2.61577, "zh": 5.93614, "my": 2.73038}, "gini_cpc": 0.308735, "gini_cps": 0.309355, "bpic": 5.313613, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9650790994465578, "vi": 1.3748539722659836, "tr": 1.1096230082941467, "fi": 0.6643365640520429, "ru": 2.0197497333408188, "el": 1.6328037769767392, "hi": 4.120908918466061, "ta": 3.601754422902927, "th": 3.949847221211781, "zh": 3.1222103738715052, "my": 3.1745531443278137}, "t": 200.4} |
| {"event": "probe", "step": 500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3536.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203183, "cps": 27.625, "mean_p": 0.336703, "boundary_entropy": 0.557503, "bpb": 3.373583}, "de": {"chunks": 4179.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.203526, "cps": 32.648438, "mean_p": 0.358444, "boundary_entropy": 0.579228, "bpb": 3.621711}, "vi": {"chunks": 5030.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.272851, "cps": 39.296875, "mean_p": 0.343046, "boundary_entropy": 0.547643, "bpb": 2.977317}, "tr": {"chunks": 3838.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212009, "cps": 29.984375, "mean_p": 0.372923, "boundary_entropy": 0.581958, "bpb": 3.589301}, "fi": {"chunks": 3182.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.169156, "cps": 24.859375, "mean_p": 0.366524, "boundary_entropy": 0.599041, "bpb": 3.70489}, "ru": {"chunks": 6441.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.339894, "cps": 50.320312, "mean_p": 0.338307, "boundary_entropy": 0.539872, "bpb": 2.582721}, "el": {"chunks": 7580.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.369342, "cps": 59.21875, "mean_p": 0.344701, "boundary_entropy": 0.587796, "bpb": 2.636492}, "hi": {"chunks": 9673.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.574918, "cps": 75.570312, "mean_p": 0.34305, "boundary_entropy": 0.598678, "bpb": 2.076444}, "ta": {"chunks": 8177.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.43739, "cps": 63.882812, "mean_p": 0.337589, "boundary_entropy": 0.595206, "bpb": 1.946092}, "th": {"chunks": 3347.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.208082, "cps": 26.148438, "mean_p": 0.446281, "boundary_entropy": 0.679464, "bpb": 2.03983}, "zh": {"chunks": 4845.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.820352, "cps": 37.851562, "mean_p": 0.407263, "boundary_entropy": 0.624731, "bpb": 4.797028}, "my": {"chunks": 17344.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.915347, "cps": 135.5, "mean_p": 0.388829, "boundary_entropy": 0.6435, "bpb": 2.141068}}, "bpb": {"en": 3.37358, "de": 3.62171, "vi": 2.97732, "tr": 3.5893, "fi": 3.70489, "ru": 2.58272, "el": 2.63649, "hi": 2.07644, "ta": 1.94609, "th": 2.03983, "zh": 4.79703, "my": 2.14107}, "gini_cpc": 0.318235, "gini_cps": 0.292511, "bpic": 4.97306, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0016865164409219, "vi": 1.3428784576357924, "tr": 1.0434371206192419, "fi": 0.8325305052810392, "ru": 1.6728459478981363, "el": 1.8177754105701516, "hi": 2.8295539490227455, "ta": 2.1526845413212086, "th": 1.0241097738681804, "zh": 4.037496906665972, "my": 4.505030676956318}, "t": 276.8} |
| {"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 2.12856, "bpb": 3.07086, "chunk_loss": 0.031479, "F": 0.20253, "G": 0.35965, "chunks": 207.39, "bytes_per_chunk": 4.938, "boundary_entropy": 0.58223, "router_noise": 0.0, "grad_norm": 1.861, "lr": 7.51e-05, "router_trainable": true, "sec_per_step": 0.527, "alpha": 0.03, "t": 277.5} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3757.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.215882, "cps": 29.351562, "mean_p": 0.257325, "boundary_entropy": 0.361446, "bpb": 2.70729}, "de": {"chunks": 4477.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.218039, "cps": 34.976562, "mean_p": 0.256124, "boundary_entropy": 0.367526, "bpb": 2.994007}, "vi": {"chunks": 4996.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.271006, "cps": 39.03125, "mean_p": 0.272364, "boundary_entropy": 0.373463, "bpb": 2.350762}, "tr": {"chunks": 3968.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.21919, "cps": 31.0, "mean_p": 0.271849, "boundary_entropy": 0.385545, "bpb": 2.978347}, "fi": {"chunks": 3894.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.207007, "cps": 30.421875, "mean_p": 0.255319, "boundary_entropy": 0.376368, "bpb": 3.204155}, "ru": {"chunks": 5774.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.304697, "cps": 45.109375, "mean_p": 0.242164, "boundary_entropy": 0.408573, "bpb": 1.948608}, "el": {"chunks": 6522.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.31779, "cps": 50.953125, "mean_p": 0.248989, "boundary_entropy": 0.425567, "bpb": 2.075192}, "hi": {"chunks": 7987.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.47471, "cps": 62.398438, "mean_p": 0.291351, "boundary_entropy": 0.47578, "bpb": 1.555635}, "ta": {"chunks": 8366.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.447499, "cps": 65.359375, "mean_p": 0.293391, "boundary_entropy": 0.520471, "bpb": 1.295388}, "th": {"chunks": 6941.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.43152, "cps": 54.226562, "mean_p": 0.24657, "boundary_entropy": 0.449066, "bpb": 1.614604}, "zh": {"chunks": 3827.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.647985, "cps": 29.898438, "mean_p": 0.282732, "boundary_entropy": 0.240769, "bpb": 3.700737}, "my": {"chunks": 10153.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.535835, "cps": 79.320312, "mean_p": 0.212509, "boundary_entropy": 0.385783, "bpb": 1.639663}}, "bpb": {"en": 2.70729, "de": 2.99401, "vi": 2.35076, "tr": 2.97835, "fi": 3.20416, "ru": 1.94861, "el": 2.07519, "hi": 1.55564, "ta": 1.29539, "th": 1.6146, "zh": 3.70074, "my": 1.63966}, "gini_cpc": 0.216787, "gini_cps": 0.190621, "bpic": 5.431222, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0099912524203105, "vi": 1.2553424440366592, "tr": 1.015322562515318, "fi": 0.95888602434706, "ru": 1.4114012260666633, "el": 1.4720510604052186, "hi": 2.198930669684049, "ta": 2.0728855107445687, "th": 1.9988670283608614, "zh": 3.0015663277966467, "my": 2.4820694846503963}, "t": 427.4} |
| {"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.73962, "bpb": 2.50974, "chunk_loss": 0.030662, "F": 0.19742, "G": 0.26183, "chunks": 202.16, "bytes_per_chunk": 5.065, "boundary_entropy": 0.38005, "router_noise": 0.0, "grad_norm": 1.3753, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.301, "alpha": 0.03, "t": 427.9} |
| {"event": "probe", "step": 1500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3493.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.200713, "cps": 27.289062, "mean_p": 0.218051, "boundary_entropy": 0.256012, "bpb": 2.418616}, "de": {"chunks": 4250.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.206984, "cps": 33.203125, "mean_p": 0.228861, "boundary_entropy": 0.274639, "bpb": 2.691271}, "vi": {"chunks": 3886.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.210795, "cps": 30.359375, "mean_p": 0.205761, "boundary_entropy": 0.258428, "bpb": 2.146125}, "tr": {"chunks": 3875.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.214053, "cps": 30.273438, "mean_p": 0.223044, "boundary_entropy": 0.290047, "bpb": 2.68937}, "fi": {"chunks": 3839.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.204083, "cps": 29.992188, "mean_p": 0.225217, "boundary_entropy": 0.267595, "bpb": 2.945862}, "ru": {"chunks": 5243.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.276675, "cps": 40.960938, "mean_p": 0.220614, "boundary_entropy": 0.36609, "bpb": 1.731339}, "el": {"chunks": 5658.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.275691, "cps": 44.203125, "mean_p": 0.269916, "boundary_entropy": 0.434391, "bpb": 1.799867}, "hi": {"chunks": 7653.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.454859, "cps": 59.789062, "mean_p": 0.230585, "boundary_entropy": 0.344307, "bpb": 1.42348}, "ta": {"chunks": 7747.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.414389, "cps": 60.523438, "mean_p": 0.230266, "boundary_entropy": 0.40235, "bpb": 1.137399}, "th": {"chunks": 5540.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.34442, "cps": 43.28125, "mean_p": 0.216645, "boundary_entropy": 0.397828, "bpb": 1.420908}, "zh": {"chunks": 3918.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.663393, "cps": 30.609375, "mean_p": 0.261574, "boundary_entropy": 0.162619, "bpb": 3.492596}, "my": {"chunks": 1309.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.069084, "cps": 10.226562, "mean_p": 0.163086, "boundary_entropy": 0.338336, "bpb": 1.35983}}, "bpb": {"en": 2.41862, "de": 2.69127, "vi": 2.14613, "tr": 2.68937, "fi": 2.94586, "ru": 1.73134, "el": 1.79987, "hi": 1.42348, "ta": 1.1374, "th": 1.42091, "zh": 3.4926, "my": 1.35983}, "gini_cpc": 0.265037, "gini_cps": 0.197537, "bpic": 6.803301, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0312454786216998, "vi": 1.0502318597441307, "tr": 1.0664652036690705, "fi": 1.0167911633773559, "ru": 1.3784663786676254, "el": 1.3735599919533348, "hi": 2.2662205582891533, "ta": 2.0645890566139733, "th": 1.7159879508561953, "zh": 3.3051907113535277, "my": 0.34419281882320685}, "t": 572.5} |
| {"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.57556, "bpb": 2.27305, "chunk_loss": 0.030229, "F": 0.17889, "G": 0.22415, "chunks": 183.19, "bytes_per_chunk": 5.59, "boundary_entropy": 0.29752, "router_noise": 0.0, "grad_norm": 0.9624, "lr": 0.0002251, "router_trainable": true, "sec_per_step": 0.29, "alpha": 0.03, "t": 573.1} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3259.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187267, "cps": 25.460938, "mean_p": 0.205154, "boundary_entropy": 0.220538, "bpb": 2.270917}, "de": {"chunks": 3607.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.175668, "cps": 28.179688, "mean_p": 0.201089, "boundary_entropy": 0.232119, "bpb": 2.5461}, "vi": {"chunks": 3873.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.21009, "cps": 30.257812, "mean_p": 0.197339, "boundary_entropy": 0.227365, "bpb": 2.058639}, "tr": {"chunks": 3225.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178147, "cps": 25.195312, "mean_p": 0.191614, "boundary_entropy": 0.272418, "bpb": 2.513693}, "fi": {"chunks": 3839.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.204083, "cps": 29.992188, "mean_p": 0.20785, "boundary_entropy": 0.230153, "bpb": 2.726325}, "ru": {"chunks": 5397.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.284802, "cps": 42.164062, "mean_p": 0.224628, "boundary_entropy": 0.350502, "bpb": 1.579401}, "el": {"chunks": 5014.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.244311, "cps": 39.171875, "mean_p": 0.256744, "boundary_entropy": 0.417195, "bpb": 1.669191}, "hi": {"chunks": 8432.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.501159, "cps": 65.875, "mean_p": 0.226345, "boundary_entropy": 0.303563, "bpb": 1.327047}, "ta": {"chunks": 8899.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.47601, "cps": 69.523438, "mean_p": 0.247344, "boundary_entropy": 0.397366, "bpb": 1.094566}, "th": {"chunks": 7612.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.473236, "cps": 59.46875, "mean_p": 0.200346, "boundary_entropy": 0.318798, "bpb": 1.275931}, "zh": {"chunks": 3997.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.676769, "cps": 31.226562, "mean_p": 0.255429, "boundary_entropy": 0.131099, "bpb": 3.347453}, "my": {"chunks": 6575.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.347002, "cps": 51.367188, "mean_p": 0.196653, "boundary_entropy": 0.366799, "bpb": 1.248263}}, "bpb": {"en": 2.27092, "de": 2.5461, "vi": 2.05864, "tr": 2.51369, "fi": 2.72633, "ru": 1.5794, "el": 1.66919, "hi": 1.32705, "ta": 1.09457, "th": 1.27593, "zh": 3.34745, "my": 1.24826}, "gini_cpc": 0.257616, "gini_cps": 0.204348, "bpic": 6.022078, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.938066215025896, "vi": 1.1218740816880388, "tr": 0.951303133511511, "fi": 1.0897979544882184, "ru": 1.520838028402775, "el": 1.3046176333613653, "hi": 2.6761797778423175, "ta": 2.5418826512151713, "th": 2.527071175558123, "zh": 3.613936067221491, "my": 1.8529860117478543}, "t": 720.0} |
| {"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.49509, "bpb": 2.15696, "chunk_loss": 0.030322, "F": 0.18546, "G": 0.21694, "chunks": 189.91, "bytes_per_chunk": 5.392, "boundary_entropy": 0.24621, "router_noise": 0.0, "grad_norm": 0.7537, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.295, "alpha": 0.03, "t": 720.6} |
| {"event": "probe", "step": 2500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3045.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17497, "cps": 23.789062, "mean_p": 0.197989, "boundary_entropy": 0.193322, "bpb": 2.15338}, "de": {"chunks": 3610.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.175815, "cps": 28.203125, "mean_p": 0.197133, "boundary_entropy": 0.193545, "bpb": 2.395655}, "vi": {"chunks": 4099.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.222349, "cps": 32.023438, "mean_p": 0.204792, "boundary_entropy": 0.222432, "bpb": 1.984562}, "tr": {"chunks": 3302.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.182401, "cps": 25.796875, "mean_p": 0.194851, "boundary_entropy": 0.240272, "bpb": 2.365917}, "fi": {"chunks": 4052.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.215406, "cps": 31.65625, "mean_p": 0.218466, "boundary_entropy": 0.220263, "bpb": 2.56788}, "ru": {"chunks": 5471.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.288707, "cps": 42.742188, "mean_p": 0.218159, "boundary_entropy": 0.322382, "bpb": 1.46692}, "el": {"chunks": 6297.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.306826, "cps": 49.195312, "mean_p": 0.269626, "boundary_entropy": 0.4137, "bpb": 1.559272}, "hi": {"chunks": 6598.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.392155, "cps": 51.546875, "mean_p": 0.193403, "boundary_entropy": 0.27812, "bpb": 1.275608}, "ta": {"chunks": 8119.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.434287, "cps": 63.429688, "mean_p": 0.25619, "boundary_entropy": 0.414979, "bpb": 1.043164}, "th": {"chunks": 6540.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.40659, "cps": 51.09375, "mean_p": 0.193581, "boundary_entropy": 0.33231, "bpb": 1.22592}, "zh": {"chunks": 3878.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.65662, "cps": 30.296875, "mean_p": 0.250759, "boundary_entropy": 0.117979, "bpb": 3.215996}, "my": {"chunks": 5014.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.264619, "cps": 39.171875, "mean_p": 0.198233, "boundary_entropy": 0.376563, "bpb": 1.158142}}, "bpb": {"en": 2.15338, "de": 2.39565, "vi": 1.98456, "tr": 2.36592, "fi": 2.56788, "ru": 1.46692, "el": 1.55927, "hi": 1.27561, "ta": 1.04316, "th": 1.22592, "zh": 3.216, "my": 1.15814}, "gini_cpc": 0.23192, "gini_cps": 0.171116, "bpic": 6.393686, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0048277445486649, "vi": 1.2707835946027917, "tr": 1.0424694597879838, "fi": 1.231102962761805, "ru": 1.65003943486229, "el": 1.753596505566625, "hi": 2.2412693988391075, "ta": 2.482069255365822, "th": 2.323771956554695, "zh": 3.7527634083398533, "my": 1.5123690348800087}, "t": 864.8} |
| {"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.36651, "bpb": 1.97145, "chunk_loss": 0.030146, "F": 0.17082, "G": 0.20694, "chunks": 174.92, "bytes_per_chunk": 5.854, "boundary_entropy": 0.24137, "router_noise": 0.0, "grad_norm": 0.5728, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.29, "alpha": 0.03, "t": 865.4} |
| {"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3023.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.173706, "cps": 23.617188, "mean_p": 0.19546, "boundary_entropy": 0.178562, "bpb": 2.079467}, "de": {"chunks": 3521.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.17148, "cps": 27.507812, "mean_p": 0.194002, "boundary_entropy": 0.187469, "bpb": 2.27448}, "vi": {"chunks": 3970.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.215351, "cps": 31.015625, "mean_p": 0.193194, "boundary_entropy": 0.20675, "bpb": 1.927555}, "tr": {"chunks": 3341.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184555, "cps": 26.101562, "mean_p": 0.196744, "boundary_entropy": 0.230606, "bpb": 2.264926}, "fi": {"chunks": 4018.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.213598, "cps": 31.390625, "mean_p": 0.21534, "boundary_entropy": 0.209338, "bpb": 2.458203}, "ru": {"chunks": 5358.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.282744, "cps": 41.859375, "mean_p": 0.207149, "boundary_entropy": 0.297851, "bpb": 1.384785}, "el": {"chunks": 6610.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.322078, "cps": 51.640625, "mean_p": 0.286986, "boundary_entropy": 0.432328, "bpb": 1.473628}, "hi": {"chunks": 7447.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.442615, "cps": 58.179688, "mean_p": 0.200081, "boundary_entropy": 0.277899, "bpb": 1.203344}, "ta": {"chunks": 7670.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.41027, "cps": 59.921875, "mean_p": 0.228646, "boundary_entropy": 0.375316, "bpb": 0.993007}, "th": {"chunks": 8671.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.539074, "cps": 67.742188, "mean_p": 0.217597, "boundary_entropy": 0.330308, "bpb": 1.138645}, "zh": {"chunks": 3873.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.655774, "cps": 30.257812, "mean_p": 0.244169, "boundary_entropy": 0.122926, "bpb": 3.099831}, "my": {"chunks": 4211.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.22224, "cps": 32.898438, "mean_p": 0.183349, "boundary_entropy": 0.348561, "bpb": 1.125078}}, "bpb": {"en": 2.07947, "de": 2.27448, "vi": 1.92755, "tr": 2.26493, "fi": 2.4582, "ru": 1.38478, "el": 1.47363, "hi": 1.20334, "ta": 0.99301, "th": 1.13864, "zh": 3.09983, "my": 1.12508}, "gini_cpc": 0.256207, "gini_cps": 0.197977, "bpic": 6.218803, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9871873712115296, "vi": 1.2397477758664452, "tr": 1.0624583088695116, "fi": 1.2296571007458579, "ru": 1.6277191313387163, "el": 1.8541573512586322, "hi": 2.548075276274663, "ta": 2.3618693343013364, "th": 3.103373833456504, "zh": 3.7752005479158037, "my": 1.2794043953083358}, "t": 1009.7} |
| {"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.46964, "bpb": 2.12024, "chunk_loss": 0.030181, "F": 0.17717, "G": 0.2051, "chunks": 181.42, "bytes_per_chunk": 5.644, "boundary_entropy": 0.21217, "router_noise": 0.0, "grad_norm": 0.4782, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.29, "alpha": 0.03, "t": 1010.3} |
| {"event": "probe", "step": 3500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3087.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.177383, "cps": 24.117188, "mean_p": 0.199597, "boundary_entropy": 0.169046, "bpb": 2.007689}, "de": {"chunks": 3787.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.184435, "cps": 29.585938, "mean_p": 0.208409, "boundary_entropy": 0.181193, "bpb": 2.193428}, "vi": {"chunks": 4043.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.219311, "cps": 31.585938, "mean_p": 0.199515, "boundary_entropy": 0.20489, "bpb": 1.866139}, "tr": {"chunks": 3416.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.188698, "cps": 26.6875, "mean_p": 0.204692, "boundary_entropy": 0.215405, "bpb": 2.17036}, "fi": {"chunks": 3647.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.193876, "cps": 28.492188, "mean_p": 0.198635, "boundary_entropy": 0.18124, "bpb": 2.38504}, "ru": {"chunks": 5734.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.302586, "cps": 44.796875, "mean_p": 0.217339, "boundary_entropy": 0.295825, "bpb": 1.336342}, "el": {"chunks": 6757.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.32924, "cps": 52.789062, "mean_p": 0.265185, "boundary_entropy": 0.402966, "bpb": 1.412859}, "hi": {"chunks": 6980.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.414859, "cps": 54.53125, "mean_p": 0.196739, "boundary_entropy": 0.270897, "bpb": 1.145463}, "ta": {"chunks": 8256.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.441615, "cps": 64.5, "mean_p": 0.220695, "boundary_entropy": 0.345335, "bpb": 0.966737}, "th": {"chunks": 8218.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.510911, "cps": 64.203125, "mean_p": 0.209616, "boundary_entropy": 0.317654, "bpb": 1.090723}, "zh": {"chunks": 3876.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.656282, "cps": 30.28125, "mean_p": 0.245179, "boundary_entropy": 0.109671, "bpb": 3.014994}, "my": {"chunks": 8466.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.446802, "cps": 66.140625, "mean_p": 0.215178, "boundary_entropy": 0.374071, "bpb": 1.089571}}, "bpb": {"en": 2.00769, "de": 2.19343, "vi": 1.86614, "tr": 2.17036, "fi": 2.38504, "ru": 1.33634, "el": 1.41286, "hi": 1.14546, "ta": 0.96674, "th": 1.09072, "zh": 3.01499, "my": 1.08957}, "gini_cpc": 0.243173, "gini_cps": 0.202495, "bpic": 5.791435, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.039753494540562, "vi": 1.2363689510830873, "tr": 1.0637873002595502, "fi": 1.0929778747924666, "ru": 1.7058308525068526, "el": 1.8560965542132515, "hi": 2.3387717557836125, "ta": 2.4896122113690926, "th": 2.880265764994216, "zh": 3.6997963231540987, "my": 2.5188504244545915}, "t": 1151.8} |
| {"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.26995, "bpb": 1.83215, "chunk_loss": 0.030215, "F": 0.17929, "G": 0.21231, "chunks": 183.59, "bytes_per_chunk": 5.578, "boundary_entropy": 0.23065, "router_noise": 0.0, "grad_norm": 0.4299, "lr": 0.0002997, "router_trainable": true, "sec_per_step": 0.284, "alpha": 0.03, "t": 1152.4} |
| {"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3032.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174223, "cps": 23.6875, "mean_p": 0.194976, "boundary_entropy": 0.152738, "bpb": 1.919248}, "de": {"chunks": 3565.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.173623, "cps": 27.851562, "mean_p": 0.190967, "boundary_entropy": 0.154057, "bpb": 2.098795}, "vi": {"chunks": 4306.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.233577, "cps": 33.640625, "mean_p": 0.209083, "boundary_entropy": 0.20347, "bpb": 1.782837}, "tr": {"chunks": 3322.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183505, "cps": 25.953125, "mean_p": 0.199753, "boundary_entropy": 0.196507, "bpb": 2.024954}, "fi": {"chunks": 3535.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187922, "cps": 27.617188, "mean_p": 0.195139, "boundary_entropy": 0.165775, "bpb": 2.245642}, "ru": {"chunks": 5686.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.300053, "cps": 44.421875, "mean_p": 0.21823, "boundary_entropy": 0.286744, "bpb": 1.297658}, "el": {"chunks": 5636.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.274619, "cps": 44.03125, "mean_p": 0.255113, "boundary_entropy": 0.395864, "bpb": 1.363269}, "hi": {"chunks": 7811.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.46425, "cps": 61.023438, "mean_p": 0.198056, "boundary_entropy": 0.2442, "bpb": 1.100977}, "ta": {"chunks": 9500.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.508157, "cps": 74.21875, "mean_p": 0.244013, "boundary_entropy": 0.359074, "bpb": 0.930254}, "th": {"chunks": 8181.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.508611, "cps": 63.914062, "mean_p": 0.207232, "boundary_entropy": 0.307034, "bpb": 1.050361}, "zh": {"chunks": 3779.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.639858, "cps": 29.523438, "mean_p": 0.236781, "boundary_entropy": 0.112963, "bpb": 2.941199}, "my": {"chunks": 7722.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.407536, "cps": 60.328125, "mean_p": 0.221023, "boundary_entropy": 0.392515, "bpb": 1.058291}}, "bpb": {"en": 1.91925, "de": 2.09879, "vi": 1.78284, "tr": 2.02495, "fi": 2.24564, "ru": 1.29766, "el": 1.36327, "hi": 1.10098, "ta": 0.93025, "th": 1.05036, "zh": 2.9412, "my": 1.05829}, "gini_cpc": 0.253529, "gini_cps": 0.217741, "bpic": 5.808263, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9965567847728741, "vi": 1.3406820870304494, "tr": 1.053280393328297, "fi": 1.0786299074995276, "ru": 1.7222356082177095, "el": 1.576249865489739, "hi": 2.6646887484464865, "ta": 2.9167087131928238, "th": 2.9193102400423867, "zh": 3.6726401061663507, "my": 2.339167624711542}, "t": 1292.2} |
| {"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.9227, "de": 2.09873, "vi": 1.79185, "tr": 2.01882, "fi": 2.25981, "ru": 1.29376, "el": 1.35971, "hi": 1.10975, "ta": 0.93087, "th": 1.04952, "zh": 2.94888, "my": 1.06754}, "bytes_per_chunk": {"en": 5.728, "de": 5.798, "vi": 5.65, "tr": 5.916, "fi": 5.375, "ru": 6.047, "el": 6.455, "hi": 5.503, "ta": 5.276, "th": 5.81, "zh": 4.401, "my": 5.923}, "bpb_mean": 1.65433, "t": 1301.3} |
| {"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.21765, "bpb": 1.75669, "chunk_loss": 0.030123, "F": 0.17465, "G": 0.20323, "chunks": 178.84, "bytes_per_chunk": 5.726, "boundary_entropy": 0.2073, "router_noise": 0.0, "grad_norm": 0.4106, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.298, "alpha": 0.03, "t": 1301.6} |
| {"event": "probe", "step": 4500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3100.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17813, "cps": 24.21875, "mean_p": 0.19875, "boundary_entropy": 0.145012, "bpb": 1.872176}, "de": {"chunks": 3767.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.183461, "cps": 29.429688, "mean_p": 0.202838, "boundary_entropy": 0.152688, "bpb": 2.029918}, "vi": {"chunks": 4108.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.222837, "cps": 32.09375, "mean_p": 0.201998, "boundary_entropy": 0.203989, "bpb": 1.721972}, "tr": {"chunks": 3369.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.186102, "cps": 26.320312, "mean_p": 0.202019, "boundary_entropy": 0.19497, "bpb": 1.951226}, "fi": {"chunks": 3452.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18351, "cps": 26.96875, "mean_p": 0.187447, "boundary_entropy": 0.158878, "bpb": 2.179985}, "ru": {"chunks": 6000.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.316623, "cps": 46.875, "mean_p": 0.231344, "boundary_entropy": 0.290139, "bpb": 1.269487}, "el": {"chunks": 6921.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.337231, "cps": 54.070312, "mean_p": 0.282694, "boundary_entropy": 0.404147, "bpb": 1.3218}, "hi": {"chunks": 7552.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.448856, "cps": 59.0, "mean_p": 0.197479, "boundary_entropy": 0.246904, "bpb": 1.087816}, "ta": {"chunks": 9375.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.501471, "cps": 73.242188, "mean_p": 0.23099, "boundary_entropy": 0.323451, "bpb": 0.910011}, "th": {"chunks": 7873.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.489462, "cps": 61.507812, "mean_p": 0.203218, "boundary_entropy": 0.301218, "bpb": 1.037265}, "zh": {"chunks": 3743.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.633762, "cps": 29.242188, "mean_p": 0.235572, "boundary_entropy": 0.106664, "bpb": 2.886529}, "my": {"chunks": 6103.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.322092, "cps": 47.679688, "mean_p": 0.197327, "boundary_entropy": 0.361448, "bpb": 1.051088}}, "bpb": {"en": 1.87218, "de": 2.02992, "vi": 1.72197, "tr": 1.95123, "fi": 2.17999, "ru": 1.26949, "el": 1.3218, "hi": 1.08782, "ta": 0.91001, "th": 1.03727, "zh": 2.88653, "my": 1.05109}, "gini_cpc": 0.24308, "gini_cps": 0.207879, "bpic": 5.871533, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.029925093044556, "vi": 1.250978135909079, "tr": 1.0447512174955853, "fi": 1.0301994886313945, "ru": 1.777478934377394, "el": 1.8931735597983694, "hi": 2.519818933039352, "ta": 2.8151933844654, "th": 2.747777813430666, "zh": 3.5578596397321482, "my": 1.8081831600236982}, "t": 1441.2} |
| {"event": "train", "step": 4500, "phase": 1, "objective": "global", "lm_loss": 1.25849, "bpb": 1.81562, "chunk_loss": 0.030187, "F": 0.17799, "G": 0.20669, "chunks": 182.27, "bytes_per_chunk": 5.618, "boundary_entropy": 0.19988, "router_noise": 0.0, "grad_norm": 0.4133, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.281, "alpha": 0.03, "t": 1441.9} |
| {"event": "probe", "step": 5000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3025.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.173821, "cps": 23.632812, "mean_p": 0.189724, "boundary_entropy": 0.128075, "bpb": 1.843925}, "de": {"chunks": 3616.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176107, "cps": 28.25, "mean_p": 0.191798, "boundary_entropy": 0.136734, "bpb": 1.990689}, "vi": {"chunks": 4201.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227882, "cps": 32.820312, "mean_p": 0.207637, "boundary_entropy": 0.200278, "bpb": 1.671986}, "tr": {"chunks": 3242.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179086, "cps": 25.328125, "mean_p": 0.196079, "boundary_entropy": 0.196392, "bpb": 1.918493}, "fi": {"chunks": 3486.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.185317, "cps": 27.234375, "mean_p": 0.188876, "boundary_entropy": 0.149176, "bpb": 2.123261}, "ru": {"chunks": 5733.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.302533, "cps": 44.789062, "mean_p": 0.213784, "boundary_entropy": 0.257316, "bpb": 1.237821}, "el": {"chunks": 5795.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.282366, "cps": 45.273438, "mean_p": 0.254049, "boundary_entropy": 0.382034, "bpb": 1.292744}, "hi": {"chunks": 6900.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.410104, "cps": 53.90625, "mean_p": 0.18121, "boundary_entropy": 0.245128, "bpb": 1.058507}, "ta": {"chunks": 7894.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.422252, "cps": 61.671875, "mean_p": 0.213842, "boundary_entropy": 0.324035, "bpb": 0.890253}, "th": {"chunks": 7956.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.494622, "cps": 62.15625, "mean_p": 0.201275, "boundary_entropy": 0.297175, "bpb": 1.013977}, "zh": {"chunks": 3678.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.622757, "cps": 28.734375, "mean_p": 0.232724, "boundary_entropy": 0.097822, "bpb": 2.744969}, "my": {"chunks": 8392.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.442896, "cps": 65.5625, "mean_p": 0.227207, "boundary_entropy": 0.398971, "bpb": 1.018456}}, "bpb": {"en": 1.84393, "de": 1.99069, "vi": 1.67199, "tr": 1.91849, "fi": 2.12326, "ru": 1.23782, "el": 1.29274, "hi": 1.05851, "ta": 0.89025, "th": 1.01398, "zh": 2.74497, "my": 1.01846}, "gini_cpc": 0.242429, "gini_cps": 0.204197, "bpic": 6.004271, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0131523493928138, "vi": 1.3110168718772905, "tr": 1.0302940702490753, "fi": 1.0661400068800961, "ru": 1.740489744652086, "el": 1.624468657015733, "hi": 2.359352105411811, "ta": 2.4292398329866782, "th": 2.845590815322525, "zh": 3.5827542798610748, "my": 2.5480083115686782}, "t": 1580.5} |
| {"event": "train", "step": 5000, "phase": 1, "objective": "global", "lm_loss": 1.14038, "bpb": 1.64522, "chunk_loss": 0.030055, "F": 0.1721, "G": 0.20135, "chunks": 176.23, "bytes_per_chunk": 5.81, "boundary_entropy": 0.1972, "router_noise": 0.0, "grad_norm": 0.3616, "lr": 0.0002989, "router_trainable": true, "sec_per_step": 0.278, "alpha": 0.03, "t": 1581.1} |
| {"event": "probe", "step": 5500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3037.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17451, "cps": 23.726562, "mean_p": 0.191098, "boundary_entropy": 0.125661, "bpb": 1.819429}, "de": {"chunks": 3692.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.179808, "cps": 28.84375, "mean_p": 0.19438, "boundary_entropy": 0.131603, "bpb": 1.961573}, "vi": {"chunks": 4218.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.228804, "cps": 32.953125, "mean_p": 0.202876, "boundary_entropy": 0.182198, "bpb": 1.649245}, "tr": {"chunks": 3373.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.186323, "cps": 26.351562, "mean_p": 0.20783, "boundary_entropy": 0.200346, "bpb": 1.874027}, "fi": {"chunks": 3474.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.184679, "cps": 27.140625, "mean_p": 0.188085, "boundary_entropy": 0.142237, "bpb": 2.100338}, "ru": {"chunks": 5491.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.289763, "cps": 42.898438, "mean_p": 0.201099, "boundary_entropy": 0.241823, "bpb": 1.209141}, "el": {"chunks": 6078.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.296156, "cps": 47.484375, "mean_p": 0.244905, "boundary_entropy": 0.354413, "bpb": 1.261203}, "hi": {"chunks": 7648.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.454562, "cps": 59.75, "mean_p": 0.188635, "boundary_entropy": 0.233526, "bpb": 1.029622}, "ta": {"chunks": 8152.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.436052, "cps": 63.6875, "mean_p": 0.204502, "boundary_entropy": 0.29727, "bpb": 0.87642}, "th": {"chunks": 7425.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.46161, "cps": 58.007812, "mean_p": 0.1817, "boundary_entropy": 0.266841, "bpb": 0.993173}, "zh": {"chunks": 3673.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.62191, "cps": 28.695312, "mean_p": 0.23021, "boundary_entropy": 0.093293, "bpb": 2.685511}, "my": {"chunks": 3983.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.210207, "cps": 31.117188, "mean_p": 0.194153, "boundary_entropy": 0.366423, "bpb": 0.997342}}, "bpb": {"en": 1.81943, "de": 1.96157, "vi": 1.64925, "tr": 1.87403, "fi": 2.10034, "ru": 1.20914, "el": 1.2612, "hi": 1.02962, "ta": 0.87642, "th": 0.99317, "zh": 2.68551, "my": 0.99734}, "gini_cpc": 0.244207, "gini_cps": 0.193732, "bpic": 6.370444, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0303591056654289, "vi": 1.311120964572282, "tr": 1.067689859644291, "fi": 1.0582718862941203, "ru": 1.6604337706544523, "el": 1.697067743698209, "hi": 2.6047865136325097, "ta": 2.4987225132697444, "th": 2.6451768976325454, "zh": 3.563746583646484, "my": 1.2045539569710535}, "t": 1721.8} |
| {"event": "train", "step": 5500, "phase": 1, "objective": "global", "lm_loss": 1.11752, "bpb": 1.61224, "chunk_loss": 0.030064, "F": 0.16928, "G": 0.19807, "chunks": 173.34, "bytes_per_chunk": 5.907, "boundary_entropy": 0.19182, "router_noise": 0.0, "grad_norm": 0.3519, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.283, "alpha": 0.03, "t": 1722.4} |
| {"event": "probe", "step": 6000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3028.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.173993, "cps": 23.65625, "mean_p": 0.18889, "boundary_entropy": 0.118737, "bpb": 1.797267}, "de": {"chunks": 3605.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.175571, "cps": 28.164062, "mean_p": 0.188926, "boundary_entropy": 0.126477, "bpb": 1.930584}, "vi": {"chunks": 4068.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.220667, "cps": 31.78125, "mean_p": 0.200727, "boundary_entropy": 0.184291, "bpb": 1.614109}, "tr": {"chunks": 3293.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181904, "cps": 25.726562, "mean_p": 0.204421, "boundary_entropy": 0.203091, "bpb": 1.853145}, "fi": {"chunks": 3474.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.184679, "cps": 27.140625, "mean_p": 0.187511, "boundary_entropy": 0.140889, "bpb": 2.065925}, "ru": {"chunks": 5670.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299208, "cps": 44.296875, "mean_p": 0.208373, "boundary_entropy": 0.238267, "bpb": 1.185949}, "el": {"chunks": 5868.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.285923, "cps": 45.84375, "mean_p": 0.250641, "boundary_entropy": 0.354556, "bpb": 1.23771}, "hi": {"chunks": 7937.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.471738, "cps": 62.007812, "mean_p": 0.194944, "boundary_entropy": 0.224323, "bpb": 1.008567}, "ta": {"chunks": 8903.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.476224, "cps": 69.554688, "mean_p": 0.221382, "boundary_entropy": 0.30366, "bpb": 0.858845}, "th": {"chunks": 8190.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.50917, "cps": 63.984375, "mean_p": 0.19636, "boundary_entropy": 0.275495, "bpb": 0.966701}, "zh": {"chunks": 3706.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.627497, "cps": 28.953125, "mean_p": 0.230636, "boundary_entropy": 0.089988, "bpb": 2.657501}, "my": {"chunks": 14994.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.791324, "cps": 117.140625, "mean_p": 0.242496, "boundary_entropy": 0.391507, "bpb": 0.981878}}, "bpb": {"en": 1.79727, "de": 1.93058, "vi": 1.61411, "tr": 1.85314, "fi": 2.06592, "ru": 1.18595, "el": 1.23771, "hi": 1.00857, "ta": 0.85885, "th": 0.9667, "zh": 2.6575, "my": 0.98188}, "gini_cpc": 0.291296, "gini_cps": 0.280553, "bpic": 5.276356, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0090695739262008, "vi": 1.268253447884835, "tr": 1.0454648327845995, "fi": 1.061417344344532, "ru": 1.7196580377340076, "el": 1.643302475149349, "hi": 2.7112499484749564, "ta": 2.737027451373876, "th": 2.9263824653457173, "zh": 3.606452542943812, "my": 4.548020146356196}, "t": 1862.5} |
| {"event": "train", "step": 6000, "phase": 1, "objective": "global", "lm_loss": 1.20065, "bpb": 1.73217, "chunk_loss": 0.03027, "F": 0.185, "G": 0.2007, "chunks": 189.44, "bytes_per_chunk": 5.405, "boundary_entropy": 0.17838, "router_noise": 0.0, "grad_norm": 0.3479, "lr": 0.0002981, "router_trainable": true, "sec_per_step": 0.282, "alpha": 0.03, "t": 1863.2} |
| {"event": "probe", "step": 6500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3005.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.172671, "cps": 23.476562, "mean_p": 0.184331, "boundary_entropy": 0.109155, "bpb": 1.783276}, "de": {"chunks": 3577.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174207, "cps": 27.945312, "mean_p": 0.186886, "boundary_entropy": 0.117879, "bpb": 1.916136}, "vi": {"chunks": 4307.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.233632, "cps": 33.648438, "mean_p": 0.207723, "boundary_entropy": 0.181811, "bpb": 1.588987}, "tr": {"chunks": 3226.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178203, "cps": 25.203125, "mean_p": 0.192396, "boundary_entropy": 0.178376, "bpb": 1.833693}, "fi": {"chunks": 3460.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.183935, "cps": 27.03125, "mean_p": 0.185777, "boundary_entropy": 0.1351, "bpb": 2.054594}, "ru": {"chunks": 5373.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.283536, "cps": 41.976562, "mean_p": 0.194781, "boundary_entropy": 0.226135, "bpb": 1.172052}, "el": {"chunks": 6387.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.311212, "cps": 49.898438, "mean_p": 0.243418, "boundary_entropy": 0.347889, "bpb": 1.216799}, "hi": {"chunks": 7371.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.438098, "cps": 57.585938, "mean_p": 0.191457, "boundary_entropy": 0.227315, "bpb": 0.988814}, "ta": {"chunks": 8820.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.471784, "cps": 68.90625, "mean_p": 0.211187, "boundary_entropy": 0.286814, "bpb": 0.837746}, "th": {"chunks": 8202.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.509916, "cps": 64.078125, "mean_p": 0.188693, "boundary_entropy": 0.255824, "bpb": 0.957691}, "zh": {"chunks": 3693.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.625296, "cps": 28.851562, "mean_p": 0.230676, "boundary_entropy": 0.080228, "bpb": 2.626775}, "my": {"chunks": 11584.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.611357, "cps": 90.5, "mean_p": 0.228947, "boundary_entropy": 0.377896, "bpb": 0.964423}}, "bpb": {"en": 1.78328, "de": 1.91614, "vi": 1.58899, "tr": 1.83369, "fi": 2.05459, "ru": 1.17205, "el": 1.2168, "hi": 0.98881, "ta": 0.83775, "th": 0.95769, "zh": 2.62677, "my": 0.96442}, "gini_cpc": 0.264585, "gini_cps": 0.248807, "bpic": 5.56164, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0088954811835302, "vi": 1.3530422986370694, "tr": 1.0320326936738313, "fi": 1.0652311534715544, "ru": 1.6420533762989564, "el": 1.8023358233767675, "hi": 2.5371782640621254, "ta": 2.732264626264775, "th": 2.9531012916509987, "zh": 3.6213083736159337, "my": 3.5405833005788323}, "t": 2003.9} |
| {"event": "train", "step": 6500, "phase": 1, "objective": "global", "lm_loss": 1.09932, "bpb": 1.58599, "chunk_loss": 0.03009, "F": 0.1712, "G": 0.19252, "chunks": 175.31, "bytes_per_chunk": 5.841, "boundary_entropy": 0.17331, "router_noise": 0.0, "grad_norm": 0.3422, "lr": 0.0002976, "router_trainable": true, "sec_per_step": 0.283, "alpha": 0.03, "t": 2004.6} |
| {"event": "probe", "step": 7000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3051.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.175315, "cps": 23.835938, "mean_p": 0.18913, "boundary_entropy": 0.106181, "bpb": 1.768828}, "de": {"chunks": 3583.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.1745, "cps": 27.992188, "mean_p": 0.187123, "boundary_entropy": 0.113918, "bpb": 1.899216}, "vi": {"chunks": 4079.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.221264, "cps": 31.867188, "mean_p": 0.195672, "boundary_entropy": 0.169578, "bpb": 1.558597}, "tr": {"chunks": 3381.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.186765, "cps": 26.414062, "mean_p": 0.202604, "boundary_entropy": 0.175887, "bpb": 1.811653}, "fi": {"chunks": 3602.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.191484, "cps": 28.140625, "mean_p": 0.191505, "boundary_entropy": 0.124935, "bpb": 2.011717}, "ru": {"chunks": 5622.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.296675, "cps": 43.921875, "mean_p": 0.203475, "boundary_entropy": 0.220073, "bpb": 1.150016}, "el": {"chunks": 5756.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.280466, "cps": 44.96875, "mean_p": 0.229796, "boundary_entropy": 0.323376, "bpb": 1.198004}, "hi": {"chunks": 7423.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.441189, "cps": 57.992188, "mean_p": 0.184179, "boundary_entropy": 0.196377, "bpb": 0.9663}, "ta": {"chunks": 8155.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.436213, "cps": 63.710938, "mean_p": 0.215233, "boundary_entropy": 0.301616, "bpb": 0.832853}, "th": {"chunks": 8263.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.513708, "cps": 64.554688, "mean_p": 0.190997, "boundary_entropy": 0.261818, "bpb": 0.940361}, "zh": {"chunks": 3671.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.621571, "cps": 28.679688, "mean_p": 0.228423, "boundary_entropy": 0.082815, "bpb": 2.598364}, "my": {"chunks": 5362.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.282985, "cps": 41.890625, "mean_p": 0.212499, "boundary_entropy": 0.39011, "bpb": 0.946732}}, "bpb": {"en": 1.76883, "de": 1.89922, "vi": 1.5586, "tr": 1.81165, "fi": 2.01172, "ru": 1.15002, "el": 1.198, "hi": 0.9663, "ta": 0.83285, "th": 0.94036, "zh": 2.59836, "my": 0.94673}, "gini_cpc": 0.24412, "gini_cps": 0.195169, "bpic": 6.195212, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9953511293736763, "vi": 1.2620962487722283, "tr": 1.0653112955576882, "fi": 1.0922290860276684, "ru": 1.6922461686942039, "el": 1.5997858555317295, "hi": 2.516554268373753, "ta": 2.488172056583941, "th": 2.9302090143653636, "zh": 3.5454621585435007, "my": 1.6141554103773947}, "t": 2142.2} |
| {"event": "train", "step": 7000, "phase": 1, "objective": "global", "lm_loss": 1.15527, "bpb": 1.66671, "chunk_loss": 0.030104, "F": 0.17598, "G": 0.19758, "chunks": 180.2, "bytes_per_chunk": 5.682, "boundary_entropy": 0.16139, "router_noise": 0.0, "grad_norm": 0.3529, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 2142.8} |
| {"event": "probe", "step": 7500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3022.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.173648, "cps": 23.609375, "mean_p": 0.185425, "boundary_entropy": 0.103165, "bpb": 1.751385}, "de": {"chunks": 3596.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.175133, "cps": 28.09375, "mean_p": 0.185754, "boundary_entropy": 0.109917, "bpb": 1.888749}, "vi": {"chunks": 4034.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.218823, "cps": 31.515625, "mean_p": 0.192399, "boundary_entropy": 0.167503, "bpb": 1.553296}, "tr": {"chunks": 3309.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.182787, "cps": 25.851562, "mean_p": 0.208715, "boundary_entropy": 0.19704, "bpb": 1.793566}, "fi": {"chunks": 3426.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.182127, "cps": 26.765625, "mean_p": 0.18324, "boundary_entropy": 0.118041, "bpb": 1.999876}, "ru": {"chunks": 5706.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.301108, "cps": 44.578125, "mean_p": 0.210691, "boundary_entropy": 0.234507, "bpb": 1.136731}, "el": {"chunks": 7203.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.350972, "cps": 56.273438, "mean_p": 0.265575, "boundary_entropy": 0.352972, "bpb": 1.175088}, "hi": {"chunks": 7274.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.432333, "cps": 56.828125, "mean_p": 0.183833, "boundary_entropy": 0.198897, "bpb": 0.957243}, "ta": {"chunks": 8948.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.478631, "cps": 69.90625, "mean_p": 0.228177, "boundary_entropy": 0.309178, "bpb": 0.819022}, "th": {"chunks": 8581.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.533478, "cps": 67.039062, "mean_p": 0.191205, "boundary_entropy": 0.249763, "bpb": 0.932519}, "zh": {"chunks": 3711.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.628344, "cps": 28.992188, "mean_p": 0.233603, "boundary_entropy": 0.077355, "bpb": 2.582683}, "my": {"chunks": 10998.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.580431, "cps": 85.921875, "mean_p": 0.244851, "boundary_entropy": 0.414174, "bpb": 0.950169}}, "bpb": {"en": 1.75139, "de": 1.88875, "vi": 1.5533, "tr": 1.79357, "fi": 1.99988, "ru": 1.13673, "el": 1.17509, "hi": 0.95724, "ta": 0.81902, "th": 0.93252, "zh": 2.58268, "my": 0.95017}, "gini_cpc": 0.261109, "gini_cps": 0.24485, "bpic": 5.497665, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.008548844376132, "vi": 1.2601504884979637, "tr": 1.0526303215371795, "fi": 1.048830082768714, "ru": 1.7340124574579732, "el": 2.0211671442990164, "hi": 2.4897049571099705, "ta": 2.7563233621181147, "th": 3.07217886365184, "zh": 3.6184882932095443, "my": 3.3425660629313683}, "t": 2279.2} |
| {"event": "train", "step": 7500, "phase": 1, "objective": "global", "lm_loss": 1.15287, "bpb": 1.66324, "chunk_loss": 0.030155, "F": 0.17622, "G": 0.19705, "chunks": 180.45, "bytes_per_chunk": 5.675, "boundary_entropy": 0.16222, "router_noise": 0.0, "grad_norm": 0.3158, "lr": 0.0002964, "router_trainable": true, "sec_per_step": 0.274, "alpha": 0.03, "t": 2279.8} |
| {"event": "probe", "step": 8000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3052.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.175372, "cps": 23.84375, "mean_p": 0.18807, "boundary_entropy": 0.103131, "bpb": 1.74297}, "de": {"chunks": 3622.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176399, "cps": 28.296875, "mean_p": 0.189154, "boundary_entropy": 0.10845, "bpb": 1.870653}, "vi": {"chunks": 4178.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226634, "cps": 32.640625, "mean_p": 0.201936, "boundary_entropy": 0.167655, "bpb": 1.53554}, "tr": {"chunks": 3272.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180744, "cps": 25.5625, "mean_p": 0.204938, "boundary_entropy": 0.187629, "bpb": 1.776918}, "fi": {"chunks": 3527.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187497, "cps": 27.554688, "mean_p": 0.188247, "boundary_entropy": 0.104241, "bpb": 1.981326}, "ru": {"chunks": 5506.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.290554, "cps": 43.015625, "mean_p": 0.207562, "boundary_entropy": 0.228195, "bpb": 1.128082}, "el": {"chunks": 5883.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.286654, "cps": 45.960938, "mean_p": 0.239457, "boundary_entropy": 0.337881, "bpb": 1.174146}, "hi": {"chunks": 7434.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.441842, "cps": 58.078125, "mean_p": 0.182662, "boundary_entropy": 0.196616, "bpb": 0.944227}, "ta": {"chunks": 8643.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.462316, "cps": 67.523438, "mean_p": 0.212422, "boundary_entropy": 0.285047, "bpb": 0.81291}, "th": {"chunks": 8143.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.506248, "cps": 63.617188, "mean_p": 0.183283, "boundary_entropy": 0.246345, "bpb": 0.923674}, "zh": {"chunks": 3709.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.628005, "cps": 28.976562, "mean_p": 0.230522, "boundary_entropy": 0.071653, "bpb": 2.565759}, "my": {"chunks": 7454.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.393392, "cps": 58.234375, "mean_p": 0.230522, "boundary_entropy": 0.407239, "bpb": 0.935276}}, "bpb": {"en": 1.74297, "de": 1.87065, "vi": 1.53554, "tr": 1.77692, "fi": 1.98133, "ru": 1.12808, "el": 1.17415, "hi": 0.94423, "ta": 0.81291, "th": 0.92367, "zh": 2.56576, "my": 0.93528}, "gini_cpc": 0.246367, "gini_cps": 0.206628, "bpic": 5.957205, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0058555804966705, "vi": 1.292304583316893, "tr": 1.0306289429157705, "fi": 1.0691365262496295, "ru": 1.6567866393247042, "el": 1.6345476823017797, "hi": 2.5194577204629414, "ta": 2.636201691847856, "th": 2.8867086957354258, "zh": 3.5809889558522574, "my": 2.243187639704277}, "t": 2414.2} |
| {"event": "eval", "step": 8000, "phase": 1, "bpb": {"en": 1.74273, "de": 1.87733, "vi": 1.55162, "tr": 1.76919, "fi": 1.99434, "ru": 1.12732, "el": 1.17108, "hi": 0.9498, "ta": 0.81884, "th": 0.92382, "zh": 2.57206, "my": 0.94233}, "bytes_per_chunk": {"en": 5.816, "de": 5.804, "vi": 6.054, "tr": 6.064, "fi": 5.377, "ru": 6.184, "el": 6.089, "hi": 5.841, "ta": 5.98, "th": 5.991, "zh": 4.526, "my": 6.24}, "bpb_mean": 1.45337, "t": 2422.8} |
| {"event": "train", "step": 8000, "phase": 1, "objective": "global", "lm_loss": 1.05645, "bpb": 1.52414, "chunk_loss": 0.03009, "F": 0.17262, "G": 0.19652, "chunks": 176.77, "bytes_per_chunk": 5.793, "boundary_entropy": 0.16899, "router_noise": 0.0, "grad_norm": 0.3363, "lr": 0.0002957, "router_trainable": true, "sec_per_step": 0.287, "alpha": 0.03, "t": 2423.1} |
| {"event": "train", "step": 8500, "phase": 1, "objective": "global", "lm_loss": 1.11748, "bpb": 1.61219, "chunk_loss": 0.030114, "F": 0.1743, "G": 0.19105, "chunks": 178.48, "bytes_per_chunk": 5.737, "boundary_entropy": 0.14556, "router_noise": 0.0, "grad_norm": 0.3212, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.269, "alpha": 0.03, "t": 2557.5} |
| {"event": "train", "step": 9000, "phase": 1, "objective": "global", "lm_loss": 1.12908, "bpb": 1.62892, "chunk_loss": 0.030136, "F": 0.17815, "G": 0.1973, "chunks": 182.42, "bytes_per_chunk": 5.613, "boundary_entropy": 0.14568, "router_noise": 0.0, "grad_norm": 0.346, "lr": 0.0002941, "router_trainable": true, "sec_per_step": 0.269, "alpha": 0.03, "t": 2691.9} |
| {"event": "train", "step": 9500, "phase": 1, "objective": "global", "lm_loss": 1.11436, "bpb": 1.60768, "chunk_loss": 0.030118, "F": 0.17802, "G": 0.19875, "chunks": 182.3, "bytes_per_chunk": 5.617, "boundary_entropy": 0.14751, "router_noise": 0.0, "grad_norm": 0.3308, "lr": 0.0002933, "router_trainable": true, "sec_per_step": 0.271, "alpha": 0.03, "t": 2827.6} |
| {"event": "train", "step": 10000, "phase": 1, "objective": "global", "lm_loss": 1.10378, "bpb": 1.59242, "chunk_loss": 0.0301, "F": 0.17525, "G": 0.19699, "chunks": 179.45, "bytes_per_chunk": 5.706, "boundary_entropy": 0.14514, "router_noise": 0.0, "grad_norm": 0.2827, "lr": 0.0002924, "router_trainable": true, "sec_per_step": 0.271, "alpha": 0.03, "t": 2963.0} |
| {"event": "train", "step": 10500, "phase": 1, "objective": "global", "lm_loss": 1.02298, "bpb": 1.47585, "chunk_loss": 0.030042, "F": 0.17162, "G": 0.20042, "chunks": 175.73, "bytes_per_chunk": 5.827, "boundary_entropy": 0.17247, "router_noise": 0.0, "grad_norm": 0.2929, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.273, "alpha": 0.03, "t": 3099.7} |
| {"event": "train", "step": 11000, "phase": 1, "objective": "global", "lm_loss": 1.09764, "bpb": 1.58356, "chunk_loss": 0.03006, "F": 0.17236, "G": 0.19124, "chunks": 176.5, "bytes_per_chunk": 5.802, "boundary_entropy": 0.13645, "router_noise": 0.0, "grad_norm": 0.3146, "lr": 0.0002903, "router_trainable": true, "sec_per_step": 0.273, "alpha": 0.03, "t": 3235.9} |
| {"event": "train", "step": 11500, "phase": 1, "objective": "global", "lm_loss": 1.07573, "bpb": 1.55195, "chunk_loss": 0.030139, "F": 0.17842, "G": 0.19974, "chunks": 182.7, "bytes_per_chunk": 5.605, "boundary_entropy": 0.14776, "router_noise": 0.0, "grad_norm": 0.2966, "lr": 0.0002893, "router_trainable": true, "sec_per_step": 0.27, "alpha": 0.03, "t": 3371.1} |
| {"event": "probe", "step": 12000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3016.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.173303, "cps": 23.5625, "mean_p": 0.180037, "boundary_entropy": 0.073028, "bpb": 1.67837}, "de": {"chunks": 3593.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174987, "cps": 28.070312, "mean_p": 0.182049, "boundary_entropy": 0.083134, "bpb": 1.79762}, "vi": {"chunks": 4288.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.232601, "cps": 33.5, "mean_p": 0.199092, "boundary_entropy": 0.13559, "bpb": 1.45991}, "tr": {"chunks": 3137.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.173286, "cps": 24.507812, "mean_p": 0.206343, "boundary_entropy": 0.194178, "bpb": 1.701287}, "fi": {"chunks": 3518.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187018, "cps": 27.484375, "mean_p": 0.185876, "boundary_entropy": 0.078365, "bpb": 1.883078}, "ru": {"chunks": 5574.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.294142, "cps": 43.546875, "mean_p": 0.203645, "boundary_entropy": 0.20248, "bpb": 1.070076}, "el": {"chunks": 6351.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.309458, "cps": 49.617188, "mean_p": 0.226797, "boundary_entropy": 0.282087, "bpb": 1.094927}, "hi": {"chunks": 6910.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.410698, "cps": 53.984375, "mean_p": 0.177668, "boundary_entropy": 0.177614, "bpb": 0.896967}, "ta": {"chunks": 8463.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.452688, "cps": 66.117188, "mean_p": 0.207462, "boundary_entropy": 0.269276, "bpb": 0.76751}, "th": {"chunks": 7763.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.482624, "cps": 60.648438, "mean_p": 0.174131, "boundary_entropy": 0.229017, "bpb": 0.87997}, "zh": {"chunks": 3733.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.632069, "cps": 29.164062, "mean_p": 0.230834, "boundary_entropy": 0.055375, "bpb": 2.477029}, "my": {"chunks": 5927.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.312803, "cps": 46.304688, "mean_p": 0.252684, "boundary_entropy": 0.448378, "bpb": 0.882383}}, "bpb": {"en": 1.67837, "de": 1.79762, "vi": 1.45991, "tr": 1.70129, "fi": 1.88308, "ru": 1.07008, "el": 1.09493, "hi": 0.89697, "ta": 0.76751, "th": 0.87997, "zh": 2.47703, "my": 0.88238}, "gini_cpc": 0.242215, "gini_cps": 0.196652, "bpic": 6.16288, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0097121751735352, "vi": 1.3421603900434462, "tr": 0.9999004187796007, "fi": 1.0791373761641527, "ru": 1.6972684294142761, "el": 1.7856405946049274, "hi": 2.3698221668683317, "ta": 2.6121111582481027, "th": 2.7848467689471303, "zh": 3.647181113752952, "my": 1.8049465023358637}, "t": 3513.3} |
| {"event": "eval", "step": 12000, "phase": 1, "bpb": {"en": 1.67841, "de": 1.80316, "vi": 1.47647, "tr": 1.69461, "fi": 1.89752, "ru": 1.06866, "el": 1.09401, "hi": 0.90357, "ta": 0.77265, "th": 0.87912, "zh": 2.48251, "my": 0.8878}, "bytes_per_chunk": {"en": 5.825, "de": 5.823, "vi": 5.833, "tr": 6.318, "fi": 5.409, "ru": 6.077, "el": 5.752, "hi": 6.245, "ta": 6.052, "th": 6.118, "zh": 4.525, "my": 8.111}, "bpb_mean": 1.38654, "t": 3522.3} |
| {"event": "train", "step": 12000, "phase": 1, "objective": "global", "lm_loss": 1.1194, "bpb": 1.61495, "chunk_loss": 0.03007, "F": 0.17482, "G": 0.19864, "chunks": 179.02, "bytes_per_chunk": 5.72, "boundary_entropy": 0.14425, "router_noise": 0.0, "grad_norm": 0.3091, "lr": 0.0002881, "router_trainable": true, "sec_per_step": 0.303, "alpha": 0.03, "t": 3522.6} |
| {"event": "train", "step": 12500, "phase": 1, "objective": "global", "lm_loss": 1.11669, "bpb": 1.61104, "chunk_loss": 0.030103, "F": 0.17659, "G": 0.19733, "chunks": 180.83, "bytes_per_chunk": 5.663, "boundary_entropy": 0.13294, "router_noise": 0.0, "grad_norm": 0.2936, "lr": 0.0002869, "router_trainable": true, "sec_per_step": 0.275, "alpha": 0.03, "t": 3659.9} |
| {"event": "train", "step": 13000, "phase": 1, "objective": "global", "lm_loss": 1.01235, "bpb": 1.46051, "chunk_loss": 0.03007, "F": 0.17529, "G": 0.19808, "chunks": 179.5, "bytes_per_chunk": 5.705, "boundary_entropy": 0.13597, "router_noise": 0.0, "grad_norm": 0.3023, "lr": 0.0002857, "router_trainable": true, "sec_per_step": 0.272, "alpha": 0.03, "t": 3795.7} |
| {"event": "train", "step": 13500, "phase": 1, "objective": "global", "lm_loss": 1.06648, "bpb": 1.53861, "chunk_loss": 0.030184, "F": 0.17868, "G": 0.19871, "chunks": 182.97, "bytes_per_chunk": 5.597, "boundary_entropy": 0.12713, "router_noise": 0.0, "grad_norm": 0.4369, "lr": 0.0002844, "router_trainable": true, "sec_per_step": 0.28, "alpha": 0.03, "t": 3935.9} |
| {"event": "train", "step": 14000, "phase": 1, "objective": "global", "lm_loss": 0.98412, "bpb": 1.41979, "chunk_loss": 0.030068, "F": 0.17348, "G": 0.19614, "chunks": 177.64, "bytes_per_chunk": 5.764, "boundary_entropy": 0.14273, "router_noise": 0.0, "grad_norm": 0.2899, "lr": 0.000283, "router_trainable": true, "sec_per_step": 0.281, "alpha": 0.03, "t": 4076.5} |
| {"event": "train", "step": 14500, "phase": 1, "objective": "global", "lm_loss": 1.05031, "bpb": 1.51528, "chunk_loss": 0.030078, "F": 0.17085, "G": 0.19347, "chunks": 174.95, "bytes_per_chunk": 5.853, "boundary_entropy": 0.13769, "router_noise": 0.0, "grad_norm": 0.2777, "lr": 0.0002816, "router_trainable": true, "sec_per_step": 0.281, "alpha": 0.03, "t": 4217.1} |
| {"event": "train", "step": 15000, "phase": 1, "objective": "global", "lm_loss": 1.05217, "bpb": 1.51796, "chunk_loss": 0.030111, "F": 0.17604, "G": 0.19119, "chunks": 180.27, "bytes_per_chunk": 5.681, "boundary_entropy": 0.1145, "router_noise": 0.0, "grad_norm": 0.2866, "lr": 0.0002801, "router_trainable": true, "sec_per_step": 0.284, "alpha": 0.03, "t": 4359.0} |
| {"event": "train", "step": 15500, "phase": 1, "objective": "global", "lm_loss": 1.01747, "bpb": 1.46791, "chunk_loss": 0.030125, "F": 0.17883, "G": 0.19128, "chunks": 183.12, "bytes_per_chunk": 5.592, "boundary_entropy": 0.11759, "router_noise": 0.0, "grad_norm": 0.2883, "lr": 0.0002786, "router_trainable": true, "sec_per_step": 0.283, "alpha": 0.03, "t": 4500.4} |
| {"event": "probe", "step": 16000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3032.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174223, "cps": 23.6875, "mean_p": 0.180883, "boundary_entropy": 0.066554, "bpb": 1.643347}, "de": {"chunks": 3679.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.179175, "cps": 28.742188, "mean_p": 0.185545, "boundary_entropy": 0.074963, "bpb": 1.752577}, "vi": {"chunks": 4124.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.223705, "cps": 32.21875, "mean_p": 0.194313, "boundary_entropy": 0.129219, "bpb": 1.415806}, "tr": {"chunks": 3283.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181351, "cps": 25.648438, "mean_p": 0.220332, "boundary_entropy": 0.204412, "bpb": 1.656998}, "fi": {"chunks": 3586.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190633, "cps": 28.015625, "mean_p": 0.189527, "boundary_entropy": 0.069264, "bpb": 1.827727}, "ru": {"chunks": 5647.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.297995, "cps": 44.117188, "mean_p": 0.208326, "boundary_entropy": 0.197457, "bpb": 1.032042}, "el": {"chunks": 6052.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.294889, "cps": 47.28125, "mean_p": 0.224524, "boundary_entropy": 0.264021, "bpb": 1.065575}, "hi": {"chunks": 7590.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.451114, "cps": 59.296875, "mean_p": 0.180686, "boundary_entropy": 0.156007, "bpb": 0.863917}, "ta": {"chunks": 8495.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.4544, "cps": 66.367188, "mean_p": 0.198751, "boundary_entropy": 0.245036, "bpb": 0.745142}, "th": {"chunks": 7949.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.494187, "cps": 62.101562, "mean_p": 0.172881, "boundary_entropy": 0.21012, "bpb": 0.853128}, "zh": {"chunks": 3710.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.628175, "cps": 28.984375, "mean_p": 0.229109, "boundary_entropy": 0.051385, "bpb": 2.423854}, "my": {"chunks": 7688.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.405742, "cps": 60.0625, "mean_p": 0.291285, "boundary_entropy": 0.479061, "bpb": 0.851251}}, "bpb": {"en": 1.64335, "de": 1.75258, "vi": 1.41581, "tr": 1.657, "fi": 1.82773, "ru": 1.03204, "el": 1.06558, "hi": 0.86392, "ta": 0.74514, "th": 0.85313, "zh": 2.42385, "my": 0.85125}, "gini_cpc": 0.242973, "gini_cps": 0.204243, "bpic": 5.919349, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.028424238760001, "vi": 1.2840160071791855, "tr": 1.0409149702880192, "fi": 1.0941914705214444, "ru": 1.710422877172952, "el": 1.6925947810404363, "hi": 2.589295557637797, "ta": 2.608151633534004, "th": 2.8365232976527235, "zh": 3.6055821100495273, "my": 2.328868259360572}, "t": 4646.1} |
| {"event": "eval", "step": 16000, "phase": 1, "bpb": {"en": 1.64055, "de": 1.75936, "vi": 1.43269, "tr": 1.64965, "fi": 1.844, "ru": 1.03119, "el": 1.06345, "hi": 0.87023, "ta": 0.74995, "th": 0.85255, "zh": 2.42852, "my": 0.85599}, "bytes_per_chunk": {"en": 5.799, "de": 5.693, "vi": 5.996, "tr": 6.022, "fi": 5.311, "ru": 5.995, "el": 6.002, "hi": 5.705, "ta": 6.07, "th": 5.85, "zh": 4.535, "my": 6.721}, "bpb_mean": 1.34818, "t": 4654.3} |
| {"event": "train", "step": 16000, "phase": 1, "objective": "global", "lm_loss": 0.98726, "bpb": 1.42431, "chunk_loss": 0.03023, "F": 0.17726, "G": 0.20291, "chunks": 181.52, "bytes_per_chunk": 5.641, "boundary_entropy": 0.15058, "router_noise": 0.0, "grad_norm": 0.2853, "lr": 0.000277, "router_trainable": true, "sec_per_step": 0.308, "alpha": 0.03, "t": 4654.7} |
| {"event": "train", "step": 16500, "phase": 1, "objective": "global", "lm_loss": 1.09252, "bpb": 1.57617, "chunk_loss": 0.030172, "F": 0.17938, "G": 0.19552, "chunks": 183.69, "bytes_per_chunk": 5.575, "boundary_entropy": 0.11532, "router_noise": 0.0, "grad_norm": 0.2903, "lr": 0.0002754, "router_trainable": true, "sec_per_step": 0.282, "alpha": 0.03, "t": 4795.7} |
| {"event": "train", "step": 17000, "phase": 1, "objective": "global", "lm_loss": 1.02751, "bpb": 1.48239, "chunk_loss": 0.030141, "F": 0.17719, "G": 0.19986, "chunks": 181.44, "bytes_per_chunk": 5.644, "boundary_entropy": 0.13264, "router_noise": 0.0, "grad_norm": 0.278, "lr": 0.0002737, "router_trainable": true, "sec_per_step": 0.283, "alpha": 0.03, "t": 4937.4} |
| {"event": "train", "step": 17500, "phase": 1, "objective": "global", "lm_loss": 1.08114, "bpb": 1.55975, "chunk_loss": 0.030122, "F": 0.17703, "G": 0.19496, "chunks": 181.28, "bytes_per_chunk": 5.649, "boundary_entropy": 0.12247, "router_noise": 0.0, "grad_norm": 0.2913, "lr": 0.000272, "router_trainable": true, "sec_per_step": 0.272, "alpha": 0.03, "t": 5073.6} |
| {"event": "train", "step": 18000, "phase": 1, "objective": "global", "lm_loss": 0.99879, "bpb": 1.44095, "chunk_loss": 0.030034, "F": 0.16986, "G": 0.19058, "chunks": 173.94, "bytes_per_chunk": 5.887, "boundary_entropy": 0.1271, "router_noise": 0.0, "grad_norm": 0.2731, "lr": 0.0002703, "router_trainable": true, "sec_per_step": 0.278, "alpha": 0.03, "t": 5212.3} |
| {"event": "train", "step": 18500, "phase": 1, "objective": "global", "lm_loss": 1.03426, "bpb": 1.49212, "chunk_loss": 0.03012, "F": 0.17674, "G": 0.19667, "chunks": 180.98, "bytes_per_chunk": 5.658, "boundary_entropy": 0.11876, "router_noise": 0.0, "grad_norm": 0.2908, "lr": 0.0002685, "router_trainable": true, "sec_per_step": 0.275, "alpha": 0.03, "t": 5349.6} |
| {"event": "train", "step": 19000, "phase": 1, "objective": "global", "lm_loss": 0.97027, "bpb": 1.3998, "chunk_loss": 0.030088, "F": 0.17267, "G": 0.18932, "chunks": 176.81, "bytes_per_chunk": 5.791, "boundary_entropy": 0.11419, "router_noise": 0.0, "grad_norm": 0.2732, "lr": 0.0002666, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 5487.6} |
| {"event": "train", "step": 19500, "phase": 1, "objective": "global", "lm_loss": 1.03663, "bpb": 1.49554, "chunk_loss": 0.03009, "F": 0.17339, "G": 0.18988, "chunks": 177.55, "bytes_per_chunk": 5.767, "boundary_entropy": 0.11951, "router_noise": 0.0, "grad_norm": 0.2683, "lr": 0.0002647, "router_trainable": true, "sec_per_step": 0.274, "alpha": 0.03, "t": 5624.7} |
| {"event": "probe", "step": 20000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3031.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174165, "cps": 23.679688, "mean_p": 0.17873, "boundary_entropy": 0.056377, "bpb": 1.614746}, "de": {"chunks": 3638.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177178, "cps": 28.421875, "mean_p": 0.181868, "boundary_entropy": 0.066448, "bpb": 1.723075}, "vi": {"chunks": 4133.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.224193, "cps": 32.289062, "mean_p": 0.192365, "boundary_entropy": 0.119286, "bpb": 1.394603}, "tr": {"chunks": 3299.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.182235, "cps": 25.773438, "mean_p": 0.228008, "boundary_entropy": 0.22463, "bpb": 1.625768}, "fi": {"chunks": 3606.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.191696, "cps": 28.171875, "mean_p": 0.189509, "boundary_entropy": 0.063756, "bpb": 1.793448}, "ru": {"chunks": 5844.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.308391, "cps": 45.65625, "mean_p": 0.20322, "boundary_entropy": 0.16931, "bpb": 1.010969}, "el": {"chunks": 5896.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.287287, "cps": 46.0625, "mean_p": 0.204791, "boundary_entropy": 0.22689, "bpb": 1.032958}, "hi": {"chunks": 7391.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.439287, "cps": 57.742188, "mean_p": 0.176213, "boundary_entropy": 0.135798, "bpb": 0.841866}, "ta": {"chunks": 8771.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.469163, "cps": 68.523438, "mean_p": 0.196197, "boundary_entropy": 0.219047, "bpb": 0.721054}, "th": {"chunks": 8538.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.530805, "cps": 66.703125, "mean_p": 0.184452, "boundary_entropy": 0.196146, "bpb": 0.831549}, "zh": {"chunks": 3715.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.629021, "cps": 29.023438, "mean_p": 0.228048, "boundary_entropy": 0.045941, "bpb": 2.395088}, "my": {"chunks": 7549.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.398406, "cps": 58.976562, "mean_p": 0.28942, "boundary_entropy": 0.481978, "bpb": 0.825547}}, "bpb": {"en": 1.61475, "de": 1.72307, "vi": 1.3946, "tr": 1.62577, "fi": 1.79345, "ru": 1.01097, "el": 1.03296, "hi": 0.84187, "ta": 0.72105, "th": 0.83155, "zh": 2.39509, "my": 0.82555}, "gini_cpc": 0.24877, "gini_cps": 0.210096, "bpic": 5.867224, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0172986575528578, "vi": 1.2872427283140662, "tr": 1.046333061011832, "fi": 1.1006570577970192, "ru": 1.770676309620291, "el": 1.6495094711756713, "hi": 2.5222394445117144, "ta": 2.693778146458248, "th": 3.047707396586941, "zh": 3.6116325568180607, "my": 2.2875164883704358}, "t": 5766.3} |
| {"event": "eval", "step": 20000, "phase": 1, "bpb": {"en": 1.61317, "de": 1.73032, "vi": 1.41149, "tr": 1.61721, "fi": 1.81019, "ru": 1.00904, "el": 1.03153, "hi": 0.84852, "ta": 0.72564, "th": 0.83102, "zh": 2.39932, "my": 0.83126}, "bytes_per_chunk": {"en": 5.806, "de": 5.802, "vi": 5.973, "tr": 6.068, "fi": 5.335, "ru": 5.843, "el": 6.197, "hi": 5.876, "ta": 5.923, "th": 5.516, "zh": 4.531, "my": 7.245}, "bpb_mean": 1.32156, "t": 5774.5} |
| {"event": "train", "step": 20000, "phase": 1, "objective": "global", "lm_loss": 1.08083, "bpb": 1.5593, "chunk_loss": 0.030126, "F": 0.17752, "G": 0.19354, "chunks": 181.78, "bytes_per_chunk": 5.633, "boundary_entropy": 0.10524, "router_noise": 0.0, "grad_norm": 0.3004, "lr": 0.0002628, "router_trainable": true, "sec_per_step": 0.3, "alpha": 0.03, "t": 5774.7} |
| {"event": "train", "step": 20500, "phase": 1, "objective": "global", "lm_loss": 0.97235, "bpb": 1.40281, "chunk_loss": 0.030076, "F": 0.17438, "G": 0.19427, "chunks": 178.56, "bytes_per_chunk": 5.735, "boundary_entropy": 0.12531, "router_noise": 0.0, "grad_norm": 0.29, "lr": 0.0002608, "router_trainable": true, "sec_per_step": 0.275, "alpha": 0.03, "t": 5912.2} |
| {"event": "train", "step": 21000, "phase": 1, "objective": "global", "lm_loss": 1.01904, "bpb": 1.47017, "chunk_loss": 0.030149, "F": 0.17937, "G": 0.20296, "chunks": 183.67, "bytes_per_chunk": 5.575, "boundary_entropy": 0.1373, "router_noise": 0.0, "grad_norm": 0.2722, "lr": 0.0002587, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 6050.2} |
| {"event": "train", "step": 21500, "phase": 1, "objective": "global", "lm_loss": 0.93813, "bpb": 1.35344, "chunk_loss": 0.03007, "F": 0.1722, "G": 0.19064, "chunks": 176.33, "bytes_per_chunk": 5.807, "boundary_entropy": 0.12204, "router_noise": 0.0, "grad_norm": 0.2738, "lr": 0.0002567, "router_trainable": true, "sec_per_step": 0.275, "alpha": 0.03, "t": 6187.4} |
| {"event": "train", "step": 22000, "phase": 1, "objective": "global", "lm_loss": 1.00157, "bpb": 1.44496, "chunk_loss": 0.030114, "F": 0.17702, "G": 0.19118, "chunks": 181.27, "bytes_per_chunk": 5.649, "boundary_entropy": 0.10811, "router_noise": 0.0, "grad_norm": 0.276, "lr": 0.0002545, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 6325.4} |
| {"event": "train", "step": 22500, "phase": 1, "objective": "global", "lm_loss": 0.97585, "bpb": 1.40786, "chunk_loss": 0.030138, "F": 0.17862, "G": 0.1936, "chunks": 182.91, "bytes_per_chunk": 5.598, "boundary_entropy": 0.11458, "router_noise": 0.0, "grad_norm": 0.2754, "lr": 0.0002524, "router_trainable": true, "sec_per_step": 0.274, "alpha": 0.03, "t": 6462.7} |
| {"event": "train", "step": 23000, "phase": 1, "objective": "global", "lm_loss": 1.01515, "bpb": 1.46455, "chunk_loss": 0.030122, "F": 0.1779, "G": 0.19288, "chunks": 182.17, "bytes_per_chunk": 5.621, "boundary_entropy": 0.10713, "router_noise": 0.0, "grad_norm": 0.2604, "lr": 0.0002502, "router_trainable": true, "sec_per_step": 0.274, "alpha": 0.03, "t": 6599.8} |
| {"event": "train", "step": 23500, "phase": 1, "objective": "global", "lm_loss": 0.97443, "bpb": 1.4058, "chunk_loss": 0.030071, "F": 0.17462, "G": 0.19583, "chunks": 178.81, "bytes_per_chunk": 5.727, "boundary_entropy": 0.12828, "router_noise": 0.0, "grad_norm": 0.2733, "lr": 0.000248, "router_trainable": true, "sec_per_step": 0.273, "alpha": 0.03, "t": 6736.5} |
| {"event": "probe", "step": 24000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2954.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.169741, "cps": 23.078125, "mean_p": 0.175032, "boundary_entropy": 0.049366, "bpb": 1.594428}, "de": {"chunks": 3579.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174305, "cps": 27.960938, "mean_p": 0.177416, "boundary_entropy": 0.057215, "bpb": 1.704394}, "vi": {"chunks": 4039.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.219094, "cps": 31.554688, "mean_p": 0.183809, "boundary_entropy": 0.102381, "bpb": 1.3676}, "tr": {"chunks": 3294.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181959, "cps": 25.734375, "mean_p": 0.232462, "boundary_entropy": 0.228405, "bpb": 1.602741}, "fi": {"chunks": 3520.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187125, "cps": 27.5, "mean_p": 0.183947, "boundary_entropy": 0.051908, "bpb": 1.756044}, "ru": {"chunks": 5944.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.313668, "cps": 46.4375, "mean_p": 0.197161, "boundary_entropy": 0.145226, "bpb": 0.992459}, "el": {"chunks": 5812.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.283194, "cps": 45.40625, "mean_p": 0.201928, "boundary_entropy": 0.209552, "bpb": 1.018668}, "hi": {"chunks": 7643.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.454264, "cps": 59.710938, "mean_p": 0.179614, "boundary_entropy": 0.13613, "bpb": 0.820899}, "ta": {"chunks": 8612.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.460658, "cps": 67.28125, "mean_p": 0.198673, "boundary_entropy": 0.222914, "bpb": 0.7096}, "th": {"chunks": 8233.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.511843, "cps": 64.320312, "mean_p": 0.173287, "boundary_entropy": 0.188484, "bpb": 0.819743}, "zh": {"chunks": 3716.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.629191, "cps": 29.03125, "mean_p": 0.229888, "boundary_entropy": 0.037708, "bpb": 2.362099}, "my": {"chunks": 9256.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.488495, "cps": 72.3125, "mean_p": 0.273749, "boundary_entropy": 0.43996, "bpb": 0.809083}}, "bpb": {"en": 1.59443, "de": 1.70439, "vi": 1.3676, "tr": 1.60274, "fi": 1.75604, "ru": 0.99246, "el": 1.01867, "hi": 0.8209, "ta": 0.7096, "th": 0.81974, "zh": 2.3621, "my": 0.80908}, "gini_cpc": 0.252467, "gini_cps": 0.223907, "bpic": 5.762304, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0268876255509032, "vi": 1.290756558524443, "tr": 1.0719799749653407, "fi": 1.1024132115325858, "ru": 1.8479202119392695, "el": 1.6683931178029006, "hi": 2.6762237216316356, "ta": 2.7138896257729512, "th": 3.0154399160350955, "zh": 3.7067721544091468, "my": 2.8778860835534776}, "t": 6877.4} |
| {"event": "eval", "step": 24000, "phase": 1, "bpb": {"en": 1.59256, "de": 1.7109, "vi": 1.38559, "tr": 1.59492, "fi": 1.77059, "ru": 0.99123, "el": 1.0176, "hi": 0.82782, "ta": 0.71481, "th": 0.81911, "zh": 2.36567, "my": 0.81333}, "bytes_per_chunk": {"en": 5.831, "de": 5.843, "vi": 6.106, "tr": 6.019, "fi": 5.409, "ru": 5.768, "el": 6.207, "hi": 5.664, "ta": 5.909, "th": 5.66, "zh": 4.53, "my": 5.828}, "bpb_mean": 1.30034, "t": 6885.9} |
| {"event": "train", "step": 24000, "phase": 1, "objective": "global", "lm_loss": 1.07242, "bpb": 1.54717, "chunk_loss": 0.030142, "F": 0.18068, "G": 0.19474, "chunks": 185.02, "bytes_per_chunk": 5.535, "boundary_entropy": 0.09928, "router_noise": 0.0, "grad_norm": 0.2863, "lr": 0.0002457, "router_trainable": true, "sec_per_step": 0.299, "alpha": 0.03, "t": 6886.2} |
| {"event": "train", "step": 24500, "phase": 1, "objective": "global", "lm_loss": 1.00894, "bpb": 1.4556, "chunk_loss": 0.030165, "F": 0.17828, "G": 0.19398, "chunks": 182.56, "bytes_per_chunk": 5.609, "boundary_entropy": 0.1097, "router_noise": 0.0, "grad_norm": 0.2988, "lr": 0.0002434, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 7024.2} |
| {"event": "train", "step": 25000, "phase": 1, "objective": "global", "lm_loss": 0.94114, "bpb": 1.35778, "chunk_loss": 0.03009, "F": 0.17055, "G": 0.18565, "chunks": 174.64, "bytes_per_chunk": 5.863, "boundary_entropy": 0.10599, "router_noise": 0.0, "grad_norm": 0.2778, "lr": 0.000241, "router_trainable": true, "sec_per_step": 0.277, "alpha": 0.03, "t": 7162.9} |
| {"event": "train", "step": 25500, "phase": 1, "objective": "global", "lm_loss": 0.95294, "bpb": 1.3748, "chunk_loss": 0.03008, "F": 0.17387, "G": 0.18537, "chunks": 178.05, "bytes_per_chunk": 5.751, "boundary_entropy": 0.09353, "router_noise": 0.0, "grad_norm": 0.2671, "lr": 0.0002387, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 7300.9} |
| {"event": "train", "step": 26000, "phase": 1, "objective": "global", "lm_loss": 0.92845, "bpb": 1.33948, "chunk_loss": 0.030127, "F": 0.17799, "G": 0.19343, "chunks": 182.27, "bytes_per_chunk": 5.618, "boundary_entropy": 0.11637, "router_noise": 0.0, "grad_norm": 0.2571, "lr": 0.0002363, "router_trainable": true, "sec_per_step": 0.273, "alpha": 0.03, "t": 7437.5} |
| {"event": "train", "step": 26500, "phase": 1, "objective": "global", "lm_loss": 0.98133, "bpb": 1.41576, "chunk_loss": 0.030089, "F": 0.17574, "G": 0.18538, "chunks": 179.95, "bytes_per_chunk": 5.69, "boundary_entropy": 0.08916, "router_noise": 0.0, "grad_norm": 0.2756, "lr": 0.0002338, "router_trainable": true, "sec_per_step": 0.274, "alpha": 0.03, "t": 7574.5} |
| {"event": "train", "step": 27000, "phase": 1, "objective": "global", "lm_loss": 0.90821, "bpb": 1.31027, "chunk_loss": 0.030078, "F": 0.17216, "G": 0.1868, "chunks": 176.3, "bytes_per_chunk": 5.808, "boundary_entropy": 0.11016, "router_noise": 0.0, "grad_norm": 0.267, "lr": 0.0002313, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 7712.5} |
| {"event": "train", "step": 27500, "phase": 1, "objective": "global", "lm_loss": 1.06211, "bpb": 1.5323, "chunk_loss": 0.030169, "F": 0.181, "G": 0.19291, "chunks": 185.34, "bytes_per_chunk": 5.525, "boundary_entropy": 0.09488, "router_noise": 0.0, "grad_norm": 0.2743, "lr": 0.0002288, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 7850.6} |
| {"event": "probe", "step": 28000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3008.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.172844, "cps": 23.5, "mean_p": 0.175661, "boundary_entropy": 0.043657, "bpb": 1.574679}, "de": {"chunks": 3661.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178298, "cps": 28.601562, "mean_p": 0.180054, "boundary_entropy": 0.052217, "bpb": 1.675446}, "vi": {"chunks": 4119.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.223434, "cps": 32.179688, "mean_p": 0.18554, "boundary_entropy": 0.093486, "bpb": 1.350343}, "tr": {"chunks": 3297.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.182125, "cps": 25.757812, "mean_p": 0.22574, "boundary_entropy": 0.211927, "bpb": 1.578835}, "fi": {"chunks": 3555.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188985, "cps": 27.773438, "mean_p": 0.185224, "boundary_entropy": 0.047702, "bpb": 1.732154}, "ru": {"chunks": 5784.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.305224, "cps": 45.1875, "mean_p": 0.185833, "boundary_entropy": 0.118758, "bpb": 0.981729}, "el": {"chunks": 6163.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.300297, "cps": 48.148438, "mean_p": 0.192965, "boundary_entropy": 0.180099, "bpb": 1.001387}, "hi": {"chunks": 7502.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.445884, "cps": 58.609375, "mean_p": 0.172477, "boundary_entropy": 0.12638, "bpb": 0.812734}, "ta": {"chunks": 8742.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.467612, "cps": 68.296875, "mean_p": 0.186656, "boundary_entropy": 0.186317, "bpb": 0.701066}, "th": {"chunks": 7917.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.492198, "cps": 61.851562, "mean_p": 0.166246, "boundary_entropy": 0.180598, "bpb": 0.810821}, "zh": {"chunks": 3706.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.627497, "cps": 28.953125, "mean_p": 0.228547, "boundary_entropy": 0.035785, "bpb": 2.345425}, "my": {"chunks": 7509.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.396295, "cps": 58.664062, "mean_p": 0.235416, "boundary_entropy": 0.401061, "bpb": 0.792951}}, "bpb": {"en": 1.57468, "de": 1.67545, "vi": 1.35034, "tr": 1.57884, "fi": 1.73215, "ru": 0.98173, "el": 1.00139, "hi": 0.81273, "ta": 0.70107, "th": 0.81082, "zh": 2.34543, "my": 0.79295}, "gini_cpc": 0.243523, "gini_cps": 0.20598, "bpic": 5.907686, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.031557901473601, "vi": 1.2926916343507089, "tr": 1.0536944292470627, "fi": 1.0933872616689873, "ru": 1.765896957278392, "el": 1.737391173602839, "hi": 2.5796944848408208, "ta": 2.705400842471249, "th": 2.847645135715183, "zh": 3.630431615702747, "my": 2.292793894348025}, "t": 7995.1} |
| {"event": "eval", "step": 28000, "phase": 1, "bpb": {"en": 1.5714, "de": 1.68335, "vi": 1.36739, "tr": 1.57156, "fi": 1.74821, "ru": 0.97875, "el": 0.99982, "hi": 0.81978, "ta": 0.70707, "th": 0.80862, "zh": 2.34677, "my": 0.79801}, "bytes_per_chunk": {"en": 5.767, "de": 5.732, "vi": 5.97, "tr": 6.014, "fi": 5.373, "ru": 5.853, "el": 5.891, "hi": 5.74, "ta": 5.889, "th": 5.865, "zh": 4.535, "my": 7.145}, "bpb_mean": 1.28339, "t": 8004.7} |
| {"event": "train", "step": 28000, "phase": 1, "objective": "global", "lm_loss": 0.99073, "bpb": 1.42933, "chunk_loss": 0.030148, "F": 0.17685, "G": 0.19211, "chunks": 181.09, "bytes_per_chunk": 5.655, "boundary_entropy": 0.10541, "router_noise": 0.0, "grad_norm": 0.3069, "lr": 0.0002263, "router_trainable": true, "sec_per_step": 0.309, "alpha": 0.03, "t": 8005.0} |
| {"event": "train", "step": 28500, "phase": 1, "objective": "global", "lm_loss": 1.00115, "bpb": 1.44436, "chunk_loss": 0.030154, "F": 0.17477, "G": 0.18563, "chunks": 178.97, "bytes_per_chunk": 5.722, "boundary_entropy": 0.09257, "router_noise": 0.0, "grad_norm": 0.2895, "lr": 0.0002238, "router_trainable": true, "sec_per_step": 0.279, "alpha": 0.03, "t": 8144.8} |
| {"event": "train", "step": 29000, "phase": 1, "objective": "global", "lm_loss": 1.02671, "bpb": 1.48123, "chunk_loss": 0.030096, "F": 0.1736, "G": 0.18556, "chunks": 177.77, "bytes_per_chunk": 5.76, "boundary_entropy": 0.09349, "router_noise": 0.0, "grad_norm": 0.2827, "lr": 0.0002212, "router_trainable": true, "sec_per_step": 0.279, "alpha": 0.03, "t": 8284.4} |
| {"event": "train", "step": 29500, "phase": 1, "objective": "global", "lm_loss": 0.95633, "bpb": 1.3797, "chunk_loss": 0.030106, "F": 0.17419, "G": 0.1837, "chunks": 178.38, "bytes_per_chunk": 5.741, "boundary_entropy": 0.08449, "router_noise": 0.0, "grad_norm": 0.278, "lr": 0.0002186, "router_trainable": true, "sec_per_step": 0.275, "alpha": 0.03, "t": 8422.0} |
| {"event": "train", "step": 30000, "phase": 1, "objective": "global", "lm_loss": 0.94722, "bpb": 1.36654, "chunk_loss": 0.030145, "F": 0.17943, "G": 0.1952, "chunks": 183.73, "bytes_per_chunk": 5.573, "boundary_entropy": 0.10927, "router_noise": 0.0, "grad_norm": 0.2745, "lr": 0.0002159, "router_trainable": true, "sec_per_step": 0.278, "alpha": 0.03, "t": 8561.2} |
| {"event": "train", "step": 30500, "phase": 1, "objective": "global", "lm_loss": 0.96556, "bpb": 1.39301, "chunk_loss": 0.030077, "F": 0.17166, "G": 0.18401, "chunks": 175.78, "bytes_per_chunk": 5.825, "boundary_entropy": 0.09614, "router_noise": 0.0, "grad_norm": 0.2844, "lr": 0.0002133, "router_trainable": true, "sec_per_step": 0.281, "alpha": 0.03, "t": 8702.0} |
| {"event": "phase_transition", "step": 30520, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 8707.0} |
| {"event": "train", "step": 31000, "phase": 2, "objective": "parity_a", "lm_loss": 0.97496, "bpb": 1.40658, "chunk_loss": 0.030515, "F": 0.15569, "G": 0.15852, "chunks": 159.42, "bytes_per_chunk": 6.423, "boundary_entropy": 0.05267, "router_noise": 0.0, "grad_norm": 0.2925, "lr": 0.0002106, "router_trainable": true, "sec_per_step": 0.279, "n_target_mean": 9.3311, "alpha": 0.03, "t": 8841.7} |
| {"event": "train", "step": 31500, "phase": 2, "objective": "parity_a", "lm_loss": 0.98854, "bpb": 1.42617, "chunk_loss": 0.030279, "F": 0.15498, "G": 0.1568, "chunks": 158.7, "bytes_per_chunk": 6.452, "boundary_entropy": 0.04057, "router_noise": 0.0, "grad_norm": 0.2772, "lr": 0.0002079, "router_trainable": true, "sec_per_step": 0.283, "n_target_mean": 8.707, "alpha": 0.03, "t": 8983.2} |
| {"event": "probe", "step": 32000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3021.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.173591, "cps": 23.601562, "mean_p": 0.172936, "boundary_entropy": 0.024462, "bpb": 1.558769}, "de": {"chunks": 3606.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.17562, "cps": 28.171875, "mean_p": 0.172559, "boundary_entropy": 0.024057, "bpb": 1.660817}, "vi": {"chunks": 3994.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.216653, "cps": 31.203125, "mean_p": 0.164081, "boundary_entropy": 0.033752, "bpb": 1.334357}, "tr": {"chunks": 3151.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.17406, "cps": 24.617188, "mean_p": 0.166274, "boundary_entropy": 0.055112, "bpb": 1.561049}, "fi": {"chunks": 3499.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186008, "cps": 27.335938, "mean_p": 0.178458, "boundary_entropy": 0.030762, "bpb": 1.706635}, "ru": {"chunks": 4353.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.22971, "cps": 34.007812, "mean_p": 0.120151, "boundary_entropy": 0.051983, "bpb": 1.007704}, "el": {"chunks": 3647.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.177703, "cps": 28.492188, "mean_p": 0.099394, "boundary_entropy": 0.047422, "bpb": 1.024881}, "hi": {"chunks": 4214.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.250461, "cps": 32.921875, "mean_p": 0.098432, "boundary_entropy": 0.067257, "bpb": 0.856224}, "ta": {"chunks": 3213.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.171864, "cps": 25.101562, "mean_p": 0.074531, "boundary_entropy": 0.090944, "bpb": 0.767264}, "th": {"chunks": 2915.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.181225, "cps": 22.773438, "mean_p": 0.070482, "boundary_entropy": 0.074512, "bpb": 0.858396}, "zh": {"chunks": 3715.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.629021, "cps": 29.023438, "mean_p": 0.229267, "boundary_entropy": 0.017916, "bpb": 2.31806}, "my": {"chunks": 3243.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.171153, "cps": 25.335938, "mean_p": 0.082561, "boundary_entropy": 0.132189, "bpb": 0.795101}}, "bpb": {"en": 1.55877, "de": 1.66082, "vi": 1.33436, "tr": 1.56105, "fi": 1.70663, "ru": 1.0077, "el": 1.02488, "hi": 0.85622, "ta": 0.76726, "th": 0.8584, "zh": 2.31806, "my": 0.7951}, "gini_cpc": 0.194381, "gini_cps": 0.071028, "bpic": 9.015081, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0116882596189891, "vi": 1.2480681877252506, "tr": 1.0027005350219018, "fi": 1.0715327617105017, "ru": 1.3232833490107507, "el": 1.0236896516568401, "hi": 1.4428223239699516, "ta": 0.9900534716465883, "th": 1.0439768990396414, "zh": 3.6235876463805172, "my": 0.9859547134592369}, "t": 9127.5} |
| {"event": "eval", "step": 32000, "phase": 2, "bpb": {"en": 1.55572, "de": 1.6688, "vi": 1.35305, "tr": 1.5535, "fi": 1.7241, "ru": 1.00601, "el": 1.02382, "hi": 0.86277, "ta": 0.77285, "th": 0.85741, "zh": 2.32278, "my": 0.80034}, "bytes_per_chunk": {"en": 5.798, "de": 5.821, "vi": 6.206, "tr": 6.289, "fi": 5.495, "ru": 7.924, "el": 9.895, "hi": 10.445, "ta": 16.088, "th": 15.522, "zh": 4.529, "my": 17.461}, "bpb_mean": 1.29176, "t": 9136.6} |
| {"event": "train", "step": 32000, "phase": 2, "objective": "parity_a", "lm_loss": 0.99706, "bpb": 1.43846, "chunk_loss": 0.030251, "F": 0.15981, "G": 0.15975, "chunks": 163.64, "bytes_per_chunk": 6.258, "boundary_entropy": 0.03498, "router_noise": 0.0, "grad_norm": 0.2877, "lr": 0.0002052, "router_trainable": true, "sec_per_step": 0.307, "n_target_mean": 8.1654, "alpha": 0.03, "t": 9136.9} |
| {"event": "train", "step": 32500, "phase": 2, "objective": "parity_a", "lm_loss": 0.97023, "bpb": 1.39975, "chunk_loss": 0.03024, "F": 0.1526, "G": 0.15259, "chunks": 156.27, "bytes_per_chunk": 6.553, "boundary_entropy": 0.03587, "router_noise": 0.0, "grad_norm": 0.312, "lr": 0.0002025, "router_trainable": true, "sec_per_step": 0.279, "n_target_mean": 8.9954, "alpha": 0.03, "t": 9276.4} |
| {"event": "train", "step": 33000, "phase": 2, "objective": "parity_a", "lm_loss": 0.94283, "bpb": 1.36022, "chunk_loss": 0.030262, "F": 0.14607, "G": 0.14598, "chunks": 149.58, "bytes_per_chunk": 6.846, "boundary_entropy": 0.03303, "router_noise": 0.0, "grad_norm": 0.2837, "lr": 0.0001997, "router_trainable": true, "sec_per_step": 0.28, "n_target_mean": 9.2673, "alpha": 0.03, "t": 9416.6} |
| {"event": "train", "step": 33500, "phase": 2, "objective": "parity_a", "lm_loss": 0.97445, "bpb": 1.40584, "chunk_loss": 0.030223, "F": 0.15031, "G": 0.15066, "chunks": 153.92, "bytes_per_chunk": 6.653, "boundary_entropy": 0.03212, "router_noise": 0.0, "grad_norm": 0.2802, "lr": 0.0001969, "router_trainable": true, "sec_per_step": 0.284, "n_target_mean": 8.8451, "alpha": 0.03, "t": 9558.8} |
| {"event": "train", "step": 34000, "phase": 2, "objective": "parity_a", "lm_loss": 0.99179, "bpb": 1.43085, "chunk_loss": 0.030195, "F": 0.14929, "G": 0.14863, "chunks": 152.88, "bytes_per_chunk": 6.698, "boundary_entropy": 0.0287, "router_noise": 0.0, "grad_norm": 0.288, "lr": 0.0001942, "router_trainable": true, "sec_per_step": 0.28, "n_target_mean": 8.7307, "alpha": 0.03, "t": 9698.9} |
| {"event": "train", "step": 34500, "phase": 2, "objective": "parity_a", "lm_loss": 1.01084, "bpb": 1.45833, "chunk_loss": 0.030163, "F": 0.15565, "G": 0.15521, "chunks": 159.39, "bytes_per_chunk": 6.424, "boundary_entropy": 0.02652, "router_noise": 0.0, "grad_norm": 0.2878, "lr": 0.0001914, "router_trainable": true, "sec_per_step": 0.283, "n_target_mean": 8.208, "alpha": 0.03, "t": 9840.6} |
| {"event": "train", "step": 35000, "phase": 2, "objective": "parity_a", "lm_loss": 0.97272, "bpb": 1.40334, "chunk_loss": 0.030306, "F": 0.15227, "G": 0.1511, "chunks": 155.92, "bytes_per_chunk": 6.567, "boundary_entropy": 0.03008, "router_noise": 0.0, "grad_norm": 0.2905, "lr": 0.0001886, "router_trainable": true, "sec_per_step": 0.273, "n_target_mean": 9.0422, "alpha": 0.03, "t": 9977.2} |
| {"event": "train", "step": 35500, "phase": 2, "objective": "parity_a", "lm_loss": 0.91242, "bpb": 1.31635, "chunk_loss": 0.030251, "F": 0.14005, "G": 0.13927, "chunks": 143.41, "bytes_per_chunk": 7.141, "boundary_entropy": 0.02971, "router_noise": 0.0, "grad_norm": 0.2871, "lr": 0.0001858, "router_trainable": true, "sec_per_step": 0.281, "n_target_mean": 9.7517, "alpha": 0.03, "t": 10117.5} |
| {"event": "probe", "step": 36000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 2976.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.171005, "cps": 23.25, "mean_p": 0.17136, "boundary_entropy": 0.018999, "bpb": 1.5437}, "de": {"chunks": 3611.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.175863, "cps": 28.210938, "mean_p": 0.171496, "boundary_entropy": 0.018864, "bpb": 1.640683}, "vi": {"chunks": 3841.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.208354, "cps": 30.007812, "mean_p": 0.157855, "boundary_entropy": 0.025406, "bpb": 1.319721}, "tr": {"chunks": 3155.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.174281, "cps": 24.648438, "mean_p": 0.15992, "boundary_entropy": 0.028365, "bpb": 1.544018}, "fi": {"chunks": 3526.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187444, "cps": 27.546875, "mean_p": 0.179544, "boundary_entropy": 0.021197, "bpb": 1.678283}, "ru": {"chunks": 4162.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.219631, "cps": 32.515625, "mean_p": 0.115207, "boundary_entropy": 0.038583, "bpb": 0.991302}, "el": {"chunks": 3477.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.16942, "cps": 27.164062, "mean_p": 0.094017, "boundary_entropy": 0.028953, "bpb": 1.005333}, "hi": {"chunks": 3947.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.234591, "cps": 30.835938, "mean_p": 0.091705, "boundary_entropy": 0.056435, "bpb": 0.836994}, "ta": {"chunks": 2994.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.16015, "cps": 23.390625, "mean_p": 0.062237, "boundary_entropy": 0.064171, "bpb": 0.742162}, "th": {"chunks": 2935.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.182468, "cps": 22.929688, "mean_p": 0.068013, "boundary_entropy": 0.056453, "bpb": 0.830418}, "zh": {"chunks": 3718.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.629529, "cps": 29.046875, "mean_p": 0.229396, "boundary_entropy": 0.013235, "bpb": 2.292198}, "my": {"chunks": 2800.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.147773, "cps": 21.875, "mean_p": 0.059025, "boundary_entropy": 0.065191, "bpb": 0.778275}}, "bpb": {"en": 1.5437, "de": 1.64068, "vi": 1.31972, "tr": 1.54402, "fi": 1.67828, "ru": 0.9913, "el": 1.00533, "hi": 0.83699, "ta": 0.74216, "th": 0.83042, "zh": 2.2922, "my": 0.77827}, "gini_cpc": 0.205274, "gini_cps": 0.071261, "bpic": 9.328205, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0284099612268527, "vi": 1.2184069245037052, "tr": 1.019154450875189, "fi": 1.0961288736343353, "ru": 1.2843519661247764, "el": 0.9907293724743128, "hi": 1.3718393208071704, "ta": 0.9365209992321561, "th": 1.0670339401566276, "zh": 3.6813502261232425, "my": 0.8641434622430147}, "t": 10261.0} |
| {"event": "eval", "step": 36000, "phase": 2, "bpb": {"en": 1.54084, "de": 1.64932, "vi": 1.34002, "tr": 1.5365, "fi": 1.69623, "ru": 0.98946, "el": 1.00371, "hi": 0.8427, "ta": 0.74778, "th": 0.83224, "zh": 2.29644, "my": 0.78372}, "bytes_per_chunk": {"en": 5.787, "de": 5.808, "vi": 6.336, "tr": 6.298, "fi": 5.424, "ru": 8.377, "el": 10.275, "hi": 10.973, "ta": 17.056, "th": 15.082, "zh": 4.53, "my": 19.615}, "bpb_mean": 1.27158, "t": 10269.1} |
| {"event": "train", "step": 36000, "phase": 2, "objective": "parity_a", "lm_loss": 0.9474, "bpb": 1.3668, "chunk_loss": 0.030145, "F": 0.14726, "G": 0.1471, "chunks": 150.8, "bytes_per_chunk": 6.791, "boundary_entropy": 0.02801, "router_noise": 0.0, "grad_norm": 0.289, "lr": 0.0001829, "router_trainable": true, "sec_per_step": 0.304, "n_target_mean": 9.0326, "alpha": 0.03, "t": 10269.4} |
| {"event": "train", "step": 36500, "phase": 2, "objective": "parity_a", "lm_loss": 0.95921, "bpb": 1.38384, "chunk_loss": 0.030244, "F": 0.14566, "G": 0.14521, "chunks": 149.16, "bytes_per_chunk": 6.865, "boundary_entropy": 0.02783, "router_noise": 0.0, "grad_norm": 0.2958, "lr": 0.0001801, "router_trainable": true, "sec_per_step": 0.279, "n_target_mean": 9.6785, "alpha": 0.03, "t": 10408.7} |
| {"event": "train", "step": 37000, "phase": 2, "objective": "parity_a", "lm_loss": 0.98391, "bpb": 1.41948, "chunk_loss": 0.030197, "F": 0.15547, "G": 0.15414, "chunks": 159.2, "bytes_per_chunk": 6.432, "boundary_entropy": 0.02462, "router_noise": 0.0, "grad_norm": 0.2984, "lr": 0.0001773, "router_trainable": true, "sec_per_step": 0.279, "n_target_mean": 8.4222, "alpha": 0.03, "t": 10548.1} |
| {"event": "train", "step": 37500, "phase": 2, "objective": "parity_a", "lm_loss": 0.96657, "bpb": 1.39447, "chunk_loss": 0.030192, "F": 0.14484, "G": 0.14409, "chunks": 148.31, "bytes_per_chunk": 6.904, "boundary_entropy": 0.02577, "router_noise": 0.0, "grad_norm": 0.2945, "lr": 0.0001744, "router_trainable": true, "sec_per_step": 0.278, "n_target_mean": 8.9795, "alpha": 0.03, "t": 10687.2} |
| {"event": "train", "step": 38000, "phase": 2, "objective": "parity_a", "lm_loss": 0.84981, "bpb": 1.22601, "chunk_loss": 0.030192, "F": 0.13216, "G": 0.1308, "chunks": 135.33, "bytes_per_chunk": 7.567, "boundary_entropy": 0.02967, "router_noise": 0.0, "grad_norm": 0.2781, "lr": 0.0001716, "router_trainable": true, "sec_per_step": 0.279, "n_target_mean": 10.1645, "alpha": 0.03, "t": 10826.7} |
| {"event": "train", "step": 38500, "phase": 2, "objective": "parity_a", "lm_loss": 0.98334, "bpb": 1.41866, "chunk_loss": 0.030168, "F": 0.15038, "G": 0.14917, "chunks": 153.98, "bytes_per_chunk": 6.65, "boundary_entropy": 0.0245, "router_noise": 0.0, "grad_norm": 0.294, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.279, "n_target_mean": 8.5988, "alpha": 0.03, "t": 10966.0} |
| {"event": "train", "step": 39000, "phase": 2, "objective": "parity_a", "lm_loss": 0.8844, "bpb": 1.27591, "chunk_loss": 0.030194, "F": 0.14072, "G": 0.13969, "chunks": 144.09, "bytes_per_chunk": 7.106, "boundary_entropy": 0.02558, "router_noise": 0.0, "grad_norm": 0.2891, "lr": 0.0001659, "router_trainable": true, "sec_per_step": 0.278, "n_target_mean": 9.1998, "alpha": 0.03, "t": 11105.0} |
| {"event": "train", "step": 39500, "phase": 2, "objective": "parity_a", "lm_loss": 1.06015, "bpb": 1.52947, "chunk_loss": 0.030226, "F": 0.16176, "G": 0.16129, "chunks": 165.64, "bytes_per_chunk": 6.182, "boundary_entropy": 0.02163, "router_noise": 0.0, "grad_norm": 0.3171, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.275, "n_target_mean": 8.2192, "alpha": 0.03, "t": 11242.4} |
| {"event": "probe", "step": 40000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 2999.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.172327, "cps": 23.429688, "mean_p": 0.171782, "boundary_entropy": 0.017791, "bpb": 1.527981}, "de": {"chunks": 3569.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.173818, "cps": 27.882812, "mean_p": 0.169083, "boundary_entropy": 0.01807, "bpb": 1.628471}, "vi": {"chunks": 3887.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.210849, "cps": 30.367188, "mean_p": 0.1598, "boundary_entropy": 0.019929, "bpb": 1.304042}, "tr": {"chunks": 3151.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.17406, "cps": 24.617188, "mean_p": 0.159133, "boundary_entropy": 0.022556, "bpb": 1.528048}, "fi": {"chunks": 3485.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.185264, "cps": 27.226562, "mean_p": 0.177356, "boundary_entropy": 0.017999, "bpb": 1.664036}, "ru": {"chunks": 4052.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.213826, "cps": 31.65625, "mean_p": 0.112148, "boundary_entropy": 0.03251, "bpb": 0.978322}, "el": {"chunks": 3327.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.162111, "cps": 25.992188, "mean_p": 0.088821, "boundary_entropy": 0.023204, "bpb": 0.996307}, "hi": {"chunks": 3713.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.220684, "cps": 29.007812, "mean_p": 0.086977, "boundary_entropy": 0.05637, "bpb": 0.827534}, "ta": {"chunks": 2682.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.143461, "cps": 20.953125, "mean_p": 0.056161, "boundary_entropy": 0.052634, "bpb": 0.738436}, "th": {"chunks": 2811.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.174759, "cps": 21.960938, "mean_p": 0.061955, "boundary_entropy": 0.041203, "bpb": 0.817557}, "zh": {"chunks": 3715.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.629021, "cps": 29.023438, "mean_p": 0.229206, "boundary_entropy": 0.011081, "bpb": 2.272762}, "my": {"chunks": 3088.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.162972, "cps": 24.125, "mean_p": 0.063712, "boundary_entropy": 0.061787, "bpb": 0.77328}}, "bpb": {"en": 1.52798, "de": 1.62847, "vi": 1.30404, "tr": 1.52805, "fi": 1.66404, "ru": 0.97832, "el": 0.99631, "hi": 0.82753, "ta": 0.73844, "th": 0.81756, "zh": 2.27276, "my": 0.77328}, "gini_cpc": 0.204968, "gini_cps": 0.070619, "bpic": 9.48099, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0086530247659462, "vi": 1.2235424826990065, "tr": 1.010056124141769, "fi": 1.0750744817053028, "ru": 1.2408174052740535, "el": 0.9407183381645657, "hi": 1.2806118929192345, "ta": 0.8324937054439381, "th": 1.0141155330711729, "zh": 3.6501694830662026, "my": 0.9457178147412869}, "t": 11384.9} |
| {"event": "eval", "step": 40000, "phase": 2, "bpb": {"en": 1.52541, "de": 1.63608, "vi": 1.32293, "tr": 1.52084, "fi": 1.68202, "ru": 0.97786, "el": 0.9952, "hi": 0.83302, "ta": 0.7438, "th": 0.81845, "zh": 2.27607, "my": 0.77872}, "bytes_per_chunk": {"en": 5.806, "de": 5.908, "vi": 6.352, "tr": 6.33, "fi": 5.48, "ru": 8.642, "el": 10.697, "hi": 11.412, "ta": 18.888, "th": 16.444, "zh": 4.536, "my": 17.63}, "bpb_mean": 1.2592, "t": 11392.5} |
| {"event": "train", "step": 40000, "phase": 2, "objective": "parity_a", "lm_loss": 0.98622, "bpb": 1.42281, "chunk_loss": 0.030177, "F": 0.15211, "G": 0.15139, "chunks": 155.77, "bytes_per_chunk": 6.574, "boundary_entropy": 0.02307, "router_noise": 0.0, "grad_norm": 0.303, "lr": 0.0001601, "router_trainable": true, "sec_per_step": 0.301, "n_target_mean": 8.6344, "alpha": 0.03, "t": 11392.8} |
| {"event": "train", "step": 40500, "phase": 2, "objective": "parity_a", "lm_loss": 0.93401, "bpb": 1.34749, "chunk_loss": 0.030241, "F": 0.14461, "G": 0.14374, "chunks": 148.08, "bytes_per_chunk": 6.915, "boundary_entropy": 0.0242, "router_noise": 0.0, "grad_norm": 0.2873, "lr": 0.0001573, "router_trainable": true, "sec_per_step": 0.277, "n_target_mean": 8.9806, "alpha": 0.03, "t": 11531.1} |
| {"event": "train", "step": 41000, "phase": 2, "objective": "parity_a", "lm_loss": 0.96899, "bpb": 1.39795, "chunk_loss": 0.030186, "F": 0.14384, "G": 0.14293, "chunks": 147.3, "bytes_per_chunk": 6.952, "boundary_entropy": 0.02635, "router_noise": 0.0, "grad_norm": 0.3079, "lr": 0.0001545, "router_trainable": true, "sec_per_step": 0.277, "n_target_mean": 9.707, "alpha": 0.03, "t": 11669.5} |
| {"event": "train", "step": 41500, "phase": 2, "objective": "parity_a", "lm_loss": 0.91203, "bpb": 1.31578, "chunk_loss": 0.030164, "F": 0.14778, "G": 0.1469, "chunks": 151.33, "bytes_per_chunk": 6.767, "boundary_entropy": 0.02222, "router_noise": 0.0, "grad_norm": 0.2989, "lr": 0.0001516, "router_trainable": true, "sec_per_step": 0.281, "n_target_mean": 8.9677, "alpha": 0.03, "t": 11809.9} |
| {"event": "train", "step": 42000, "phase": 2, "objective": "parity_a", "lm_loss": 0.90772, "bpb": 1.30956, "chunk_loss": 0.030173, "F": 0.13463, "G": 0.13385, "chunks": 137.86, "bytes_per_chunk": 7.428, "boundary_entropy": 0.02514, "router_noise": 0.0, "grad_norm": 0.302, "lr": 0.0001488, "router_trainable": true, "sec_per_step": 0.275, "n_target_mean": 10.1977, "alpha": 0.03, "t": 11947.4} |
| {"event": "train", "step": 42500, "phase": 2, "objective": "parity_a", "lm_loss": 0.95757, "bpb": 1.38148, "chunk_loss": 0.03024, "F": 0.14996, "G": 0.14839, "chunks": 153.56, "bytes_per_chunk": 6.668, "boundary_entropy": 0.02437, "router_noise": 0.0, "grad_norm": 0.3218, "lr": 0.0001459, "router_trainable": true, "sec_per_step": 0.28, "n_target_mean": 9.1641, "alpha": 0.03, "t": 12087.5} |
| {"event": "train", "step": 43000, "phase": 2, "objective": "parity_a", "lm_loss": 0.92929, "bpb": 1.34068, "chunk_loss": 0.030188, "F": 0.15109, "G": 0.14988, "chunks": 154.72, "bytes_per_chunk": 6.618, "boundary_entropy": 0.02285, "router_noise": 0.0, "grad_norm": 0.3025, "lr": 0.0001431, "router_trainable": true, "sec_per_step": 0.277, "n_target_mean": 8.5046, "alpha": 0.03, "t": 12226.2} |
| {"event": "train", "step": 43500, "phase": 2, "objective": "parity_a", "lm_loss": 0.95896, "bpb": 1.38349, "chunk_loss": 0.030245, "F": 0.15038, "G": 0.1493, "chunks": 153.98, "bytes_per_chunk": 6.65, "boundary_entropy": 0.02385, "router_noise": 0.0, "grad_norm": 0.3122, "lr": 0.0001403, "router_trainable": true, "sec_per_step": 0.278, "n_target_mean": 9.3288, "alpha": 0.03, "t": 12365.0} |
| {"event": "probe", "step": 44000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3008.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.172844, "cps": 23.5, "mean_p": 0.171042, "boundary_entropy": 0.017438, "bpb": 1.513561}, "de": {"chunks": 3585.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174597, "cps": 28.007812, "mean_p": 0.169637, "boundary_entropy": 0.01714, "bpb": 1.611626}, "vi": {"chunks": 3837.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.208137, "cps": 29.976562, "mean_p": 0.158191, "boundary_entropy": 0.018895, "bpb": 1.291032}, "tr": {"chunks": 3139.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.173397, "cps": 24.523438, "mean_p": 0.158474, "boundary_entropy": 0.020045, "bpb": 1.512276}, "fi": {"chunks": 3493.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.185689, "cps": 27.289062, "mean_p": 0.177882, "boundary_entropy": 0.017442, "bpb": 1.637752}, "ru": {"chunks": 4018.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.212032, "cps": 31.390625, "mean_p": 0.110402, "boundary_entropy": 0.03375, "bpb": 0.963148}, "el": {"chunks": 3363.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.163865, "cps": 26.273438, "mean_p": 0.089817, "boundary_entropy": 0.02436, "bpb": 0.976415}, "hi": {"chunks": 3649.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.21688, "cps": 28.507812, "mean_p": 0.084213, "boundary_entropy": 0.048356, "bpb": 0.813898}, "ta": {"chunks": 3043.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.162771, "cps": 23.773438, "mean_p": 0.060637, "boundary_entropy": 0.039643, "bpb": 0.719432}, "th": {"chunks": 2814.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.174946, "cps": 21.984375, "mean_p": 0.062024, "boundary_entropy": 0.042119, "bpb": 0.807835}, "zh": {"chunks": 3694.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.625466, "cps": 28.859375, "mean_p": 0.228285, "boundary_entropy": 0.010622, "bpb": 2.251946}, "my": {"chunks": 2926.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.154423, "cps": 22.859375, "mean_p": 0.057073, "boundary_entropy": 0.047994, "bpb": 0.763323}}, "bpb": {"en": 1.51356, "de": 1.61163, "vi": 1.29103, "tr": 1.51228, "fi": 1.63775, "ru": 0.96315, "el": 0.97641, "hi": 0.8139, "ta": 0.71943, "th": 0.80784, "zh": 2.25195, "my": 0.76332}, "gini_cpc": 0.197435, "gini_cps": 0.06315, "bpic": 9.459957, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0101434244148755, "vi": 1.204189803593996, "tr": 1.0031989121645524, "fi": 1.0743183417749007, "ru": 1.2267244077359232, "el": 0.9480523311417082, "hi": 1.2547727506243875, "ta": 0.9417221189247323, "th": 1.0121603400154762, "zh": 3.6186763055601587, "my": 0.8934232167881637}, "t": 12508.1} |
| {"event": "eval", "step": 44000, "phase": 2, "bpb": {"en": 1.5108, "de": 1.61891, "vi": 1.30967, "tr": 1.50351, "fi": 1.65649, "ru": 0.96238, "el": 0.97555, "hi": 0.81955, "ta": 0.72561, "th": 0.80793, "zh": 2.2563, "my": 0.76975}, "bytes_per_chunk": {"en": 5.8, "de": 5.854, "vi": 6.367, "tr": 6.318, "fi": 5.451, "ru": 8.714, "el": 10.652, "hi": 11.451, "ta": 17.144, "th": 15.633, "zh": 4.542, "my": 17.598}, "bpb_mean": 1.24304, "t": 12516.0} |
| {"event": "train", "step": 44000, "phase": 2, "objective": "parity_a", "lm_loss": 0.88325, "bpb": 1.27426, "chunk_loss": 0.030211, "F": 0.13815, "G": 0.13644, "chunks": 141.47, "bytes_per_chunk": 7.238, "boundary_entropy": 0.02488, "router_noise": 0.0, "grad_norm": 0.3013, "lr": 0.0001375, "router_trainable": true, "sec_per_step": 0.303, "n_target_mean": 9.8917, "alpha": 0.03, "t": 12516.2} |
| {"event": "train", "step": 44500, "phase": 2, "objective": "parity_a", "lm_loss": 0.92919, "bpb": 1.34054, "chunk_loss": 0.030159, "F": 0.14676, "G": 0.14559, "chunks": 150.28, "bytes_per_chunk": 6.814, "boundary_entropy": 0.02369, "router_noise": 0.0, "grad_norm": 0.3334, "lr": 0.0001347, "router_trainable": true, "sec_per_step": 0.28, "n_target_mean": 9.0233, "alpha": 0.03, "t": 12656.3} |
| {"event": "train", "step": 45000, "phase": 2, "objective": "parity_a", "lm_loss": 0.90664, "bpb": 1.308, "chunk_loss": 0.030142, "F": 0.13918, "G": 0.13749, "chunks": 142.52, "bytes_per_chunk": 7.185, "boundary_entropy": 0.02302, "router_noise": 0.0, "grad_norm": 0.3085, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.282, "n_target_mean": 9.5583, "alpha": 0.03, "t": 12797.1} |
| {"event": "train", "step": 45500, "phase": 2, "objective": "parity_a", "lm_loss": 0.88668, "bpb": 1.27922, "chunk_loss": 0.030175, "F": 0.14493, "G": 0.14394, "chunks": 148.41, "bytes_per_chunk": 6.9, "boundary_entropy": 0.02302, "router_noise": 0.0, "grad_norm": 0.3137, "lr": 0.0001292, "router_trainable": true, "sec_per_step": 0.281, "n_target_mean": 9.6359, "alpha": 0.03, "t": 12937.5} |
| {"event": "train", "step": 46000, "phase": 2, "objective": "parity_a", "lm_loss": 0.992, "bpb": 1.43115, "chunk_loss": 0.030211, "F": 0.15901, "G": 0.1574, "chunks": 162.83, "bytes_per_chunk": 6.289, "boundary_entropy": 0.02036, "router_noise": 0.0, "grad_norm": 0.3303, "lr": 0.0001264, "router_trainable": true, "sec_per_step": 0.282, "n_target_mean": 8.2977, "alpha": 0.03, "t": 13078.7} |
| {"event": "train", "step": 46500, "phase": 2, "objective": "parity_a", "lm_loss": 0.90555, "bpb": 1.30643, "chunk_loss": 0.030252, "F": 0.14423, "G": 0.143, "chunks": 147.69, "bytes_per_chunk": 6.934, "boundary_entropy": 0.02455, "router_noise": 0.0, "grad_norm": 0.2957, "lr": 0.0001237, "router_trainable": true, "sec_per_step": 0.277, "n_target_mean": 9.4783, "alpha": 0.03, "t": 13217.3} |
| {"event": "train", "step": 47000, "phase": 2, "objective": "parity_a", "lm_loss": 0.84191, "bpb": 1.21462, "chunk_loss": 0.030219, "F": 0.13356, "G": 0.13243, "chunks": 136.77, "bytes_per_chunk": 7.487, "boundary_entropy": 0.02509, "router_noise": 0.0, "grad_norm": 0.3005, "lr": 0.000121, "router_trainable": true, "sec_per_step": 0.275, "n_target_mean": 10.3482, "alpha": 0.03, "t": 13355.0} |
| {"event": "train", "step": 47500, "phase": 2, "objective": "parity_a", "lm_loss": 0.97071, "bpb": 1.40044, "chunk_loss": 0.03016, "F": 0.15218, "G": 0.15129, "chunks": 155.83, "bytes_per_chunk": 6.571, "boundary_entropy": 0.02178, "router_noise": 0.0, "grad_norm": 0.3154, "lr": 0.0001183, "router_trainable": true, "sec_per_step": 0.273, "n_target_mean": 8.4947, "alpha": 0.03, "t": 13491.6} |
| {"event": "probe", "step": 48000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3006.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.172729, "cps": 23.484375, "mean_p": 0.170892, "boundary_entropy": 0.016762, "bpb": 1.501013}, "de": {"chunks": 3598.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.17523, "cps": 28.109375, "mean_p": 0.169475, "boundary_entropy": 0.019091, "bpb": 1.594938}, "vi": {"chunks": 3845.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.208571, "cps": 30.039062, "mean_p": 0.158177, "boundary_entropy": 0.019208, "bpb": 1.276796}, "tr": {"chunks": 3137.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.173286, "cps": 24.507812, "mean_p": 0.158034, "boundary_entropy": 0.018335, "bpb": 1.49573}, "fi": {"chunks": 3528.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18755, "cps": 27.5625, "mean_p": 0.17899, "boundary_entropy": 0.017722, "bpb": 1.620074}, "ru": {"chunks": 3998.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.210976, "cps": 31.234375, "mean_p": 0.110138, "boundary_entropy": 0.032589, "bpb": 0.951935}, "el": {"chunks": 3246.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.158164, "cps": 25.359375, "mean_p": 0.086274, "boundary_entropy": 0.02405, "bpb": 0.963843}, "hi": {"chunks": 3683.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.2189, "cps": 28.773438, "mean_p": 0.084089, "boundary_entropy": 0.048273, "bpb": 0.80325}, "ta": {"chunks": 2921.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.156245, "cps": 22.820312, "mean_p": 0.05699, "boundary_entropy": 0.042651, "bpb": 0.711667}, "th": {"chunks": 3071.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.190923, "cps": 23.992188, "mean_p": 0.066271, "boundary_entropy": 0.041519, "bpb": 0.796509}, "zh": {"chunks": 3719.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.629699, "cps": 29.054688, "mean_p": 0.229341, "boundary_entropy": 0.010311, "bpb": 2.238177}, "my": {"chunks": 2747.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.144976, "cps": 21.460938, "mean_p": 0.053749, "boundary_entropy": 0.04769, "bpb": 0.755705}}, "bpb": {"en": 1.50101, "de": 1.59494, "vi": 1.2768, "tr": 1.49573, "fi": 1.62007, "ru": 0.95193, "el": 0.96384, "hi": 0.80325, "ta": 0.71167, "th": 0.79651, "zh": 2.23818, "my": 0.75571}, "gini_cpc": 0.206489, "gini_cps": 0.065419, "bpic": 9.476308, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.014480948977222, "vi": 1.207503355101568, "tr": 1.0032267674781354, "fi": 1.0858049994317887, "ru": 1.221430384613359, "el": 0.9156780723015305, "hi": 1.2673068721398215, "ta": 0.9045680236211183, "th": 1.1053349316288157, "zh": 3.64559046485424, "my": 0.8393255182689298}, "t": 13638.0} |
| {"event": "eval", "step": 48000, "phase": 2, "bpb": {"en": 1.49764, "de": 1.60322, "vi": 1.29586, "tr": 1.48702, "fi": 1.63831, "ru": 0.9505, "el": 0.96281, "hi": 0.80972, "ta": 0.71727, "th": 0.79658, "zh": 2.24233, "my": 0.76088}, "bytes_per_chunk": {"en": 5.795, "de": 5.834, "vi": 6.358, "tr": 6.343, "fi": 5.437, "ru": 8.746, "el": 10.943, "hi": 11.442, "ta": 18.103, "th": 15.161, "zh": 4.533, "my": 19.775}, "bpb_mean": 1.23018, "t": 13645.8} |
| {"event": "train", "step": 48000, "phase": 2, "objective": "parity_a", "lm_loss": 1.00522, "bpb": 1.45022, "chunk_loss": 0.030124, "F": 0.14999, "G": 0.14881, "chunks": 153.59, "bytes_per_chunk": 6.667, "boundary_entropy": 0.02124, "router_noise": 0.0, "grad_norm": 0.3115, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.309, "n_target_mean": 8.6966, "alpha": 0.03, "t": 13646.0} |
| {"event": "train", "step": 48500, "phase": 2, "objective": "parity_a", "lm_loss": 0.83621, "bpb": 1.2064, "chunk_loss": 0.030211, "F": 0.14018, "G": 0.13893, "chunks": 143.55, "bytes_per_chunk": 7.134, "boundary_entropy": 0.02456, "router_noise": 0.0, "grad_norm": 0.2979, "lr": 0.000113, "router_trainable": true, "sec_per_step": 0.284, "n_target_mean": 9.6682, "alpha": 0.03, "t": 13787.8} |
| {"event": "train", "step": 49000, "phase": 2, "objective": "parity_a", "lm_loss": 0.94117, "bpb": 1.35782, "chunk_loss": 0.030135, "F": 0.14163, "G": 0.13991, "chunks": 145.03, "bytes_per_chunk": 7.061, "boundary_entropy": 0.02297, "router_noise": 0.0, "grad_norm": 0.3299, "lr": 0.0001104, "router_trainable": true, "sec_per_step": 0.272, "n_target_mean": 9.3887, "alpha": 0.03, "t": 13923.7} |
| {"event": "train", "step": 49500, "phase": 2, "objective": "parity_a", "lm_loss": 0.85383, "bpb": 1.23182, "chunk_loss": 0.03018, "F": 0.13539, "G": 0.13388, "chunks": 138.64, "bytes_per_chunk": 7.386, "boundary_entropy": 0.02485, "router_noise": 0.0, "grad_norm": 0.3011, "lr": 0.0001078, "router_trainable": true, "sec_per_step": 0.218, "n_target_mean": 10.2123, "alpha": 0.03, "t": 14032.8} |
| {"event": "train", "step": 50000, "phase": 2, "objective": "parity_a", "lm_loss": 0.96044, "bpb": 1.38562, "chunk_loss": 0.030203, "F": 0.14883, "G": 0.14752, "chunks": 152.41, "bytes_per_chunk": 6.719, "boundary_entropy": 0.02179, "router_noise": 0.0, "grad_norm": 0.3093, "lr": 0.0001052, "router_trainable": true, "sec_per_step": 0.217, "n_target_mean": 9.0008, "alpha": 0.03, "t": 14141.1} |
| {"event": "train", "step": 50500, "phase": 2, "objective": "parity_a", "lm_loss": 0.89498, "bpb": 1.29119, "chunk_loss": 0.030212, "F": 0.13623, "G": 0.13533, "chunks": 139.5, "bytes_per_chunk": 7.341, "boundary_entropy": 0.02284, "router_noise": 0.0, "grad_norm": 0.3237, "lr": 0.0001027, "router_trainable": true, "sec_per_step": 0.217, "n_target_mean": 9.8433, "alpha": 0.03, "t": 14249.6} |
| {"event": "train", "step": 51000, "phase": 2, "objective": "parity_a", "lm_loss": 0.85323, "bpb": 1.23096, "chunk_loss": 0.030245, "F": 0.14542, "G": 0.14365, "chunks": 148.91, "bytes_per_chunk": 6.877, "boundary_entropy": 0.02334, "router_noise": 0.0, "grad_norm": 0.3303, "lr": 0.0001002, "router_trainable": true, "sec_per_step": 0.218, "n_target_mean": 9.6274, "alpha": 0.03, "t": 14358.7} |
| {"event": "train", "step": 51500, "phase": 2, "objective": "parity_a", "lm_loss": 0.97697, "bpb": 1.40947, "chunk_loss": 0.030267, "F": 0.15753, "G": 0.15534, "chunks": 161.31, "bytes_per_chunk": 6.348, "boundary_entropy": 0.0214, "router_noise": 0.0, "grad_norm": 0.3243, "lr": 9.77e-05, "router_trainable": true, "sec_per_step": 0.21, "n_target_mean": 8.3009, "alpha": 0.03, "t": 14463.7} |
| {"event": "probe", "step": 52000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3019.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.173476, "cps": 23.585938, "mean_p": 0.171257, "boundary_entropy": 0.017162, "bpb": 1.487304}, "de": {"chunks": 3588.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174743, "cps": 28.03125, "mean_p": 0.16887, "boundary_entropy": 0.019309, "bpb": 1.583999}, "vi": {"chunks": 3852.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.20895, "cps": 30.09375, "mean_p": 0.158696, "boundary_entropy": 0.019385, "bpb": 1.259957}, "tr": {"chunks": 3157.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.174391, "cps": 24.664062, "mean_p": 0.159141, "boundary_entropy": 0.019596, "bpb": 1.481465}, "fi": {"chunks": 3508.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186487, "cps": 27.40625, "mean_p": 0.177777, "boundary_entropy": 0.018708, "bpb": 1.603296}, "ru": {"chunks": 3980.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.210026, "cps": 31.09375, "mean_p": 0.108681, "boundary_entropy": 0.035189, "bpb": 0.93918}, "el": {"chunks": 3358.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.163621, "cps": 26.234375, "mean_p": 0.089519, "boundary_entropy": 0.022708, "bpb": 0.950781}, "hi": {"chunks": 3658.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.217415, "cps": 28.578125, "mean_p": 0.08365, "boundary_entropy": 0.044018, "bpb": 0.793994}, "ta": {"chunks": 2765.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.147901, "cps": 21.601562, "mean_p": 0.054862, "boundary_entropy": 0.038481, "bpb": 0.702002}, "th": {"chunks": 3098.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.192602, "cps": 24.203125, "mean_p": 0.066553, "boundary_entropy": 0.043372, "bpb": 0.789544}, "zh": {"chunks": 3708.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.627836, "cps": 28.96875, "mean_p": 0.228486, "boundary_entropy": 0.010913, "bpb": 2.218784}, "my": {"chunks": 2717.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.143392, "cps": 21.226562, "mean_p": 0.052491, "boundary_entropy": 0.0439, "bpb": 0.746215}}, "bpb": {"en": 1.4873, "de": 1.584, "vi": 1.25996, "tr": 1.48147, "fi": 1.6033, "ru": 0.93918, "el": 0.95078, "hi": 0.79399, "ta": 0.702, "th": 0.78954, "zh": 2.21878, "my": 0.74622}, "gini_cpc": 0.207411, "gini_cps": 0.067524, "bpic": 9.497649, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0073051036837477, "vi": 1.2044926207790045, "tr": 1.00527535957478, "fi": 1.0750006044154392, "ru": 1.2106953236363194, "el": 0.9431936187026579, "hi": 1.253284404319941, "ta": 0.8525712300300824, "th": 1.1102514661315555, "zh": 3.6191558889870867, "my": 0.8265845234771296}, "t": 14556.4} |
| {"event": "eval", "step": 52000, "phase": 2, "bpb": {"en": 1.48481, "de": 1.59066, "vi": 1.2793, "tr": 1.47276, "fi": 1.62269, "ru": 0.93843, "el": 0.95042, "hi": 0.80069, "ta": 0.70842, "th": 0.78908, "zh": 2.22177, "my": 0.75269}, "bytes_per_chunk": {"en": 5.798, "de": 5.845, "vi": 6.334, "tr": 6.314, "fi": 5.465, "ru": 8.777, "el": 10.634, "hi": 11.23, "ta": 18.352, "th": 14.78, "zh": 4.534, "my": 20.167}, "bpb_mean": 1.21764, "t": 14561.5} |
| {"event": "train", "step": 52000, "phase": 2, "objective": "parity_a", "lm_loss": 0.97208, "bpb": 1.40242, "chunk_loss": 0.030308, "F": 0.15173, "G": 0.15013, "chunks": 155.38, "bytes_per_chunk": 6.591, "boundary_entropy": 0.02202, "router_noise": 0.0, "grad_norm": 0.3291, "lr": 9.52e-05, "router_trainable": true, "sec_per_step": 0.196, "n_target_mean": 9.2047, "alpha": 0.03, "t": 14561.6} |
| {"event": "train", "step": 52500, "phase": 2, "objective": "parity_a", "lm_loss": 0.90804, "bpb": 1.31002, "chunk_loss": 0.03026, "F": 0.14566, "G": 0.14381, "chunks": 149.16, "bytes_per_chunk": 6.865, "boundary_entropy": 0.02444, "router_noise": 0.0, "grad_norm": 0.3259, "lr": 9.28e-05, "router_trainable": true, "sec_per_step": 0.18, "n_target_mean": 9.5906, "alpha": 0.03, "t": 14651.5} |
| {"event": "train", "step": 53000, "phase": 2, "objective": "parity_a", "lm_loss": 0.89561, "bpb": 1.29209, "chunk_loss": 0.030245, "F": 0.14752, "G": 0.14588, "chunks": 151.06, "bytes_per_chunk": 6.779, "boundary_entropy": 0.02139, "router_noise": 0.0, "grad_norm": 0.3383, "lr": 9.04e-05, "router_trainable": true, "sec_per_step": 0.179, "n_target_mean": 9.2575, "alpha": 0.03, "t": 14741.0} |
| {"event": "phase_transition", "step": 53410, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 295680, "t": 14811.5} |
| {"event": "train", "step": 53500, "phase": 3, "objective": "none", "lm_loss": 0.91636, "bpb": 1.32203, "chunk_loss": 0.0, "F": 0.15083, "G": 0.14937, "chunks": 154.45, "bytes_per_chunk": 6.63, "boundary_entropy": 0.02185, "router_noise": 0.0, "grad_norm": 0.3276, "lr": 8.8e-05, "router_trainable": false, "sec_per_step": 0.174, "alpha": 0.0, "t": 14828.2} |
| {"event": "train", "step": 54000, "phase": 3, "objective": "none", "lm_loss": 0.96872, "bpb": 1.39757, "chunk_loss": 0.0, "F": 0.15729, "G": 0.15648, "chunks": 161.06, "bytes_per_chunk": 6.358, "boundary_entropy": 0.02047, "router_noise": 0.0, "grad_norm": 0.3431, "lr": 8.57e-05, "router_trainable": false, "sec_per_step": 0.184, "alpha": 0.0, "t": 14920.2} |
| {"event": "train", "step": 54500, "phase": 3, "objective": "none", "lm_loss": 0.96207, "bpb": 1.38797, "chunk_loss": 0.0, "F": 0.161, "G": 0.1597, "chunks": 164.86, "bytes_per_chunk": 6.211, "boundary_entropy": 0.02164, "router_noise": 0.0, "grad_norm": 0.3374, "lr": 8.34e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 15010.4} |
| {"event": "train", "step": 55000, "phase": 3, "objective": "none", "lm_loss": 0.86922, "bpb": 1.25402, "chunk_loss": 0.0, "F": 0.15808, "G": 0.15658, "chunks": 161.88, "bytes_per_chunk": 6.326, "boundary_entropy": 0.0225, "router_noise": 0.0, "grad_norm": 0.3374, "lr": 8.11e-05, "router_trainable": false, "sec_per_step": 0.183, "alpha": 0.0, "t": 15101.8} |
| {"event": "train", "step": 55500, "phase": 3, "objective": "none", "lm_loss": 0.89299, "bpb": 1.28832, "chunk_loss": 0.0, "F": 0.15121, "G": 0.15017, "chunks": 154.84, "bytes_per_chunk": 6.613, "boundary_entropy": 0.02519, "router_noise": 0.0, "grad_norm": 0.3298, "lr": 7.89e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 15191.7} |
| {"event": "probe", "step": 56000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 2973.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.170833, "cps": 23.226562, "mean_p": 0.171043, "boundary_entropy": 0.016841, "bpb": 1.476766}, "de": {"chunks": 3595.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.175084, "cps": 28.085938, "mean_p": 0.17005, "boundary_entropy": 0.018255, "bpb": 1.571065}, "vi": {"chunks": 3847.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.208679, "cps": 30.054688, "mean_p": 0.15889, "boundary_entropy": 0.016973, "bpb": 1.249085}, "tr": {"chunks": 3156.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.174336, "cps": 24.65625, "mean_p": 0.159871, "boundary_entropy": 0.019803, "bpb": 1.468547}, "fi": {"chunks": 3512.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186699, "cps": 27.4375, "mean_p": 0.178252, "boundary_entropy": 0.017623, "bpb": 1.583259}, "ru": {"chunks": 4226.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.223008, "cps": 33.015625, "mean_p": 0.118166, "boundary_entropy": 0.032533, "bpb": 0.925534}, "el": {"chunks": 3625.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.176631, "cps": 28.320312, "mean_p": 0.09736, "boundary_entropy": 0.025168, "bpb": 0.939738}, "hi": {"chunks": 7105.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.422288, "cps": 55.507812, "mean_p": 0.161408, "boundary_entropy": 0.054211, "bpb": 0.778828}, "ta": {"chunks": 4500.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.240706, "cps": 35.15625, "mean_p": 0.088204, "boundary_entropy": 0.062625, "bpb": 0.69061}, "th": {"chunks": 4305.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.267641, "cps": 33.632812, "mean_p": 0.092241, "boundary_entropy": 0.053663, "bpb": 0.775515}, "zh": {"chunks": 3714.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.628852, "cps": 29.015625, "mean_p": 0.229275, "boundary_entropy": 0.00933, "bpb": 2.204456}, "my": {"chunks": 4092.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.215959, "cps": 31.96875, "mean_p": 0.079461, "boundary_entropy": 0.053386, "bpb": 0.73556}}, "bpb": {"en": 1.47677, "de": 1.57106, "vi": 1.24909, "tr": 1.46855, "fi": 1.58326, "ru": 0.92553, "el": 0.93974, "hi": 0.77883, "ta": 0.69061, "th": 0.77551, "zh": 2.20446, "my": 0.73556}, "gini_cpc": 0.225252, "gini_cps": 0.116115, "bpic": 7.888613, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0248863253428324, "vi": 1.221541582405468, "tr": 1.0205062153314537, "fi": 1.092878380572674, "ru": 1.3054176773015185, "el": 1.033942474916651, "hi": 2.471941680173568, "ta": 1.4090170723459516, "th": 1.5666836153921193, "zh": 3.681100442417244, "my": 1.2641582986587967}, "t": 15284.9} |
| {"event": "eval", "step": 56000, "phase": 3, "bpb": {"en": 1.47306, "de": 1.5783, "vi": 1.26733, "tr": 1.45965, "fi": 1.60229, "ru": 0.92482, "el": 0.93922, "hi": 0.78438, "ta": 0.69722, "th": 0.77566, "zh": 2.20827, "my": 0.74085}, "bytes_per_chunk": {"en": 5.775, "de": 5.798, "vi": 6.329, "tr": 6.278, "fi": 5.444, "ru": 8.242, "el": 9.789, "hi": 6.01, "ta": 11.202, "th": 10.673, "zh": 4.532, "my": 13.328}, "bpb_mean": 1.20426, "t": 15290.3} |
| {"event": "train", "step": 56000, "phase": 3, "objective": "none", "lm_loss": 0.88369, "bpb": 1.2749, "chunk_loss": 0.0, "F": 0.15739, "G": 0.15605, "chunks": 161.17, "bytes_per_chunk": 6.353, "boundary_entropy": 0.02317, "router_noise": 0.0, "grad_norm": 0.3213, "lr": 7.68e-05, "router_trainable": false, "sec_per_step": 0.198, "alpha": 0.0, "t": 15290.6} |
| {"event": "train", "step": 56500, "phase": 3, "objective": "none", "lm_loss": 0.91082, "bpb": 1.31403, "chunk_loss": 0.0, "F": 0.15857, "G": 0.15738, "chunks": 162.38, "bytes_per_chunk": 6.306, "boundary_entropy": 0.02206, "router_noise": 0.0, "grad_norm": 0.337, "lr": 7.46e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 15380.7} |
| {"event": "train", "step": 57000, "phase": 3, "objective": "none", "lm_loss": 0.86877, "bpb": 1.25337, "chunk_loss": 0.0, "F": 0.15327, "G": 0.15235, "chunks": 156.95, "bytes_per_chunk": 6.524, "boundary_entropy": 0.02435, "router_noise": 0.0, "grad_norm": 0.3272, "lr": 7.25e-05, "router_trainable": false, "sec_per_step": 0.178, "alpha": 0.0, "t": 15469.9} |
| {"event": "train", "step": 57500, "phase": 3, "objective": "none", "lm_loss": 0.85809, "bpb": 1.23797, "chunk_loss": 0.0, "F": 0.14964, "G": 0.14834, "chunks": 153.23, "bytes_per_chunk": 6.683, "boundary_entropy": 0.02672, "router_noise": 0.0, "grad_norm": 0.3394, "lr": 7.05e-05, "router_trainable": false, "sec_per_step": 0.179, "alpha": 0.0, "t": 15559.4} |
| {"event": "train", "step": 58000, "phase": 3, "objective": "none", "lm_loss": 0.85793, "bpb": 1.23774, "chunk_loss": 0.0, "F": 0.15269, "G": 0.15145, "chunks": 156.36, "bytes_per_chunk": 6.549, "boundary_entropy": 0.02627, "router_noise": 0.0, "grad_norm": 0.32, "lr": 6.84e-05, "router_trainable": false, "sec_per_step": 0.178, "alpha": 0.0, "t": 15648.3} |
| {"event": "train", "step": 58500, "phase": 3, "objective": "none", "lm_loss": 0.93884, "bpb": 1.35447, "chunk_loss": 0.0, "F": 0.15852, "G": 0.15748, "chunks": 162.33, "bytes_per_chunk": 6.308, "boundary_entropy": 0.02569, "router_noise": 0.0, "grad_norm": 0.3437, "lr": 6.65e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 15738.6} |
| {"event": "train", "step": 59000, "phase": 3, "objective": "none", "lm_loss": 0.94783, "bpb": 1.36743, "chunk_loss": 0.0, "F": 0.16145, "G": 0.16083, "chunks": 165.33, "bytes_per_chunk": 6.194, "boundary_entropy": 0.02062, "router_noise": 0.0, "grad_norm": 0.3368, "lr": 6.45e-05, "router_trainable": false, "sec_per_step": 0.176, "alpha": 0.0, "t": 15826.7} |
| {"event": "train", "step": 59500, "phase": 3, "objective": "none", "lm_loss": 0.92838, "bpb": 1.33937, "chunk_loss": 0.0, "F": 0.16031, "G": 0.15878, "chunks": 164.16, "bytes_per_chunk": 6.238, "boundary_entropy": 0.02244, "router_noise": 0.0, "grad_norm": 0.3441, "lr": 6.27e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 15916.9} |
| {"event": "probe", "step": 60000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3030.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174108, "cps": 23.671875, "mean_p": 0.17147, "boundary_entropy": 0.016319, "bpb": 1.466294}, "de": {"chunks": 3611.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.175863, "cps": 28.210938, "mean_p": 0.170343, "boundary_entropy": 0.018089, "bpb": 1.559943}, "vi": {"chunks": 3876.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.210252, "cps": 30.28125, "mean_p": 0.159812, "boundary_entropy": 0.017995, "bpb": 1.239483}, "tr": {"chunks": 3185.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175938, "cps": 24.882812, "mean_p": 0.159995, "boundary_entropy": 0.019495, "bpb": 1.457679}, "fi": {"chunks": 3514.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186806, "cps": 27.453125, "mean_p": 0.178544, "boundary_entropy": 0.017517, "bpb": 1.567316}, "ru": {"chunks": 4373.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.230765, "cps": 34.164062, "mean_p": 0.122888, "boundary_entropy": 0.031393, "bpb": 0.914273}, "el": {"chunks": 3681.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.17936, "cps": 28.757812, "mean_p": 0.098997, "boundary_entropy": 0.025859, "bpb": 0.928309}, "hi": {"chunks": 7249.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.430847, "cps": 56.632812, "mean_p": 0.164669, "boundary_entropy": 0.055748, "bpb": 0.763589}, "ta": {"chunks": 5979.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.319818, "cps": 46.710938, "mean_p": 0.116321, "boundary_entropy": 0.065208, "bpb": 0.680721}, "th": {"chunks": 5555.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.345353, "cps": 43.398438, "mean_p": 0.118816, "boundary_entropy": 0.063423, "bpb": 0.768676}, "zh": {"chunks": 3714.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.628852, "cps": 29.015625, "mean_p": 0.229112, "boundary_entropy": 0.009698, "bpb": 2.189199}, "my": {"chunks": 4236.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.223559, "cps": 33.09375, "mean_p": 0.081243, "boundary_entropy": 0.055532, "bpb": 0.729867}}, "bpb": {"en": 1.46629, "de": 1.55994, "vi": 1.23948, "tr": 1.45768, "fi": 1.56732, "ru": 0.91427, "el": 0.92831, "hi": 0.76359, "ta": 0.68072, "th": 0.76868, "zh": 2.1892, "my": 0.72987}, "gini_cpc": 0.236944, "gini_cps": 0.146593, "bpic": 7.379978, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0100818629079584, "vi": 1.207597257691595, "tr": 1.010509444580872, "fi": 1.0729299425078334, "ru": 1.3254146137568903, "el": 1.0301642253101204, "hi": 2.474597207714828, "ta": 1.83689603033585, "th": 1.9835561081753186, "zh": 3.61185201825296, "my": 1.28402673706256}, "t": 16009.7} |
| {"event": "eval", "step": 60000, "phase": 3, "bpb": {"en": 1.46281, "de": 1.56677, "vi": 1.25814, "tr": 1.44873, "fi": 1.58835, "ru": 0.91347, "el": 0.92795, "hi": 0.77059, "ta": 0.68739, "th": 0.76856, "zh": 2.19339, "my": 0.73627}, "bytes_per_chunk": {"en": 5.785, "de": 5.808, "vi": 6.327, "tr": 6.286, "fi": 5.442, "ru": 7.926, "el": 9.83, "hi": 5.884, "ta": 8.391, "th": 8.337, "zh": 4.532, "my": 12.648}, "bpb_mean": 1.19354, "t": 16014.8} |
| {"event": "train", "step": 60000, "phase": 3, "objective": "none", "lm_loss": 0.88259, "bpb": 1.27331, "chunk_loss": 0.0, "F": 0.157, "G": 0.15593, "chunks": 160.77, "bytes_per_chunk": 6.37, "boundary_entropy": 0.0241, "router_noise": 0.0, "grad_norm": 0.3464, "lr": 6.08e-05, "router_trainable": false, "sec_per_step": 0.196, "alpha": 0.0, "t": 16014.9} |
| {"event": "train", "step": 60500, "phase": 3, "objective": "none", "lm_loss": 0.89381, "bpb": 1.28949, "chunk_loss": 0.0, "F": 0.15578, "G": 0.15448, "chunks": 159.52, "bytes_per_chunk": 6.419, "boundary_entropy": 0.02664, "router_noise": 0.0, "grad_norm": 0.3444, "lr": 5.9e-05, "router_trainable": false, "sec_per_step": 0.179, "alpha": 0.0, "t": 16104.4} |
| {"event": "train", "step": 61000, "phase": 3, "objective": "none", "lm_loss": 0.84704, "bpb": 1.22202, "chunk_loss": 0.0, "F": 0.15414, "G": 0.15293, "chunks": 157.84, "bytes_per_chunk": 6.487, "boundary_entropy": 0.02662, "router_noise": 0.0, "grad_norm": 0.3398, "lr": 5.73e-05, "router_trainable": false, "sec_per_step": 0.179, "alpha": 0.0, "t": 16194.0} |
| {"event": "train", "step": 61500, "phase": 3, "objective": "none", "lm_loss": 0.86022, "bpb": 1.24103, "chunk_loss": 0.0, "F": 0.15776, "G": 0.15668, "chunks": 161.55, "bytes_per_chunk": 6.339, "boundary_entropy": 0.03013, "router_noise": 0.0, "grad_norm": 0.3488, "lr": 5.56e-05, "router_trainable": false, "sec_per_step": 0.179, "alpha": 0.0, "t": 16283.6} |
| {"event": "train", "step": 62000, "phase": 3, "objective": "none", "lm_loss": 0.88379, "bpb": 1.27503, "chunk_loss": 0.0, "F": 0.16261, "G": 0.16145, "chunks": 166.52, "bytes_per_chunk": 6.15, "boundary_entropy": 0.0233, "router_noise": 0.0, "grad_norm": 0.3418, "lr": 5.39e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 16373.8} |
| {"event": "train", "step": 62500, "phase": 3, "objective": "none", "lm_loss": 0.92888, "bpb": 1.34009, "chunk_loss": 0.0, "F": 0.16008, "G": 0.15907, "chunks": 163.92, "bytes_per_chunk": 6.247, "boundary_entropy": 0.02458, "router_noise": 0.0, "grad_norm": 0.3594, "lr": 5.23e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 16464.1} |
| {"event": "train", "step": 63000, "phase": 3, "objective": "none", "lm_loss": 0.9674, "bpb": 1.39567, "chunk_loss": 0.0, "F": 0.16164, "G": 0.16068, "chunks": 165.52, "bytes_per_chunk": 6.187, "boundary_entropy": 0.02047, "router_noise": 0.0, "grad_norm": 0.3702, "lr": 5.08e-05, "router_trainable": false, "sec_per_step": 0.182, "alpha": 0.0, "t": 16555.0} |
| {"event": "train", "step": 63500, "phase": 3, "objective": "none", "lm_loss": 0.90842, "bpb": 1.31057, "chunk_loss": 0.0, "F": 0.16319, "G": 0.16192, "chunks": 167.11, "bytes_per_chunk": 6.128, "boundary_entropy": 0.02619, "router_noise": 0.0, "grad_norm": 0.3491, "lr": 4.93e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 16645.0} |
| {"event": "probe", "step": 64000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 2975.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.170948, "cps": 23.242188, "mean_p": 0.170985, "boundary_entropy": 0.016445, "bpb": 1.459649}, "de": {"chunks": 3598.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.17523, "cps": 28.109375, "mean_p": 0.170442, "boundary_entropy": 0.017733, "bpb": 1.548754}, "vi": {"chunks": 3887.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.210849, "cps": 30.367188, "mean_p": 0.16064, "boundary_entropy": 0.017654, "bpb": 1.230097}, "tr": {"chunks": 3179.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175606, "cps": 24.835938, "mean_p": 0.160176, "boundary_entropy": 0.019512, "bpb": 1.446464}, "fi": {"chunks": 3500.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186061, "cps": 27.34375, "mean_p": 0.177579, "boundary_entropy": 0.017065, "bpb": 1.554985}, "ru": {"chunks": 4543.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.239736, "cps": 35.492188, "mean_p": 0.127818, "boundary_entropy": 0.032957, "bpb": 0.905932}, "el": {"chunks": 3796.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.184963, "cps": 29.65625, "mean_p": 0.102227, "boundary_entropy": 0.027889, "bpb": 0.919243}, "hi": {"chunks": 7546.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.448499, "cps": 58.953125, "mean_p": 0.171325, "boundary_entropy": 0.057716, "bpb": 0.752993}, "ta": {"chunks": 6557.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.350735, "cps": 51.226562, "mean_p": 0.126739, "boundary_entropy": 0.066658, "bpb": 0.675679}, "th": {"chunks": 6513.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.404911, "cps": 50.882812, "mean_p": 0.137067, "boundary_entropy": 0.068941, "bpb": 0.758655}, "zh": {"chunks": 3708.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.627836, "cps": 28.96875, "mean_p": 0.228942, "boundary_entropy": 0.009874, "bpb": 2.178639}, "my": {"chunks": 4609.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.243245, "cps": 36.007812, "mean_p": 0.088463, "boundary_entropy": 0.055491, "bpb": 0.722779}}, "bpb": {"en": 1.45965, "de": 1.54875, "vi": 1.2301, "tr": 1.44646, "fi": 1.55498, "ru": 0.90593, "el": 0.91924, "hi": 0.75299, "ta": 0.67568, "th": 0.75865, "zh": 2.17864, "my": 0.72278}, "gini_cpc": 0.246679, "gini_cps": 0.16759, "bpic": 7.053372, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0250520109665646, "vi": 1.2334130775174188, "tr": 1.0272523101932594, "fi": 1.0884119742203404, "ru": 1.4023960266956388, "el": 1.0819881608541677, "hi": 2.62360758674941, "ta": 2.051714200309704, "th": 2.368629659137513, "zh": 3.672682900454459, "my": 1.4229199951038047}, "t": 16737.6} |
| {"event": "eval", "step": 64000, "phase": 3, "bpb": {"en": 1.45506, "de": 1.55721, "vi": 1.24819, "tr": 1.4382, "fi": 1.57623, "ru": 0.90509, "el": 0.91986, "hi": 0.75974, "ta": 0.68159, "th": 0.75868, "zh": 2.18234, "my": 0.72888}, "bytes_per_chunk": {"en": 5.778, "de": 5.807, "vi": 6.279, "tr": 6.287, "fi": 5.462, "ru": 7.662, "el": 9.544, "hi": 5.668, "ta": 7.629, "th": 6.939, "zh": 4.535, "my": 11.583}, "bpb_mean": 1.18426, "t": 16743.1} |
| {"event": "train", "step": 64000, "phase": 3, "objective": "none", "lm_loss": 0.88386, "bpb": 1.27514, "chunk_loss": 0.0, "F": 0.16708, "G": 0.16587, "chunks": 171.09, "bytes_per_chunk": 5.985, "boundary_entropy": 0.02669, "router_noise": 0.0, "grad_norm": 0.3533, "lr": 4.78e-05, "router_trainable": false, "sec_per_step": 0.197, "alpha": 0.0, "t": 16743.3} |
| {"event": "train", "step": 64500, "phase": 3, "objective": "none", "lm_loss": 0.9194, "bpb": 1.32641, "chunk_loss": 0.0, "F": 0.16249, "G": 0.16105, "chunks": 166.39, "bytes_per_chunk": 6.154, "boundary_entropy": 0.0261, "router_noise": 0.0, "grad_norm": 0.3653, "lr": 4.65e-05, "router_trainable": false, "sec_per_step": 0.178, "alpha": 0.0, "t": 16832.1} |
| {"event": "train", "step": 65000, "phase": 3, "objective": "none", "lm_loss": 0.8582, "bpb": 1.23812, "chunk_loss": 0.0, "F": 0.15611, "G": 0.15475, "chunks": 159.86, "bytes_per_chunk": 6.406, "boundary_entropy": 0.02826, "router_noise": 0.0, "grad_norm": 0.3559, "lr": 4.51e-05, "router_trainable": false, "sec_per_step": 0.177, "alpha": 0.0, "t": 16920.6} |
| {"event": "train", "step": 65500, "phase": 3, "objective": "none", "lm_loss": 0.91535, "bpb": 1.32057, "chunk_loss": 0.0, "F": 0.16052, "G": 0.15948, "chunks": 164.38, "bytes_per_chunk": 6.23, "boundary_entropy": 0.02596, "router_noise": 0.0, "grad_norm": 0.3639, "lr": 4.38e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 17010.4} |
| {"event": "train", "step": 66000, "phase": 3, "objective": "none", "lm_loss": 0.86194, "bpb": 1.24352, "chunk_loss": 0.0, "F": 0.16321, "G": 0.16172, "chunks": 167.12, "bytes_per_chunk": 6.127, "boundary_entropy": 0.02665, "router_noise": 0.0, "grad_norm": 0.3596, "lr": 4.26e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 17100.7} |
| {"event": "train", "step": 66500, "phase": 3, "objective": "none", "lm_loss": 0.90768, "bpb": 1.3095, "chunk_loss": 0.0, "F": 0.16292, "G": 0.16186, "chunks": 166.83, "bytes_per_chunk": 6.138, "boundary_entropy": 0.02478, "router_noise": 0.0, "grad_norm": 0.3619, "lr": 4.14e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 17191.1} |
| {"event": "train", "step": 67000, "phase": 3, "objective": "none", "lm_loss": 0.88972, "bpb": 1.28359, "chunk_loss": 0.0, "F": 0.16982, "G": 0.16856, "chunks": 173.89, "bytes_per_chunk": 5.889, "boundary_entropy": 0.02509, "router_noise": 0.0, "grad_norm": 0.3621, "lr": 4.03e-05, "router_trainable": false, "sec_per_step": 0.184, "alpha": 0.0, "t": 17283.0} |
| {"event": "train", "step": 67500, "phase": 3, "objective": "none", "lm_loss": 0.83791, "bpb": 1.20885, "chunk_loss": 0.0, "F": 0.1545, "G": 0.15345, "chunks": 158.2, "bytes_per_chunk": 6.473, "boundary_entropy": 0.02613, "router_noise": 0.0, "grad_norm": 0.3551, "lr": 3.92e-05, "router_trainable": false, "sec_per_step": 0.183, "alpha": 0.0, "t": 17374.3} |
| {"event": "probe", "step": 68000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3029.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17405, "cps": 23.664062, "mean_p": 0.171803, "boundary_entropy": 0.01631, "bpb": 1.451441}, "de": {"chunks": 3613.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.175961, "cps": 28.226562, "mean_p": 0.171003, "boundary_entropy": 0.017728, "bpb": 1.54024}, "vi": {"chunks": 3912.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.212205, "cps": 30.5625, "mean_p": 0.161397, "boundary_entropy": 0.018308, "bpb": 1.222938}, "tr": {"chunks": 3199.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.176711, "cps": 24.992188, "mean_p": 0.161045, "boundary_entropy": 0.019424, "bpb": 1.438489}, "fi": {"chunks": 3553.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188879, "cps": 27.757812, "mean_p": 0.180757, "boundary_entropy": 0.016863, "bpb": 1.543139}, "ru": {"chunks": 4627.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.244169, "cps": 36.148438, "mean_p": 0.130716, "boundary_entropy": 0.031926, "bpb": 0.899194}, "el": {"chunks": 3795.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.184914, "cps": 29.648438, "mean_p": 0.102192, "boundary_entropy": 0.026223, "bpb": 0.912059}, "hi": {"chunks": 7757.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.46104, "cps": 60.601562, "mean_p": 0.176664, "boundary_entropy": 0.058687, "bpb": 0.746482}, "ta": {"chunks": 6547.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.350201, "cps": 51.148438, "mean_p": 0.125809, "boundary_entropy": 0.069035, "bpb": 0.668546}, "th": {"chunks": 6793.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.422319, "cps": 53.070312, "mean_p": 0.142556, "boundary_entropy": 0.068629, "bpb": 0.752159}, "zh": {"chunks": 3708.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.627836, "cps": 28.96875, "mean_p": 0.229091, "boundary_entropy": 0.009879, "bpb": 2.169677}, "my": {"chunks": 5445.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.287365, "cps": 42.539062, "mean_p": 0.103746, "boundary_entropy": 0.063927, "bpb": 0.71648}}, "bpb": {"en": 1.45144, "de": 1.54024, "vi": 1.22294, "tr": 1.43849, "fi": 1.54314, "ru": 0.89919, "el": 0.91206, "hi": 0.74648, "ta": 0.66855, "th": 0.75216, "zh": 2.16968, "my": 0.71648}, "gini_cpc": 0.247944, "gini_cps": 0.174691, "bpic": 6.855926, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0109749651805786, "vi": 1.219215712715784, "tr": 1.0152863223314401, "fi": 1.0851959665711914, "ru": 1.4028625834174657, "el": 1.062418885143269, "hi": 2.648887814033437, "ta": 2.012063664508662, "th": 2.4264167549149027, "zh": 3.6072075367619725, "my": 1.6510466901020764}, "t": 17469.8} |
| {"event": "eval", "step": 68000, "phase": 3, "bpb": {"en": 1.44678, "de": 1.5481, "vi": 1.24098, "tr": 1.42945, "fi": 1.56438, "ru": 0.89735, "el": 0.91285, "hi": 0.75357, "ta": 0.67434, "th": 0.75217, "zh": 2.17257, "my": 0.72292}, "bytes_per_chunk": {"en": 5.767, "de": 5.792, "vi": 6.284, "tr": 6.262, "fi": 5.397, "ru": 7.45, "el": 9.521, "hi": 5.498, "ta": 7.687, "th": 6.685, "zh": 4.529, "my": 9.871}, "bpb_mean": 1.17629, "t": 17475.6} |
| {"event": "train", "step": 68000, "phase": 3, "objective": "none", "lm_loss": 0.94636, "bpb": 1.36531, "chunk_loss": 0.0, "F": 0.16461, "G": 0.16364, "chunks": 168.56, "bytes_per_chunk": 6.075, "boundary_entropy": 0.02169, "router_noise": 0.0, "grad_norm": 0.3735, "lr": 3.82e-05, "router_trainable": false, "sec_per_step": 0.203, "alpha": 0.0, "t": 17475.9} |
| {"event": "train", "step": 68500, "phase": 3, "objective": "none", "lm_loss": 0.86398, "bpb": 1.24646, "chunk_loss": 0.0, "F": 0.16206, "G": 0.16082, "chunks": 165.95, "bytes_per_chunk": 6.17, "boundary_entropy": 0.02772, "router_noise": 0.0, "grad_norm": 0.3452, "lr": 3.73e-05, "router_trainable": false, "sec_per_step": 0.183, "alpha": 0.0, "t": 17567.3} |
| {"event": "train", "step": 69000, "phase": 3, "objective": "none", "lm_loss": 0.80153, "bpb": 1.15637, "chunk_loss": 0.0, "F": 0.15411, "G": 0.15287, "chunks": 157.81, "bytes_per_chunk": 6.489, "boundary_entropy": 0.029, "router_noise": 0.0, "grad_norm": 0.3571, "lr": 3.64e-05, "router_trainable": false, "sec_per_step": 0.179, "alpha": 0.0, "t": 17656.9} |
| {"event": "train", "step": 69500, "phase": 3, "objective": "none", "lm_loss": 0.94318, "bpb": 1.36072, "chunk_loss": 0.0, "F": 0.1689, "G": 0.16784, "chunks": 172.95, "bytes_per_chunk": 5.921, "boundary_entropy": 0.02147, "router_noise": 0.0, "grad_norm": 0.3755, "lr": 3.55e-05, "router_trainable": false, "sec_per_step": 0.179, "alpha": 0.0, "t": 17746.4} |
| {"event": "train", "step": 70000, "phase": 3, "objective": "none", "lm_loss": 0.93096, "bpb": 1.34309, "chunk_loss": 0.0, "F": 0.16652, "G": 0.16578, "chunks": 170.52, "bytes_per_chunk": 6.005, "boundary_entropy": 0.02248, "router_noise": 0.0, "grad_norm": 0.3673, "lr": 3.48e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 17836.2} |
| {"event": "train", "step": 70500, "phase": 3, "objective": "none", "lm_loss": 0.96392, "bpb": 1.39064, "chunk_loss": 0.0, "F": 0.16698, "G": 0.16611, "chunks": 170.98, "bytes_per_chunk": 5.989, "boundary_entropy": 0.0205, "router_noise": 0.0, "grad_norm": 0.3803, "lr": 3.4e-05, "router_trainable": false, "sec_per_step": 0.183, "alpha": 0.0, "t": 17927.5} |
| {"event": "train", "step": 71000, "phase": 3, "objective": "none", "lm_loss": 0.80249, "bpb": 1.15775, "chunk_loss": 0.0, "F": 0.15919, "G": 0.15782, "chunks": 163.02, "bytes_per_chunk": 6.282, "boundary_entropy": 0.02786, "router_noise": 0.0, "grad_norm": 0.3501, "lr": 3.34e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 18017.3} |
| {"event": "train", "step": 71500, "phase": 3, "objective": "none", "lm_loss": 0.84363, "bpb": 1.2171, "chunk_loss": 0.0, "F": 0.16663, "G": 0.16491, "chunks": 170.62, "bytes_per_chunk": 6.001, "boundary_entropy": 0.02883, "router_noise": 0.0, "grad_norm": 0.3604, "lr": 3.28e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 18107.1} |
| {"event": "probe", "step": 72000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3032.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174223, "cps": 23.6875, "mean_p": 0.172303, "boundary_entropy": 0.016557, "bpb": 1.446011}, "de": {"chunks": 3615.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176058, "cps": 28.242188, "mean_p": 0.171198, "boundary_entropy": 0.017297, "bpb": 1.535501}, "vi": {"chunks": 3909.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.212042, "cps": 30.539062, "mean_p": 0.161545, "boundary_entropy": 0.017909, "bpb": 1.216627}, "tr": {"chunks": 3207.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177153, "cps": 25.054688, "mean_p": 0.161789, "boundary_entropy": 0.019478, "bpb": 1.431555}, "fi": {"chunks": 3545.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188454, "cps": 27.695312, "mean_p": 0.180175, "boundary_entropy": 0.016638, "bpb": 1.533923}, "ru": {"chunks": 4694.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.247704, "cps": 36.671875, "mean_p": 0.132315, "boundary_entropy": 0.032659, "bpb": 0.894034}, "el": {"chunks": 3908.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.190421, "cps": 30.53125, "mean_p": 0.105467, "boundary_entropy": 0.028652, "bpb": 0.905778}, "hi": {"chunks": 7965.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.473403, "cps": 62.226562, "mean_p": 0.179735, "boundary_entropy": 0.060632, "bpb": 0.739129}, "ta": {"chunks": 6696.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.358171, "cps": 52.3125, "mean_p": 0.128366, "boundary_entropy": 0.067025, "bpb": 0.662939}, "th": {"chunks": 7266.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.451725, "cps": 56.765625, "mean_p": 0.152449, "boundary_entropy": 0.072163, "bpb": 0.746582}, "zh": {"chunks": 3723.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.630376, "cps": 29.085938, "mean_p": 0.229992, "boundary_entropy": 0.010167, "bpb": 2.163139}, "my": {"chunks": 4944.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.260925, "cps": 38.625, "mean_p": 0.094728, "boundary_entropy": 0.05792, "bpb": 0.710828}}, "bpb": {"en": 1.44601, "de": 1.5355, "vi": 1.21663, "tr": 1.43155, "fi": 1.53392, "ru": 0.89403, "el": 0.90578, "hi": 0.73913, "ta": 0.66294, "th": 0.74658, "zh": 2.16314, "my": 0.71083}, "gini_cpc": 0.252109, "gini_cps": 0.180533, "bpic": 6.792103, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.010533738276, "vi": 1.2170753084537915, "tr": 1.0168182484659387, "fi": 1.08168119436657, "ru": 1.4217681929254182, "el": 1.0929709855099183, "hi": 2.7172251800507317, "ta": 2.0558191098462264, "th": 2.5928013939797068, "zh": 3.618216225259943, "my": 1.4976488910352064}, "t": 18200.4} |
| {"event": "eval", "step": 72000, "phase": 3, "bpb": {"en": 1.44169, "de": 1.54355, "vi": 1.2345, "tr": 1.42268, "fi": 1.55481, "ru": 0.89258, "el": 0.90633, "hi": 0.74624, "ta": 0.66896, "th": 0.74655, "zh": 2.16639, "my": 0.71758}, "bytes_per_chunk": {"en": 5.767, "de": 5.798, "vi": 6.264, "tr": 6.219, "fi": 5.419, "ru": 7.412, "el": 9.259, "hi": 5.417, "ta": 7.534, "th": 6.252, "zh": 4.512, "my": 10.725}, "bpb_mean": 1.17016, "t": 18205.7} |
| {"event": "train", "step": 72000, "phase": 3, "objective": "none", "lm_loss": 0.85085, "bpb": 1.22752, "chunk_loss": 0.0, "F": 0.16368, "G": 0.16234, "chunks": 167.61, "bytes_per_chunk": 6.109, "boundary_entropy": 0.0263, "router_noise": 0.0, "grad_norm": 0.3697, "lr": 3.22e-05, "router_trainable": false, "sec_per_step": 0.197, "alpha": 0.0, "t": 18205.9} |
| {"event": "train", "step": 72500, "phase": 3, "objective": "none", "lm_loss": 0.87502, "bpb": 1.26238, "chunk_loss": 0.0, "F": 0.16446, "G": 0.16364, "chunks": 168.41, "bytes_per_chunk": 6.081, "boundary_entropy": 0.02376, "router_noise": 0.0, "grad_norm": 0.3673, "lr": 3.17e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 18295.7} |
| {"event": "train", "step": 73000, "phase": 3, "objective": "none", "lm_loss": 0.88116, "bpb": 1.27124, "chunk_loss": 0.0, "F": 0.1656, "G": 0.16419, "chunks": 169.58, "bytes_per_chunk": 6.039, "boundary_entropy": 0.02487, "router_noise": 0.0, "grad_norm": 0.3785, "lr": 3.13e-05, "router_trainable": false, "sec_per_step": 0.187, "alpha": 0.0, "t": 18389.1} |
| {"event": "train", "step": 73500, "phase": 3, "objective": "none", "lm_loss": 0.91429, "bpb": 1.31905, "chunk_loss": 0.0, "F": 0.16475, "G": 0.16334, "chunks": 168.7, "bytes_per_chunk": 6.07, "boundary_entropy": 0.02711, "router_noise": 0.0, "grad_norm": 0.382, "lr": 3.09e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 18479.6} |
| {"event": "train", "step": 74000, "phase": 3, "objective": "none", "lm_loss": 0.89359, "bpb": 1.28918, "chunk_loss": 0.0, "F": 0.1599, "G": 0.15892, "chunks": 163.73, "bytes_per_chunk": 6.254, "boundary_entropy": 0.02731, "router_noise": 0.0, "grad_norm": 0.3749, "lr": 3.06e-05, "router_trainable": false, "sec_per_step": 0.184, "alpha": 0.0, "t": 18571.8} |
| {"event": "train", "step": 74500, "phase": 3, "objective": "none", "lm_loss": 0.88967, "bpb": 1.28352, "chunk_loss": 0.0, "F": 0.1624, "G": 0.16099, "chunks": 166.3, "bytes_per_chunk": 6.158, "boundary_entropy": 0.02583, "router_noise": 0.0, "grad_norm": 0.3919, "lr": 3.04e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 18662.4} |
| {"event": "train", "step": 75000, "phase": 3, "objective": "none", "lm_loss": 0.90361, "bpb": 1.30364, "chunk_loss": 0.0, "F": 0.16864, "G": 0.16741, "chunks": 172.69, "bytes_per_chunk": 5.93, "boundary_entropy": 0.02235, "router_noise": 0.0, "grad_norm": 0.3969, "lr": 3.02e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 18752.7} |
| {"event": "train", "step": 75500, "phase": 3, "objective": "none", "lm_loss": 0.87673, "bpb": 1.26485, "chunk_loss": 0.0, "F": 0.16174, "G": 0.16068, "chunks": 165.62, "bytes_per_chunk": 6.183, "boundary_entropy": 0.02408, "router_noise": 0.0, "grad_norm": 0.3941, "lr": 3.01e-05, "router_trainable": false, "sec_per_step": 0.187, "alpha": 0.0, "t": 18846.1} |
| {"event": "probe", "step": 76000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3032.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174223, "cps": 23.6875, "mean_p": 0.171812, "boundary_entropy": 0.016386, "bpb": 1.443628}, "de": {"chunks": 3615.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176058, "cps": 28.242188, "mean_p": 0.170983, "boundary_entropy": 0.017831, "bpb": 1.532854}, "vi": {"chunks": 3906.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.21188, "cps": 30.515625, "mean_p": 0.161372, "boundary_entropy": 0.017725, "bpb": 1.212725}, "tr": {"chunks": 3193.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.17638, "cps": 24.945312, "mean_p": 0.16066, "boundary_entropy": 0.01958, "bpb": 1.427362}, "fi": {"chunks": 3556.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189038, "cps": 27.78125, "mean_p": 0.180343, "boundary_entropy": 0.016829, "bpb": 1.52813}, "ru": {"chunks": 4746.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.250449, "cps": 37.078125, "mean_p": 0.133817, "boundary_entropy": 0.032899, "bpb": 0.889068}, "el": {"chunks": 3911.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.190567, "cps": 30.554688, "mean_p": 0.105119, "boundary_entropy": 0.027321, "bpb": 0.903688}, "hi": {"chunks": 8003.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.475661, "cps": 62.523438, "mean_p": 0.181102, "boundary_entropy": 0.061414, "bpb": 0.735045}, "ta": {"chunks": 6836.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.365659, "cps": 53.40625, "mean_p": 0.130775, "boundary_entropy": 0.068151, "bpb": 0.660429}, "th": {"chunks": 7133.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.443457, "cps": 55.726562, "mean_p": 0.149956, "boundary_entropy": 0.072271, "bpb": 0.742461}, "zh": {"chunks": 3861.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.653742, "cps": 30.164062, "mean_p": 0.238326, "boundary_entropy": 0.011172, "bpb": 2.156862}, "my": {"chunks": 5192.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.274013, "cps": 40.5625, "mean_p": 0.098519, "boundary_entropy": 0.060448, "bpb": 0.708055}}, "bpb": {"en": 1.44363, "de": 1.53285, "vi": 1.21272, "tr": 1.42736, "fi": 1.52813, "ru": 0.88907, "el": 0.90369, "hi": 0.73504, "ta": 0.66043, "th": 0.74246, "zh": 2.15686, "my": 0.70806}, "gini_cpc": 0.256267, "gini_cps": 0.18082, "bpic": 6.73489, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.010533738276, "vi": 1.216141252192507, "tr": 1.0123793786566082, "fi": 1.085037609920317, "ru": 1.4375185009850948, "el": 1.0938100113432165, "hi": 2.730188715121909, "ta": 2.098802185619594, "th": 2.545341638213219, "zh": 3.7523322174935916, "my": 1.572773673595225}, "t": 18939.3} |
| {"event": "eval", "step": 76000, "phase": 3, "bpb": {"en": 1.4386, "de": 1.54061, "vi": 1.23108, "tr": 1.41889, "fi": 1.54953, "ru": 0.88805, "el": 0.90345, "hi": 0.74204, "ta": 0.6663, "th": 0.74223, "zh": 2.16141, "my": 0.71459}, "bytes_per_chunk": {"en": 5.763, "de": 5.79, "vi": 6.278, "tr": 6.271, "fi": 5.411, "ru": 7.333, "el": 9.224, "hi": 5.356, "ta": 7.336, "th": 6.352, "zh": 4.357, "my": 10.334}, "bpb_mean": 1.1664, "t": 18944.7} |
| {"event": "train", "step": 76000, "phase": 3, "objective": "none", "lm_loss": 0.85508, "bpb": 1.23362, "chunk_loss": 0.0, "F": 0.16083, "G": 0.15953, "chunks": 164.69, "bytes_per_chunk": 6.218, "boundary_entropy": 0.0247, "router_noise": 0.0, "grad_norm": 0.3888, "lr": 3e-05, "router_trainable": false, "sec_per_step": 0.198, "alpha": 0.0, "t": 18944.8} |
| {"event": "probe", "step": 76300, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3034.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174338, "cps": 23.703125, "mean_p": 0.171874, "boundary_entropy": 0.016675, "bpb": 1.44305}, "de": {"chunks": 3608.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.175717, "cps": 28.1875, "mean_p": 0.170557, "boundary_entropy": 0.017518, "bpb": 1.532304}, "vi": {"chunks": 3903.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211717, "cps": 30.492188, "mean_p": 0.161252, "boundary_entropy": 0.017443, "bpb": 1.212965}, "tr": {"chunks": 3186.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175993, "cps": 24.890625, "mean_p": 0.160058, "boundary_entropy": 0.019561, "bpb": 1.428253}, "fi": {"chunks": 3542.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188294, "cps": 27.671875, "mean_p": 0.179721, "boundary_entropy": 0.016935, "bpb": 1.529141}, "ru": {"chunks": 4746.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.250449, "cps": 37.078125, "mean_p": 0.133907, "boundary_entropy": 0.032546, "bpb": 0.888616}, "el": {"chunks": 3905.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.190274, "cps": 30.507812, "mean_p": 0.105091, "boundary_entropy": 0.02815, "bpb": 0.903701}, "hi": {"chunks": 7903.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.469718, "cps": 61.742188, "mean_p": 0.17818, "boundary_entropy": 0.060582, "bpb": 0.734805}, "ta": {"chunks": 6792.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.363306, "cps": 53.0625, "mean_p": 0.130717, "boundary_entropy": 0.070117, "bpb": 0.659136}, "th": {"chunks": 7238.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.449984, "cps": 56.546875, "mean_p": 0.152029, "boundary_entropy": 0.073223, "bpb": 0.742212}, "zh": {"chunks": 3867.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.654758, "cps": 30.210938, "mean_p": 0.238565, "boundary_entropy": 0.010726, "bpb": 2.156795}, "my": {"chunks": 5233.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.276177, "cps": 40.882812, "mean_p": 0.100763, "boundary_entropy": 0.062241, "bpb": 0.707793}}, "bpb": {"en": 1.44305, "de": 1.5323, "vi": 1.21296, "tr": 1.42825, "fi": 1.52914, "ru": 0.88862, "el": 0.9037, "hi": 0.7348, "ta": 0.65914, "th": 0.74221, "zh": 2.15679, "my": 0.70779}, "gini_cpc": 0.256467, "gini_cps": 0.180752, "bpic": 6.738083, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0079121151054136, "vi": 1.214406136474445, "tr": 1.009494050578738, "fi": 1.0800533720452137, "ru": 1.4365708948539246, "el": 1.0914120308963453, "hi": 2.6942969084403505, "ta": 2.083918602455579, "th": 2.5811072890824516, "zh": 3.755685982167113, "my": 1.584148543653247}, "t": 19001.5} |
| {"event": "eval", "step": 76300, "phase": 3, "bpb": {"en": 1.43839, "de": 1.53977, "vi": 1.23096, "tr": 1.41892, "fi": 1.54965, "ru": 0.88789, "el": 0.90362, "hi": 0.74194, "ta": 0.66521, "th": 0.74213, "zh": 2.16056, "my": 0.71385}, "bytes_per_chunk": {"en": 5.766, "de": 5.799, "vi": 6.272, "tr": 6.286, "fi": 5.427, "ru": 7.325, "el": 9.229, "hi": 5.451, "ta": 7.376, "th": 6.275, "zh": 4.356, "my": 10.338}, "bpb_mean": 1.16607, "t": 19006.8} |
| {"event": "probe", "step": 76300, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3034.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174338, "cps": 23.703125, "mean_p": 0.171874, "boundary_entropy": 0.016675, "bpb": 1.44305}, "de": {"chunks": 3608.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.175717, "cps": 28.1875, "mean_p": 0.170557, "boundary_entropy": 0.017518, "bpb": 1.532304}, "vi": {"chunks": 3903.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211717, "cps": 30.492188, "mean_p": 0.161252, "boundary_entropy": 0.017443, "bpb": 1.212965}, "tr": {"chunks": 3186.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175993, "cps": 24.890625, "mean_p": 0.160058, "boundary_entropy": 0.019561, "bpb": 1.428253}, "fi": {"chunks": 3542.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188294, "cps": 27.671875, "mean_p": 0.179721, "boundary_entropy": 0.016935, "bpb": 1.529141}, "ru": {"chunks": 4746.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.250449, "cps": 37.078125, "mean_p": 0.133907, "boundary_entropy": 0.032546, "bpb": 0.888616}, "el": {"chunks": 3905.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.190274, "cps": 30.507812, "mean_p": 0.105091, "boundary_entropy": 0.02815, "bpb": 0.903701}, "hi": {"chunks": 7903.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.469718, "cps": 61.742188, "mean_p": 0.17818, "boundary_entropy": 0.060582, "bpb": 0.734805}, "ta": {"chunks": 6792.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.363306, "cps": 53.0625, "mean_p": 0.130717, "boundary_entropy": 0.070117, "bpb": 0.659136}, "th": {"chunks": 7238.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.449984, "cps": 56.546875, "mean_p": 0.152029, "boundary_entropy": 0.073223, "bpb": 0.742212}, "zh": {"chunks": 3867.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.654758, "cps": 30.210938, "mean_p": 0.238565, "boundary_entropy": 0.010726, "bpb": 2.156795}, "my": {"chunks": 5233.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.276177, "cps": 40.882812, "mean_p": 0.100763, "boundary_entropy": 0.062241, "bpb": 0.707793}}, "bpb": {"en": 1.44305, "de": 1.5323, "vi": 1.21296, "tr": 1.42825, "fi": 1.52914, "ru": 0.88862, "el": 0.9037, "hi": 0.7348, "ta": 0.65914, "th": 0.74221, "zh": 2.15679, "my": 0.70779}, "gini_cpc": 0.256467, "gini_cps": 0.180752, "bpic": 6.738083, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0079121151054136, "vi": 1.214406136474445, "tr": 1.009494050578738, "fi": 1.0800533720452137, "ru": 1.4365708948539246, "el": 1.0914120308963453, "hi": 2.6942969084403505, "ta": 2.083918602455579, "th": 2.5811072890824516, "zh": 3.755685982167113, "my": 1.584148543653247}, "t": 19010.0} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_B_s1/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_B_s1/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_B_s1/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_B_s1/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_B_s1/th0.70"], "t": 19013.1} |
| {"event": "final", "run": "specfull_B_s1", "condition": "B", "step": 76300, "params": 82470528, "per_lang": {"en": {"cpc": 0.17433775785784061, "cps": 23.703125, "bpb": 1.436288310932275, "bytes_per_chunk": 5.777150916784203}, "de": {"cpc": 0.17571713826523158, "cps": 28.1875, "bpb": 1.536096034603394, "bytes_per_chunk": 5.770252199800794}, "vi": {"cpc": 0.21171684296175752, "cps": 30.4921875, "bpb": 1.229617931022247, "bytes_per_chunk": 6.250602064903789}, "tr": {"cpc": 0.17599292934872673, "cps": 24.890625, "bpb": 1.4171520485958309, "bytes_per_chunk": 6.315163122361344}, "fi": {"cpc": 0.18829408324916272, "cps": 27.671875, "bpb": 1.5550818121203331, "bytes_per_chunk": 5.438672199170124}, "ru": {"cpc": 0.25044854881266493, "cps": 37.078125, "bpb": 0.8910775694185785, "bytes_per_chunk": 7.320975500771492}, "el": {"cpc": 0.1902743263655411, "cps": 30.5078125, "bpb": 0.9020325364366112, "bytes_per_chunk": 9.231885333943758}, "hi": {"cpc": 0.4697176820208024, "cps": 61.7421875, "bpb": 0.7435096509146639, "bytes_per_chunk": 5.452926737945667}, "ta": {"cpc": 0.36330569671035035, "cps": 53.0625, "bpb": 0.6632081941299469, "bytes_per_chunk": 7.376909168935509}, "th": {"cpc": 0.44998445756916383, "cps": 56.546875, "bpb": 0.7442883722343973, "bytes_per_chunk": 6.265204791921848}, "zh": {"cpc": 0.6547578733491365, "cps": 30.2109375, "bpb": 2.1613313945784967, "bytes_per_chunk": 4.3705052913250615}, "my": {"cpc": 0.27617690521427063, "cps": 40.8828125, "bpb": 0.7134491124016794, "bytes_per_chunk": 10.283786434427837}}, "gini_cpc": 0.25646730539030815, "gini_cps": 0.18075185373761493, "bpic": 6.7380831153326195, "seed": 1, "mean_boundary_entropy": 0.029838798567652702, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 30520, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 53410, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 295680}], "t": 19023.2} |
|
|
