| {"event": "start", "run": "parity_B_ste_s1", "arch": "hnet", "objective": "parity", "condition": "B", "params": 22516992, "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}, "device": "cuda", "t": 1.5} |
| {"event": "probe_schedule", "n": 40, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 200, 256, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000, 2200, 2400, 2600, 2800, 3000, 3200, 3400, 3600, 3800, 4000, 4200, 4400, 4600, 4800, 5000, 5200, 5400, 5600, 5800, 6000], "t": 1.5} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2674.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.153652, "cps": 20.890625, "mean_p": 0.478126, "boundary_entropy": 0.685438, "bpb": 7.975542}, "de": {"chunks": 2056.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.100131, "cps": 16.0625, "mean_p": 0.469682, "boundary_entropy": 0.685273, "bpb": 7.980804}, "vi": {"chunks": 14185.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.76946, "cps": 110.820312, "mean_p": 0.507219, "boundary_entropy": 0.688283, "bpb": 7.810944}, "tr": {"chunks": 5645.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.311827, "cps": 44.101562, "mean_p": 0.486063, "boundary_entropy": 0.686375, "bpb": 7.940182}, "fi": {"chunks": 4732.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251555, "cps": 36.96875, "mean_p": 0.48451, "boundary_entropy": 0.686595, "bpb": 7.978067}, "ru": {"chunks": 17548.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.926016, "cps": 137.09375, "mean_p": 0.502342, "boundary_entropy": 0.690066, "bpb": 8.056575}, "el": {"chunks": 31473.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.533548, "cps": 245.882812, "mean_p": 0.517965, "boundary_entropy": 0.689704, "bpb": 7.856999}, "hi": {"chunks": 1659.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.098603, "cps": 12.960938, "mean_p": 0.476241, "boundary_entropy": 0.689492, "bpb": 8.113177}, "ta": {"chunks": 39693.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.123188, "cps": 310.101562, "mean_p": 0.514597, "boundary_entropy": 0.690542, "bpb": 8.099115}, "th": {"chunks": 44424.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.761828, "cps": 347.0625, "mean_p": 0.52078, "boundary_entropy": 0.690224, "bpb": 8.255005}, "zh": {"chunks": 12933.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.189807, "cps": 101.039062, "mean_p": 0.528309, "boundary_entropy": 0.684752, "bpb": 8.116607}, "my": {"chunks": 7048.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.371965, "cps": 55.0625, "mean_p": 0.491145, "boundary_entropy": 0.691021, "bpb": 8.11144}}, "bpb": {"en": 7.97554, "de": 7.9808, "vi": 7.81094, "tr": 7.94018, "fi": 7.97807, "ru": 8.05657, "el": 7.857, "hi": 8.11318, "ta": 8.09911, "th": 8.25501, "zh": 8.11661, "my": 8.11144}, "gini_cpc": 0.507116, "gini_cps": 0.500863, "bpic": 2.084973, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.6516785410034256, "vi": 5.0078223656304, "tr": 2.029439514461233, "fi": 1.6371767549398106, "ru": 6.026721581613623, "el": 9.980677296913099, "hi": 0.6417324941848254, "ta": 13.818190381639045, "th": 17.9746032262692, "zh": 14.25176170627287, "my": 2.4208352616119475}, "t": 3.9} |
| {"event": "train", "step": 0, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 5.54764, "bpb": 8.00355, "chunk_loss": 0.043969, "F": 0.34537, "G": 0.49096, "chunks": 353.66, "bytes_per_chunk": 2.895, "boundary_entropy": 0.6904, "router_noise": 0.0, "grad_norm": 6.9667, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.008, "alpha": 0.03, "t": 4.4} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2670.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.153422, "cps": 20.859375, "mean_p": 0.478155, "boundary_entropy": 0.685445, "bpb": 7.949553}, "de": {"chunks": 2047.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.099693, "cps": 15.992188, "mean_p": 0.469689, "boundary_entropy": 0.685277, "bpb": 7.958538}, "vi": {"chunks": 14152.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.76767, "cps": 110.5625, "mean_p": 0.507131, "boundary_entropy": 0.688286, "bpb": 7.793496}, "tr": {"chunks": 5644.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.311772, "cps": 44.09375, "mean_p": 0.486068, "boundary_entropy": 0.686378, "bpb": 7.922246}, "fi": {"chunks": 4721.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25097, "cps": 36.882812, "mean_p": 0.484469, "boundary_entropy": 0.686593, "bpb": 7.955682}, "ru": {"chunks": 17499.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.92343, "cps": 136.710938, "mean_p": 0.502295, "boundary_entropy": 0.690066, "bpb": 8.046968}, "el": {"chunks": 31467.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.533255, "cps": 245.835938, "mean_p": 0.517949, "boundary_entropy": 0.689705, "bpb": 7.846303}, "hi": {"chunks": 1639.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.097415, "cps": 12.804688, "mean_p": 0.476156, "boundary_entropy": 0.689484, "bpb": 8.093431}, "ta": {"chunks": 39589.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.117625, "cps": 309.289062, "mean_p": 0.514537, "boundary_entropy": 0.690545, "bpb": 8.084843}, "th": {"chunks": 44416.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.76133, "cps": 347.0, "mean_p": 0.520791, "boundary_entropy": 0.690223, "bpb": 8.239268}, "zh": {"chunks": 12935.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.190146, "cps": 101.054688, "mean_p": 0.528311, "boundary_entropy": 0.684752, "bpb": 8.112216}, "my": {"chunks": 7053.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.372229, "cps": 55.101562, "mean_p": 0.491149, "boundary_entropy": 0.691021, "bpb": 8.097779}}, "bpb": {"en": 7.94955, "de": 7.95854, "vi": 7.7935, "tr": 7.92225, "fi": 7.95568, "ru": 8.04697, "el": 7.8463, "hi": 8.09343, "ta": 8.08484, "th": 8.23927, "zh": 8.11222, "my": 8.09778}, "gini_cpc": 0.507446, "gini_cps": 0.501046, "bpic": 2.087672, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.6497978863293234, "vi": 5.0036570641458145, "tr": 2.032119823705426, "fi": 1.635817974063705, "ru": 6.018896504698942, "el": 9.993724059660112, "hi": 0.6349459238359183, "ta": 13.802632380826454, "th": 17.998289703925423, "zh": 14.275319899397678, "my": 2.426181945620539}, "t": 5.8} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2682.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.154111, "cps": 20.953125, "mean_p": 0.47818, "boundary_entropy": 0.685452, "bpb": 7.904539}, "de": {"chunks": 2041.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.099401, "cps": 15.945312, "mean_p": 0.469689, "boundary_entropy": 0.685281, "bpb": 7.92117}, "vi": {"chunks": 14085.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.764036, "cps": 110.039062, "mean_p": 0.506995, "boundary_entropy": 0.688292, "bpb": 7.765661}, "tr": {"chunks": 5631.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.311053, "cps": 43.992188, "mean_p": 0.486066, "boundary_entropy": 0.686383, "bpb": 7.892266}, "fi": {"chunks": 4696.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.249641, "cps": 36.6875, "mean_p": 0.484376, "boundary_entropy": 0.686588, "bpb": 7.91984}, "ru": {"chunks": 17212.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.908285, "cps": 134.46875, "mean_p": 0.502012, "boundary_entropy": 0.690066, "bpb": 8.010384}, "el": {"chunks": 31499.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.534815, "cps": 246.085938, "mean_p": 0.517994, "boundary_entropy": 0.689703, "bpb": 7.826382}, "hi": {"chunks": 1623.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.096464, "cps": 12.679688, "mean_p": 0.476103, "boundary_entropy": 0.689479, "bpb": 8.058881}, "ta": {"chunks": 39512.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.113506, "cps": 308.6875, "mean_p": 0.514487, "boundary_entropy": 0.690547, "bpb": 8.052299}, "th": {"chunks": 44427.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.762014, "cps": 347.085938, "mean_p": 0.520845, "boundary_entropy": 0.690219, "bpb": 8.18595}, "zh": {"chunks": 12929.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.18913, "cps": 101.007812, "mean_p": 0.5283, "boundary_entropy": 0.684753, "bpb": 8.103229}, "my": {"chunks": 7056.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.372388, "cps": 55.125, "mean_p": 0.491144, "boundary_entropy": 0.69102, "bpb": 8.08449}}, "bpb": {"en": 7.90454, "de": 7.92117, "vi": 7.76566, "tr": 7.89227, "fi": 7.91984, "ru": 8.01038, "el": 7.82638, "hi": 8.05888, "ta": 8.0523, "th": 8.18595, "zh": 8.10323, "my": 8.08449}, "gini_cpc": 0.50811, "gini_cps": 0.501481, "bpic": 2.09267, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.6449944003492604, "vi": 4.957686447758586, "tr": 2.0183678629408823, "fi": 1.6198751813191317, "ru": 5.893692455714732, "el": 9.959126940135151, "hi": 0.6259343618177649, "ta": 13.714149843268817, "th": 17.92219792920579, "zh": 14.204856131594836, "my": 2.41635390677134}, "t": 7.5} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2690.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.154571, "cps": 21.015625, "mean_p": 0.478427, "boundary_entropy": 0.685496, "bpb": 7.752629}, "de": {"chunks": 2063.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.100472, "cps": 16.117188, "mean_p": 0.469875, "boundary_entropy": 0.685324, "bpb": 7.790503}, "vi": {"chunks": 13906.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.754326, "cps": 108.640625, "mean_p": 0.506488, "boundary_entropy": 0.688312, "bpb": 7.673623}, "tr": {"chunks": 5608.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.309783, "cps": 43.8125, "mean_p": 0.48619, "boundary_entropy": 0.686408, "bpb": 7.789637}, "fi": {"chunks": 4647.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.247036, "cps": 36.304688, "mean_p": 0.484202, "boundary_entropy": 0.686582, "bpb": 7.797632}, "ru": {"chunks": 15924.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.840317, "cps": 124.40625, "mean_p": 0.500754, "boundary_entropy": 0.690065, "bpb": 7.864284}, "el": {"chunks": 31546.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.537105, "cps": 246.453125, "mean_p": 0.518035, "boundary_entropy": 0.689701, "bpb": 7.754592}, "hi": {"chunks": 1562.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.092838, "cps": 12.203125, "mean_p": 0.475794, "boundary_entropy": 0.689447, "bpb": 7.934496}, "ta": {"chunks": 39125.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.092806, "cps": 305.664062, "mean_p": 0.514223, "boundary_entropy": 0.690559, "bpb": 7.939972}, "th": {"chunks": 44461.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.764128, "cps": 347.351562, "mean_p": 0.520978, "boundary_entropy": 0.690208, "bpb": 8.037415}, "zh": {"chunks": 12928.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.18896, "cps": 101.0, "mean_p": 0.528344, "boundary_entropy": 0.684745, "bpb": 8.079776}, "my": {"chunks": 7068.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.373021, "cps": 55.21875, "mean_p": 0.491111, "boundary_entropy": 0.691017, "bpb": 8.002272}}, "bpb": {"en": 7.75263, "de": 7.7905, "vi": 7.67362, "tr": 7.78964, "fi": 7.79763, "ru": 7.86428, "el": 7.75459, "hi": 7.9345, "ta": 7.93997, "th": 8.03742, "zh": 8.07978, "my": 8.00227}, "gini_cpc": 0.51047, "gini_cps": 0.502484, "bpic": 2.11417, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.650007938983705, "vi": 4.880124742514391, "tr": 2.004145711436027, "fi": 1.5982055306957743, "ru": 5.436442447842591, "el": 9.94432464655214, "hi": 0.6006172439887979, "ta": 13.53944059948836, "th": 17.88257278653251, "zh": 14.161515792017946, "my": 2.4132649481696333}, "t": 9.9} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2574.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.147906, "cps": 20.109375, "mean_p": 0.478202, "boundary_entropy": 0.685534, "bpb": 7.305504}, "de": {"chunks": 1967.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.095797, "cps": 15.367188, "mean_p": 0.469429, "boundary_entropy": 0.685328, "bpb": 7.395849}, "vi": {"chunks": 12943.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.702088, "cps": 101.117188, "mean_p": 0.504312, "boundary_entropy": 0.688379, "bpb": 7.427704}, "tr": {"chunks": 5394.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.297962, "cps": 42.140625, "mean_p": 0.485774, "boundary_entropy": 0.686434, "bpb": 7.488489}, "fi": {"chunks": 4219.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.224284, "cps": 32.960938, "mean_p": 0.482628, "boundary_entropy": 0.686483, "bpb": 7.435758}, "ru": {"chunks": 11182.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.590079, "cps": 87.359375, "mean_p": 0.495904, "boundary_entropy": 0.690001, "bpb": 7.346737}, "el": {"chunks": 31645.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.541929, "cps": 247.226562, "mean_p": 0.518153, "boundary_entropy": 0.689696, "bpb": 7.59713}, "hi": {"chunks": 1363.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.08101, "cps": 10.648438, "mean_p": 0.474732, "boundary_entropy": 0.689336, "bpb": 7.571274}, "ta": {"chunks": 36847.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.970955, "cps": 287.867188, "mean_p": 0.512711, "boundary_entropy": 0.690623, "bpb": 7.569567}, "th": {"chunks": 44514.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.767423, "cps": 347.765625, "mean_p": 0.521088, "boundary_entropy": 0.690198, "bpb": 7.640126}, "zh": {"chunks": 12946.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.192008, "cps": 101.140625, "mean_p": 0.528411, "boundary_entropy": 0.684736, "bpb": 7.985642}, "my": {"chunks": 7212.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.380621, "cps": 56.34375, "mean_p": 0.491105, "boundary_entropy": 0.691013, "bpb": 7.753102}}, "bpb": {"en": 7.3055, "de": 7.39585, "vi": 7.4277, "tr": 7.48849, "fi": 7.43576, "ru": 7.34674, "el": 7.59713, "hi": 7.57127, "ta": 7.56957, "th": 7.64013, "zh": 7.98564, "my": 7.7531}, "gini_cpc": 0.524529, "gini_cps": 0.512294, "bpic": 2.220878, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.6476905049203301, "vi": 4.746870532956782, "tr": 2.014540340066496, "fi": 1.5163980580226395, "ru": 3.9895678112564656, "el": 10.42509046930915, "hi": 0.5477171763353041, "ta": 13.325767831170346, "th": 18.7107473192024, "zh": 14.820325345723246, "my": 2.5734036915164213}, "t": 11.8} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 1446.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.083089, "cps": 11.296875, "mean_p": 0.470282, "boundary_entropy": 0.684696, "bpb": 6.68746}, "de": {"chunks": 1188.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.057858, "cps": 9.28125, "mean_p": 0.462621, "boundary_entropy": 0.684468, "bpb": 6.814656}, "vi": {"chunks": 8388.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.455004, "cps": 65.53125, "mean_p": 0.493561, "boundary_entropy": 0.688339, "bpb": 7.082002}, "tr": {"chunks": 3855.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212948, "cps": 30.117188, "mean_p": 0.479441, "boundary_entropy": 0.686012, "bpb": 7.045844}, "fi": {"chunks": 2239.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.119026, "cps": 17.492188, "mean_p": 0.472436, "boundary_entropy": 0.685496, "bpb": 6.902305}, "ru": {"chunks": 3220.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.169921, "cps": 25.15625, "mean_p": 0.480149, "boundary_entropy": 0.689126, "bpb": 6.554243}, "el": {"chunks": 32743.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.59543, "cps": 255.804688, "mean_p": 0.519825, "boundary_entropy": 0.689606, "bpb": 6.952021}, "hi": {"chunks": 1539.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.091471, "cps": 12.023438, "mean_p": 0.476253, "boundary_entropy": 0.689506, "bpb": 6.76741}, "ta": {"chunks": 34046.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.821129, "cps": 265.984375, "mean_p": 0.511619, "boundary_entropy": 0.690646, "bpb": 6.706853}, "th": {"chunks": 45594.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.834566, "cps": 356.203125, "mean_p": 0.529295, "boundary_entropy": 0.689396, "bpb": 6.582024}, "zh": {"chunks": 13107.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.219269, "cps": 102.398438, "mean_p": 0.529121, "boundary_entropy": 0.684694, "bpb": 7.72386}, "my": {"chunks": 5420.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.286046, "cps": 42.34375, "mean_p": 0.488594, "boundary_entropy": 0.690887, "bpb": 6.989671}}, "bpb": {"en": 6.68746, "de": 6.81466, "vi": 7.082, "tr": 7.04584, "fi": 6.9023, "ru": 6.55424, "el": 6.95202, "hi": 6.76741, "ta": 6.70685, "th": 6.58202, "zh": 7.72386, "my": 6.98967}, "gini_cpc": 0.588085, "gini_cps": 0.584538, "bpic": 2.511902, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.6963376230914995, "vi": 5.476096681510792, "tr": 2.5628881804281307, "fi": 1.4325112370245097, "ru": 2.0450431907509388, "el": 19.201424552920383, "hi": 1.1008784596129302, "ta": 21.917774406585735, "th": 34.11477072657317, "zh": 26.70949544242932, "my": 3.442641008324158}, "t": 14.4} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4222.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.242602, "cps": 32.984375, "mean_p": 0.490933, "boundary_entropy": 0.687024, "bpb": 6.233184}, "de": {"chunks": 3992.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.194419, "cps": 31.1875, "mean_p": 0.486778, "boundary_entropy": 0.687632, "bpb": 6.334594}, "vi": {"chunks": 7850.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.42582, "cps": 61.328125, "mean_p": 0.493644, "boundary_entropy": 0.688535, "bpb": 6.757948}, "tr": {"chunks": 6991.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.386179, "cps": 54.617188, "mean_p": 0.494589, "boundary_entropy": 0.687436, "bpb": 6.652625}, "fi": {"chunks": 4170.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.221679, "cps": 32.578125, "mean_p": 0.486685, "boundary_entropy": 0.687214, "bpb": 6.419069}, "ru": {"chunks": 874.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.046121, "cps": 6.828125, "mean_p": 0.47542, "boundary_entropy": 0.688856, "bpb": 5.834916}, "el": {"chunks": 27729.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.351118, "cps": 216.632812, "mean_p": 0.511652, "boundary_entropy": 0.690113, "bpb": 6.130558}, "hi": {"chunks": 497.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.029539, "cps": 3.882812, "mean_p": 0.47101, "boundary_entropy": 0.689014, "bpb": 6.04926}, "ta": {"chunks": 18498.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.989462, "cps": 144.515625, "mean_p": 0.495337, "boundary_entropy": 0.69098, "bpb": 6.132973}, "th": {"chunks": 43689.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.716133, "cps": 341.320312, "mean_p": 0.515881, "boundary_entropy": 0.690576, "bpb": 5.571372}, "zh": {"chunks": 9875.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.672028, "cps": 77.148438, "mean_p": 0.511273, "boundary_entropy": 0.686209, "bpb": 7.355606}, "my": {"chunks": 1210.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.063859, "cps": 9.453125, "mean_p": 0.476106, "boundary_entropy": 0.689903, "bpb": 6.048668}}, "bpb": {"en": 6.23318, "de": 6.33459, "vi": 6.75795, "tr": 6.65263, "fi": 6.41907, "ru": 5.83492, "el": 6.13056, "hi": 6.04926, "ta": 6.13297, "th": 5.57137, "zh": 7.35561, "my": 6.04867}, "gini_cpc": 0.578886, "gini_cps": 0.56653, "bpic": 2.961342, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8013901804914194, "vi": 1.7552234238185889, "tr": 1.5918225105548325, "fi": 0.9137556235490281, "ru": 0.19011137789365667, "el": 5.56928257612463, "hi": 0.12176071986896626, "ta": 4.078544421919646, "th": 11.19584636450689, "zh": 6.89206798664631, "my": 0.2632254552512863}, "t": 17.9} |
| {"event": "train", "step": 50, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 4.18801, "bpb": 6.04202, "chunk_loss": 0.029644, "F": 0.15939, "G": 0.48521, "chunks": 163.22, "bytes_per_chunk": 6.274, "boundary_entropy": 0.69148, "router_noise": 0.0, "grad_norm": 1.6366, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.315, "alpha": 0.03, "t": 20.2} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5180.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.29765, "cps": 40.46875, "mean_p": 0.491163, "boundary_entropy": 0.686727, "bpb": 5.669381}, "de": {"chunks": 5313.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.258754, "cps": 41.507812, "mean_p": 0.489465, "boundary_entropy": 0.687651, "bpb": 5.795432}, "vi": {"chunks": 4326.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.234662, "cps": 33.796875, "mean_p": 0.480328, "boundary_entropy": 0.687209, "bpb": 6.166696}, "tr": {"chunks": 6182.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.34149, "cps": 48.296875, "mean_p": 0.490481, "boundary_entropy": 0.687227, "bpb": 6.14367}, "fi": {"chunks": 4924.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.261762, "cps": 38.46875, "mean_p": 0.488231, "boundary_entropy": 0.687199, "bpb": 5.87142}, "ru": {"chunks": 10969.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.578839, "cps": 85.695312, "mean_p": 0.495605, "boundary_entropy": 0.690218, "bpb": 5.330443}, "el": {"chunks": 16402.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.799201, "cps": 128.140625, "mean_p": 0.500021, "boundary_entropy": 0.690534, "bpb": 5.60343}, "hi": {"chunks": 4085.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.242793, "cps": 31.914062, "mean_p": 0.489157, "boundary_entropy": 0.69057, "bpb": 5.170544}, "ta": {"chunks": 30092.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.609628, "cps": 235.09375, "mean_p": 0.505456, "boundary_entropy": 0.690986, "bpb": 5.181075}, "th": {"chunks": 9486.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.589742, "cps": 74.109375, "mean_p": 0.492806, "boundary_entropy": 0.69087, "bpb": 4.961438}, "zh": {"chunks": 5435.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.920251, "cps": 42.460938, "mean_p": 0.496066, "boundary_entropy": 0.686721, "bpb": 6.965186}, "my": {"chunks": 1958.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.103335, "cps": 15.296875, "mean_p": 0.484012, "boundary_entropy": 0.690659, "bpb": 5.075704}}, "bpb": {"en": 5.66938, "de": 5.79543, "vi": 6.1667, "tr": 6.14367, "fi": 5.87142, "ru": 5.33044, "el": 5.60343, "hi": 5.17054, "ta": 5.18107, "th": 4.96144, "zh": 6.96519, "my": 5.0757}, "gini_cpc": 0.391203, "gini_cps": 0.393513, "bpic": 3.677754, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8693241992784194, "vi": 0.7883838761462847, "tr": 1.1472889473939025, "fi": 0.8794284704443656, "ru": 1.9446980674606007, "el": 2.685037297827826, "hi": 0.8157016643049333, "ta": 5.407791565684051, "th": 1.9813281757266838, "zh": 3.0917222129971877, "my": 0.34717118502742317}, "t": 22.6} |
| {"event": "train", "step": 100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 3.24706, "bpb": 4.68452, "chunk_loss": 0.036504, "F": 0.2525, "G": 0.46418, "chunks": 258.56, "bytes_per_chunk": 3.96, "boundary_entropy": 0.679, "router_noise": 0.0, "grad_norm": 1.9247, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.113, "alpha": 0.03, "t": 25.8} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4543.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.261047, "cps": 35.492188, "mean_p": 0.441906, "boundary_entropy": 0.659426, "bpb": 4.398604}, "de": {"chunks": 6426.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.31296, "cps": 50.203125, "mean_p": 0.449813, "boundary_entropy": 0.664673, "bpb": 4.590288}, "vi": {"chunks": 8046.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.436452, "cps": 62.859375, "mean_p": 0.447827, "boundary_entropy": 0.659097, "bpb": 4.676044}, "tr": {"chunks": 3991.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.220461, "cps": 31.179688, "mean_p": 0.446766, "boundary_entropy": 0.667945, "bpb": 4.86121}, "fi": {"chunks": 3312.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.176067, "cps": 25.875, "mean_p": 0.429029, "boundary_entropy": 0.66248, "bpb": 4.679272}, "ru": {"chunks": 7752.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.409077, "cps": 60.5625, "mean_p": 0.490277, "boundary_entropy": 0.689467, "bpb": 3.596948}, "el": {"chunks": 12848.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.626029, "cps": 100.375, "mean_p": 0.495836, "boundary_entropy": 0.689964, "bpb": 4.030393}, "hi": {"chunks": 20995.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.247845, "cps": 164.023438, "mean_p": 0.496596, "boundary_entropy": 0.690502, "bpb": 3.324}, "ta": {"chunks": 25067.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.34084, "cps": 195.835938, "mean_p": 0.510342, "boundary_entropy": 0.689443, "bpb": 3.361885}, "th": {"chunks": 17267.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.073485, "cps": 134.898438, "mean_p": 0.515226, "boundary_entropy": 0.685923, "bpb": 3.114318}, "zh": {"chunks": 2918.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.494074, "cps": 22.796875, "mean_p": 0.48189, "boundary_entropy": 0.68545, "bpb": 6.121489}, "my": {"chunks": 32295.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.704402, "cps": 252.304688, "mean_p": 0.491496, "boundary_entropy": 0.687899, "bpb": 3.554134}}, "bpb": {"en": 4.3986, "de": 4.59029, "vi": 4.67604, "tr": 4.86121, "fi": 4.67927, "ru": 3.59695, "el": 4.03039, "hi": 3.324, "ta": 3.36188, "th": 3.11432, "zh": 6.12149, "my": 3.55413}, "gini_cpc": 0.38455, "gini_cps": 0.413143, "bpic": 2.638395, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1988633877878723, "vi": 1.671930689602491, "tr": 0.8445250961951537, "fi": 0.6744656372056135, "ru": 1.5670611111530568, "el": 2.3981484660392245, "hi": 4.780157314667602, "ta": 5.136393348583683, "th": 4.112228201612457, "zh": 1.8926627219199552, "my": 6.529099637173294}, "t": 29.3} |
| {"event": "train", "step": 150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.83168, "bpb": 4.08525, "chunk_loss": 0.037362, "F": 0.27649, "G": 0.45225, "chunks": 283.12, "bytes_per_chunk": 3.617, "boundary_entropy": 0.67074, "router_noise": 0.0, "grad_norm": 1.6216, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 31.4} |
| {"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 1867.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.10728, "cps": 14.585938, "mean_p": 0.411964, "boundary_entropy": 0.650559, "bpb": 3.841715}, "de": {"chunks": 1882.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.091657, "cps": 14.703125, "mean_p": 0.411959, "boundary_entropy": 0.655982, "bpb": 4.055772}, "vi": {"chunks": 4203.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.22799, "cps": 32.835938, "mean_p": 0.412703, "boundary_entropy": 0.641192, "bpb": 3.627295}, "tr": {"chunks": 3528.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194885, "cps": 27.5625, "mean_p": 0.435845, "boundary_entropy": 0.665576, "bpb": 4.09747}, "fi": {"chunks": 1884.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.100154, "cps": 14.71875, "mean_p": 0.399264, "boundary_entropy": 0.651806, "bpb": 4.103911}, "ru": {"chunks": 9455.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.498945, "cps": 73.867188, "mean_p": 0.494025, "boundary_entropy": 0.689918, "bpb": 2.812815}, "el": {"chunks": 12616.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.614725, "cps": 98.5625, "mean_p": 0.494844, "boundary_entropy": 0.690139, "bpb": 2.966961}, "hi": {"chunks": 12908.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.767192, "cps": 100.84375, "mean_p": 0.491084, "boundary_entropy": 0.68966, "bpb": 2.408151}, "ta": {"chunks": 17362.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.928698, "cps": 135.640625, "mean_p": 0.490947, "boundary_entropy": 0.687012, "bpb": 2.382921}, "th": {"chunks": 16499.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.025738, "cps": 128.898438, "mean_p": 0.482586, "boundary_entropy": 0.688464, "bpb": 2.316978}, "zh": {"chunks": 480.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.081273, "cps": 3.75, "mean_p": 0.370508, "boundary_entropy": 0.598514, "bpb": 4.951356}, "my": {"chunks": 14296.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.754486, "cps": 111.6875, "mean_p": 0.417777, "boundary_entropy": 0.649538, "bpb": 2.420051}}, "bpb": {"en": 3.84171, "de": 4.05577, "vi": 3.62729, "tr": 4.09747, "fi": 4.10391, "ru": 2.81281, "el": 2.96696, "hi": 2.40815, "ta": 2.38292, "th": 2.31698, "zh": 4.95136, "my": 2.42005}, "gini_cpc": 0.422741, "gini_cps": 0.422702, "bpic": 3.957321, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.854372014209115, "vi": 2.125181615685564, "tr": 1.8165937986054388, "fi": 0.933574148642985, "ru": 4.650847733089502, "el": 5.730079367235595, "hi": 7.151278409475277, "ta": 8.65673423330826, "th": 9.561287109738887, "zh": 0.7575784233295383, "my": 7.032843700782193}, "t": 36.2} |
| {"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.84826, "de": 4.05912, "vi": 3.63351, "tr": 4.10768, "fi": 4.10823, "ru": 2.79531, "el": 2.95962, "hi": 2.40137, "ta": 2.40084, "th": 2.35185, "zh": 4.96119, "my": 2.43715}, "bytes_per_chunk": {"en": 8.31, "de": 10.405, "vi": 6.196, "tr": 5.074, "fi": 10.949, "ru": 3.655, "el": 3.343, "hi": 3.712, "ta": 2.992, "th": 3.029, "zh": 49.508, "my": 3.949}, "bpb_mean": 3.33868, "frozen_inventory": false, "t": 37.6} |
| {"event": "train", "step": 200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.50216, "bpb": 3.60985, "chunk_loss": 0.030748, "F": 0.16623, "G": 0.43324, "chunks": 170.22, "bytes_per_chunk": 6.016, "boundary_entropy": 0.66241, "router_noise": 0.0, "grad_norm": 1.8931, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.13, "alpha": 0.03, "t": 37.9} |
| {"event": "train", "step": 250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.16632, "bpb": 3.12534, "chunk_loss": 0.033787, "F": 0.21848, "G": 0.43951, "chunks": 223.72, "bytes_per_chunk": 4.577, "boundary_entropy": 0.66327, "router_noise": 0.0, "grad_norm": 1.2926, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 41.6} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2178.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.125151, "cps": 17.015625, "mean_p": 0.419178, "boundary_entropy": 0.655559, "bpb": 3.617365}, "de": {"chunks": 2056.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.100131, "cps": 16.0625, "mean_p": 0.42136, "boundary_entropy": 0.662899, "bpb": 3.82866}, "vi": {"chunks": 7142.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.387415, "cps": 55.796875, "mean_p": 0.431717, "boundary_entropy": 0.653488, "bpb": 3.19241}, "tr": {"chunks": 2338.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.12915, "cps": 18.265625, "mean_p": 0.435506, "boundary_entropy": 0.668557, "bpb": 3.861206}, "fi": {"chunks": 2360.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.125459, "cps": 18.4375, "mean_p": 0.41834, "boundary_entropy": 0.661701, "bpb": 3.874961}, "ru": {"chunks": 10148.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.535515, "cps": 79.28125, "mean_p": 0.493459, "boundary_entropy": 0.689881, "bpb": 2.708302}, "el": {"chunks": 11773.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.573649, "cps": 91.976562, "mean_p": 0.494734, "boundary_entropy": 0.689883, "bpb": 2.739234}, "hi": {"chunks": 11083.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.658722, "cps": 86.585938, "mean_p": 0.489558, "boundary_entropy": 0.686693, "bpb": 2.127075}, "ta": {"chunks": 17060.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.912543, "cps": 133.28125, "mean_p": 0.494161, "boundary_entropy": 0.680761, "bpb": 1.931513}, "th": {"chunks": 18714.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.163444, "cps": 146.203125, "mean_p": 0.478018, "boundary_entropy": 0.686908, "bpb": 2.093925}, "zh": {"chunks": 8052.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.363359, "cps": 62.90625, "mean_p": 0.40142, "boundary_entropy": 0.601174, "bpb": 4.46093}, "my": {"chunks": 29697.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.567289, "cps": 232.007812, "mean_p": 0.434093, "boundary_entropy": 0.655347, "bpb": 2.229369}}, "bpb": {"en": 3.61737, "de": 3.82866, "vi": 3.19241, "tr": 3.86121, "fi": 3.87496, "ru": 2.7083, "el": 2.73923, "hi": 2.12707, "ta": 1.93151, "th": 2.09392, "zh": 4.46093, "my": 2.22937}, "gini_cpc": 0.428754, "gini_cps": 0.419323, "bpic": 3.130325, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8000865099371717, "vi": 3.0955865332816295, "tr": 1.0319536706775385, "fi": 1.0024584115179076, "ru": 4.278952731924669, "el": 4.583661561462521, "hi": 5.2634258019885465, "ta": 7.291549058137111, "th": 9.29633583981791, "zh": 10.893729027491304, "my": 12.523203783473518}, "t": 43.2} |
| {"event": "train", "step": 300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.18254, "bpb": 3.14874, "chunk_loss": 0.034483, "F": 0.24069, "G": 0.45243, "chunks": 246.47, "bytes_per_chunk": 4.155, "boundary_entropy": 0.67133, "router_noise": 0.0, "grad_norm": 1.1135, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.099, "alpha": 0.03, "t": 46.6} |
| {"event": "train", "step": 350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.38939, "bpb": 3.44717, "chunk_loss": 0.039626, "F": 0.32141, "G": 0.44159, "chunks": 329.12, "bytes_per_chunk": 3.111, "boundary_entropy": 0.65586, "router_noise": 0.0, "grad_norm": 1.3769, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.069, "alpha": 0.03, "t": 50.0} |
| {"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2844.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.16342, "cps": 22.21875, "mean_p": 0.434075, "boundary_entropy": 0.663898, "bpb": 3.191527}, "de": {"chunks": 2806.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.136658, "cps": 21.921875, "mean_p": 0.437508, "boundary_entropy": 0.67181, "bpb": 3.472812}, "vi": {"chunks": 5600.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.30377, "cps": 43.75, "mean_p": 0.43481, "boundary_entropy": 0.664542, "bpb": 2.775847}, "tr": {"chunks": 3212.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177429, "cps": 25.09375, "mean_p": 0.451892, "boundary_entropy": 0.671433, "bpb": 3.45222}, "fi": {"chunks": 3744.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.199032, "cps": 29.25, "mean_p": 0.442041, "boundary_entropy": 0.672448, "bpb": 3.626641}, "ru": {"chunks": 10300.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.543536, "cps": 80.46875, "mean_p": 0.43388, "boundary_entropy": 0.641809, "bpb": 2.456633}, "el": {"chunks": 14274.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.695512, "cps": 111.515625, "mean_p": 0.487658, "boundary_entropy": 0.687921, "bpb": 2.578424}, "hi": {"chunks": 16249.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.965765, "cps": 126.945312, "mean_p": 0.467334, "boundary_entropy": 0.668171, "bpb": 2.070531}, "ta": {"chunks": 13750.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.735491, "cps": 107.421875, "mean_p": 0.455782, "boundary_entropy": 0.672199, "bpb": 1.765085}, "th": {"chunks": 3508.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.218091, "cps": 27.40625, "mean_p": 0.464644, "boundary_entropy": 0.684829, "bpb": 1.943058}, "zh": {"chunks": 2867.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.485439, "cps": 22.398438, "mean_p": 0.405345, "boundary_entropy": 0.633155, "bpb": 4.156641}, "my": {"chunks": 18980.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.001689, "cps": 148.28125, "mean_p": 0.416597, "boundary_entropy": 0.646703, "bpb": 1.966009}}, "bpb": {"en": 3.19153, "de": 3.47281, "vi": 2.77585, "tr": 3.45222, "fi": 3.62664, "ru": 2.45663, "el": 2.57842, "hi": 2.07053, "ta": 1.76509, "th": 1.94306, "zh": 4.15664, "my": 1.96601}, "gini_cpc": 0.359699, "gini_cps": 0.384958, "bpic": 3.910785, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8362377862044038, "vi": 1.8588288879047035, "tr": 1.08572418819549, "fi": 1.2179192217857988, "ru": 3.326002248867217, "el": 4.255978010469267, "hi": 5.909708967341898, "ta": 4.500613896402726, "th": 1.3345444625007705, "zh": 2.970494676800108, "my": 6.129532613798012}, "t": 55.6} |
| {"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 3.19963, "de": 3.46928, "vi": 2.78236, "tr": 3.45986, "fi": 3.61907, "ru": 2.4409, "el": 2.57302, "hi": 2.06761, "ta": 1.77837, "th": 1.9719, "zh": 4.1642, "my": 1.9805}, "bytes_per_chunk": {"en": 5.508, "de": 7.213, "vi": 4.653, "tr": 6.448, "fi": 5.129, "ru": 3.42, "el": 2.599, "hi": 2.601, "ta": 3.67, "th": 10.272, "zh": 5.454, "my": 2.835}, "bpb_mean": 2.79223, "frozen_inventory": false, "t": 57.9} |
| {"event": "train", "step": 400, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.01864, "bpb": 2.91229, "chunk_loss": 0.032418, "F": 0.207, "G": 0.43639, "chunks": 211.97, "bytes_per_chunk": 4.831, "boundary_entropy": 0.66559, "router_noise": 0.0, "grad_norm": 1.1409, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 58.3} |
| {"event": "train", "step": 450, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.06079, "bpb": 2.9731, "chunk_loss": 0.032322, "F": 0.20349, "G": 0.43151, "chunks": 208.38, "bytes_per_chunk": 4.914, "boundary_entropy": 0.6605, "router_noise": 0.0, "grad_norm": 0.9018, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 63.8} |
| {"event": "train", "step": 500, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.07236, "bpb": 2.98978, "chunk_loss": 0.030376, "F": 0.17194, "G": 0.42915, "chunks": 176.06, "bytes_per_chunk": 5.816, "boundary_entropy": 0.66274, "router_noise": 0.0, "grad_norm": 0.9117, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.097, "alpha": 0.03, "t": 68.6} |
| {"event": "train", "step": 550, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.97346, "bpb": 2.84711, "chunk_loss": 0.032083, "F": 0.20203, "G": 0.43026, "chunks": 206.88, "bytes_per_chunk": 4.95, "boundary_entropy": 0.66171, "router_noise": 0.0, "grad_norm": 0.9046, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 74.1} |
| {"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2637.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.151526, "cps": 20.601562, "mean_p": 0.43678, "boundary_entropy": 0.66772, "bpb": 2.914}, "de": {"chunks": 2394.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.116593, "cps": 18.703125, "mean_p": 0.436936, "boundary_entropy": 0.672112, "bpb": 3.199782}, "vi": {"chunks": 5092.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.276214, "cps": 39.78125, "mean_p": 0.438396, "boundary_entropy": 0.665565, "bpb": 2.564904}, "tr": {"chunks": 4163.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.229962, "cps": 32.523438, "mean_p": 0.454858, "boundary_entropy": 0.670987, "bpb": 3.138593}, "fi": {"chunks": 3182.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.169156, "cps": 24.859375, "mean_p": 0.434964, "boundary_entropy": 0.669612, "bpb": 3.379228}, "ru": {"chunks": 7877.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.415673, "cps": 61.539062, "mean_p": 0.412917, "boundary_entropy": 0.632179, "bpb": 2.193933}, "el": {"chunks": 7680.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.374214, "cps": 60.0, "mean_p": 0.448975, "boundary_entropy": 0.675212, "bpb": 2.424858}, "hi": {"chunks": 13742.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.816761, "cps": 107.359375, "mean_p": 0.477023, "boundary_entropy": 0.653994, "bpb": 1.736695}, "ta": {"chunks": 17503.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.93624, "cps": 136.742188, "mean_p": 0.46784, "boundary_entropy": 0.653159, "bpb": 1.489678}, "th": {"chunks": 11680.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.726142, "cps": 91.25, "mean_p": 0.422406, "boundary_entropy": 0.665176, "bpb": 1.869881}, "zh": {"chunks": 3195.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.540975, "cps": 24.960938, "mean_p": 0.423148, "boundary_entropy": 0.652442, "bpb": 3.944113}, "my": {"chunks": 10971.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.579006, "cps": 85.710938, "mean_p": 0.426811, "boundary_entropy": 0.660712, "bpb": 1.861773}}, "bpb": {"en": 2.914, "de": 3.19978, "vi": 2.5649, "tr": 3.13859, "fi": 3.37923, "ru": 2.19393, "el": 2.42486, "hi": 1.7367, "ta": 1.48968, "th": 1.86988, "zh": 3.94411, "my": 1.86177}, "gini_cpc": 0.335728, "gini_cps": 0.349771, "bpic": 4.258744, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.769459435065737, "vi": 1.8228848831780817, "tr": 1.5176437924386286, "fi": 1.1163548982456408, "ru": 2.7432514761017104, "el": 2.469644063706764, "hi": 5.390249422297051, "ta": 6.178755551742069, "th": 4.792209192985359, "zh": 3.570190666286035, "my": 3.8211741349236026}, "t": 81.7} |
| {"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.92274, "de": 3.19555, "vi": 2.57133, "tr": 3.15008, "fi": 3.37892, "ru": 2.17914, "el": 2.41973, "hi": 1.73208, "ta": 1.50088, "th": 1.89692, "zh": 3.95755, "my": 1.87561}, "bytes_per_chunk": {"en": 6.281, "de": 8.606, "vi": 4.876, "tr": 4.904, "fi": 5.803, "ru": 4.302, "el": 4.68, "hi": 3.027, "ta": 2.892, "th": 3.656, "zh": 4.66, "my": 4.521}, "bpb_mean": 2.56505, "frozen_inventory": false, "t": 84.9} |
| {"event": "train", "step": 600, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.89044, "bpb": 2.72733, "chunk_loss": 0.03283, "F": 0.21518, "G": 0.4302, "chunks": 220.34, "bytes_per_chunk": 4.647, "boundary_entropy": 0.65972, "router_noise": 0.0, "grad_norm": 0.7026, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 85.3} |
| {"event": "train", "step": 650, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.91121, "bpb": 2.75729, "chunk_loss": 0.03171, "F": 0.19769, "G": 0.42739, "chunks": 202.44, "bytes_per_chunk": 5.058, "boundary_entropy": 0.66222, "router_noise": 0.0, "grad_norm": 0.8875, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.126, "alpha": 0.03, "t": 91.6} |
| {"event": "train", "step": 700, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.84727, "bpb": 2.66504, "chunk_loss": 0.034259, "F": 0.23822, "G": 0.43947, "chunks": 243.94, "bytes_per_chunk": 4.198, "boundary_entropy": 0.66503, "router_noise": 0.0, "grad_norm": 0.8829, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.132, "alpha": 0.03, "t": 98.2} |
| {"event": "train", "step": 750, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.91808, "bpb": 2.7672, "chunk_loss": 0.031058, "F": 0.18439, "G": 0.43267, "chunks": 188.81, "bytes_per_chunk": 5.423, "boundary_entropy": 0.66554, "router_noise": 0.0, "grad_norm": 0.8609, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 103.9} |
| {"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3351.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192553, "cps": 26.179688, "mean_p": 0.445024, "boundary_entropy": 0.670462, "bpb": 2.710833}, "de": {"chunks": 3534.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.172113, "cps": 27.609375, "mean_p": 0.444924, "boundary_entropy": 0.67358, "bpb": 3.039707}, "vi": {"chunks": 5276.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.286195, "cps": 41.21875, "mean_p": 0.443327, "boundary_entropy": 0.667902, "bpb": 2.413169}, "tr": {"chunks": 4118.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.227476, "cps": 32.171875, "mean_p": 0.4573, "boundary_entropy": 0.672167, "bpb": 2.982912}, "fi": {"chunks": 3433.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.1825, "cps": 26.820312, "mean_p": 0.441194, "boundary_entropy": 0.672033, "bpb": 3.220385}, "ru": {"chunks": 9440.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.498153, "cps": 73.75, "mean_p": 0.42285, "boundary_entropy": 0.649208, "bpb": 2.023938}, "el": {"chunks": 6493.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.316377, "cps": 50.726562, "mean_p": 0.441579, "boundary_entropy": 0.673851, "bpb": 2.127673}, "hi": {"chunks": 11466.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.681486, "cps": 89.578125, "mean_p": 0.46962, "boundary_entropy": 0.660632, "bpb": 1.604685}, "ta": {"chunks": 16014.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.856593, "cps": 125.109375, "mean_p": 0.470557, "boundary_entropy": 0.656937, "bpb": 1.350584}, "th": {"chunks": 11508.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.715449, "cps": 89.90625, "mean_p": 0.425361, "boundary_entropy": 0.666262, "bpb": 1.701311}, "zh": {"chunks": 3396.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.575008, "cps": 26.53125, "mean_p": 0.420833, "boundary_entropy": 0.650075, "bpb": 3.80605}, "my": {"chunks": 22011.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.161653, "cps": 171.960938, "mean_p": 0.445189, "boundary_entropy": 0.67004, "bpb": 1.607992}}, "bpb": {"en": 2.71083, "de": 3.03971, "vi": 2.41317, "tr": 2.98291, "fi": 3.22038, "ru": 2.02394, "el": 2.12767, "hi": 1.60469, "ta": 1.35058, "th": 1.70131, "zh": 3.80605, "my": 1.60799}, "gini_cpc": 0.339984, "gini_cps": 0.360014, "bpic": 3.836275, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8938483244279434, "vi": 1.4863166308880265, "tr": 1.1813687626820188, "fi": 0.9477888873093597, "ru": 2.587095570258632, "el": 1.6430631292506679, "hi": 3.5392118296062076, "ta": 4.448607062962128, "th": 3.715595945777092, "zh": 2.986234656287811, "my": 6.032899492726549}, "t": 112.1} |
| {"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.7214, "de": 3.03828, "vi": 2.4186, "tr": 2.99542, "fi": 3.21678, "ru": 2.00566, "el": 2.11576, "hi": 1.60022, "ta": 1.3619, "th": 1.73194, "zh": 3.81918, "my": 1.61676}, "bytes_per_chunk": {"en": 4.75, "de": 5.44, "vi": 4.639, "tr": 4.834, "fi": 5.58, "ru": 3.589, "el": 5.567, "hi": 3.748, "ta": 3.134, "th": 3.647, "zh": 4.631, "my": 2.356}, "bpb_mean": 2.38683, "frozen_inventory": false, "t": 114.9} |
| {"event": "train", "step": 800, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.64902, "bpb": 2.37903, "chunk_loss": 0.034058, "F": 0.23456, "G": 0.44223, "chunks": 240.19, "bytes_per_chunk": 4.263, "boundary_entropy": 0.66787, "router_noise": 0.0, "grad_norm": 0.8126, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 115.2} |
| {"event": "train", "step": 850, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.72533, "bpb": 2.48912, "chunk_loss": 0.032103, "F": 0.20273, "G": 0.43639, "chunks": 207.59, "bytes_per_chunk": 4.933, "boundary_entropy": 0.66518, "router_noise": 0.0, "grad_norm": 0.8174, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.114, "alpha": 0.03, "t": 120.9} |
| {"event": "train", "step": 900, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.6381, "bpb": 2.36328, "chunk_loss": 0.032401, "F": 0.20752, "G": 0.43406, "chunks": 212.5, "bytes_per_chunk": 4.819, "boundary_entropy": 0.6648, "router_noise": 0.0, "grad_norm": 0.7517, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.116, "alpha": 0.03, "t": 126.8} |
| {"event": "train", "step": 950, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.74284, "bpb": 2.51439, "chunk_loss": 0.032733, "F": 0.21463, "G": 0.43312, "chunks": 219.78, "bytes_per_chunk": 4.659, "boundary_entropy": 0.6655, "router_noise": 0.0, "grad_norm": 0.7766, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.114, "alpha": 0.03, "t": 132.4} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3183.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.1829, "cps": 24.867188, "mean_p": 0.441286, "boundary_entropy": 0.668422, "bpb": 2.589307}, "de": {"chunks": 2697.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.13135, "cps": 21.070312, "mean_p": 0.434268, "boundary_entropy": 0.670006, "bpb": 2.931329}, "vi": {"chunks": 5453.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.295796, "cps": 42.601562, "mean_p": 0.441581, "boundary_entropy": 0.665246, "bpb": 2.316205}, "tr": {"chunks": 4911.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.271281, "cps": 38.367188, "mean_p": 0.458826, "boundary_entropy": 0.671648, "bpb": 2.888796}, "fi": {"chunks": 2988.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.158843, "cps": 23.34375, "mean_p": 0.437033, "boundary_entropy": 0.670385, "bpb": 3.158931}, "ru": {"chunks": 9108.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.480633, "cps": 71.15625, "mean_p": 0.427312, "boundary_entropy": 0.652307, "bpb": 1.94295}, "el": {"chunks": 5887.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.286849, "cps": 45.992188, "mean_p": 0.432875, "boundary_entropy": 0.672095, "bpb": 2.07579}, "hi": {"chunks": 13156.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.781932, "cps": 102.78125, "mean_p": 0.456269, "boundary_entropy": 0.652354, "bpb": 1.566536}, "ta": {"chunks": 16650.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.890612, "cps": 130.078125, "mean_p": 0.451938, "boundary_entropy": 0.65126, "bpb": 1.316598}, "th": {"chunks": 12668.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.787566, "cps": 98.96875, "mean_p": 0.453902, "boundary_entropy": 0.677558, "bpb": 1.622519}, "zh": {"chunks": 3267.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.553166, "cps": 25.523438, "mean_p": 0.427054, "boundary_entropy": 0.657007, "bpb": 3.700254}, "my": {"chunks": 5892.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.310956, "cps": 46.03125, "mean_p": 0.444235, "boundary_entropy": 0.674811, "bpb": 1.460412}}, "bpb": {"en": 2.58931, "de": 2.93133, "vi": 2.31621, "tr": 2.8888, "fi": 3.15893, "ru": 1.94295, "el": 2.07579, "hi": 1.56654, "ta": 1.3166, "th": 1.62252, "zh": 3.70025, "my": 1.46041}, "gini_cpc": 0.32817, "gini_cps": 0.33385, "bpic": 4.469846, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7181514155760298, "vi": 1.6172599706485864, "tr": 1.4832244102406362, "fi": 0.8684727402214327, "ru": 2.62785434162524, "el": 1.5683416362405838, "hi": 4.2751983956372985, "ta": 4.869408953035722, "th": 4.306004417507486, "zh": 3.0244274660539467, "my": 1.7001484493400958}, "t": 139.5} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.60119, "de": 2.93395, "vi": 2.32403, "tr": 2.89835, "fi": 3.15529, "ru": 1.92484, "el": 2.06516, "hi": 1.56108, "ta": 1.32666, "th": 1.64637, "zh": 3.71347, "my": 1.46892}, "bytes_per_chunk": {"en": 5.076, "de": 7.256, "vi": 4.699, "tr": 3.92, "fi": 6.593, "ru": 3.679, "el": 6.028, "hi": 3.503, "ta": 3.091, "th": 3.421, "zh": 4.626, "my": 8.427}, "bpb_mean": 2.30161, "frozen_inventory": false, "t": 142.4} |
| {"event": "train", "step": 1000, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.75833, "bpb": 2.53674, "chunk_loss": 0.032859, "F": 0.215, "G": 0.43611, "chunks": 220.16, "bytes_per_chunk": 4.651, "boundary_entropy": 0.66781, "router_noise": 0.0, "grad_norm": 0.8005, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.215, "alpha": 0.03, "t": 143.2} |
| {"event": "train", "step": 1050, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.53171, "bpb": 2.20979, "chunk_loss": 0.032319, "F": 0.20578, "G": 0.43974, "chunks": 210.72, "bytes_per_chunk": 4.86, "boundary_entropy": 0.66935, "router_noise": 0.0, "grad_norm": 0.8091, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 148.7} |
| {"event": "train", "step": 1100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.77736, "bpb": 2.56419, "chunk_loss": 0.033278, "F": 0.22165, "G": 0.4402, "chunks": 226.97, "bytes_per_chunk": 4.512, "boundary_entropy": 0.66872, "router_noise": 0.0, "grad_norm": 0.834, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.111, "alpha": 0.03, "t": 154.2} |
| {"event": "train", "step": 1150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.68643, "bpb": 2.43301, "chunk_loss": 0.033415, "F": 0.22525, "G": 0.43487, "chunks": 230.66, "bytes_per_chunk": 4.44, "boundary_entropy": 0.66613, "router_noise": 0.0, "grad_norm": 0.7852, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.104, "alpha": 0.03, "t": 159.5} |
| {"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3089.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.177498, "cps": 24.132812, "mean_p": 0.442138, "boundary_entropy": 0.6696, "bpb": 2.495232}, "de": {"chunks": 2788.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.135781, "cps": 21.78125, "mean_p": 0.438932, "boundary_entropy": 0.672237, "bpb": 2.801711}, "vi": {"chunks": 5815.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.315433, "cps": 45.429688, "mean_p": 0.446144, "boundary_entropy": 0.667837, "bpb": 2.247141}, "tr": {"chunks": 4099.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.226427, "cps": 32.023438, "mean_p": 0.452678, "boundary_entropy": 0.669663, "bpb": 2.800626}, "fi": {"chunks": 3328.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.176918, "cps": 26.0, "mean_p": 0.440579, "boundary_entropy": 0.671356, "bpb": 3.057437}, "ru": {"chunks": 8413.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.443958, "cps": 65.726562, "mean_p": 0.420678, "boundary_entropy": 0.654318, "bpb": 1.884427}, "el": {"chunks": 6117.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.298056, "cps": 47.789062, "mean_p": 0.439348, "boundary_entropy": 0.674488, "bpb": 1.956421}, "hi": {"chunks": 12222.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.726419, "cps": 95.484375, "mean_p": 0.447326, "boundary_entropy": 0.653805, "bpb": 1.571452}, "ta": {"chunks": 14028.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.750361, "cps": 109.59375, "mean_p": 0.443594, "boundary_entropy": 0.654427, "bpb": 1.233454}, "th": {"chunks": 8626.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.536276, "cps": 67.390625, "mean_p": 0.438198, "boundary_entropy": 0.674065, "bpb": 1.518474}, "zh": {"chunks": 3615.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.612089, "cps": 28.242188, "mean_p": 0.42459, "boundary_entropy": 0.655681, "bpb": 3.627417}, "my": {"chunks": 5473.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.288843, "cps": 42.757812, "mean_p": 0.436813, "boundary_entropy": 0.671429, "bpb": 1.445273}}, "bpb": {"en": 2.49523, "de": 2.80171, "vi": 2.24714, "tr": 2.80063, "fi": 3.05744, "ru": 1.88443, "el": 1.95642, "hi": 1.57145, "ta": 1.23345, "th": 1.51847, "zh": 3.62742, "my": 1.44527}, "gini_cpc": 0.298275, "gini_cps": 0.293453, "bpic": 4.944803, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7649738231393954, "vi": 1.7771037733041022, "tr": 1.2756560079657846, "fi": 0.996730266323891, "ru": 2.5011969273898105, "el": 1.6792054968781955, "hi": 4.092544575320471, "ta": 4.227430725682787, "th": 3.0213052160075873, "zh": 3.4484272705755887, "my": 1.627302471270813}, "t": 166.7} |
| {"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.50536, "de": 2.80386, "vi": 2.25401, "tr": 2.8108, "fi": 3.05899, "ru": 1.86692, "el": 1.94456, "hi": 1.56657, "ta": 1.24281, "th": 1.54037, "zh": 3.63566, "my": 1.45287}, "bytes_per_chunk": {"en": 5.179, "de": 7.132, "vi": 4.222, "tr": 4.691, "fi": 5.682, "ru": 4.084, "el": 6.111, "hi": 3.597, "ta": 3.622, "th": 5.13, "zh": 4.184, "my": 8.383}, "bpb_mean": 2.22356, "frozen_inventory": false, "t": 169.5} |
| {"event": "train", "step": 1200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.6268, "bpb": 2.34698, "chunk_loss": 0.033075, "F": 0.21967, "G": 0.43643, "chunks": 224.94, "bytes_per_chunk": 4.552, "boundary_entropy": 0.66666, "router_noise": 0.0, "grad_norm": 0.7815, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.207, "alpha": 0.03, "t": 169.8} |
| {"event": "train", "step": 1250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.66455, "bpb": 2.40143, "chunk_loss": 0.032415, "F": 0.2074, "G": 0.43804, "chunks": 212.38, "bytes_per_chunk": 4.822, "boundary_entropy": 0.66896, "router_noise": 0.0, "grad_norm": 0.974, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 175.2} |
| {"event": "train", "step": 1300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.5951, "bpb": 2.30124, "chunk_loss": 0.03338, "F": 0.22437, "G": 0.43889, "chunks": 229.75, "bytes_per_chunk": 4.457, "boundary_entropy": 0.66745, "router_noise": 0.0, "grad_norm": 0.7679, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.105, "alpha": 0.03, "t": 180.4} |
| {"event": "train", "step": 1350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.72621, "bpb": 2.4904, "chunk_loss": 0.032394, "F": 0.20724, "G": 0.43883, "chunks": 212.22, "bytes_per_chunk": 4.825, "boundary_entropy": 0.67069, "router_noise": 0.0, "grad_norm": 0.8788, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 185.8} |
| {"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3026.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.173878, "cps": 23.640625, "mean_p": 0.441359, "boundary_entropy": 0.669189, "bpb": 2.438514}, "de": {"chunks": 3311.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.161253, "cps": 25.867188, "mean_p": 0.443221, "boundary_entropy": 0.673246, "bpb": 2.752361}, "vi": {"chunks": 5281.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.286466, "cps": 41.257812, "mean_p": 0.439014, "boundary_entropy": 0.665575, "bpb": 2.214083}, "tr": {"chunks": 4403.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.243219, "cps": 34.398438, "mean_p": 0.454351, "boundary_entropy": 0.670896, "bpb": 2.741091}, "fi": {"chunks": 3199.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.17006, "cps": 24.992188, "mean_p": 0.439123, "boundary_entropy": 0.670733, "bpb": 2.973207}, "ru": {"chunks": 8709.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.459578, "cps": 68.039062, "mean_p": 0.428503, "boundary_entropy": 0.657722, "bpb": 1.824193}, "el": {"chunks": 6049.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.294742, "cps": 47.257812, "mean_p": 0.436137, "boundary_entropy": 0.673955, "bpb": 1.905414}, "hi": {"chunks": 11830.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.70312, "cps": 92.421875, "mean_p": 0.437428, "boundary_entropy": 0.645449, "bpb": 1.470816}, "ta": {"chunks": 14842.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.793902, "cps": 115.953125, "mean_p": 0.443684, "boundary_entropy": 0.651946, "bpb": 1.164866}, "th": {"chunks": 7271.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.452036, "cps": 56.804688, "mean_p": 0.448841, "boundary_entropy": 0.678752, "bpb": 1.443448}, "zh": {"chunks": 2813.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.476295, "cps": 21.976562, "mean_p": 0.41997, "boundary_entropy": 0.654368, "bpb": 3.585516}, "my": {"chunks": 6985.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.36864, "cps": 54.570312, "mean_p": 0.434785, "boundary_entropy": 0.672732, "bpb": 1.560585}}, "bpb": {"en": 2.43851, "de": 2.75236, "vi": 2.21408, "tr": 2.74109, "fi": 2.97321, "ru": 1.82419, "el": 1.90541, "hi": 1.47082, "ta": 1.16487, "th": 1.44345, "zh": 3.58552, "my": 1.56059}, "gini_cpc": 0.2823, "gini_cps": 0.298553, "bpic": 4.938059, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9273890636094673, "vi": 1.6475106172326948, "tr": 1.3987926142975866, "fi": 0.9780421082951879, "ru": 2.6431041265932715, "el": 1.695110194906822, "hi": 4.043755309359498, "ta": 4.565855409841108, "th": 2.5997301801134545, "zh": 2.7392488376501745, "my": 2.120109201361056}, "t": 193.0} |
| {"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.44704, "de": 2.75802, "vi": 2.22265, "tr": 2.7499, "fi": 2.97596, "ru": 1.80512, "el": 1.8937, "hi": 1.4677, "ta": 1.17389, "th": 1.46593, "zh": 3.59128, "my": 1.56931}, "bytes_per_chunk": {"en": 5.547, "de": 6.097, "vi": 4.68, "tr": 4.372, "fi": 5.895, "ru": 3.917, "el": 6.024, "hi": 3.698, "ta": 3.475, "th": 6.324, "zh": 5.395, "my": 7.567}, "bpb_mean": 2.17671, "frozen_inventory": false, "t": 195.7} |
| {"event": "train", "step": 1400, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.59159, "bpb": 2.29618, "chunk_loss": 0.031112, "F": 0.18622, "G": 0.43563, "chunks": 190.69, "bytes_per_chunk": 5.37, "boundary_entropy": 0.67064, "router_noise": 0.0, "grad_norm": 0.9913, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.206, "alpha": 0.03, "t": 196.1} |
| {"event": "train", "step": 1450, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.73714, "bpb": 2.50616, "chunk_loss": 0.031333, "F": 0.18985, "G": 0.43404, "chunks": 194.41, "bytes_per_chunk": 5.267, "boundary_entropy": 0.66917, "router_noise": 0.0, "grad_norm": 0.9263, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 201.1} |
| {"event": "train", "step": 1500, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.57568, "bpb": 2.27322, "chunk_loss": 0.033602, "F": 0.22922, "G": 0.43481, "chunks": 234.72, "bytes_per_chunk": 4.363, "boundary_entropy": 0.66513, "router_noise": 0.0, "grad_norm": 0.8451, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 205.8} |
| {"event": "train", "step": 1550, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.63293, "bpb": 2.35582, "chunk_loss": 0.031556, "F": 0.19366, "G": 0.43408, "chunks": 198.31, "bytes_per_chunk": 5.164, "boundary_entropy": 0.6688, "router_noise": 0.0, "grad_norm": 0.8502, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 211.3} |
| {"event": "probe", "step": 1600, "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": 0.441219, "boundary_entropy": 0.668899, "bpb": 2.382952}, "de": {"chunks": 2468.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.120197, "cps": 19.28125, "mean_p": 0.434457, "boundary_entropy": 0.670604, "bpb": 2.681996}, "vi": {"chunks": 6341.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.343965, "cps": 49.539062, "mean_p": 0.445913, "boundary_entropy": 0.666795, "bpb": 2.148348}, "tr": {"chunks": 4031.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.22267, "cps": 31.492188, "mean_p": 0.451538, "boundary_entropy": 0.669859, "bpb": 2.659664}, "fi": {"chunks": 2838.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.150869, "cps": 22.171875, "mean_p": 0.434498, "boundary_entropy": 0.66929, "bpb": 2.930002}, "ru": {"chunks": 9419.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.497045, "cps": 73.585938, "mean_p": 0.428349, "boundary_entropy": 0.657522, "bpb": 1.756423}, "el": {"chunks": 7287.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.355065, "cps": 56.929688, "mean_p": 0.440777, "boundary_entropy": 0.674213, "bpb": 1.862366}, "hi": {"chunks": 11367.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.675602, "cps": 88.804688, "mean_p": 0.437135, "boundary_entropy": 0.650294, "bpb": 1.423662}, "ta": {"chunks": 15345.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.820808, "cps": 119.882812, "mean_p": 0.439892, "boundary_entropy": 0.651624, "bpb": 1.167924}, "th": {"chunks": 4493.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.279329, "cps": 35.101562, "mean_p": 0.442798, "boundary_entropy": 0.67684, "bpb": 1.386679}, "zh": {"chunks": 3687.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.62428, "cps": 28.804688, "mean_p": 0.421064, "boundary_entropy": 0.652194, "bpb": 3.538924}, "my": {"chunks": 3339.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.176219, "cps": 26.085938, "mean_p": 0.440355, "boundary_entropy": 0.67513, "bpb": 1.315862}}, "bpb": {"en": 2.38295, "de": 2.682, "vi": 2.14835, "tr": 2.65966, "fi": 2.93, "ru": 1.75642, "el": 1.86237, "hi": 1.42366, "ta": 1.16792, "th": 1.38668, "zh": 3.53892, "my": 1.31586}, "gini_cpc": 0.331215, "gini_cps": 0.329641, "bpic": 5.211371, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.690813788751696, "vi": 1.9768916207074085, "tr": 1.2797657883251503, "fi": 0.8670991431346523, "ru": 2.8566947191210965, "el": 2.0406859469007, "hi": 3.882925307334614, "ta": 4.717475709461094, "th": 1.6054012718923452, "zh": 3.5879629049740513, "my": 1.012794402339825}, "t": 218.7} |
| {"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.39247, "de": 2.68683, "vi": 2.1538, "tr": 2.66786, "fi": 2.93094, "ru": 1.7385, "el": 1.85141, "hi": 1.42247, "ta": 1.17516, "th": 1.40734, "zh": 3.54646, "my": 1.32372}, "bytes_per_chunk": {"en": 5.645, "de": 8.48, "vi": 3.766, "tr": 4.962, "fi": 6.942, "ru": 3.598, "el": 4.853, "hi": 3.769, "ta": 3.312, "th": 10.109, "zh": 4.326, "my": 15.534}, "bpb_mean": 2.10808, "frozen_inventory": false, "t": 221.3} |
| {"event": "train", "step": 1600, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.60156, "bpb": 2.31056, "chunk_loss": 0.032607, "F": 0.21259, "G": 0.43093, "chunks": 217.69, "bytes_per_chunk": 4.704, "boundary_entropy": 0.66455, "router_noise": 0.0, "grad_norm": 0.8238, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.209, "alpha": 0.03, "t": 221.7} |
| {"event": "train", "step": 1650, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.53471, "bpb": 2.21411, "chunk_loss": 0.032394, "F": 0.20667, "G": 0.4343, "chunks": 211.62, "bytes_per_chunk": 4.839, "boundary_entropy": 0.66711, "router_noise": 0.0, "grad_norm": 0.7992, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 227.0} |
| {"event": "train", "step": 1700, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.45831, "bpb": 2.10389, "chunk_loss": 0.031893, "F": 0.2002, "G": 0.43211, "chunks": 205.0, "bytes_per_chunk": 4.995, "boundary_entropy": 0.66684, "router_noise": 0.0, "grad_norm": 0.7425, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 232.5} |
| {"event": "train", "step": 1750, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.64834, "bpb": 2.37805, "chunk_loss": 0.031682, "F": 0.19589, "G": 0.43555, "chunks": 200.59, "bytes_per_chunk": 5.105, "boundary_entropy": 0.66943, "router_noise": 0.0, "grad_norm": 0.8684, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 237.9} |
| {"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2951.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.169568, "cps": 23.054688, "mean_p": 0.438015, "boundary_entropy": 0.667594, "bpb": 2.345981}, "de": {"chunks": 2521.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.122778, "cps": 19.695312, "mean_p": 0.435138, "boundary_entropy": 0.671397, "bpb": 2.614252}, "vi": {"chunks": 5255.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.285056, "cps": 41.054688, "mean_p": 0.440263, "boundary_entropy": 0.665767, "bpb": 2.126229}, "tr": {"chunks": 4333.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.239353, "cps": 33.851562, "mean_p": 0.453949, "boundary_entropy": 0.670675, "bpb": 2.637499}, "fi": {"chunks": 2582.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.13726, "cps": 20.171875, "mean_p": 0.435443, "boundary_entropy": 0.670893, "bpb": 2.870265}, "ru": {"chunks": 9220.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.486544, "cps": 72.03125, "mean_p": 0.429201, "boundary_entropy": 0.658915, "bpb": 1.708437}, "el": {"chunks": 6733.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.328071, "cps": 52.601562, "mean_p": 0.438903, "boundary_entropy": 0.673777, "bpb": 1.813667}, "hi": {"chunks": 9694.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.576166, "cps": 75.734375, "mean_p": 0.424915, "boundary_entropy": 0.651173, "bpb": 1.390472}, "ta": {"chunks": 12276.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.656646, "cps": 95.90625, "mean_p": 0.430493, "boundary_entropy": 0.649421, "bpb": 1.11549}, "th": {"chunks": 9206.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.572334, "cps": 71.921875, "mean_p": 0.452494, "boundary_entropy": 0.678543, "bpb": 1.350187}, "zh": {"chunks": 3588.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.607518, "cps": 28.03125, "mean_p": 0.423554, "boundary_entropy": 0.654491, "bpb": 3.505304}, "my": {"chunks": 22666.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.196221, "cps": 177.078125, "mean_p": 0.472754, "boundary_entropy": 0.679156, "bpb": 1.333646}}, "bpb": {"en": 2.34598, "de": 2.61425, "vi": 2.12623, "tr": 2.6375, "fi": 2.87027, "ru": 1.70844, "el": 1.81367, "hi": 1.39047, "ta": 1.11549, "th": 1.35019, "zh": 3.5053, "my": 1.33365}, "gini_cpc": 0.344837, "gini_cps": 0.366473, "bpic": 4.216215, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7240613225153851, "vi": 1.6810649339262194, "tr": 1.411539540682835, "fi": 0.8094672178317116, "ru": 2.8693043545902333, "el": 1.9347403091304047, "hi": 3.39783944093296, "ta": 3.8724544759737567, "th": 3.375241218421241, "zh": 3.582728532620452, "my": 7.054502944333122}, "t": 244.9} |
| {"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.35339, "de": 2.6206, "vi": 2.13205, "tr": 2.64918, "fi": 2.86855, "ru": 1.68898, "el": 1.80365, "hi": 1.38951, "ta": 1.12367, "th": 1.36866, "zh": 3.5112, "my": 1.34124}, "bytes_per_chunk": {"en": 5.728, "de": 8.183, "vi": 4.693, "tr": 4.629, "fi": 7.58, "ru": 3.822, "el": 5.365, "hi": 4.396, "ta": 4.092, "th": 4.841, "zh": 4.465, "my": 2.341}, "bpb_mean": 2.07089, "frozen_inventory": false, "t": 247.0} |
| {"event": "train", "step": 1800, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.59197, "bpb": 2.29673, "chunk_loss": 0.030632, "F": 0.17734, "G": 0.43535, "chunks": 181.59, "bytes_per_chunk": 5.639, "boundary_entropy": 0.67086, "router_noise": 0.0, "grad_norm": 1.1281, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 247.3} |
| {"event": "train", "step": 1850, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.57135, "bpb": 2.26699, "chunk_loss": 0.032423, "F": 0.20746, "G": 0.43875, "chunks": 212.44, "bytes_per_chunk": 4.82, "boundary_entropy": 0.67105, "router_noise": 0.0, "grad_norm": 0.8366, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 252.7} |
| {"event": "train", "step": 1900, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.61595, "bpb": 2.33132, "chunk_loss": 0.031841, "F": 0.19919, "G": 0.43225, "chunks": 203.97, "bytes_per_chunk": 5.02, "boundary_entropy": 0.66831, "router_noise": 0.0, "grad_norm": 0.7741, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 258.1} |
| {"event": "train", "step": 1950, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.62397, "bpb": 2.34289, "chunk_loss": 0.032353, "F": 0.20779, "G": 0.43457, "chunks": 212.78, "bytes_per_chunk": 4.812, "boundary_entropy": 0.6683, "router_noise": 0.0, "grad_norm": 0.8627, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.099, "alpha": 0.03, "t": 263.0} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3188.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.183187, "cps": 24.90625, "mean_p": 0.441875, "boundary_entropy": 0.668678, "bpb": 2.29189}, "de": {"chunks": 3127.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.152291, "cps": 24.429688, "mean_p": 0.439907, "boundary_entropy": 0.672101, "bpb": 2.574321}, "vi": {"chunks": 5333.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.289287, "cps": 41.664062, "mean_p": 0.439678, "boundary_entropy": 0.665161, "bpb": 2.099169}, "tr": {"chunks": 4437.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.245097, "cps": 34.664062, "mean_p": 0.453844, "boundary_entropy": 0.67041, "bpb": 2.564586}, "fi": {"chunks": 3706.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.197012, "cps": 28.953125, "mean_p": 0.443482, "boundary_entropy": 0.672084, "bpb": 2.795764}, "ru": {"chunks": 9227.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.486913, "cps": 72.085938, "mean_p": 0.431337, "boundary_entropy": 0.660771, "bpb": 1.674747}, "el": {"chunks": 5747.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.280027, "cps": 44.898438, "mean_p": 0.435147, "boundary_entropy": 0.673051, "bpb": 1.790358}, "hi": {"chunks": 11047.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.656582, "cps": 86.304688, "mean_p": 0.428746, "boundary_entropy": 0.649153, "bpb": 1.356087}, "ta": {"chunks": 14790.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.791121, "cps": 115.546875, "mean_p": 0.423644, "boundary_entropy": 0.645071, "bpb": 1.083087}, "th": {"chunks": 11484.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.713957, "cps": 89.71875, "mean_p": 0.45365, "boundary_entropy": 0.67756, "bpb": 1.300915}, "zh": {"chunks": 3914.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.662716, "cps": 30.578125, "mean_p": 0.421503, "boundary_entropy": 0.651479, "bpb": 3.478287}, "my": {"chunks": 11999.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.633259, "cps": 93.742188, "mean_p": 0.462498, "boundary_entropy": 0.679954, "bpb": 1.269183}}, "bpb": {"en": 2.29189, "de": 2.57432, "vi": 2.09917, "tr": 2.56459, "fi": 2.79576, "ru": 1.67475, "el": 1.79036, "hi": 1.35609, "ta": 1.08309, "th": 1.30091, "zh": 3.47829, "my": 1.26918}, "gini_cpc": 0.289441, "gini_cps": 0.295919, "bpic": 4.361197, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8313449854966156, "vi": 1.5791895053970697, "tr": 1.337964790352376, "fi": 1.075472572142878, "ru": 2.6580130800528368, "el": 1.528643308107787, "hi": 3.584223546322484, "ta": 4.318655005750033, "th": 3.897426430868945, "zh": 3.61770530022634, "my": 3.4569053182334324}, "t": 270.0} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.30253, "de": 2.58274, "vi": 2.10241, "tr": 2.57327, "fi": 2.79773, "ru": 1.65407, "el": 1.77957, "hi": 1.35465, "ta": 1.09345, "th": 1.3184, "zh": 3.48549, "my": 1.27729}, "bytes_per_chunk": {"en": 5.37, "de": 6.656, "vi": 4.583, "tr": 4.55, "fi": 5.234, "ru": 3.766, "el": 6.771, "hi": 3.818, "ta": 3.494, "th": 4.08, "zh": 4.236, "my": 4.564}, "bpb_mean": 2.0268, "frozen_inventory": false, "t": 272.8} |
| {"event": "train", "step": 2000, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.53529, "bpb": 2.21495, "chunk_loss": 0.032915, "F": 0.21658, "G": 0.43694, "chunks": 221.78, "bytes_per_chunk": 4.617, "boundary_entropy": 0.66813, "router_noise": 0.0, "grad_norm": 0.7485, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 273.1} |
| {"event": "train", "step": 2050, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.42187, "bpb": 2.05133, "chunk_loss": 0.031677, "F": 0.1962, "G": 0.43105, "chunks": 200.91, "bytes_per_chunk": 5.097, "boundary_entropy": 0.66631, "router_noise": 0.0, "grad_norm": 0.7808, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.099, "alpha": 0.03, "t": 278.1} |
| {"event": "train", "step": 2100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.57985, "bpb": 2.27925, "chunk_loss": 0.031541, "F": 0.19299, "G": 0.43588, "chunks": 197.62, "bytes_per_chunk": 5.182, "boundary_entropy": 0.66975, "router_noise": 0.0, "grad_norm": 0.9156, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.101, "alpha": 0.03, "t": 283.1} |
| {"event": "train", "step": 2150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.49799, "bpb": 2.16114, "chunk_loss": 0.031241, "F": 0.18866, "G": 0.43063, "chunks": 193.19, "bytes_per_chunk": 5.301, "boundary_entropy": 0.66618, "router_noise": 0.0, "grad_norm": 0.7797, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 288.3} |
| {"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3224.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185255, "cps": 25.1875, "mean_p": 0.439871, "boundary_entropy": 0.667737, "bpb": 2.261881}, "de": {"chunks": 2834.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.138022, "cps": 22.140625, "mean_p": 0.436633, "boundary_entropy": 0.671201, "bpb": 2.537601}, "vi": {"chunks": 4701.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.255004, "cps": 36.726562, "mean_p": 0.431422, "boundary_entropy": 0.66202, "bpb": 2.080908}, "tr": {"chunks": 4085.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.225653, "cps": 31.914062, "mean_p": 0.450398, "boundary_entropy": 0.669017, "bpb": 2.52107}, "fi": {"chunks": 2275.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.12094, "cps": 17.773438, "mean_p": 0.435928, "boundary_entropy": 0.67143, "bpb": 2.760704}, "ru": {"chunks": 8405.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.443536, "cps": 65.664062, "mean_p": 0.43147, "boundary_entropy": 0.662824, "bpb": 1.655208}, "el": {"chunks": 6506.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.31701, "cps": 50.828125, "mean_p": 0.439618, "boundary_entropy": 0.6747, "bpb": 1.748286}, "hi": {"chunks": 9506.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.564993, "cps": 74.265625, "mean_p": 0.409689, "boundary_entropy": 0.640024, "bpb": 1.338583}, "ta": {"chunks": 13140.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.702862, "cps": 102.65625, "mean_p": 0.42336, "boundary_entropy": 0.643115, "bpb": 1.07482}, "th": {"chunks": 10195.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.63382, "cps": 79.648438, "mean_p": 0.451494, "boundary_entropy": 0.6772, "bpb": 1.282453}, "zh": {"chunks": 3121.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.528446, "cps": 24.382812, "mean_p": 0.415638, "boundary_entropy": 0.651009, "bpb": 3.430509}, "my": {"chunks": 15542.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.820245, "cps": 121.421875, "mean_p": 0.462095, "boundary_entropy": 0.678346, "bpb": 1.251823}}, "bpb": {"en": 2.26188, "de": 2.5376, "vi": 2.08091, "tr": 2.52107, "fi": 2.7607, "ru": 1.65521, "el": 1.74829, "hi": 1.33858, "ta": 1.07482, "th": 1.28245, "zh": 3.43051, "my": 1.25182}, "gini_cpc": 0.313348, "gini_cps": 0.333581, "bpic": 4.594309, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7450347434421065, "vi": 1.3764999384195427, "tr": 1.2180653707643976, "fi": 0.6528277466341759, "ru": 2.394184365936217, "el": 1.711206025703293, "hi": 3.0498032586212616, "ta": 3.7940144689102926, "th": 3.421332256290442, "zh": 2.852524696254555, "my": 4.427643194565144}, "t": 295.2} |
| {"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.26952, "de": 2.54497, "vi": 2.08516, "tr": 2.52717, "fi": 2.76134, "ru": 1.63385, "el": 1.73902, "hi": 1.33627, "ta": 1.08212, "th": 1.30438, "zh": 3.43942, "my": 1.25986}, "bytes_per_chunk": {"en": 5.209, "de": 7.505, "vi": 5.466, "tr": 4.749, "fi": 8.416, "ru": 4.09, "el": 5.54, "hi": 4.576, "ta": 3.865, "th": 4.463, "zh": 5.079, "my": 3.492}, "bpb_mean": 1.99859, "frozen_inventory": false, "t": 297.9} |
| {"event": "train", "step": 2200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.51199, "bpb": 2.18134, "chunk_loss": 0.031905, "F": 0.19968, "G": 0.43274, "chunks": 204.47, "bytes_per_chunk": 5.008, "boundary_entropy": 0.66793, "router_noise": 0.0, "grad_norm": 0.7407, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 298.4} |
| {"event": "train", "step": 2250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.73191, "bpb": 2.49862, "chunk_loss": 0.031856, "F": 0.19907, "G": 0.43496, "chunks": 203.84, "bytes_per_chunk": 5.023, "boundary_entropy": 0.66913, "router_noise": 0.0, "grad_norm": 0.8151, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.1, "alpha": 0.03, "t": 303.4} |
| {"event": "train", "step": 2300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.50899, "bpb": 2.17701, "chunk_loss": 0.031921, "F": 0.20056, "G": 0.43142, "chunks": 205.38, "bytes_per_chunk": 4.986, "boundary_entropy": 0.6668, "router_noise": 0.0, "grad_norm": 0.7458, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 308.3} |
| {"event": "train", "step": 2350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.52105, "bpb": 2.19441, "chunk_loss": 0.031544, "F": 0.19354, "G": 0.43202, "chunks": 198.19, "bytes_per_chunk": 5.167, "boundary_entropy": 0.66816, "router_noise": 0.0, "grad_norm": 0.8064, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.102, "alpha": 0.03, "t": 313.4} |
| {"event": "probe", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 2973.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.170833, "cps": 23.226562, "mean_p": 0.438101, "boundary_entropy": 0.667432, "bpb": 2.238396}, "de": {"chunks": 3108.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.151366, "cps": 24.28125, "mean_p": 0.439854, "boundary_entropy": 0.672388, "bpb": 2.493206}, "vi": {"chunks": 4713.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.255655, "cps": 36.820312, "mean_p": 0.434048, "boundary_entropy": 0.662407, "bpb": 2.066605}, "tr": {"chunks": 4486.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.247804, "cps": 35.046875, "mean_p": 0.452332, "boundary_entropy": 0.66988, "bpb": 2.477179}, "fi": {"chunks": 3250.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.172771, "cps": 25.390625, "mean_p": 0.43994, "boundary_entropy": 0.671841, "bpb": 2.719543}, "ru": {"chunks": 7895.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.416623, "cps": 61.679688, "mean_p": 0.427575, "boundary_entropy": 0.660917, "bpb": 1.633734}, "el": {"chunks": 6643.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.323686, "cps": 51.898438, "mean_p": 0.438773, "boundary_entropy": 0.674374, "bpb": 1.731062}, "hi": {"chunks": 9392.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.558217, "cps": 73.375, "mean_p": 0.40953, "boundary_entropy": 0.641228, "bpb": 1.307749}, "ta": {"chunks": 13103.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.700883, "cps": 102.367188, "mean_p": 0.418927, "boundary_entropy": 0.6451, "bpb": 1.055353}, "th": {"chunks": 10566.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.656885, "cps": 82.546875, "mean_p": 0.449798, "boundary_entropy": 0.676461, "bpb": 1.262606}, "zh": {"chunks": 3372.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.570945, "cps": 26.34375, "mean_p": 0.414475, "boundary_entropy": 0.64753, "bpb": 3.424866}, "my": {"chunks": 7098.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.374604, "cps": 55.453125, "mean_p": 0.451244, "boundary_entropy": 0.677948, "bpb": 1.258301}}, "bpb": {"en": 2.2384, "de": 2.49321, "vi": 2.0666, "tr": 2.47718, "fi": 2.71954, "ru": 1.63373, "el": 1.73106, "hi": 1.30775, "ta": 1.05535, "th": 1.26261, "zh": 3.42487, "my": 1.2583}, "gini_cpc": 0.277017, "gini_cps": 0.276251, "bpic": 5.010261, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8860491513673221, "vi": 1.4965233891024097, "tr": 1.4505674530978776, "fi": 1.0113481596985168, "ru": 2.4387772272350894, "el": 1.8947530650679485, "hi": 3.267625089400443, "ta": 4.102744599766446, "th": 3.8451983926209365, "zh": 3.3421299654902925, "my": 2.1928141749462706}, "t": 320.1} |
| {"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.24458, "de": 2.50121, "vi": 2.0686, "tr": 2.48689, "fi": 2.71946, "ru": 1.61208, "el": 1.7234, "hi": 1.30455, "ta": 1.0623, "th": 1.28047, "zh": 3.4299, "my": 1.26633}, "bytes_per_chunk": {"en": 5.984, "de": 7.068, "vi": 5.242, "tr": 4.41, "fi": 6.078, "ru": 4.427, "el": 5.449, "hi": 4.566, "ta": 4.012, "th": 4.347, "zh": 4.929, "my": 7.596}, "bpb_mean": 1.97498, "frozen_inventory": false, "t": 322.1} |
| {"event": "phase_transition", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 322.4} |
| {"event": "train", "step": 2400, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.53522, "bpb": 2.21486, "chunk_loss": 0.042122, "F": 0.1947, "G": 0.43003, "chunks": 199.38, "bytes_per_chunk": 5.136, "boundary_entropy": 0.66522, "router_noise": 0.0, "grad_norm": 0.7999, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.18, "n_target_mean": 10.0837, "alpha": 0.03, "t": 322.4} |
| {"event": "train", "step": 2450, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.33765, "bpb": 1.92981, "chunk_loss": 0.039204, "F": 0.18405, "G": 0.42482, "chunks": 188.47, "bytes_per_chunk": 5.433, "boundary_entropy": 0.66337, "router_noise": 0.0, "grad_norm": 0.6707, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.1, "n_target_mean": 10.1484, "alpha": 0.03, "t": 327.4} |
| {"event": "train", "step": 2500, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.43797, "bpb": 2.07455, "chunk_loss": 0.035871, "F": 0.18091, "G": 0.42773, "chunks": 185.25, "bytes_per_chunk": 5.528, "boundary_entropy": 0.66621, "router_noise": 0.0, "grad_norm": 0.8007, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.101, "n_target_mean": 8.8824, "alpha": 0.03, "t": 332.5} |
| {"event": "train", "step": 2550, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.60214, "bpb": 2.3114, "chunk_loss": 0.035404, "F": 0.19296, "G": 0.42895, "chunks": 197.59, "bytes_per_chunk": 5.182, "boundary_entropy": 0.66569, "router_noise": 0.0, "grad_norm": 0.7333, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.089, "n_target_mean": 7.9249, "alpha": 0.03, "t": 337.0} |
| {"event": "probe", "step": 2600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3284.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188703, "cps": 25.65625, "mean_p": 0.438198, "boundary_entropy": 0.66653, "bpb": 2.211309}, "de": {"chunks": 2967.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.144499, "cps": 23.179688, "mean_p": 0.436677, "boundary_entropy": 0.67066, "bpb": 2.463702}, "vi": {"chunks": 4764.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.258421, "cps": 37.21875, "mean_p": 0.428789, "boundary_entropy": 0.659419, "bpb": 2.050972}, "tr": {"chunks": 4456.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.246147, "cps": 34.8125, "mean_p": 0.448579, "boundary_entropy": 0.667704, "bpb": 2.452674}, "fi": {"chunks": 2970.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157886, "cps": 23.203125, "mean_p": 0.438045, "boundary_entropy": 0.671148, "bpb": 2.681042}, "ru": {"chunks": 7548.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.398311, "cps": 58.96875, "mean_p": 0.417981, "boundary_entropy": 0.656176, "bpb": 1.589376}, "el": {"chunks": 5617.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.273693, "cps": 43.882812, "mean_p": 0.429443, "boundary_entropy": 0.670847, "bpb": 1.707041}, "hi": {"chunks": 7477.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.444398, "cps": 58.414062, "mean_p": 0.366033, "boundary_entropy": 0.615759, "bpb": 1.307382}, "ta": {"chunks": 8727.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.466809, "cps": 68.179688, "mean_p": 0.354474, "boundary_entropy": 0.60965, "bpb": 1.084213}, "th": {"chunks": 7727.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.480385, "cps": 60.367188, "mean_p": 0.426273, "boundary_entropy": 0.666585, "bpb": 1.243539}, "zh": {"chunks": 3743.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.633762, "cps": 29.242188, "mean_p": 0.417378, "boundary_entropy": 0.649457, "bpb": 3.379345}, "my": {"chunks": 7710.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.406903, "cps": 60.234375, "mean_p": 0.434839, "boundary_entropy": 0.672738, "bpb": 1.22775}}, "bpb": {"en": 2.21131, "de": 2.4637, "vi": 2.05097, "tr": 2.45267, "fi": 2.68104, "ru": 1.58938, "el": 1.70704, "hi": 1.30738, "ta": 1.08421, "th": 1.24354, "zh": 3.37934, "my": 1.22775}, "gini_cpc": 0.239689, "gini_cps": 0.206618, "bpic": 5.72893, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7657484226840688, "vi": 1.3694607282987565, "tr": 1.3044144539538707, "fi": 0.836691848026568, "ru": 2.1107832972853107, "el": 1.4503891774553406, "hi": 2.355012840861994, "ta": 2.4737766062626583, "th": 2.545721079814642, "zh": 3.358515494266035, "my": 2.156313856732587}, "t": 343.3} |
| {"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.21957, "de": 2.47184, "vi": 2.05193, "tr": 2.45925, "fi": 2.68311, "ru": 1.56573, "el": 1.69819, "hi": 1.30223, "ta": 1.09295, "th": 1.26571, "zh": 3.38499, "my": 1.23578}, "bytes_per_chunk": {"en": 5.302, "de": 7.211, "vi": 5.132, "tr": 4.378, "fi": 6.355, "ru": 4.63, "el": 6.653, "hi": 5.783, "ta": 5.898, "th": 6.09, "zh": 4.56, "my": 6.839}, "bpb_mean": 1.95261, "frozen_inventory": false, "t": 345.4} |
| {"event": "train", "step": 2600, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.69898, "bpb": 2.4511, "chunk_loss": 0.034691, "F": 0.19394, "G": 0.42624, "chunks": 198.59, "bytes_per_chunk": 5.156, "boundary_entropy": 0.66465, "router_noise": 0.0, "grad_norm": 0.7568, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.177, "n_target_mean": 7.6058, "alpha": 0.03, "t": 345.8} |
| {"event": "train", "step": 2650, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.4494, "bpb": 2.09105, "chunk_loss": 0.036825, "F": 0.1954, "G": 0.42201, "chunks": 200.09, "bytes_per_chunk": 5.118, "boundary_entropy": 0.6615, "router_noise": 0.0, "grad_norm": 0.7348, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.098, "n_target_mean": 9.0748, "alpha": 0.03, "t": 350.7} |
| {"event": "train", "step": 2700, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.39366, "bpb": 2.01063, "chunk_loss": 0.037379, "F": 0.18225, "G": 0.41631, "chunks": 186.62, "bytes_per_chunk": 5.487, "boundary_entropy": 0.65825, "router_noise": 0.0, "grad_norm": 0.7629, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.104, "n_target_mean": 9.9458, "alpha": 0.03, "t": 355.9} |
| {"event": "train", "step": 2750, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.36638, "bpb": 1.97128, "chunk_loss": 0.035632, "F": 0.17661, "G": 0.42271, "chunks": 180.84, "bytes_per_chunk": 5.662, "boundary_entropy": 0.66349, "router_noise": 0.0, "grad_norm": 0.7723, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.098, "n_target_mean": 9.1192, "alpha": 0.03, "t": 360.8} |
| {"event": "probe", "step": 2800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3121.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.179337, "cps": 24.382812, "mean_p": 0.437809, "boundary_entropy": 0.666751, "bpb": 2.1947}, "de": {"chunks": 3004.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.146301, "cps": 23.46875, "mean_p": 0.437432, "boundary_entropy": 0.67121, "bpb": 2.44917}, "vi": {"chunks": 4053.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.219854, "cps": 31.664062, "mean_p": 0.42427, "boundary_entropy": 0.658671, "bpb": 2.037192}, "tr": {"chunks": 3609.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.199359, "cps": 28.195312, "mean_p": 0.443009, "boundary_entropy": 0.667441, "bpb": 2.427494}, "fi": {"chunks": 2248.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.119505, "cps": 17.5625, "mean_p": 0.432566, "boundary_entropy": 0.670272, "bpb": 2.657305}, "ru": {"chunks": 5075.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.26781, "cps": 39.648438, "mean_p": 0.4013, "boundary_entropy": 0.651038, "bpb": 1.601845}, "el": {"chunks": 3554.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.173172, "cps": 27.765625, "mean_p": 0.422162, "boundary_entropy": 0.669044, "bpb": 1.682067}, "hi": {"chunks": 7342.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.436374, "cps": 57.359375, "mean_p": 0.359085, "boundary_entropy": 0.607838, "bpb": 1.323041}, "ta": {"chunks": 8700.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.465365, "cps": 67.96875, "mean_p": 0.333285, "boundary_entropy": 0.589591, "bpb": 1.046158}, "th": {"chunks": 4298.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.267205, "cps": 33.578125, "mean_p": 0.407983, "boundary_entropy": 0.659782, "bpb": 1.22542}, "zh": {"chunks": 4389.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.743143, "cps": 34.289062, "mean_p": 0.420002, "boundary_entropy": 0.647787, "bpb": 3.359224}, "my": {"chunks": 3888.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.205193, "cps": 30.375, "mean_p": 0.429575, "boundary_entropy": 0.671, "bpb": 1.206936}}, "bpb": {"en": 2.1947, "de": 2.44917, "vi": 2.03719, "tr": 2.42749, "fi": 2.65731, "ru": 1.60185, "el": 1.68207, "hi": 1.32304, "ta": 1.04616, "th": 1.22542, "zh": 3.35922, "my": 1.20694}, "gini_cpc": 0.297886, "gini_cps": 0.203479, "bpic": 7.202962, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8157890519482138, "vi": 1.2259247508087814, "tr": 1.1116464416661802, "fi": 0.666368984408916, "ru": 1.4933347930733927, "el": 0.9656215064184365, "hi": 2.4332663979284272, "ta": 2.594920963931452, "th": 1.4899637330000617, "zh": 4.143835338491651, "my": 1.144177048204983}, "t": 367.5} |
| {"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.20244, "de": 2.45636, "vi": 2.03915, "tr": 2.43483, "fi": 2.66089, "ru": 1.58119, "el": 1.6735, "hi": 1.31969, "ta": 1.05384, "th": 1.24297, "zh": 3.36372, "my": 1.21649}, "bytes_per_chunk": {"en": 5.735, "de": 7.176, "vi": 6.1, "tr": 5.375, "fi": 8.741, "ru": 6.858, "el": 10.657, "hi": 5.9, "ta": 5.919, "th": 11.264, "zh": 3.966, "my": 13.441}, "bpb_mean": 1.93709, "frozen_inventory": false, "t": 370.1} |
| {"event": "train", "step": 2800, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.49603, "bpb": 2.15832, "chunk_loss": 0.033035, "F": 0.15576, "G": 0.41742, "chunks": 159.5, "bytes_per_chunk": 6.42, "boundary_entropy": 0.66075, "router_noise": 0.0, "grad_norm": 0.8181, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.194, "n_target_mean": 8.8073, "alpha": 0.03, "t": 370.5} |
| {"event": "train", "step": 2850, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.42408, "bpb": 2.05452, "chunk_loss": 0.036579, "F": 0.19083, "G": 0.41901, "chunks": 195.41, "bytes_per_chunk": 5.24, "boundary_entropy": 0.6595, "router_noise": 0.0, "grad_norm": 0.7495, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.097, "n_target_mean": 9.5124, "alpha": 0.03, "t": 375.3} |
| {"event": "train", "step": 2900, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.50118, "bpb": 2.16575, "chunk_loss": 0.033399, "F": 0.16306, "G": 0.42004, "chunks": 166.97, "bytes_per_chunk": 6.133, "boundary_entropy": 0.66166, "router_noise": 0.0, "grad_norm": 0.7585, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.093, "n_target_mean": 9.031, "alpha": 0.03, "t": 380.0} |
| {"event": "train", "step": 2950, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.39757, "bpb": 2.01626, "chunk_loss": 0.03634, "F": 0.17258, "G": 0.41527, "chunks": 176.72, "bytes_per_chunk": 5.795, "boundary_entropy": 0.65803, "router_noise": 0.0, "grad_norm": 0.7972, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.1, "n_target_mean": 9.8764, "alpha": 0.03, "t": 385.0} |
| {"event": "probe", "step": 3000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3219.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184968, "cps": 25.148438, "mean_p": 0.435829, "boundary_entropy": 0.665642, "bpb": 2.177223}, "de": {"chunks": 2826.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.137632, "cps": 22.078125, "mean_p": 0.435192, "boundary_entropy": 0.670381, "bpb": 2.410171}, "vi": {"chunks": 4286.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.232493, "cps": 33.484375, "mean_p": 0.421974, "boundary_entropy": 0.656264, "bpb": 2.017891}, "tr": {"chunks": 3827.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.211401, "cps": 29.898438, "mean_p": 0.442765, "boundary_entropy": 0.666625, "bpb": 2.392559}, "fi": {"chunks": 2977.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.158258, "cps": 23.257812, "mean_p": 0.438342, "boundary_entropy": 0.67111, "bpb": 2.611705}, "ru": {"chunks": 6352.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.335198, "cps": 49.625, "mean_p": 0.404631, "boundary_entropy": 0.651269, "bpb": 1.555893}, "el": {"chunks": 4852.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.236418, "cps": 37.90625, "mean_p": 0.423773, "boundary_entropy": 0.668272, "bpb": 1.647973}, "hi": {"chunks": 6787.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.403388, "cps": 53.023438, "mean_p": 0.352276, "boundary_entropy": 0.608588, "bpb": 1.269669}, "ta": {"chunks": 8043.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.430222, "cps": 62.835938, "mean_p": 0.315635, "boundary_entropy": 0.57694, "bpb": 1.043511}, "th": {"chunks": 5360.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.33323, "cps": 41.875, "mean_p": 0.402377, "boundary_entropy": 0.654361, "bpb": 1.204869}, "zh": {"chunks": 4093.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.693024, "cps": 31.976562, "mean_p": 0.418136, "boundary_entropy": 0.64817, "bpb": 3.333184}, "my": {"chunks": 2464.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.13004, "cps": 19.25, "mean_p": 0.419924, "boundary_entropy": 0.667773, "bpb": 1.200857}}, "bpb": {"en": 2.17722, "de": 2.41017, "vi": 2.01789, "tr": 2.39256, "fi": 2.6117, "ru": 1.55589, "el": 1.64797, "hi": 1.26967, "ta": 1.04351, "th": 1.20487, "zh": 3.33318, "my": 1.20086}, "gini_cpc": 0.278774, "gini_cps": 0.203645, "bpic": 6.966943, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7440855896781345, "vi": 1.256933115040351, "tr": 1.1429074254039444, "fi": 0.8555986580198386, "ru": 1.8121928752517418, "el": 1.2781537264648728, "hi": 2.180850623642036, "ta": 2.325925192930016, "th": 1.8015522731203437, "zh": 3.74672178513392, "my": 0.703040705311894}, "t": 391.6} |
| {"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.18573, "de": 2.41883, "vi": 2.0188, "tr": 2.39815, "fi": 2.61595, "ru": 1.53295, "el": 1.64201, "hi": 1.2661, "ta": 1.05167, "th": 1.22275, "zh": 3.33993, "my": 1.20969}, "bytes_per_chunk": {"en": 5.511, "de": 7.576, "vi": 5.649, "tr": 5.276, "fi": 6.47, "ru": 5.354, "el": 7.706, "hi": 6.425, "ta": 6.41, "th": 8.909, "zh": 4.233, "my": 22.285}, "bpb_mean": 1.90855, "frozen_inventory": false, "t": 394.0} |
| {"event": "train", "step": 3000, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.50112, "bpb": 2.16565, "chunk_loss": 0.032633, "F": 0.16074, "G": 0.42161, "chunks": 164.59, "bytes_per_chunk": 6.221, "boundary_entropy": 0.66373, "router_noise": 0.0, "grad_norm": 0.81, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.191, "n_target_mean": 8.834, "alpha": 0.03, "t": 394.5} |
| {"event": "train", "step": 3050, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.40342, "bpb": 2.02471, "chunk_loss": 0.034099, "F": 0.17398, "G": 0.42177, "chunks": 178.16, "bytes_per_chunk": 5.748, "boundary_entropy": 0.66214, "router_noise": 0.0, "grad_norm": 0.7784, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.1, "n_target_mean": 8.4055, "alpha": 0.03, "t": 399.5} |
| {"event": "train", "step": 3100, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.44335, "bpb": 2.08231, "chunk_loss": 0.033658, "F": 0.17474, "G": 0.42453, "chunks": 178.94, "bytes_per_chunk": 5.723, "boundary_entropy": 0.66406, "router_noise": 0.0, "grad_norm": 0.783, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.099, "n_target_mean": 8.0087, "alpha": 0.03, "t": 404.5} |
| {"event": "train", "step": 3150, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.39469, "bpb": 2.01211, "chunk_loss": 0.033996, "F": 0.15628, "G": 0.42284, "chunks": 160.03, "bytes_per_chunk": 6.399, "boundary_entropy": 0.66533, "router_noise": 0.0, "grad_norm": 0.8107, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.092, "n_target_mean": 9.8608, "alpha": 0.03, "t": 409.1} |
| {"event": "probe", "step": 3200, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 2974.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17089, "cps": 23.234375, "mean_p": 0.435272, "boundary_entropy": 0.66547, "bpb": 2.177183}, "de": {"chunks": 2689.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.13096, "cps": 21.007812, "mean_p": 0.436084, "boundary_entropy": 0.670874, "bpb": 2.396347}, "vi": {"chunks": 3850.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.208842, "cps": 30.078125, "mean_p": 0.416665, "boundary_entropy": 0.653637, "bpb": 2.013102}, "tr": {"chunks": 4174.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.23057, "cps": 32.609375, "mean_p": 0.447082, "boundary_entropy": 0.668269, "bpb": 2.384507}, "fi": {"chunks": 2374.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.126203, "cps": 18.546875, "mean_p": 0.432704, "boundary_entropy": 0.669929, "bpb": 2.595183}, "ru": {"chunks": 6254.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.330026, "cps": 48.859375, "mean_p": 0.400535, "boundary_entropy": 0.648694, "bpb": 1.53708}, "el": {"chunks": 4407.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.214735, "cps": 34.429688, "mean_p": 0.420411, "boundary_entropy": 0.667284, "bpb": 1.629364}, "hi": {"chunks": 6711.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.398871, "cps": 52.429688, "mean_p": 0.349966, "boundary_entropy": 0.605487, "bpb": 1.265976}, "ta": {"chunks": 8026.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.429313, "cps": 62.703125, "mean_p": 0.307614, "boundary_entropy": 0.567108, "bpb": 1.0404}, "th": {"chunks": 5166.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.321169, "cps": 40.359375, "mean_p": 0.402062, "boundary_entropy": 0.654545, "bpb": 1.194989}, "zh": {"chunks": 3877.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.656451, "cps": 30.289062, "mean_p": 0.417421, "boundary_entropy": 0.647952, "bpb": 3.325342}, "my": {"chunks": 2608.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.13764, "cps": 20.375, "mean_p": 0.426591, "boundary_entropy": 0.670676, "bpb": 1.179852}}, "bpb": {"en": 2.17718, "de": 2.39635, "vi": 2.0131, "tr": 2.38451, "fi": 2.59518, "ru": 1.53708, "el": 1.62936, "hi": 1.26598, "ta": 1.0404, "th": 1.19499, "zh": 3.32534, "my": 1.17985}, "gini_cpc": 0.285463, "gini_cps": 0.214934, "bpic": 7.226153, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7663400995612194, "vi": 1.2220831146858342, "tr": 1.3492270802942188, "fi": 0.7385025295155431, "ru": 1.9312203033147437, "el": 1.2565661646662047, "hi": 2.334077767596535, "ta": 2.5122152170914083, "th": 1.8793881863728732, "zh": 3.8413644633442425, "my": 0.8054291936065704}, "t": 415.7} |
| {"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.18361, "de": 2.40374, "vi": 2.01686, "tr": 2.39127, "fi": 2.59889, "ru": 1.51416, "el": 1.62351, "hi": 1.26185, "ta": 1.04901, "th": 1.21345, "zh": 3.32701, "my": 1.18852}, "bytes_per_chunk": {"en": 5.975, "de": 7.877, "vi": 6.323, "tr": 4.678, "fi": 8.125, "ru": 5.449, "el": 8.479, "hi": 6.451, "ta": 6.439, "th": 9.101, "zh": 4.503, "my": 19.99}, "bpb_mean": 1.89766, "frozen_inventory": false, "t": 418.1} |
| {"event": "train", "step": 3200, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.3434, "bpb": 1.93812, "chunk_loss": 0.034823, "F": 0.16241, "G": 0.41074, "chunks": 166.31, "bytes_per_chunk": 6.157, "boundary_entropy": 0.65453, "router_noise": 0.0, "grad_norm": 0.9105, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.188, "n_target_mean": 9.495, "alpha": 0.03, "t": 418.5} |
| {"event": "train", "step": 3250, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.57821, "bpb": 2.27688, "chunk_loss": 0.033514, "F": 0.18518, "G": 0.42151, "chunks": 189.62, "bytes_per_chunk": 5.4, "boundary_entropy": 0.66109, "router_noise": 0.0, "grad_norm": 0.7859, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.099, "n_target_mean": 7.9529, "alpha": 0.03, "t": 423.4} |
| {"event": "train", "step": 3300, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.52747, "bpb": 2.20367, "chunk_loss": 0.032537, "F": 0.16788, "G": 0.42541, "chunks": 171.91, "bytes_per_chunk": 5.957, "boundary_entropy": 0.66527, "router_noise": 0.0, "grad_norm": 0.8688, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.096, "n_target_mean": 7.6934, "alpha": 0.03, "t": 428.2} |
| {"event": "train", "step": 3350, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.36783, "bpb": 1.97337, "chunk_loss": 0.035537, "F": 0.16776, "G": 0.41025, "chunks": 171.78, "bytes_per_chunk": 5.961, "boundary_entropy": 0.65469, "router_noise": 0.0, "grad_norm": 0.7402, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.097, "n_target_mean": 9.8711, "alpha": 0.03, "t": 433.1} |
| {"event": "probe", "step": 3400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3004.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.172614, "cps": 23.46875, "mean_p": 0.434424, "boundary_entropy": 0.665158, "bpb": 2.149233}, "de": {"chunks": 2459.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.119758, "cps": 19.210938, "mean_p": 0.433258, "boundary_entropy": 0.670228, "bpb": 2.379365}, "vi": {"chunks": 4184.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.22696, "cps": 32.6875, "mean_p": 0.417005, "boundary_entropy": 0.653469, "bpb": 1.989463}, "tr": {"chunks": 4205.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.232282, "cps": 32.851562, "mean_p": 0.446176, "boundary_entropy": 0.66777, "bpb": 2.353596}, "fi": {"chunks": 3011.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.160066, "cps": 23.523438, "mean_p": 0.438604, "boundary_entropy": 0.671323, "bpb": 2.566507}, "ru": {"chunks": 5510.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.290765, "cps": 43.046875, "mean_p": 0.396379, "boundary_entropy": 0.647276, "bpb": 1.531913}, "el": {"chunks": 4519.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.220192, "cps": 35.304688, "mean_p": 0.420143, "boundary_entropy": 0.666644, "bpb": 1.610812}, "hi": {"chunks": 6648.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.395126, "cps": 51.9375, "mean_p": 0.346511, "boundary_entropy": 0.602798, "bpb": 1.249669}, "ta": {"chunks": 7923.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.423803, "cps": 61.898438, "mean_p": 0.296162, "boundary_entropy": 0.556003, "bpb": 1.028238}, "th": {"chunks": 4840.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.300901, "cps": 37.8125, "mean_p": 0.402485, "boundary_entropy": 0.655557, "bpb": 1.182357}, "zh": {"chunks": 3912.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.662377, "cps": 30.5625, "mean_p": 0.419318, "boundary_entropy": 0.650189, "bpb": 3.301086}, "my": {"chunks": 3399.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.179386, "cps": 26.554688, "mean_p": 0.426759, "boundary_entropy": 0.669307, "bpb": 1.171006}}, "bpb": {"en": 2.14923, "de": 2.37936, "vi": 1.98946, "tr": 2.3536, "fi": 2.56651, "ru": 1.53191, "el": 1.61081, "hi": 1.24967, "ta": 1.02824, "th": 1.18236, "zh": 3.30109, "my": 1.17101}, "gini_cpc": 0.264084, "gini_cps": 0.185648, "bpic": 7.158224, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.69379363854736, "vi": 1.3148394492182378, "tr": 1.3456733157278078, "fi": 0.9273059874237111, "ru": 1.6844827828506586, "el": 1.2756328306398856, "hi": 2.2890755664067526, "ta": 2.455208496187129, "th": 1.7432051017253136, "zh": 3.8373339441755725, "my": 1.0392307301474328}, "t": 439.3} |
| {"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 2.1555, "de": 2.38836, "vi": 1.99225, "tr": 2.36013, "fi": 2.57244, "ru": 1.50896, "el": 1.60322, "hi": 1.24583, "ta": 1.03513, "th": 1.20011, "zh": 3.30298, "my": 1.1801}, "bytes_per_chunk": {"en": 5.811, "de": 8.655, "vi": 5.886, "tr": 4.627, "fi": 6.518, "ru": 6.29, "el": 7.935, "hi": 6.504, "ta": 6.504, "th": 9.836, "zh": 4.611, "my": 16.775}, "bpb_mean": 1.87875, "frozen_inventory": false, "t": 441.8} |
| {"event": "train", "step": 3400, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.3936, "bpb": 2.01054, "chunk_loss": 0.033091, "F": 0.15625, "G": 0.41504, "chunks": 160.0, "bytes_per_chunk": 6.4, "boundary_entropy": 0.65993, "router_noise": 0.0, "grad_norm": 0.7502, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.18, "n_target_mean": 8.9964, "alpha": 0.03, "t": 442.1} |
| {"event": "train", "step": 3450, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.39641, "bpb": 2.01459, "chunk_loss": 0.034796, "F": 0.16486, "G": 0.41679, "chunks": 168.81, "bytes_per_chunk": 6.066, "boundary_entropy": 0.66014, "router_noise": 0.0, "grad_norm": 0.8106, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.093, "n_target_mean": 9.522, "alpha": 0.03, "t": 446.7} |
| {"event": "train", "step": 3500, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.46838, "bpb": 2.11842, "chunk_loss": 0.036438, "F": 0.17941, "G": 0.41274, "chunks": 183.72, "bytes_per_chunk": 5.574, "boundary_entropy": 0.65551, "router_noise": 0.0, "grad_norm": 0.7733, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.1, "n_target_mean": 9.4057, "alpha": 0.03, "t": 451.7} |
| {"event": "train", "step": 3550, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.41825, "bpb": 2.0461, "chunk_loss": 0.034779, "F": 0.17889, "G": 0.41462, "chunks": 183.19, "bytes_per_chunk": 5.59, "boundary_entropy": 0.65589, "router_noise": 0.0, "grad_norm": 0.7462, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.097, "n_target_mean": 8.9203, "alpha": 0.03, "t": 456.6} |
| {"event": "probe", "step": 3600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 2938.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.168821, "cps": 22.953125, "mean_p": 0.435439, "boundary_entropy": 0.665903, "bpb": 2.136492}, "de": {"chunks": 2667.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.129888, "cps": 20.835938, "mean_p": 0.435117, "boundary_entropy": 0.670748, "bpb": 2.358285}, "vi": {"chunks": 3975.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.215622, "cps": 31.054688, "mean_p": 0.414286, "boundary_entropy": 0.651093, "bpb": 1.982682}, "tr": {"chunks": 3896.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.215213, "cps": 30.4375, "mean_p": 0.442862, "boundary_entropy": 0.667438, "bpb": 2.341545}, "fi": {"chunks": 2851.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.15156, "cps": 22.273438, "mean_p": 0.437396, "boundary_entropy": 0.671039, "bpb": 2.54366}, "ru": {"chunks": 5875.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.310026, "cps": 45.898438, "mean_p": 0.398476, "boundary_entropy": 0.647951, "bpb": 1.517368}, "el": {"chunks": 4636.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.225893, "cps": 36.21875, "mean_p": 0.420545, "boundary_entropy": 0.667054, "bpb": 1.59437}, "hi": {"chunks": 6417.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.381397, "cps": 50.132812, "mean_p": 0.335682, "boundary_entropy": 0.595012, "bpb": 1.23671}, "ta": {"chunks": 8161.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.436534, "cps": 63.757812, "mean_p": 0.28855, "boundary_entropy": 0.543146, "bpb": 0.998537}, "th": {"chunks": 4331.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.269257, "cps": 33.835938, "mean_p": 0.393941, "boundary_entropy": 0.650547, "bpb": 1.165688}, "zh": {"chunks": 4061.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.687606, "cps": 31.726562, "mean_p": 0.415367, "boundary_entropy": 0.645835, "bpb": 3.267564}, "my": {"chunks": 4158.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.219443, "cps": 32.484375, "mean_p": 0.427001, "boundary_entropy": 0.669699, "bpb": 1.172765}}, "bpb": {"en": 2.13649, "de": 2.35829, "vi": 1.98268, "tr": 2.34155, "fi": 2.54366, "ru": 1.51737, "el": 1.59437, "hi": 1.23671, "ta": 0.99854, "th": 1.16569, "zh": 3.26756, "my": 1.17276}, "gini_cpc": 0.264543, "gini_cps": 0.182139, "bpic": 7.111533, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7693836221802642, "vi": 1.2772217917238329, "tr": 1.2747960981310071, "fi": 0.8977546486476725, "ru": 1.8364156508026028, "el": 1.338057913539167, "hi": 2.2591719913375687, "ta": 2.5857720518421994, "th": 1.5949219947720723, "zh": 4.072976230515885, "my": 1.2998505960851905}, "t": 463.0} |
| {"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 2.14472, "de": 2.36762, "vi": 1.98529, "tr": 2.34917, "fi": 2.55, "ru": 1.49429, "el": 1.58703, "hi": 1.23281, "ta": 1.0063, "th": 1.18398, "zh": 3.27116, "my": 1.18162}, "bytes_per_chunk": {"en": 6.144, "de": 8.078, "vi": 6.336, "tr": 5.078, "fi": 6.717, "ru": 5.912, "el": 7.969, "hi": 6.752, "ta": 6.327, "th": 11.139, "zh": 4.38, "my": 13.216}, "bpb_mean": 1.86283, "frozen_inventory": false, "t": 465.3} |
| {"event": "train", "step": 3600, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.51119, "bpb": 2.18018, "chunk_loss": 0.034025, "F": 0.15491, "G": 0.41544, "chunks": 158.62, "bytes_per_chunk": 6.455, "boundary_entropy": 0.66022, "router_noise": 0.0, "grad_norm": 2.1931, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.181, "n_target_mean": 9.551, "alpha": 0.03, "t": 465.6} |
| {"event": "train", "step": 3650, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.35815, "bpb": 1.95939, "chunk_loss": 0.033774, "F": 0.16736, "G": 0.41469, "chunks": 171.38, "bytes_per_chunk": 5.975, "boundary_entropy": 0.65784, "router_noise": 0.0, "grad_norm": 0.7477, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.099, "n_target_mean": 8.7039, "alpha": 0.03, "t": 470.6} |
| {"event": "train", "step": 3700, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.60669, "bpb": 2.31797, "chunk_loss": 0.032311, "F": 0.17322, "G": 0.41655, "chunks": 177.38, "bytes_per_chunk": 5.773, "boundary_entropy": 0.65818, "router_noise": 0.0, "grad_norm": 0.8253, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.085, "n_target_mean": 7.5805, "alpha": 0.03, "t": 474.8} |
| {"event": "train", "step": 3750, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.3289, "bpb": 1.9172, "chunk_loss": 0.03416, "F": 0.15326, "G": 0.41091, "chunks": 156.94, "bytes_per_chunk": 6.525, "boundary_entropy": 0.65537, "router_noise": 0.0, "grad_norm": 0.7662, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.094, "n_target_mean": 10.2592, "alpha": 0.03, "t": 479.5} |
| {"event": "probe", "step": 3800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3196.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.183646, "cps": 24.96875, "mean_p": 0.437576, "boundary_entropy": 0.666036, "bpb": 2.121975}, "de": {"chunks": 3035.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.147811, "cps": 23.710938, "mean_p": 0.436808, "boundary_entropy": 0.670828, "bpb": 2.334127}, "vi": {"chunks": 3983.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.216056, "cps": 31.117188, "mean_p": 0.414925, "boundary_entropy": 0.651656, "bpb": 1.975174}, "tr": {"chunks": 3832.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.211678, "cps": 29.9375, "mean_p": 0.441533, "boundary_entropy": 0.666294, "bpb": 2.32096}, "fi": {"chunks": 2402.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.127691, "cps": 18.765625, "mean_p": 0.433592, "boundary_entropy": 0.670197, "bpb": 2.534873}, "ru": {"chunks": 5226.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.275778, "cps": 40.828125, "mean_p": 0.394967, "boundary_entropy": 0.64634, "bpb": 1.500684}, "el": {"chunks": 4206.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.204941, "cps": 32.859375, "mean_p": 0.418982, "boundary_entropy": 0.666604, "bpb": 1.579857}, "hi": {"chunks": 6045.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.359287, "cps": 47.226562, "mean_p": 0.333727, "boundary_entropy": 0.595755, "bpb": 1.229993}, "ta": {"chunks": 7657.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.409575, "cps": 59.820312, "mean_p": 0.282799, "boundary_entropy": 0.541233, "bpb": 0.992131}, "th": {"chunks": 5144.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.319801, "cps": 40.1875, "mean_p": 0.398671, "boundary_entropy": 0.652727, "bpb": 1.156718}, "zh": {"chunks": 4107.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.695395, "cps": 32.085938, "mean_p": 0.417131, "boundary_entropy": 0.646683, "bpb": 3.267806}, "my": {"chunks": 2714.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.143234, "cps": 21.203125, "mean_p": 0.425707, "boundary_entropy": 0.671209, "bpb": 1.156346}}, "bpb": {"en": 2.12197, "de": 2.33413, "vi": 1.97517, "tr": 2.32096, "fi": 2.53487, "ru": 1.50068, "el": 1.57986, "hi": 1.22999, "ta": 0.99213, "th": 1.15672, "zh": 3.26781, "my": 1.15635}, "gini_cpc": 0.278709, "gini_cps": 0.184471, "bpic": 7.445264, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8048661036932703, "vi": 1.17647990622941, "tr": 1.1526363154738872, "fi": 0.6953099899141643, "ru": 1.5016804977197749, "el": 1.1159526626526104, "hi": 1.9564041790722801, "ta": 2.2302344867430905, "th": 1.7413948038093934, "zh": 3.78659284357733, "my": 0.7799447100198871}, "t": 486.2} |
| {"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 2.13058, "de": 2.34226, "vi": 1.97632, "tr": 2.3294, "fi": 2.54034, "ru": 1.47855, "el": 1.5716, "hi": 1.22649, "ta": 1.00089, "th": 1.17556, "zh": 3.26795, "my": 1.16514}, "bytes_per_chunk": {"en": 5.449, "de": 6.921, "vi": 6.274, "tr": 5.101, "fi": 8.182, "ru": 6.584, "el": 8.975, "hi": 7.067, "ta": 6.736, "th": 9.189, "zh": 4.313, "my": 20.453}, "bpb_mean": 1.85042, "frozen_inventory": false, "t": 488.6} |
| {"event": "train", "step": 3800, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.39, "bpb": 2.00535, "chunk_loss": 0.033092, "F": 0.16779, "G": 0.41478, "chunks": 171.81, "bytes_per_chunk": 5.96, "boundary_entropy": 0.65742, "router_noise": 0.0, "grad_norm": 0.7996, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.187, "n_target_mean": 8.2652, "alpha": 0.03, "t": 488.9} |
| {"event": "train", "step": 3850, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.49815, "bpb": 2.16137, "chunk_loss": 0.033641, "F": 0.16934, "G": 0.41039, "chunks": 173.41, "bytes_per_chunk": 5.905, "boundary_entropy": 0.65389, "router_noise": 0.0, "grad_norm": 0.7203, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.095, "n_target_mean": 9.011, "alpha": 0.03, "t": 493.7} |
| {"event": "train", "step": 3900, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.52147, "bpb": 2.19502, "chunk_loss": 0.033958, "F": 0.17416, "G": 0.40941, "chunks": 178.34, "bytes_per_chunk": 5.742, "boundary_entropy": 0.65349, "router_noise": 0.0, "grad_norm": 0.778, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.099, "n_target_mean": 8.7756, "alpha": 0.03, "t": 498.6} |
| {"event": "train", "step": 3950, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.44425, "bpb": 2.08361, "chunk_loss": 0.033493, "F": 0.16943, "G": 0.40981, "chunks": 173.5, "bytes_per_chunk": 5.902, "boundary_entropy": 0.65344, "router_noise": 0.0, "grad_norm": 0.7921, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.102, "n_target_mean": 8.8445, "alpha": 0.03, "t": 503.7} |
| {"event": "probe", "step": 4000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3538.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203298, "cps": 27.640625, "mean_p": 0.438459, "boundary_entropy": 0.666054, "bpb": 2.106615}, "de": {"chunks": 3452.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.16812, "cps": 26.96875, "mean_p": 0.440542, "boundary_entropy": 0.672001, "bpb": 2.321631}, "vi": {"chunks": 3988.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.216328, "cps": 31.15625, "mean_p": 0.415126, "boundary_entropy": 0.652083, "bpb": 1.961371}, "tr": {"chunks": 4484.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.247694, "cps": 35.03125, "mean_p": 0.446313, "boundary_entropy": 0.66727, "bpb": 2.294744}, "fi": {"chunks": 2742.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.145766, "cps": 21.421875, "mean_p": 0.437696, "boundary_entropy": 0.671382, "bpb": 2.499271}, "ru": {"chunks": 5538.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.292243, "cps": 43.265625, "mean_p": 0.396599, "boundary_entropy": 0.646681, "bpb": 1.479576}, "el": {"chunks": 4668.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.227452, "cps": 36.46875, "mean_p": 0.4227, "boundary_entropy": 0.667833, "bpb": 1.560411}, "hi": {"chunks": 5933.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.35263, "cps": 46.351562, "mean_p": 0.333581, "boundary_entropy": 0.595179, "bpb": 1.22667}, "ta": {"chunks": 7573.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.405082, "cps": 59.164062, "mean_p": 0.277837, "boundary_entropy": 0.535096, "bpb": 0.988831}, "th": {"chunks": 4528.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.281505, "cps": 35.375, "mean_p": 0.394325, "boundary_entropy": 0.650045, "bpb": 1.149802}, "zh": {"chunks": 4182.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.708093, "cps": 32.671875, "mean_p": 0.419746, "boundary_entropy": 0.647981, "bpb": 3.239442}, "my": {"chunks": 3172.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.167406, "cps": 24.78125, "mean_p": 0.428637, "boundary_entropy": 0.672318, "bpb": 1.149928}}, "bpb": {"en": 2.10661, "de": 2.32163, "vi": 1.96137, "tr": 2.29474, "fi": 2.49927, "ru": 1.47958, "el": 1.56041, "hi": 1.22667, "ta": 0.98883, "th": 1.1498, "zh": 3.23944, "my": 1.14993}, "gini_cpc": 0.25073, "gini_cps": 0.155154, "bpic": 7.133741, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8269603203512762, "vi": 1.0640898467918465, "tr": 1.21837602411201, "fi": 0.7170044624972718, "ru": 1.4375072003770595, "el": 1.1188098824556392, "hi": 1.7345449826422865, "ta": 1.9925479390005674, "th": 1.384687094581095, "zh": 3.483027291903282, "my": 0.8234478390950187}, "t": 509.7} |
| {"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 2.11597, "de": 2.3297, "vi": 1.96097, "tr": 2.30299, "fi": 2.50488, "ru": 1.45759, "el": 1.55351, "hi": 1.22222, "ta": 0.99681, "th": 1.16783, "zh": 3.23922, "my": 1.15782}, "bytes_per_chunk": {"en": 4.99, "de": 6.185, "vi": 6.323, "tr": 4.445, "fi": 6.979, "ru": 6.297, "el": 8.202, "hi": 7.272, "ta": 6.8, "th": 10.43, "zh": 4.27, "my": 17.139}, "bpb_mean": 1.83413, "frozen_inventory": false, "t": 512.0} |
| {"event": "train", "step": 4000, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.42975, "bpb": 2.06269, "chunk_loss": 0.034115, "F": 0.17737, "G": 0.41407, "chunks": 181.62, "bytes_per_chunk": 5.638, "boundary_entropy": 0.65736, "router_noise": 0.0, "grad_norm": 0.805, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.175, "n_target_mean": 8.8695, "alpha": 0.03, "t": 512.5} |
| {"event": "train", "step": 4050, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.32306, "bpb": 1.90878, "chunk_loss": 0.034725, "F": 0.16986, "G": 0.41038, "chunks": 173.94, "bytes_per_chunk": 5.887, "boundary_entropy": 0.65408, "router_noise": 0.0, "grad_norm": 0.7735, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.098, "n_target_mean": 9.3813, "alpha": 0.03, "t": 517.4} |
| {"event": "train", "step": 4100, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.47379, "bpb": 2.12623, "chunk_loss": 0.0332, "F": 0.16931, "G": 0.41378, "chunks": 173.38, "bytes_per_chunk": 5.906, "boundary_entropy": 0.65647, "router_noise": 0.0, "grad_norm": 0.7785, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.092, "n_target_mean": 8.3701, "alpha": 0.03, "t": 522.0} |
| {"event": "train", "step": 4150, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.26498, "bpb": 1.82498, "chunk_loss": 0.034771, "F": 0.16391, "G": 0.40414, "chunks": 167.84, "bytes_per_chunk": 6.101, "boundary_entropy": 0.65057, "router_noise": 0.0, "grad_norm": 0.7721, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.099, "n_target_mean": 10.0005, "alpha": 0.03, "t": 526.9} |
| {"event": "probe", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3137.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.180256, "cps": 24.507812, "mean_p": 0.434204, "boundary_entropy": 0.664347, "bpb": 2.093839}, "de": {"chunks": 3245.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.158038, "cps": 25.351562, "mean_p": 0.438216, "boundary_entropy": 0.670988, "bpb": 2.316134}, "vi": {"chunks": 4009.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.217467, "cps": 31.320312, "mean_p": 0.414882, "boundary_entropy": 0.651871, "bpb": 1.95415}, "tr": {"chunks": 4276.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.236204, "cps": 33.40625, "mean_p": 0.444754, "boundary_entropy": 0.667021, "bpb": 2.284916}, "fi": {"chunks": 3083.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.163893, "cps": 24.085938, "mean_p": 0.438952, "boundary_entropy": 0.671407, "bpb": 2.480563}, "ru": {"chunks": 5166.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.272612, "cps": 40.359375, "mean_p": 0.391901, "boundary_entropy": 0.644988, "bpb": 1.470221}, "el": {"chunks": 3657.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.17819, "cps": 28.570312, "mean_p": 0.415977, "boundary_entropy": 0.665868, "bpb": 1.553509}, "hi": {"chunks": 5698.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.338663, "cps": 44.515625, "mean_p": 0.323821, "boundary_entropy": 0.587579, "bpb": 1.21763}, "ta": {"chunks": 7255.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.388072, "cps": 56.679688, "mean_p": 0.271892, "boundary_entropy": 0.530017, "bpb": 0.983415}, "th": {"chunks": 4514.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.280634, "cps": 35.265625, "mean_p": 0.393693, "boundary_entropy": 0.650017, "bpb": 1.139271}, "zh": {"chunks": 3462.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.586184, "cps": 27.046875, "mean_p": 0.413161, "boundary_entropy": 0.645999, "bpb": 3.220986}, "my": {"chunks": 2984.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.157484, "cps": 23.3125, "mean_p": 0.424302, "boundary_entropy": 0.670426, "bpb": 1.140819}}, "bpb": {"en": 2.09384, "de": 2.31613, "vi": 1.95415, "tr": 2.28492, "fi": 2.48056, "ru": 1.47022, "el": 1.55351, "hi": 1.21763, "ta": 0.98341, "th": 1.13927, "zh": 3.22099, "my": 1.14082}, "gini_cpc": 0.232696, "gini_cps": 0.154924, "bpic": 7.601731, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8767421689811595, "vi": 1.2064310768374422, "tr": 1.3103784632233457, "fi": 0.9092247367048496, "ru": 1.512358630457986, "el": 0.9885387925014923, "hi": 1.8787845212274907, "ta": 2.1528885538478484, "th": 1.5568619014640603, "zh": 3.2519452292698237, "my": 0.8736652142410256}, "t": 533.5} |
| {"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 2.10454, "de": 2.32477, "vi": 1.95446, "tr": 2.29131, "fi": 2.48625, "ru": 1.44717, "el": 1.54591, "hi": 1.21389, "ta": 0.99288, "th": 1.15705, "zh": 3.22275, "my": 1.15}, "bytes_per_chunk": {"en": 5.655, "de": 6.664, "vi": 6.375, "tr": 4.641, "fi": 6.341, "ru": 6.862, "el": 10.205, "hi": 7.519, "ta": 7.154, "th": 10.521, "zh": 5.178, "my": 18.559}, "bpb_mean": 1.82425, "frozen_inventory": false, "t": 535.7} |
| {"event": "phase_transition", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 131584, "t": 535.9} |
| {"event": "train", "step": 4200, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.27804, "bpb": 1.84383, "chunk_loss": 0.0, "F": 0.15097, "G": 0.40915, "chunks": 154.59, "bytes_per_chunk": 6.624, "boundary_entropy": 0.65403, "router_noise": 0.0, "grad_norm": 0.7779, "lr": 9.12e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 536.0} |
| {"event": "train", "step": 4250, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.30362, "bpb": 1.88073, "chunk_loss": 0.0, "F": 0.16602, "G": 0.40913, "chunks": 170.0, "bytes_per_chunk": 6.024, "boundary_entropy": 0.65228, "router_noise": 0.0, "grad_norm": 0.7659, "lr": 8.81e-05, "router_trainable": false, "sec_per_step": 0.097, "alpha": 0.0, "t": 540.9} |
| {"event": "train", "step": 4300, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.38116, "bpb": 1.99259, "chunk_loss": 0.0, "F": 0.16888, "G": 0.41488, "chunks": 172.94, "bytes_per_chunk": 5.921, "boundary_entropy": 0.65766, "router_noise": 0.0, "grad_norm": 0.7749, "lr": 8.5e-05, "router_trainable": false, "sec_per_step": 0.097, "alpha": 0.0, "t": 545.7} |
| {"event": "train", "step": 4350, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.35338, "bpb": 1.95252, "chunk_loss": 0.0, "F": 0.16733, "G": 0.40877, "chunks": 171.34, "bytes_per_chunk": 5.976, "boundary_entropy": 0.65157, "router_noise": 0.0, "grad_norm": 0.7497, "lr": 8.21e-05, "router_trainable": false, "sec_per_step": 0.094, "alpha": 0.0, "t": 550.4} |
| {"event": "probe", "step": 4400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3527.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.202666, "cps": 27.554688, "mean_p": 0.439542, "boundary_entropy": 0.666479, "bpb": 2.086102}, "de": {"chunks": 3000.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.146106, "cps": 23.4375, "mean_p": 0.438711, "boundary_entropy": 0.671798, "bpb": 2.299748}, "vi": {"chunks": 4560.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.247356, "cps": 35.625, "mean_p": 0.42071, "boundary_entropy": 0.653977, "bpb": 1.945252}, "tr": {"chunks": 4634.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.25598, "cps": 36.203125, "mean_p": 0.448364, "boundary_entropy": 0.66818, "bpb": 2.269316}, "fi": {"chunks": 3356.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.178406, "cps": 26.21875, "mean_p": 0.442167, "boundary_entropy": 0.672356, "bpb": 2.461648}, "ru": {"chunks": 6003.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.316781, "cps": 46.898438, "mean_p": 0.400137, "boundary_entropy": 0.648314, "bpb": 1.449898}, "el": {"chunks": 4925.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.239975, "cps": 38.476562, "mean_p": 0.425999, "boundary_entropy": 0.669203, "bpb": 1.541102}, "hi": {"chunks": 6228.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.370163, "cps": 48.65625, "mean_p": 0.340079, "boundary_entropy": 0.599475, "bpb": 1.191934}, "ta": {"chunks": 8208.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.439048, "cps": 64.125, "mean_p": 0.289063, "boundary_entropy": 0.545643, "bpb": 0.951693}, "th": {"chunks": 4985.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.309916, "cps": 38.945312, "mean_p": 0.402456, "boundary_entropy": 0.655016, "bpb": 1.132586}, "zh": {"chunks": 4319.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.73129, "cps": 33.742188, "mean_p": 0.419192, "boundary_entropy": 0.647225, "bpb": 3.203453}, "my": {"chunks": 4298.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.226831, "cps": 33.578125, "mean_p": 0.432936, "boundary_entropy": 0.672196, "bpb": 1.132506}}, "bpb": {"en": 2.0861, "de": 2.29975, "vi": 1.94525, "tr": 2.26932, "fi": 2.46165, "ru": 1.4499, "el": 1.5411, "hi": 1.19193, "ta": 0.95169, "th": 1.13259, "zh": 3.20345, "my": 1.13251}, "gini_cpc": 0.241584, "gini_cps": 0.151892, "bpic": 6.612012, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7209207205241445, "vi": 1.2205072435516047, "tr": 1.2630604563894776, "fi": 0.8802960016050307, "ru": 1.56306770312396, "el": 1.1840881918826376, "hi": 1.8264685207006364, "ta": 2.1663595540397376, "th": 1.529194607715127, "zh": 3.6083480529620515, "my": 1.1192360642132269}, "t": 556.9} |
| {"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 2.09608, "de": 2.30846, "vi": 1.94493, "tr": 2.27686, "fi": 2.46781, "ru": 1.42687, "el": 1.53324, "hi": 1.18707, "ta": 0.96039, "th": 1.14942, "zh": 3.20556, "my": 1.14127}, "bytes_per_chunk": {"en": 4.97, "de": 6.965, "vi": 5.477, "tr": 4.247, "fi": 5.695, "ru": 5.831, "el": 7.679, "hi": 6.867, "ta": 6.266, "th": 9.624, "zh": 4.098, "my": 12.141}, "bpb_mean": 1.80816, "frozen_inventory": false, "t": 559.0} |
| {"event": "train", "step": 4400, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.36371, "bpb": 1.96742, "chunk_loss": 0.0, "F": 0.17972, "G": 0.41658, "chunks": 184.03, "bytes_per_chunk": 5.564, "boundary_entropy": 0.65832, "router_noise": 0.0, "grad_norm": 0.8158, "lr": 7.92e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 559.4} |
| {"event": "train", "step": 4450, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.19507, "bpb": 1.72412, "chunk_loss": 0.0, "F": 0.17126, "G": 0.4157, "chunks": 175.38, "bytes_per_chunk": 5.839, "boundary_entropy": 0.65777, "router_noise": 0.0, "grad_norm": 1.004, "lr": 7.63e-05, "router_trainable": false, "sec_per_step": 0.091, "alpha": 0.0, "t": 564.0} |
| {"event": "train", "step": 4500, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.31642, "bpb": 1.8992, "chunk_loss": 0.0, "F": 0.1651, "G": 0.41082, "chunks": 169.06, "bytes_per_chunk": 6.057, "boundary_entropy": 0.65373, "router_noise": 0.0, "grad_norm": 0.7918, "lr": 7.36e-05, "router_trainable": false, "sec_per_step": 0.093, "alpha": 0.0, "t": 568.6} |
| {"event": "train", "step": 4550, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.327, "bpb": 1.91446, "chunk_loss": 0.0, "F": 0.18307, "G": 0.42031, "chunks": 187.47, "bytes_per_chunk": 5.462, "boundary_entropy": 0.66057, "router_noise": 0.0, "grad_norm": 0.7568, "lr": 7.09e-05, "router_trainable": false, "sec_per_step": 0.101, "alpha": 0.0, "t": 573.7} |
| {"event": "probe", "step": 4600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3536.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203183, "cps": 27.625, "mean_p": 0.440616, "boundary_entropy": 0.666746, "bpb": 2.077623}, "de": {"chunks": 3426.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.166853, "cps": 26.765625, "mean_p": 0.441792, "boundary_entropy": 0.672443, "bpb": 2.299834}, "vi": {"chunks": 4442.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.240955, "cps": 34.703125, "mean_p": 0.419762, "boundary_entropy": 0.653586, "bpb": 1.936672}, "tr": {"chunks": 4450.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.245816, "cps": 34.765625, "mean_p": 0.447502, "boundary_entropy": 0.66805, "bpb": 2.25851}, "fi": {"chunks": 3195.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.169847, "cps": 24.960938, "mean_p": 0.441106, "boundary_entropy": 0.672212, "bpb": 2.453355}, "ru": {"chunks": 6408.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.338153, "cps": 50.0625, "mean_p": 0.404578, "boundary_entropy": 0.649814, "bpb": 1.434183}, "el": {"chunks": 5039.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.245529, "cps": 39.367188, "mean_p": 0.425263, "boundary_entropy": 0.668842, "bpb": 1.52491}, "hi": {"chunks": 6190.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.367905, "cps": 48.359375, "mean_p": 0.344136, "boundary_entropy": 0.603193, "bpb": 1.182658}, "ta": {"chunks": 8055.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.430864, "cps": 62.929688, "mean_p": 0.290176, "boundary_entropy": 0.547145, "bpb": 0.949802}, "th": {"chunks": 5177.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.321853, "cps": 40.445312, "mean_p": 0.403961, "boundary_entropy": 0.655969, "bpb": 1.124827}, "zh": {"chunks": 3918.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.663393, "cps": 30.609375, "mean_p": 0.41851, "boundary_entropy": 0.648673, "bpb": 3.191909}, "my": {"chunks": 5198.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.27433, "cps": 40.609375, "mean_p": 0.433928, "boundary_entropy": 0.673313, "bpb": 1.126943}}, "bpb": {"en": 2.07762, "de": 2.29983, "vi": 1.93667, "tr": 2.25851, "fi": 2.45335, "ru": 1.43418, "el": 1.52491, "hi": 1.18266, "ta": 0.9498, "th": 1.12483, "zh": 3.19191, "my": 1.12694}, "gini_cpc": 0.220506, "gini_cps": 0.152553, "bpic": 6.501016, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8211959811741082, "vi": 1.1858978347551075, "tr": 1.209821570285468, "fi": 0.8359317927004777, "ru": 1.6642752420635394, "el": 1.2084129675280995, "hi": 1.8107039123799022, "ta": 2.120566709103869, "th": 1.584050283631781, "zh": 3.2649974985174977, "my": 1.3501585250702806}, "t": 580.3} |
| {"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 2.08647, "de": 2.30643, "vi": 1.93572, "tr": 2.26497, "fi": 2.46002, "ru": 1.41111, "el": 1.51911, "hi": 1.17802, "ta": 0.95781, "th": 1.14248, "zh": 3.19402, "my": 1.13539}, "bytes_per_chunk": {"en": 4.886, "de": 6.297, "vi": 5.636, "tr": 4.454, "fi": 6.003, "ru": 5.514, "el": 7.514, "hi": 6.916, "ta": 6.38, "th": 9.163, "zh": 4.543, "my": 10.363}, "bpb_mean": 1.7993, "frozen_inventory": false, "t": 582.9} |
| {"event": "train", "step": 4600, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.47631, "bpb": 2.12986, "chunk_loss": 0.0, "F": 0.18604, "G": 0.42276, "chunks": 190.5, "bytes_per_chunk": 5.375, "boundary_entropy": 0.66252, "router_noise": 0.0, "grad_norm": 0.7512, "lr": 6.82e-05, "router_trainable": false, "sec_per_step": 0.191, "alpha": 0.0, "t": 583.2} |
| {"event": "train", "step": 4650, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.35791, "bpb": 1.95905, "chunk_loss": 0.0, "F": 0.17819, "G": 0.41158, "chunks": 182.47, "bytes_per_chunk": 5.612, "boundary_entropy": 0.65386, "router_noise": 0.0, "grad_norm": 0.8064, "lr": 6.57e-05, "router_trainable": false, "sec_per_step": 0.1, "alpha": 0.0, "t": 588.2} |
| {"event": "train", "step": 4700, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.42589, "bpb": 2.05712, "chunk_loss": 0.0, "F": 0.16943, "G": 0.41423, "chunks": 173.5, "bytes_per_chunk": 5.902, "boundary_entropy": 0.65653, "router_noise": 0.0, "grad_norm": 0.8212, "lr": 6.32e-05, "router_trainable": false, "sec_per_step": 0.092, "alpha": 0.0, "t": 592.9} |
| {"event": "train", "step": 4750, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.39416, "bpb": 2.01135, "chunk_loss": 0.0, "F": 0.1778, "G": 0.42098, "chunks": 182.06, "bytes_per_chunk": 5.624, "boundary_entropy": 0.6619, "router_noise": 0.0, "grad_norm": 0.7628, "lr": 6.08e-05, "router_trainable": false, "sec_per_step": 0.096, "alpha": 0.0, "t": 597.7} |
| {"event": "probe", "step": 4800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3518.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.202149, "cps": 27.484375, "mean_p": 0.44054, "boundary_entropy": 0.666973, "bpb": 2.070007}, "de": {"chunks": 3329.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.162129, "cps": 26.007812, "mean_p": 0.44141, "boundary_entropy": 0.672506, "bpb": 2.282987}, "vi": {"chunks": 4351.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.236018, "cps": 33.992188, "mean_p": 0.419451, "boundary_entropy": 0.653712, "bpb": 1.927344}, "tr": {"chunks": 4426.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.24449, "cps": 34.578125, "mean_p": 0.448836, "boundary_entropy": 0.668859, "bpb": 2.246273}, "fi": {"chunks": 3265.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.173569, "cps": 25.507812, "mean_p": 0.442408, "boundary_entropy": 0.672613, "bpb": 2.435888}, "ru": {"chunks": 6422.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.338892, "cps": 50.171875, "mean_p": 0.404163, "boundary_entropy": 0.650104, "bpb": 1.422462}, "el": {"chunks": 5118.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.249379, "cps": 39.984375, "mean_p": 0.427593, "boundary_entropy": 0.669789, "bpb": 1.517383}, "hi": {"chunks": 6313.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.375215, "cps": 49.320312, "mean_p": 0.349799, "boundary_entropy": 0.608094, "bpb": 1.169261}, "ta": {"chunks": 8184.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.437764, "cps": 63.9375, "mean_p": 0.295791, "boundary_entropy": 0.552723, "bpb": 0.940084}, "th": {"chunks": 4989.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.310165, "cps": 38.976562, "mean_p": 0.405968, "boundary_entropy": 0.65729, "bpb": 1.117611}, "zh": {"chunks": 3945.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.667965, "cps": 30.820312, "mean_p": 0.419048, "boundary_entropy": 0.648567, "bpb": 3.176156}, "my": {"chunks": 3919.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.206829, "cps": 30.617188, "mean_p": 0.434612, "boundary_entropy": 0.673054, "bpb": 1.122984}}, "bpb": {"en": 2.07001, "de": 2.28299, "vi": 1.92734, "tr": 2.24627, "fi": 2.43589, "ru": 1.42246, "el": 1.51738, "hi": 1.16926, "ta": 0.94008, "th": 1.11761, "zh": 3.17616, "my": 1.12298}, "gini_cpc": 0.232138, "gini_cps": 0.159679, "bpic": 6.642223, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.802028263436606, "vi": 1.1675466079134578, "tr": 1.2094534097524252, "fi": 0.8586172062123365, "ru": 1.6764452397845384, "el": 1.233637951424707, "hi": 1.8561326128330042, "ta": 2.165551100738783, "th": 1.5343368789632537, "zh": 3.304318105117371, "my": 1.0231520352581007}, "t": 603.9} |
| {"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 2.07864, "de": 2.29067, "vi": 1.92717, "tr": 2.25384, "fi": 2.44375, "ru": 1.39833, "el": 1.50922, "hi": 1.16438, "ta": 0.94932, "th": 1.13563, "zh": 3.17927, "my": 1.13153}, "bytes_per_chunk": {"en": 4.931, "de": 6.385, "vi": 5.755, "tr": 4.462, "fi": 5.847, "ru": 5.472, "el": 7.37, "hi": 6.838, "ta": 6.291, "th": 9.416, "zh": 4.412, "my": 13.061}, "bpb_mean": 1.78848, "frozen_inventory": false, "t": 606.0} |
| {"event": "train", "step": 4800, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.27916, "bpb": 1.84544, "chunk_loss": 0.0, "F": 0.18021, "G": 0.42045, "chunks": 184.53, "bytes_per_chunk": 5.549, "boundary_entropy": 0.66143, "router_noise": 0.0, "grad_norm": 0.7643, "lr": 5.85e-05, "router_trainable": false, "sec_per_step": 0.174, "alpha": 0.0, "t": 606.3} |
| {"event": "train", "step": 4850, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.3696, "bpb": 1.97592, "chunk_loss": 0.0, "F": 0.17194, "G": 0.41429, "chunks": 176.06, "bytes_per_chunk": 5.816, "boundary_entropy": 0.65719, "router_noise": 0.0, "grad_norm": 0.8049, "lr": 5.62e-05, "router_trainable": false, "sec_per_step": 0.091, "alpha": 0.0, "t": 610.9} |
| {"event": "train", "step": 4900, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.40382, "bpb": 2.02529, "chunk_loss": 0.0, "F": 0.18411, "G": 0.42082, "chunks": 188.53, "bytes_per_chunk": 5.431, "boundary_entropy": 0.66115, "router_noise": 0.0, "grad_norm": 0.8308, "lr": 5.41e-05, "router_trainable": false, "sec_per_step": 0.096, "alpha": 0.0, "t": 615.7} |
| {"event": "train", "step": 4950, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.28174, "bpb": 1.84915, "chunk_loss": 0.0, "F": 0.17477, "G": 0.40778, "chunks": 178.97, "bytes_per_chunk": 5.722, "boundary_entropy": 0.65114, "router_noise": 0.0, "grad_norm": 0.8294, "lr": 5.2e-05, "router_trainable": false, "sec_per_step": 0.093, "alpha": 0.0, "t": 620.3} |
| {"event": "probe", "step": 5000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3401.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195426, "cps": 26.570312, "mean_p": 0.439052, "boundary_entropy": 0.666462, "bpb": 2.062234}, "de": {"chunks": 3190.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.15536, "cps": 24.921875, "mean_p": 0.439564, "boundary_entropy": 0.671809, "bpb": 2.274125}, "vi": {"chunks": 4323.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.2345, "cps": 33.773438, "mean_p": 0.419749, "boundary_entropy": 0.653507, "bpb": 1.920886}, "tr": {"chunks": 4626.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.255538, "cps": 36.140625, "mean_p": 0.450447, "boundary_entropy": 0.669453, "bpb": 2.237811}, "fi": {"chunks": 3296.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.175217, "cps": 25.75, "mean_p": 0.442768, "boundary_entropy": 0.67271, "bpb": 2.426012}, "ru": {"chunks": 6210.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.327704, "cps": 48.515625, "mean_p": 0.405438, "boundary_entropy": 0.651029, "bpb": 1.415608}, "el": {"chunks": 4789.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.233348, "cps": 37.414062, "mean_p": 0.42798, "boundary_entropy": 0.670295, "bpb": 1.514235}, "hi": {"chunks": 6337.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.376642, "cps": 49.507812, "mean_p": 0.346641, "boundary_entropy": 0.604906, "bpb": 1.16455}, "ta": {"chunks": 8519.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.455683, "cps": 66.554688, "mean_p": 0.298578, "boundary_entropy": 0.553311, "bpb": 0.927511}, "th": {"chunks": 5573.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.346472, "cps": 43.539062, "mean_p": 0.408691, "boundary_entropy": 0.658557, "bpb": 1.114993}, "zh": {"chunks": 4017.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.680156, "cps": 31.382812, "mean_p": 0.420107, "boundary_entropy": 0.649698, "bpb": 3.172905}, "my": {"chunks": 5115.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.269949, "cps": 39.960938, "mean_p": 0.436644, "boundary_entropy": 0.673748, "bpb": 1.121226}}, "bpb": {"en": 2.06223, "de": 2.27412, "vi": 1.92089, "tr": 2.23781, "fi": 2.42601, "ru": 1.41561, "el": 1.51423, "hi": 1.16455, "ta": 0.92751, "th": 1.11499, "zh": 3.1729, "my": 1.12123}, "gini_cpc": 0.23442, "gini_cps": 0.162707, "bpic": 6.461395, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.794979196933335, "vi": 1.1999401410918964, "tr": 1.3075929346306654, "fi": 0.8965877644738918, "ru": 1.6768718497353332, "el": 1.194047160433007, "hi": 1.9272858039742704, "ta": 2.3317427602394054, "th": 1.7729050050831654, "zh": 3.4803736934008676, "my": 1.3813373353104246}, "t": 626.9} |
| {"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 2.07103, "de": 2.28297, "vi": 1.92163, "tr": 2.24541, "fi": 2.43257, "ru": 1.39168, "el": 1.50629, "hi": 1.1596, "ta": 0.93644, "th": 1.13172, "zh": 3.17269, "my": 1.13014}, "bytes_per_chunk": {"en": 5.035, "de": 6.817, "vi": 5.761, "tr": 4.246, "fi": 5.813, "ru": 5.606, "el": 7.957, "hi": 6.813, "ta": 6.049, "th": 8.644, "zh": 4.424, "my": 10.457}, "bpb_mean": 1.78185, "frozen_inventory": false, "t": 629.4} |
| {"event": "train", "step": 5000, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.48157, "bpb": 2.13745, "chunk_loss": 0.0, "F": 0.19257, "G": 0.42532, "chunks": 197.19, "bytes_per_chunk": 5.193, "boundary_entropy": 0.66283, "router_noise": 0.0, "grad_norm": 0.856, "lr": 5e-05, "router_trainable": false, "sec_per_step": 0.189, "alpha": 0.0, "t": 629.7} |
| {"event": "train", "step": 5050, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.29653, "bpb": 1.8705, "chunk_loss": 0.0, "F": 0.17929, "G": 0.40912, "chunks": 183.59, "bytes_per_chunk": 5.578, "boundary_entropy": 0.65405, "router_noise": 0.0, "grad_norm": 0.7652, "lr": 4.81e-05, "router_trainable": false, "sec_per_step": 0.094, "alpha": 0.0, "t": 634.4} |
| {"event": "train", "step": 5100, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.46718, "bpb": 2.1167, "chunk_loss": 0.0, "F": 0.17923, "G": 0.41782, "chunks": 183.53, "bytes_per_chunk": 5.579, "boundary_entropy": 0.65889, "router_noise": 0.0, "grad_norm": 0.8245, "lr": 4.63e-05, "router_trainable": false, "sec_per_step": 0.094, "alpha": 0.0, "t": 639.1} |
| {"event": "train", "step": 5150, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.26713, "bpb": 1.82808, "chunk_loss": 0.0, "F": 0.18039, "G": 0.40613, "chunks": 184.72, "bytes_per_chunk": 5.544, "boundary_entropy": 0.65003, "router_noise": 0.0, "grad_norm": 0.7249, "lr": 4.45e-05, "router_trainable": false, "sec_per_step": 0.1, "alpha": 0.0, "t": 644.1} |
| {"event": "probe", "step": 5200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3293.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.18922, "cps": 25.726562, "mean_p": 0.439514, "boundary_entropy": 0.666819, "bpb": 2.056989}, "de": {"chunks": 3074.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.14971, "cps": 24.015625, "mean_p": 0.439971, "boundary_entropy": 0.672063, "bpb": 2.266259}, "vi": {"chunks": 4396.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.238459, "cps": 34.34375, "mean_p": 0.419251, "boundary_entropy": 0.653824, "bpb": 1.918988}, "tr": {"chunks": 4508.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.249019, "cps": 35.21875, "mean_p": 0.449198, "boundary_entropy": 0.669188, "bpb": 2.226728}, "fi": {"chunks": 3291.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.174951, "cps": 25.710938, "mean_p": 0.442003, "boundary_entropy": 0.672369, "bpb": 2.417667}, "ru": {"chunks": 6166.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.325383, "cps": 48.171875, "mean_p": 0.404991, "boundary_entropy": 0.650894, "bpb": 1.409796}, "el": {"chunks": 4272.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.208157, "cps": 33.375, "mean_p": 0.425372, "boundary_entropy": 0.669501, "bpb": 1.506601}, "hi": {"chunks": 6135.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.364636, "cps": 47.929688, "mean_p": 0.346613, "boundary_entropy": 0.606684, "bpb": 1.157965}, "ta": {"chunks": 8117.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.43418, "cps": 63.414062, "mean_p": 0.29465, "boundary_entropy": 0.551685, "bpb": 0.923777}, "th": {"chunks": 5302.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.329624, "cps": 41.421875, "mean_p": 0.406892, "boundary_entropy": 0.657704, "bpb": 1.1103}, "zh": {"chunks": 3635.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.615476, "cps": 28.398438, "mean_p": 0.417142, "boundary_entropy": 0.64892, "bpb": 3.15899}, "my": {"chunks": 4666.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.246253, "cps": 36.453125, "mean_p": 0.436441, "boundary_entropy": 0.67439, "bpb": 1.116817}}, "bpb": {"en": 2.05699, "de": 2.26626, "vi": 1.91899, "tr": 2.22673, "fi": 2.41767, "ru": 1.4098, "el": 1.5066, "hi": 1.15797, "ta": 0.92378, "th": 1.1103, "zh": 3.15899, "my": 1.11682}, "gini_cpc": 0.225888, "gini_cps": 0.162505, "bpic": 6.750171, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7911955673733508, "vi": 1.2602216354092823, "tr": 1.3160298664830474, "fi": 0.9245882891239809, "ru": 1.7195970664566318, "el": 1.1000762493234355, "hi": 1.9270451211516708, "ta": 2.2945761012341337, "th": 1.742011620055579, "zh": 3.2526951494046465, "my": 1.3014088264068695}, "t": 650.6} |
| {"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 2.06618, "de": 2.27449, "vi": 1.91845, "tr": 2.2348, "fi": 2.42442, "ru": 1.38454, "el": 1.49706, "hi": 1.15342, "ta": 0.93307, "th": 1.12666, "zh": 3.16011, "my": 1.12603}, "bytes_per_chunk": {"en": 5.239, "de": 6.916, "vi": 5.761, "tr": 4.427, "fi": 5.865, "ru": 5.678, "el": 8.846, "hi": 7.063, "ta": 6.361, "th": 8.78, "zh": 4.874, "my": 11.4}, "bpb_mean": 1.77494, "frozen_inventory": false, "t": 653.0} |
| {"event": "train", "step": 5200, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.37432, "bpb": 1.98273, "chunk_loss": 0.0, "F": 0.17542, "G": 0.42101, "chunks": 179.62, "bytes_per_chunk": 5.701, "boundary_entropy": 0.66132, "router_noise": 0.0, "grad_norm": 0.8057, "lr": 4.29e-05, "router_trainable": false, "sec_per_step": 0.186, "alpha": 0.0, "t": 653.4} |
| {"event": "train", "step": 5250, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.33333, "bpb": 1.9236, "chunk_loss": 0.0, "F": 0.16678, "G": 0.41815, "chunks": 170.78, "bytes_per_chunk": 5.996, "boundary_entropy": 0.65979, "router_noise": 0.0, "grad_norm": 0.813, "lr": 4.14e-05, "router_trainable": false, "sec_per_step": 0.094, "alpha": 0.0, "t": 658.1} |
| {"event": "train", "step": 5300, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.29671, "bpb": 1.87075, "chunk_loss": 0.0, "F": 0.17755, "G": 0.4142, "chunks": 181.81, "bytes_per_chunk": 5.632, "boundary_entropy": 0.65645, "router_noise": 0.0, "grad_norm": 0.7733, "lr": 3.99e-05, "router_trainable": false, "sec_per_step": 0.094, "alpha": 0.0, "t": 662.7} |
| {"event": "train", "step": 5350, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.37625, "bpb": 1.98551, "chunk_loss": 0.0, "F": 0.18851, "G": 0.42144, "chunks": 193.03, "bytes_per_chunk": 5.305, "boundary_entropy": 0.65994, "router_noise": 0.0, "grad_norm": 0.7998, "lr": 3.86e-05, "router_trainable": false, "sec_per_step": 0.087, "alpha": 0.0, "t": 667.1} |
| {"event": "probe", "step": 5400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3527.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.202666, "cps": 27.554688, "mean_p": 0.441383, "boundary_entropy": 0.667624, "bpb": 2.051562}, "de": {"chunks": 3299.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.160668, "cps": 25.773438, "mean_p": 0.441709, "boundary_entropy": 0.672833, "bpb": 2.258258}, "vi": {"chunks": 4454.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.241606, "cps": 34.796875, "mean_p": 0.421389, "boundary_entropy": 0.655163, "bpb": 1.911907}, "tr": {"chunks": 4555.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.251616, "cps": 35.585938, "mean_p": 0.44979, "boundary_entropy": 0.669553, "bpb": 2.224041}, "fi": {"chunks": 3279.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.174313, "cps": 25.617188, "mean_p": 0.44263, "boundary_entropy": 0.672796, "bpb": 2.409626}, "ru": {"chunks": 6555.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.34591, "cps": 51.210938, "mean_p": 0.408091, "boundary_entropy": 0.6523, "bpb": 1.397377}, "el": {"chunks": 5276.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.257077, "cps": 41.21875, "mean_p": 0.429865, "boundary_entropy": 0.670813, "bpb": 1.497833}, "hi": {"chunks": 6753.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.401367, "cps": 52.757812, "mean_p": 0.352251, "boundary_entropy": 0.608668, "bpb": 1.141026}, "ta": {"chunks": 8638.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.462049, "cps": 67.484375, "mean_p": 0.301947, "boundary_entropy": 0.555336, "bpb": 0.914293}, "th": {"chunks": 5591.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.347591, "cps": 43.679688, "mean_p": 0.410638, "boundary_entropy": 0.659619, "bpb": 1.108325}, "zh": {"chunks": 4160.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.704368, "cps": 32.5, "mean_p": 0.420494, "boundary_entropy": 0.649477, "bpb": 3.151117}, "my": {"chunks": 6100.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.321934, "cps": 47.65625, "mean_p": 0.441575, "boundary_entropy": 0.675561, "bpb": 1.108366}}, "bpb": {"en": 2.05156, "de": 2.25826, "vi": 1.91191, "tr": 2.22404, "fi": 2.40963, "ru": 1.39738, "el": 1.49783, "hi": 1.14103, "ta": 0.91429, "th": 1.10833, "zh": 3.15112, "my": 1.10837}, "gini_cpc": 0.233325, "gini_cps": 0.16759, "bpic": 6.171402, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7927724856697175, "vi": 1.1921358032409752, "tr": 1.2415279194764934, "fi": 0.860098506931733, "ru": 1.7067980666296112, "el": 1.2684770153041212, "hi": 1.9804338343435126, "ta": 2.279850612548155, "th": 1.7150906823942376, "zh": 3.475510048696952, "my": 1.5884923200792658}, "t": 673.1} |
| {"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 2.06001, "de": 2.2673, "vi": 1.91076, "tr": 2.22936, "fi": 2.4154, "ru": 1.37415, "el": 1.48977, "hi": 1.13697, "ta": 0.92406, "th": 1.12511, "zh": 3.15051, "my": 1.11818}, "bytes_per_chunk": {"en": 4.936, "de": 6.452, "vi": 5.645, "tr": 4.363, "fi": 5.83, "ru": 5.339, "el": 7.145, "hi": 6.353, "ta": 5.966, "th": 8.413, "zh": 4.234, "my": 8.886}, "bpb_mean": 1.7668, "frozen_inventory": false, "t": 675.7} |
| {"event": "train", "step": 5400, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.25242, "bpb": 1.80686, "chunk_loss": 0.0, "F": 0.17648, "G": 0.41612, "chunks": 180.72, "bytes_per_chunk": 5.666, "boundary_entropy": 0.65763, "router_noise": 0.0, "grad_norm": 0.7907, "lr": 3.73e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 676.1} |
| {"event": "train", "step": 5450, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.30336, "bpb": 1.88035, "chunk_loss": 0.0, "F": 0.16199, "G": 0.41681, "chunks": 165.88, "bytes_per_chunk": 6.173, "boundary_entropy": 0.65992, "router_noise": 0.0, "grad_norm": 0.7921, "lr": 3.62e-05, "router_trainable": false, "sec_per_step": 0.102, "alpha": 0.0, "t": 681.2} |
| {"event": "train", "step": 5500, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.47621, "bpb": 2.12971, "chunk_loss": 0.0, "F": 0.19275, "G": 0.42574, "chunks": 197.38, "bytes_per_chunk": 5.188, "boundary_entropy": 0.66323, "router_noise": 0.0, "grad_norm": 0.8401, "lr": 3.51e-05, "router_trainable": false, "sec_per_step": 0.098, "alpha": 0.0, "t": 686.1} |
| {"event": "train", "step": 5550, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.31696, "bpb": 1.89998, "chunk_loss": 0.0, "F": 0.18231, "G": 0.41346, "chunks": 186.69, "bytes_per_chunk": 5.485, "boundary_entropy": 0.65193, "router_noise": 0.0, "grad_norm": 0.8252, "lr": 3.41e-05, "router_trainable": false, "sec_per_step": 0.098, "alpha": 0.0, "t": 690.9} |
| {"event": "probe", "step": 5600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3691.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.21209, "cps": 28.835938, "mean_p": 0.44242, "boundary_entropy": 0.667815, "bpb": 2.044828}, "de": {"chunks": 3284.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.159938, "cps": 25.65625, "mean_p": 0.441955, "boundary_entropy": 0.672728, "bpb": 2.253292}, "vi": {"chunks": 4451.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.241443, "cps": 34.773438, "mean_p": 0.420397, "boundary_entropy": 0.654331, "bpb": 1.905204}, "tr": {"chunks": 4393.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.242667, "cps": 34.320312, "mean_p": 0.449561, "boundary_entropy": 0.669547, "bpb": 2.216002}, "fi": {"chunks": 3198.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.170007, "cps": 24.984375, "mean_p": 0.442645, "boundary_entropy": 0.672857, "bpb": 2.405046}, "ru": {"chunks": 6156.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.324855, "cps": 48.09375, "mean_p": 0.405258, "boundary_entropy": 0.651218, "bpb": 1.396268}, "el": {"chunks": 4614.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.224821, "cps": 36.046875, "mean_p": 0.426984, "boundary_entropy": 0.669991, "bpb": 1.492352}, "hi": {"chunks": 6355.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.377712, "cps": 49.648438, "mean_p": 0.346417, "boundary_entropy": 0.605252, "bpb": 1.140841}, "ta": {"chunks": 8402.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.449425, "cps": 65.640625, "mean_p": 0.297246, "boundary_entropy": 0.55254, "bpb": 0.908768}, "th": {"chunks": 5099.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.317003, "cps": 39.835938, "mean_p": 0.408032, "boundary_entropy": 0.658863, "bpb": 1.104849}, "zh": {"chunks": 4025.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.68151, "cps": 31.445312, "mean_p": 0.41905, "boundary_entropy": 0.648981, "bpb": 3.138511}, "my": {"chunks": 4022.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.212265, "cps": 31.421875, "mean_p": 0.434418, "boundary_entropy": 0.674083, "bpb": 1.110586}}, "bpb": {"en": 2.04483, "de": 2.25329, "vi": 1.9052, "tr": 2.216, "fi": 2.40505, "ru": 1.39627, "el": 1.49235, "hi": 1.14084, "ta": 0.90877, "th": 1.10485, "zh": 3.13851, "my": 1.11059}, "gini_cpc": 0.236416, "gini_cps": 0.157953, "bpic": 6.65247, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7541032565837578, "vi": 1.1383990570161757, "tr": 1.144170448958549, "fi": 0.8015795907675837, "ru": 1.5316850443459058, "el": 1.0600267380791013, "hi": 1.7809041952977462, "ta": 2.1190308658672503, "th": 1.4946655396053226, "zh": 3.2133091971089005, "my": 1.0008264286972968}, "t": 697.5} |
| {"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 2.05411, "de": 2.26308, "vi": 1.90389, "tr": 2.22289, "fi": 2.41115, "ru": 1.37278, "el": 1.48399, "hi": 1.13554, "ta": 0.91836, "th": 1.12224, "zh": 3.13974, "my": 1.11981}, "bytes_per_chunk": {"en": 4.72, "de": 6.387, "vi": 5.613, "tr": 4.498, "fi": 5.993, "ru": 5.701, "el": 8.224, "hi": 6.77, "ta": 6.143, "th": 9.267, "zh": 4.409, "my": 12.512}, "bpb_mean": 1.7623, "frozen_inventory": false, "t": 700.0} |
| {"event": "train", "step": 5600, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.33592, "bpb": 1.92733, "chunk_loss": 0.0, "F": 0.18948, "G": 0.42206, "chunks": 194.03, "bytes_per_chunk": 5.278, "boundary_entropy": 0.66103, "router_noise": 0.0, "grad_norm": 0.7902, "lr": 3.33e-05, "router_trainable": false, "sec_per_step": 0.188, "alpha": 0.0, "t": 700.3} |
| {"event": "train", "step": 5650, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.24334, "bpb": 1.79376, "chunk_loss": 0.0, "F": 0.17169, "G": 0.42011, "chunks": 175.81, "bytes_per_chunk": 5.824, "boundary_entropy": 0.66152, "router_noise": 0.0, "grad_norm": 0.7945, "lr": 3.25e-05, "router_trainable": false, "sec_per_step": 0.099, "alpha": 0.0, "t": 705.3} |
| {"event": "train", "step": 5700, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.30665, "bpb": 1.88509, "chunk_loss": 0.0, "F": 0.16949, "G": 0.4203, "chunks": 173.56, "bytes_per_chunk": 5.9, "boundary_entropy": 0.6608, "router_noise": 0.0, "grad_norm": 0.7756, "lr": 3.18e-05, "router_trainable": false, "sec_per_step": 0.094, "alpha": 0.0, "t": 710.0} |
| {"event": "train", "step": 5750, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.33303, "bpb": 1.92315, "chunk_loss": 0.0, "F": 0.16953, "G": 0.42046, "chunks": 173.59, "bytes_per_chunk": 5.899, "boundary_entropy": 0.66149, "router_noise": 0.0, "grad_norm": 0.8245, "lr": 3.13e-05, "router_trainable": false, "sec_per_step": 0.091, "alpha": 0.0, "t": 714.5} |
| {"event": "probe", "step": 5800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3379.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194162, "cps": 26.398438, "mean_p": 0.441212, "boundary_entropy": 0.667494, "bpb": 2.045104}, "de": {"chunks": 3020.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.14708, "cps": 23.59375, "mean_p": 0.440444, "boundary_entropy": 0.672436, "bpb": 2.249054}, "vi": {"chunks": 4382.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.2377, "cps": 34.234375, "mean_p": 0.420945, "boundary_entropy": 0.654861, "bpb": 1.90214}, "tr": {"chunks": 4369.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.241341, "cps": 34.132812, "mean_p": 0.448784, "boundary_entropy": 0.669306, "bpb": 2.210578}, "fi": {"chunks": 3141.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.166977, "cps": 24.539062, "mean_p": 0.441922, "boundary_entropy": 0.672532, "bpb": 2.398487}, "ru": {"chunks": 6315.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.333245, "cps": 49.335938, "mean_p": 0.405224, "boundary_entropy": 0.651199, "bpb": 1.392523}, "el": {"chunks": 4506.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.219559, "cps": 35.203125, "mean_p": 0.426403, "boundary_entropy": 0.669965, "bpb": 1.490593}, "hi": {"chunks": 6552.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.389421, "cps": 51.1875, "mean_p": 0.347902, "boundary_entropy": 0.605439, "bpb": 1.134574}, "ta": {"chunks": 8523.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.455897, "cps": 66.585938, "mean_p": 0.299009, "boundary_entropy": 0.552848, "bpb": 0.905124}, "th": {"chunks": 5150.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.320174, "cps": 40.234375, "mean_p": 0.408594, "boundary_entropy": 0.659319, "bpb": 1.100676}, "zh": {"chunks": 3800.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.643413, "cps": 29.6875, "mean_p": 0.419013, "boundary_entropy": 0.649436, "bpb": 3.135327}, "my": {"chunks": 4070.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.214798, "cps": 31.796875, "mean_p": 0.43748, "boundary_entropy": 0.675211, "bpb": 1.106636}}, "bpb": {"en": 2.0451, "de": 2.24905, "vi": 1.90214, "tr": 2.21058, "fi": 2.39849, "ru": 1.39252, "el": 1.49059, "hi": 1.13457, "ta": 0.90512, "th": 1.10068, "zh": 3.13533, "my": 1.10664}, "gini_cpc": 0.241422, "gini_cps": 0.174593, "bpic": 6.708637, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7575136521151963, "vi": 1.2242360378839194, "tr": 1.2429893903775142, "fi": 0.8599871883315111, "ru": 1.7163271361763366, "el": 1.1308012259525857, "hi": 2.0056481361367102, "ta": 2.3480262474176, "th": 1.649005454615872, "zh": 3.3137983884246704, "my": 1.106284841309275}, "t": 720.8} |
| {"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 2.05439, "de": 2.25903, "vi": 1.9016, "tr": 2.21653, "fi": 2.40514, "ru": 1.36821, "el": 1.48211, "hi": 1.12995, "ta": 0.91514, "th": 1.11706, "zh": 3.13397, "my": 1.11527}, "bytes_per_chunk": {"en": 5.06, "de": 6.971, "vi": 5.676, "tr": 4.467, "fi": 6.036, "ru": 5.579, "el": 8.357, "hi": 6.567, "ta": 6.065, "th": 9.143, "zh": 4.614, "my": 13.217}, "bpb_mean": 1.7582, "frozen_inventory": false, "t": 723.3} |
| {"event": "train", "step": 5800, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.35391, "bpb": 1.95327, "chunk_loss": 0.0, "F": 0.17267, "G": 0.42112, "chunks": 176.81, "bytes_per_chunk": 5.791, "boundary_entropy": 0.66095, "router_noise": 0.0, "grad_norm": 0.8197, "lr": 3.08e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 723.5} |
| {"event": "train", "step": 5850, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.32553, "bpb": 1.91234, "chunk_loss": 0.0, "F": 0.17514, "G": 0.41791, "chunks": 179.34, "bytes_per_chunk": 5.71, "boundary_entropy": 0.65902, "router_noise": 0.0, "grad_norm": 0.7765, "lr": 3.05e-05, "router_trainable": false, "sec_per_step": 0.084, "alpha": 0.0, "t": 727.7} |
| {"event": "train", "step": 5900, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.24073, "bpb": 1.78999, "chunk_loss": 0.0, "F": 0.16794, "G": 0.42661, "chunks": 171.97, "bytes_per_chunk": 5.955, "boundary_entropy": 0.66665, "router_noise": 0.0, "grad_norm": 0.8075, "lr": 3.02e-05, "router_trainable": false, "sec_per_step": 0.08, "alpha": 0.0, "t": 731.7} |
| {"event": "train", "step": 5950, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.26278, "bpb": 1.82181, "chunk_loss": 0.0, "F": 0.17484, "G": 0.41602, "chunks": 179.03, "bytes_per_chunk": 5.72, "boundary_entropy": 0.65745, "router_noise": 0.0, "grad_norm": 0.8675, "lr": 3.01e-05, "router_trainable": false, "sec_per_step": 0.094, "alpha": 0.0, "t": 736.4} |
| {"event": "train", "step": 5999, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.37904, "bpb": 1.98953, "chunk_loss": 0.0, "F": 0.1785, "G": 0.41566, "chunks": 182.78, "bytes_per_chunk": 5.602, "boundary_entropy": 0.65711, "router_noise": 0.0, "grad_norm": 0.8657, "lr": 3e-05, "router_trainable": false, "sec_per_step": 0.094, "alpha": 0.0, "t": 741.1} |
| {"event": "probe", "step": 6000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3372.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19376, "cps": 26.34375, "mean_p": 0.440715, "boundary_entropy": 0.667416, "bpb": 2.041441}, "de": {"chunks": 2946.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.143476, "cps": 23.015625, "mean_p": 0.439778, "boundary_entropy": 0.67235, "bpb": 2.245626}, "vi": {"chunks": 4349.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.23591, "cps": 33.976562, "mean_p": 0.420294, "boundary_entropy": 0.654572, "bpb": 1.901516}, "tr": {"chunks": 4463.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.246534, "cps": 34.867188, "mean_p": 0.449783, "boundary_entropy": 0.669565, "bpb": 2.205757}, "fi": {"chunks": 3072.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.163309, "cps": 24.0, "mean_p": 0.441165, "boundary_entropy": 0.672446, "bpb": 2.395151}, "ru": {"chunks": 6215.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.327968, "cps": 48.554688, "mean_p": 0.406566, "boundary_entropy": 0.651618, "bpb": 1.386491}, "el": {"chunks": 4451.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.216879, "cps": 34.773438, "mean_p": 0.42648, "boundary_entropy": 0.669918, "bpb": 1.486144}, "hi": {"chunks": 6258.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.371947, "cps": 48.890625, "mean_p": 0.347868, "boundary_entropy": 0.606464, "bpb": 1.133869}, "ta": {"chunks": 8511.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.455255, "cps": 66.492188, "mean_p": 0.299704, "boundary_entropy": 0.554246, "bpb": 0.904099}, "th": {"chunks": 5135.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.319242, "cps": 40.117188, "mean_p": 0.409328, "boundary_entropy": 0.659671, "bpb": 1.099364}, "zh": {"chunks": 3924.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.664409, "cps": 30.65625, "mean_p": 0.41957, "boundary_entropy": 0.649667, "bpb": 3.128802}, "my": {"chunks": 5912.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.312012, "cps": 46.1875, "mean_p": 0.442254, "boundary_entropy": 0.676312, "bpb": 1.106787}}, "bpb": {"en": 2.04144, "de": 2.24563, "vi": 1.90152, "tr": 2.20576, "fi": 2.39515, "ru": 1.38649, "el": 1.48614, "hi": 1.13387, "ta": 0.9041, "th": 1.09936, "zh": 3.1288, "my": 1.10679}, "gini_cpc": 0.239132, "gini_cps": 0.173611, "bpic": 6.54827, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7404860641831142, "vi": 1.2175388278717085, "tr": 1.2723684440733856, "fi": 0.8428414433820146, "ru": 1.692655095353008, "el": 1.119317524665929, "hi": 1.9196278415550494, "ta": 2.349587782525063, "th": 1.6476157579230446, "zh": 3.429036518746242, "my": 1.6103030057163379}, "t": 742.9} |
| {"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 2.05074, "de": 2.25475, "vi": 1.90079, "tr": 2.21221, "fi": 2.40204, "ru": 1.36338, "el": 1.47803, "hi": 1.12899, "ta": 0.91386, "th": 1.1157, "zh": 3.12775, "my": 1.11517}, "bytes_per_chunk": {"en": 5.066, "de": 7.184, "vi": 5.74, "tr": 4.394, "fi": 6.166, "ru": 5.649, "el": 8.609, "hi": 6.853, "ta": 6.06, "th": 9.24, "zh": 4.494, "my": 8.924}, "bpb_mean": 1.75528, "frozen_inventory": false, "t": 745.5} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s1/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s1/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s1/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s1/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s1/probe/reps/th0.70"], "t": 751.0} |
| {"event": "final", "run": "parity_B_ste_s1", "arch": "hnet", "objective": "parity", "condition": "B", "step": 6000, "params": 22516992, "seed": 1, "per_lang": {"en": {"cpc": 0.19375969660403378, "cps": 26.34375, "bpb": 2.060061646392393, "bytes_per_chunk": 5.072543973374098}, "de": {"cpc": 0.1434763551356353, "cps": 23.015625, "bpb": 2.246107716326134, "bytes_per_chunk": 7.176915074193725}, "vi": {"cpc": 0.23590995389205316, "cps": 33.9765625, "bpb": 1.9042289296186155, "bytes_per_chunk": 5.742789857056794}, "tr": {"cpc": 0.24653372369220572, "cps": 34.8671875, "bpb": 2.2053669718904723, "bytes_per_chunk": 4.412641501493919}, "fi": {"cpc": 0.16330870235500505, "cps": 24.0, "bpb": 2.4159184842825216, "bytes_per_chunk": 6.190744960974861}, "ru": {"cpc": 0.3279683377308707, "cps": 48.5546875, "bpb": 1.364129469492899, "bytes_per_chunk": 5.632902144484078}, "el": {"cpc": 0.21687862398284852, "cps": 34.7734375, "bpb": 1.4728984727997547, "bytes_per_chunk": 8.521959624199473}, "hi": {"cpc": 0.3719465081723626, "cps": 48.890625, "bpb": 1.1371366037843895, "bytes_per_chunk": 6.840740064194567}, "ta": {"cpc": 0.4552554158866007, "cps": 66.4921875, "bpb": 0.9155036203856824, "bytes_per_chunk": 6.04443214701576}, "th": {"cpc": 0.3192415293751943, "cps": 40.1171875, "bpb": 1.1132721836615094, "bytes_per_chunk": 9.221655468392726}, "zh": {"cpc": 0.664409075516424, "cps": 30.65625, "bpb": 3.1354258763540965, "bytes_per_chunk": 4.506515385937768}, "my": {"cpc": 0.3120118218281613, "cps": 46.1875, "bpb": 1.1255674705967522, "bytes_per_chunk": 8.84530899397702}}, "bpb_mean": 1.7579681204654352, "gini_cpc": 0.23913245300900932, "gini_cps": 0.17361111111111116, "bpic": 6.548269860769861, "mean_boundary_entropy": 0.6571149826049805, "dechunk": "ste", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 131584}], "probe_overhead_sec_total": 0.0, "probe_overhead_seq_frac": 0.0, "wall_sec": 755.5127236843109, "t": 755.5} |
|
|
