{"event": "start", "run": "parity_B_ste_s2", "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": 1.6} {"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": 1.6} {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10933.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.628225, "cps": 85.414062, "mean_p": 0.513249, "boundary_entropy": 0.686295, "bpb": 8.157895}, "de": {"chunks": 16287.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.793211, "cps": 127.242188, "mean_p": 0.524631, "boundary_entropy": 0.686391, "bpb": 8.150604}, "vi": {"chunks": 5099.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.276593, "cps": 39.835938, "mean_p": 0.481643, "boundary_entropy": 0.68728, "bpb": 7.885576}, "tr": {"chunks": 8187.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.452245, "cps": 63.960938, "mean_p": 0.498002, "boundary_entropy": 0.686826, "bpb": 8.139464}, "fi": {"chunks": 13540.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.719792, "cps": 105.78125, "mean_p": 0.517169, "boundary_entropy": 0.686661, "bpb": 8.097447}, "ru": {"chunks": 33477.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.766596, "cps": 261.539062, "mean_p": 0.530376, "boundary_entropy": 0.688543, "bpb": 8.304526}, "el": {"chunks": 20696.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.00843, "cps": 161.6875, "mean_p": 0.503648, "boundary_entropy": 0.690112, "bpb": 8.265181}, "hi": {"chunks": 875.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.052006, "cps": 6.835938, "mean_p": 0.461978, "boundary_entropy": 0.68743, "bpb": 8.053074}, "ta": {"chunks": 1261.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.067451, "cps": 9.851562, "mean_p": 0.477518, "boundary_entropy": 0.690053, "bpb": 8.127442}, "th": {"chunks": 3630.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.225676, "cps": 28.359375, "mean_p": 0.484845, "boundary_entropy": 0.690398, "bpb": 7.710377}, "zh": {"chunks": 11722.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.984761, "cps": 91.578125, "mean_p": 0.520316, "boundary_entropy": 0.68562, "bpb": 8.124258}, "my": {"chunks": 1628.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.085919, "cps": 12.71875, "mean_p": 0.474628, "boundary_entropy": 0.689685, "bpb": 7.688162}}, "bpb": {"en": 8.15789, "de": 8.1506, "vi": 7.88558, "tr": 8.13946, "fi": 8.09745, "ru": 8.30453, "el": 8.26518, "hi": 8.05307, "ta": 8.12744, "th": 7.71038, "zh": 8.12426, "my": 7.68816}, "gini_cpc": 0.489538, "gini_cps": 0.463551, "bpic": 3.013947, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2626223171148012, "vi": 0.440277652395235, "tr": 0.7198781817505053, "fi": 1.1457544508645294, "ru": 2.8120438593717987, "el": 1.6052044042347315, "hi": 0.08278235025738971, "ta": 0.10736788276924242, "th": 0.3592281253258146, "zh": 3.159315851384177, "my": 0.13676526035979295}, "t": 3.6} {"event": "train", "step": 0, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 5.65891, "bpb": 8.16408, "chunk_loss": 0.065103, "F": 0.62439, "G": 0.5091, "chunks": 639.38, "bytes_per_chunk": 1.602, "boundary_entropy": 0.69066, "router_noise": 0.0, "grad_norm": 7.3918, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.006, "alpha": 0.03, "t": 4.0} {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10929.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.627995, "cps": 85.382812, "mean_p": 0.513237, "boundary_entropy": 0.6863, "bpb": 8.134238}, "de": {"chunks": 16276.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.792675, "cps": 127.15625, "mean_p": 0.524625, "boundary_entropy": 0.686395, "bpb": 8.129312}, "vi": {"chunks": 5091.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.276159, "cps": 39.773438, "mean_p": 0.481599, "boundary_entropy": 0.687276, "bpb": 7.87308}, "tr": {"chunks": 8215.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.453792, "cps": 64.179688, "mean_p": 0.49813, "boundary_entropy": 0.686829, "bpb": 8.124966}, "fi": {"chunks": 13529.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.719207, "cps": 105.695312, "mean_p": 0.517125, "boundary_entropy": 0.686667, "bpb": 8.076833}, "ru": {"chunks": 33477.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.766596, "cps": 261.539062, "mean_p": 0.530429, "boundary_entropy": 0.688536, "bpb": 8.267661}, "el": {"chunks": 20700.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.008624, "cps": 161.71875, "mean_p": 0.503648, "boundary_entropy": 0.690112, "bpb": 8.259779}, "hi": {"chunks": 876.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.052065, "cps": 6.84375, "mean_p": 0.462019, "boundary_entropy": 0.687436, "bpb": 8.046167}, "ta": {"chunks": 1278.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.068361, "cps": 9.984375, "mean_p": 0.477596, "boundary_entropy": 0.690061, "bpb": 8.122179}, "th": {"chunks": 3630.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.225676, "cps": 28.359375, "mean_p": 0.484847, "boundary_entropy": 0.690397, "bpb": 7.707629}, "zh": {"chunks": 11708.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.982391, "cps": 91.46875, "mean_p": 0.520277, "boundary_entropy": 0.685623, "bpb": 8.120824}, "my": {"chunks": 1621.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.08555, "cps": 12.664062, "mean_p": 0.474603, "boundary_entropy": 0.689682, "bpb": 7.686739}}, "bpb": {"en": 8.13424, "de": 8.12931, "vi": 7.87308, "tr": 8.12497, "fi": 8.07683, "ru": 8.26766, "el": 8.25978, "hi": 8.04617, "ta": 8.12218, "th": 7.70763, "zh": 8.12082, "my": 7.68674}, "gini_cpc": 0.489277, "gini_cps": 0.463398, "bpic": 3.014066, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2622313666347693, "vi": 0.439747773576039, "tr": 0.722604581169554, "fi": 1.1452426351306557, "ru": 2.813073063822113, "el": 1.6061022649160093, "hi": 0.08290729151838755, "ta": 0.10885517455650125, "th": 0.3593596023595142, "zh": 3.1566974930420115, "my": 0.13622704402548524}, "t": 5.3} {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10928.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.627938, "cps": 85.375, "mean_p": 0.513223, "boundary_entropy": 0.68631, "bpb": 8.088279}, "de": {"chunks": 16285.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.793114, "cps": 127.226562, "mean_p": 0.524617, "boundary_entropy": 0.686403, "bpb": 8.087205}, "vi": {"chunks": 5075.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.275292, "cps": 39.648438, "mean_p": 0.481502, "boundary_entropy": 0.687266, "bpb": 7.851354}, "tr": {"chunks": 8276.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.457162, "cps": 64.65625, "mean_p": 0.498371, "boundary_entropy": 0.686836, "bpb": 8.094999}, "fi": {"chunks": 13521.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.718782, "cps": 105.632812, "mean_p": 0.517052, "boundary_entropy": 0.686677, "bpb": 8.035239}, "ru": {"chunks": 33486.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.767071, "cps": 261.609375, "mean_p": 0.53058, "boundary_entropy": 0.688517, "bpb": 8.193541}, "el": {"chunks": 20683.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.007796, "cps": 161.585938, "mean_p": 0.503627, "boundary_entropy": 0.690113, "bpb": 8.25456}, "hi": {"chunks": 895.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.053195, "cps": 6.992188, "mean_p": 0.462177, "boundary_entropy": 0.687462, "bpb": 8.021077}, "ta": {"chunks": 1315.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.07034, "cps": 10.273438, "mean_p": 0.477806, "boundary_entropy": 0.690081, "bpb": 8.106841}, "th": {"chunks": 3680.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.228785, "cps": 28.75, "mean_p": 0.484935, "boundary_entropy": 0.690402, "bpb": 7.694209}, "zh": {"chunks": 11694.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.98002, "cps": 91.359375, "mean_p": 0.520226, "boundary_entropy": 0.685626, "bpb": 8.11392}, "my": {"chunks": 1619.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.085444, "cps": 12.648438, "mean_p": 0.474601, "boundary_entropy": 0.689682, "bpb": 7.686075}}, "bpb": {"en": 8.08828, "de": 8.0872, "vi": 7.85135, "tr": 8.095, "fi": 8.03524, "ru": 8.19354, "el": 8.25456, "hi": 8.02108, "ta": 8.10684, "th": 7.69421, "zh": 8.11392, "my": 7.68607}, "gini_cpc": 0.488173, "gini_cps": 0.462371, "bpic": 3.011063, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2630449000819235, "vi": 0.4384058478187578, "tr": 0.7280368541845592, "fi": 1.144670164523997, "ru": 2.8140868220677824, "el": 1.6049300943219151, "hi": 0.08471326026467446, "ta": 0.11201694320910884, "th": 0.3643427963905151, "zh": 3.1532113469737455, "my": 0.1360714167324242}, "t": 7.0} {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10923.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.62765, "cps": 85.335938, "mean_p": 0.513076, "boundary_entropy": 0.686344, "bpb": 7.941114}, "de": {"chunks": 16310.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.794331, "cps": 127.421875, "mean_p": 0.524497, "boundary_entropy": 0.686437, "bpb": 7.95558}, "vi": {"chunks": 5035.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.273122, "cps": 39.335938, "mean_p": 0.48129, "boundary_entropy": 0.687243, "bpb": 7.781691}, "tr": {"chunks": 8479.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.468375, "cps": 66.242188, "mean_p": 0.49911, "boundary_entropy": 0.686855, "bpb": 7.99714}, "fi": {"chunks": 13475.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.716336, "cps": 105.273438, "mean_p": 0.51673, "boundary_entropy": 0.686715, "bpb": 7.903127}, "ru": {"chunks": 33529.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.76934, "cps": 261.945312, "mean_p": 0.531283, "boundary_entropy": 0.688427, "bpb": 7.950023}, "el": {"chunks": 20453.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.996589, "cps": 159.789062, "mean_p": 0.503358, "boundary_entropy": 0.690116, "bpb": 8.206714}, "hi": {"chunks": 940.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.055869, "cps": 7.34375, "mean_p": 0.462568, "boundary_entropy": 0.687524, "bpb": 7.939367}, "ta": {"chunks": 1398.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.074779, "cps": 10.921875, "mean_p": 0.478289, "boundary_entropy": 0.690125, "bpb": 8.032172}, "th": {"chunks": 3737.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.232328, "cps": 29.195312, "mean_p": 0.484905, "boundary_entropy": 0.690396, "bpb": 7.621773}, "zh": {"chunks": 11643.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.971385, "cps": 90.960938, "mean_p": 0.519957, "boundary_entropy": 0.685644, "bpb": 8.087027}, "my": {"chunks": 1584.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.083597, "cps": 12.375, "mean_p": 0.474426, "boundary_entropy": 0.689662, "bpb": 7.683268}}, "bpb": {"en": 7.94111, "de": 7.95558, "vi": 7.78169, "tr": 7.99714, "fi": 7.90313, "ru": 7.95002, "el": 8.20671, "hi": 7.93937, "ta": 8.03217, "th": 7.62177, "zh": 8.08703, "my": 7.68327}, "gini_cpc": 0.485835, "gini_cps": 0.460118, "bpic": 3.009905, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2655629158295338, "vi": 0.4351495307004047, "tr": 0.7462361269638627, "fi": 1.1412980545387004, "ru": 2.8189902452209843, "el": 1.587809367302657, "hi": 0.08901331337463012, "ta": 0.11914172628614751, "th": 0.37015551053662404, "zh": 3.140896614013205, "my": 0.13319072647225957}, "t": 8.8} {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10609.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.609608, "cps": 82.882812, "mean_p": 0.511399, "boundary_entropy": 0.686513, "bpb": 7.483463}, "de": {"chunks": 16140.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.786052, "cps": 126.09375, "mean_p": 0.522869, "boundary_entropy": 0.686671, "bpb": 7.541422}, "vi": {"chunks": 4836.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.262327, "cps": 37.78125, "mean_p": 0.480521, "boundary_entropy": 0.687161, "bpb": 7.560942}, "tr": {"chunks": 9124.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.504005, "cps": 71.28125, "mean_p": 0.501696, "boundary_entropy": 0.686913, "bpb": 7.662329}, "fi": {"chunks": 13005.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.691351, "cps": 101.601562, "mean_p": 0.514408, "boundary_entropy": 0.686914, "bpb": 7.489219}, "ru": {"chunks": 33815.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.784433, "cps": 264.179688, "mean_p": 0.535775, "boundary_entropy": 0.687825, "bpb": 7.383533}, "el": {"chunks": 19656.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.957755, "cps": 153.5625, "mean_p": 0.502311, "boundary_entropy": 0.690126, "bpb": 7.973843}, "hi": {"chunks": 1042.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.061932, "cps": 8.140625, "mean_p": 0.463583, "boundary_entropy": 0.687684, "bpb": 7.685674}, "ta": {"chunks": 1662.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.088901, "cps": 12.984375, "mean_p": 0.479383, "boundary_entropy": 0.690219, "bpb": 7.809914}, "th": {"chunks": 3767.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.234193, "cps": 29.429688, "mean_p": 0.484364, "boundary_entropy": 0.690344, "bpb": 7.383222}, "zh": {"chunks": 11529.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.952083, "cps": 90.070312, "mean_p": 0.519235, "boundary_entropy": 0.685688, "bpb": 8.009445}, "my": {"chunks": 1617.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.085339, "cps": 12.632812, "mean_p": 0.474682, "boundary_entropy": 0.689693, "bpb": 7.572924}}, "bpb": {"en": 7.48346, "de": 7.54142, "vi": 7.56094, "tr": 7.66233, "fi": 7.48922, "ru": 7.38353, "el": 7.97384, "hi": 7.68567, "ta": 7.80991, "th": 7.38322, "zh": 8.00944, "my": 7.57292}, "gini_cpc": 0.481392, "gini_cps": 0.454747, "bpic": 3.026616, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2894389774090844, "vi": 0.4303212779483713, "tr": 0.8267694030774883, "fi": 1.1340916265458758, "ru": 2.9271828245595226, "el": 1.5711005082122753, "hi": 0.1015926565476031, "ta": 0.14583280214100108, "th": 0.3841706883261943, "zh": 3.202195680293301, "my": 0.13998977471480206}, "t": 10.7} {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6658.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.382578, "cps": 52.015625, "mean_p": 0.496494, "boundary_entropy": 0.686792, "bpb": 6.904331}, "de": {"chunks": 11892.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.579165, "cps": 92.90625, "mean_p": 0.506573, "boundary_entropy": 0.687732, "bpb": 6.995027}, "vi": {"chunks": 3716.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.201573, "cps": 29.03125, "mean_p": 0.475661, "boundary_entropy": 0.686649, "bpb": 7.200356}, "tr": {"chunks": 9639.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.532453, "cps": 75.304688, "mean_p": 0.503096, "boundary_entropy": 0.687052, "bpb": 7.195105}, "fi": {"chunks": 8339.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.443304, "cps": 65.148438, "mean_p": 0.498593, "boundary_entropy": 0.687378, "bpb": 6.97367}, "ru": {"chunks": 34323.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.81124, "cps": 268.148438, "mean_p": 0.556801, "boundary_entropy": 0.683947, "bpb": 6.514181}, "el": {"chunks": 18482.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.900551, "cps": 144.390625, "mean_p": 0.500428, "boundary_entropy": 0.690137, "bpb": 7.390383}, "hi": {"chunks": 879.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.052244, "cps": 6.867188, "mean_p": 0.462593, "boundary_entropy": 0.687558, "bpb": 6.860809}, "ta": {"chunks": 2300.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.123028, "cps": 17.96875, "mean_p": 0.480733, "boundary_entropy": 0.690313, "bpb": 6.968968}, "th": {"chunks": 2538.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.157787, "cps": 19.828125, "mean_p": 0.479496, "boundary_entropy": 0.689919, "bpb": 6.640838}, "zh": {"chunks": 10897.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.845073, "cps": 85.132812, "mean_p": 0.51621, "boundary_entropy": 0.685848, "bpb": 7.789539}, "my": {"chunks": 1808.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.095419, "cps": 14.125, "mean_p": 0.476179, "boundary_entropy": 0.689874, "bpb": 7.15753}}, "bpb": {"en": 6.90433, "de": 6.99503, "vi": 7.20036, "tr": 7.19511, "fi": 6.97367, "ru": 6.51418, "el": 7.39038, "hi": 6.86081, "ta": 6.96897, "th": 6.64084, "zh": 7.78954, "my": 7.15753}, "gini_cpc": 0.51193, "gini_cps": 0.483189, "bpic": 3.442878, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.5138499218125918, "vi": 0.5268814308071608, "tr": 1.3917516928223892, "fi": 1.1587304498452846, "ru": 4.734306331740497, "el": 2.353902391483435, "hi": 0.13655705165600868, "ta": 0.3215753092017751, "th": 0.41243060180908475, "zh": 4.8227398716875785, "my": 0.2494108712447144}, "t": 13.1} {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2308.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.132621, "cps": 18.03125, "mean_p": 0.481644, "boundary_entropy": 0.68632, "bpb": 6.38996}, "de": {"chunks": 3065.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.149272, "cps": 23.945312, "mean_p": 0.484047, "boundary_entropy": 0.687466, "bpb": 6.478985}, "vi": {"chunks": 3833.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.20792, "cps": 29.945312, "mean_p": 0.480477, "boundary_entropy": 0.687428, "bpb": 6.791523}, "tr": {"chunks": 7636.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.421809, "cps": 59.65625, "mean_p": 0.496497, "boundary_entropy": 0.687396, "bpb": 6.743888}, "fi": {"chunks": 2150.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.114295, "cps": 16.796875, "mean_p": 0.480775, "boundary_entropy": 0.686897, "bpb": 6.485423}, "ru": {"chunks": 19179.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.012084, "cps": 149.835938, "mean_p": 0.503842, "boundary_entropy": 0.69008, "bpb": 5.908903}, "el": {"chunks": 19086.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.929981, "cps": 149.109375, "mean_p": 0.500902, "boundary_entropy": 0.690219, "bpb": 6.297631}, "hi": {"chunks": 2335.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.138782, "cps": 18.242188, "mean_p": 0.481386, "boundary_entropy": 0.690017, "bpb": 5.886247}, "ta": {"chunks": 1125.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.060177, "cps": 8.789062, "mean_p": 0.478739, "boundary_entropy": 0.69016, "bpb": 6.054761}, "th": {"chunks": 2861.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.177868, "cps": 22.351562, "mean_p": 0.480808, "boundary_entropy": 0.690068, "bpb": 5.732033}, "zh": {"chunks": 4183.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.708263, "cps": 32.679688, "mean_p": 0.486739, "boundary_entropy": 0.685656, "bpb": 7.360237}, "my": {"chunks": 3442.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.181655, "cps": 26.890625, "mean_p": 0.485214, "boundary_entropy": 0.690747, "bpb": 5.880467}}, "bpb": {"en": 6.38996, "de": 6.47898, "vi": 6.79152, "tr": 6.74389, "fi": 6.48542, "ru": 5.9089, "el": 6.29763, "hi": 5.88625, "ta": 6.05476, "th": 5.73203, "zh": 7.36024, "my": 5.88047}, "gini_cpc": 0.461197, "gini_cps": 0.469136, "bpic": 5.389955, "worst_lang": "ru", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1255541339759871, "vi": 1.5677759320183944, "tr": 3.180560668104552, "fi": 0.8618166736948318, "ru": 7.631414810479095, "el": 7.0123307147216405, "hi": 1.0464539646323532, "ta": 0.4537486737526554, "th": 1.3411739464683126, "zh": 5.340510062028979, "my": 1.3697326423607947}, "t": 16.1} {"event": "train", "step": 50, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 4.27945, "bpb": 6.17394, "chunk_loss": 0.036796, "F": 0.25555, "G": 0.48649, "chunks": 261.69, "bytes_per_chunk": 3.913, "boundary_entropy": 0.69112, "router_noise": 0.0, "grad_norm": 1.7778, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.281, "alpha": 0.03, "t": 18.0} {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5229.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.300465, "cps": 40.851562, "mean_p": 0.495998, "boundary_entropy": 0.687459, "bpb": 5.77136}, "de": {"chunks": 7084.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.345006, "cps": 55.34375, "mean_p": 0.496505, "boundary_entropy": 0.688316, "bpb": 5.907654}, "vi": {"chunks": 6214.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.337076, "cps": 48.546875, "mean_p": 0.488856, "boundary_entropy": 0.688246, "bpb": 6.209108}, "tr": {"chunks": 6076.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.335635, "cps": 47.46875, "mean_p": 0.493689, "boundary_entropy": 0.687748, "bpb": 6.209525}, "fi": {"chunks": 2866.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.152358, "cps": 22.390625, "mean_p": 0.488785, "boundary_entropy": 0.687766, "bpb": 5.897939}, "ru": {"chunks": 13583.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.716781, "cps": 106.117188, "mean_p": 0.497351, "boundary_entropy": 0.690171, "bpb": 5.427274}, "el": {"chunks": 10584.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.515714, "cps": 82.6875, "mean_p": 0.494773, "boundary_entropy": 0.690372, "bpb": 5.66795}, "hi": {"chunks": 22392.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.330877, "cps": 174.9375, "mean_p": 0.503185, "boundary_entropy": 0.690852, "bpb": 5.148084}, "ta": {"chunks": 12939.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.69211, "cps": 101.085938, "mean_p": 0.493589, "boundary_entropy": 0.691062, "bpb": 5.238655}, "th": {"chunks": 3759.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.233696, "cps": 29.367188, "mean_p": 0.474032, "boundary_entropy": 0.689113, "bpb": 5.01401}, "zh": {"chunks": 6429.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.088554, "cps": 50.226562, "mean_p": 0.498948, "boundary_entropy": 0.686933, "bpb": 6.950183}, "my": {"chunks": 18249.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.96311, "cps": 142.570312, "mean_p": 0.498661, "boundary_entropy": 0.691344, "bpb": 5.104971}}, "bpb": {"en": 5.77136, "de": 5.90765, "vi": 6.20911, "tr": 6.20952, "fi": 5.89794, "ru": 5.42727, "el": 5.66795, "hi": 5.14808, "ta": 5.23866, "th": 5.01401, "zh": 6.95018, "my": 5.10497}, "gini_cpc": 0.337992, "gini_cps": 0.326212, "bpic": 3.325543, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1482372288550295, "vi": 1.1218468822790097, "tr": 1.1170502037981236, "fi": 0.5070721764004917, "ru": 2.3855689020688864, "el": 1.7163840275588542, "hi": 4.429383575482145, "ta": 2.3034602475925943, "th": 0.777779941650344, "zh": 3.622892615403805, "my": 3.2053921830450367}, "t": 20.2} {"event": "train", "step": 100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 3.31402, "bpb": 4.78112, "chunk_loss": 0.036663, "F": 0.26346, "G": 0.47713, "chunks": 269.78, "bytes_per_chunk": 3.796, "boundary_entropy": 0.68835, "router_noise": 0.0, "grad_norm": 1.8275, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 23.2} {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4064.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.233523, "cps": 31.75, "mean_p": 0.415356, "boundary_entropy": 0.651974, "bpb": 4.378619}, "de": {"chunks": 5412.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.263576, "cps": 42.28125, "mean_p": 0.415759, "boundary_entropy": 0.654045, "bpb": 4.59777}, "vi": {"chunks": 5078.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.275454, "cps": 39.671875, "mean_p": 0.463629, "boundary_entropy": 0.680883, "bpb": 4.803996}, "tr": {"chunks": 7292.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.402806, "cps": 56.96875, "mean_p": 0.448729, "boundary_entropy": 0.666945, "bpb": 4.866306}, "fi": {"chunks": 6274.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.333528, "cps": 49.015625, "mean_p": 0.438726, "boundary_entropy": 0.666098, "bpb": 4.637786}, "ru": {"chunks": 5666.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.298997, "cps": 44.265625, "mean_p": 0.486257, "boundary_entropy": 0.688108, "bpb": 3.656264}, "el": {"chunks": 7909.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.385373, "cps": 61.789062, "mean_p": 0.486416, "boundary_entropy": 0.689627, "bpb": 4.204641}, "hi": {"chunks": 18162.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.079465, "cps": 141.890625, "mean_p": 0.493553, "boundary_entropy": 0.689882, "bpb": 3.447751}, "ta": {"chunks": 10777.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.576464, "cps": 84.195312, "mean_p": 0.482447, "boundary_entropy": 0.689311, "bpb": 3.517475}, "th": {"chunks": 13945.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.866957, "cps": 108.945312, "mean_p": 0.484644, "boundary_entropy": 0.689517, "bpb": 3.278706}, "zh": {"chunks": 5774.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.97765, "cps": 45.109375, "mean_p": 0.496082, "boundary_entropy": 0.68666, "bpb": 6.276391}, "my": {"chunks": 44620.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.354866, "cps": 348.59375, "mean_p": 0.512327, "boundary_entropy": 0.690984, "bpb": 3.64506}}, "bpb": {"en": 4.37862, "de": 4.59777, "vi": 4.804, "tr": 4.86631, "fi": 4.63779, "ru": 3.65626, "el": 4.20464, "hi": 3.44775, "ta": 3.51748, "th": 3.27871, "zh": 6.27639, "my": 3.64506}, "gini_cpc": 0.405649, "gini_cps": 0.405378, "bpic": 2.843391, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1286929222058901, "vi": 1.1795598335500985, "tr": 1.724910355485518, "fi": 1.4282461164664502, "ru": 1.2803767425674693, "el": 1.650255357346338, "hi": 4.62252234676089, "ta": 2.4685551502254386, "th": 3.7125120698601672, "zh": 4.186525663878453, "my": 10.084087625228765}, "t": 26.2} {"event": "train", "step": 150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.60031, "bpb": 3.75145, "chunk_loss": 0.039175, "F": 0.29712, "G": 0.45473, "chunks": 304.25, "bytes_per_chunk": 3.366, "boundary_entropy": 0.6741, "router_noise": 0.0, "grad_norm": 1.2548, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.099, "alpha": 0.03, "t": 28.1} {"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4062.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.233408, "cps": 31.734375, "mean_p": 0.410719, "boundary_entropy": 0.645541, "bpb": 3.783967}, "de": {"chunks": 4560.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.222082, "cps": 35.625, "mean_p": 0.404765, "boundary_entropy": 0.646344, "bpb": 4.044019}, "vi": {"chunks": 6315.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.342555, "cps": 49.335938, "mean_p": 0.440678, "boundary_entropy": 0.66864, "bpb": 3.548962}, "tr": {"chunks": 5509.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.304314, "cps": 43.039062, "mean_p": 0.430679, "boundary_entropy": 0.658935, "bpb": 4.144513}, "fi": {"chunks": 4068.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.216256, "cps": 31.78125, "mean_p": 0.411518, "boundary_entropy": 0.653545, "bpb": 4.054327}, "ru": {"chunks": 6689.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.352982, "cps": 52.257812, "mean_p": 0.492926, "boundary_entropy": 0.689453, "bpb": 2.838923}, "el": {"chunks": 12903.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.628709, "cps": 100.804688, "mean_p": 0.490079, "boundary_entropy": 0.689246, "bpb": 2.981507}, "hi": {"chunks": 4495.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.267162, "cps": 35.117188, "mean_p": 0.462044, "boundary_entropy": 0.680409, "bpb": 2.439709}, "ta": {"chunks": 10110.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.540786, "cps": 78.984375, "mean_p": 0.485445, "boundary_entropy": 0.685047, "bpb": 2.260618}, "th": {"chunks": 5908.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.367299, "cps": 46.15625, "mean_p": 0.488001, "boundary_entropy": 0.690345, "bpb": 2.33115}, "zh": {"chunks": 3497.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.59211, "cps": 27.320312, "mean_p": 0.396131, "boundary_entropy": 0.612196, "bpb": 4.907498}, "my": {"chunks": 12250.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.646506, "cps": 95.703125, "mean_p": 0.486044, "boundary_entropy": 0.689435, "bpb": 2.438838}}, "bpb": {"en": 3.78397, "de": 4.04402, "vi": 3.54896, "tr": 4.14451, "fi": 4.05433, "ru": 2.83892, "el": 2.98151, "hi": 2.43971, "ta": 2.26062, "th": 2.33115, "zh": 4.9075, "my": 2.43884}, "gini_cpc": 0.220449, "gini_cps": 0.244803, "bpic": 4.775415, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9514733676904872, "vi": 1.467622678427418, "tr": 1.3037863265765217, "fi": 0.9265167219249367, "ru": 1.5122938386409077, "el": 2.6936058913186782, "hi": 1.1446133957828986, "ta": 2.316913858198722, "th": 1.5736336091177363, "zh": 2.536800944986972, "my": 2.7698542290507273}, "t": 32.4} {"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.78817, "de": 4.04367, "vi": 3.55378, "tr": 4.15334, "fi": 4.05487, "ru": 2.82523, "el": 2.97494, "hi": 2.43544, "ta": 2.27778, "th": 2.36546, "zh": 4.91342, "my": 2.45517}, "bytes_per_chunk": {"en": 4.302, "de": 4.502, "vi": 3.881, "tr": 3.342, "fi": 4.416, "ru": 4.811, "el": 2.943, "hi": 11.094, "ta": 4.852, "th": 9.689, "zh": 5.759, "my": 4.513}, "bpb_mean": 3.32011, "frozen_inventory": false, "t": 33.7} {"event": "train", "step": 200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.45927, "bpb": 3.54798, "chunk_loss": 0.032526, "F": 0.21204, "G": 0.4356, "chunks": 217.12, "bytes_per_chunk": 4.716, "boundary_entropy": 0.65994, "router_noise": 0.0, "grad_norm": 1.7307, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.12, "alpha": 0.03, "t": 34.1} {"event": "train", "step": 250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.37827, "bpb": 3.43111, "chunk_loss": 0.033899, "F": 0.23883, "G": 0.43206, "chunks": 244.56, "bytes_per_chunk": 4.187, "boundary_entropy": 0.65592, "router_noise": 0.0, "grad_norm": 1.3553, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.065, "alpha": 0.03, "t": 37.4} {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4314.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.247888, "cps": 33.703125, "mean_p": 0.418536, "boundary_entropy": 0.650516, "bpb": 3.545253}, "de": {"chunks": 4497.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.219013, "cps": 35.132812, "mean_p": 0.415363, "boundary_entropy": 0.653615, "bpb": 3.7962}, "vi": {"chunks": 5859.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.317819, "cps": 45.773438, "mean_p": 0.451181, "boundary_entropy": 0.673971, "bpb": 3.184367}, "tr": {"chunks": 5979.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.330277, "cps": 46.710938, "mean_p": 0.444235, "boundary_entropy": 0.664132, "bpb": 3.782012}, "fi": {"chunks": 3756.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.19967, "cps": 29.34375, "mean_p": 0.419771, "boundary_entropy": 0.658907, "bpb": 3.856645}, "ru": {"chunks": 5319.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.280686, "cps": 41.554688, "mean_p": 0.456939, "boundary_entropy": 0.68304, "bpb": 2.650396}, "el": {"chunks": 16770.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.817132, "cps": 131.015625, "mean_p": 0.486496, "boundary_entropy": 0.68807, "bpb": 2.758625}, "hi": {"chunks": 19167.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.139198, "cps": 149.742188, "mean_p": 0.473806, "boundary_entropy": 0.676038, "bpb": 2.079608}, "ta": {"chunks": 14230.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.761166, "cps": 111.171875, "mean_p": 0.48652, "boundary_entropy": 0.677979, "bpb": 1.889506}, "th": {"chunks": 14457.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.898788, "cps": 112.945312, "mean_p": 0.463178, "boundary_entropy": 0.684029, "bpb": 2.082515}, "zh": {"chunks": 8268.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.399932, "cps": 64.59375, "mean_p": 0.40368, "boundary_entropy": 0.599166, "bpb": 4.513753}, "my": {"chunks": 12432.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.656111, "cps": 97.125, "mean_p": 0.480378, "boundary_entropy": 0.687487, "bpb": 2.181568}}, "bpb": {"en": 3.54525, "de": 3.7962, "vi": 3.18437, "tr": 3.78201, "fi": 3.85665, "ru": 2.6504, "el": 2.75862, "hi": 2.07961, "ta": 1.88951, "th": 2.08252, "zh": 4.51375, "my": 2.18157}, "gini_cpc": 0.348334, "gini_cps": 0.302706, "bpic": 3.335834, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8835160835771204, "vi": 1.2821071893768015, "tr": 1.332361213538017, "fi": 0.8054854123541629, "ru": 1.1323084685927758, "el": 3.29637185106001, "hi": 4.595608768162815, "ta": 3.0706011625757297, "th": 3.625777045883245, "zh": 5.6474319272685145, "my": 2.6468029183450033}, "t": 38.8} {"event": "train", "step": 300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.27353, "bpb": 3.28001, "chunk_loss": 0.035164, "F": 0.25809, "G": 0.42726, "chunks": 264.28, "bytes_per_chunk": 3.875, "boundary_entropy": 0.66144, "router_noise": 0.0, "grad_norm": 1.4142, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.092, "alpha": 0.03, "t": 41.9} {"event": "train", "step": 350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.04615, "bpb": 2.95198, "chunk_loss": 0.03441, "F": 0.24316, "G": 0.42349, "chunks": 249.0, "bytes_per_chunk": 4.112, "boundary_entropy": 0.65701, "router_noise": 0.0, "grad_norm": 0.9525, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.062, "alpha": 0.03, "t": 45.1} {"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4524.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.259955, "cps": 35.34375, "mean_p": 0.429697, "boundary_entropy": 0.657544, "bpb": 3.171848}, "de": {"chunks": 4795.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.233527, "cps": 37.460938, "mean_p": 0.429579, "boundary_entropy": 0.662062, "bpb": 3.482303}, "vi": {"chunks": 5023.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.272471, "cps": 39.242188, "mean_p": 0.445532, "boundary_entropy": 0.673614, "bpb": 2.755008}, "tr": {"chunks": 5028.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.277744, "cps": 39.28125, "mean_p": 0.443784, "boundary_entropy": 0.667289, "bpb": 3.408562}, "fi": {"chunks": 4210.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.223805, "cps": 32.890625, "mean_p": 0.437155, "boundary_entropy": 0.668633, "bpb": 3.549843}, "ru": {"chunks": 5506.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.290554, "cps": 43.015625, "mean_p": 0.404746, "boundary_entropy": 0.628874, "bpb": 2.37118}, "el": {"chunks": 4222.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.20572, "cps": 32.984375, "mean_p": 0.431322, "boundary_entropy": 0.673914, "bpb": 2.513564}, "hi": {"chunks": 5556.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.330223, "cps": 43.40625, "mean_p": 0.420199, "boundary_entropy": 0.659335, "bpb": 1.794028}, "ta": {"chunks": 11249.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.601712, "cps": 87.882812, "mean_p": 0.423678, "boundary_entropy": 0.650639, "bpb": 1.575525}, "th": {"chunks": 13296.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.826609, "cps": 103.875, "mean_p": 0.428207, "boundary_entropy": 0.669012, "bpb": 1.928887}, "zh": {"chunks": 2758.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.466983, "cps": 21.546875, "mean_p": 0.409918, "boundary_entropy": 0.641344, "bpb": 4.137728}, "my": {"chunks": 19062.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.006016, "cps": 148.921875, "mean_p": 0.451038, "boundary_entropy": 0.67678, "bpb": 1.934919}}, "bpb": {"en": 3.17185, "de": 3.4823, "vi": 2.75501, "tr": 3.40856, "fi": 3.54984, "ru": 2.37118, "el": 2.51356, "hi": 1.79403, "ta": 1.57553, "th": 1.92889, "zh": 4.13773, "my": 1.93492}, "gini_cpc": 0.300731, "gini_cps": 0.310541, "bpic": 4.502939, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8983337749247798, "vi": 1.0481454662917025, "tr": 1.068430412416073, "fi": 0.8609377209875084, "ru": 1.1177084047787351, "el": 0.791368770316696, "hi": 1.2703069907496818, "ta": 2.314674734659333, "th": 3.179812155087899, "zh": 1.7963970911207503, "my": 3.869961220125554}, "t": 49.5} {"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 3.18321, "de": 3.4804, "vi": 2.75852, "tr": 3.41885, "fi": 3.54412, "ru": 2.35659, "el": 2.50692, "hi": 1.78948, "ta": 1.5918, "th": 1.95814, "zh": 4.14357, "my": 1.95053}, "bytes_per_chunk": {"en": 3.563, "de": 4.113, "vi": 4.566, "tr": 3.832, "fi": 4.339, "ru": 6.276, "el": 8.768, "hi": 8.308, "ta": 3.983, "th": 3.575, "zh": 5.131, "my": 2.794}, "bpb_mean": 2.72351, "frozen_inventory": false, "t": 51.4} {"event": "train", "step": 400, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.19669, "bpb": 3.16915, "chunk_loss": 0.033629, "F": 0.23059, "G": 0.42252, "chunks": 236.12, "bytes_per_chunk": 4.337, "boundary_entropy": 0.65482, "router_noise": 0.0, "grad_norm": 0.783, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.136, "alpha": 0.03, "t": 51.9} {"event": "train", "step": 450, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.15786, "bpb": 3.11313, "chunk_loss": 0.034299, "F": 0.24161, "G": 0.42622, "chunks": 247.41, "bytes_per_chunk": 4.139, "boundary_entropy": 0.65962, "router_noise": 0.0, "grad_norm": 0.926, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.12, "alpha": 0.03, "t": 57.9} {"event": "train", "step": 500, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.06138, "bpb": 2.97395, "chunk_loss": 0.034435, "F": 0.24274, "G": 0.42948, "chunks": 248.56, "bytes_per_chunk": 4.12, "boundary_entropy": 0.66233, "router_noise": 0.0, "grad_norm": 1.231, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.152, "alpha": 0.03, "t": 65.5} {"event": "train", "step": 550, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.92923, "bpb": 2.78329, "chunk_loss": 0.032616, "F": 0.21173, "G": 0.42695, "chunks": 216.81, "bytes_per_chunk": 4.723, "boundary_entropy": 0.66123, "router_noise": 0.0, "grad_norm": 0.7954, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 73.9} {"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4696.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.269839, "cps": 36.6875, "mean_p": 0.437933, "boundary_entropy": 0.661751, "bpb": 2.869544}, "de": {"chunks": 5620.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.273706, "cps": 43.90625, "mean_p": 0.444191, "boundary_entropy": 0.667979, "bpb": 3.17769}, "vi": {"chunks": 5200.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.282072, "cps": 40.625, "mean_p": 0.444683, "boundary_entropy": 0.67177, "bpb": 2.53192}, "tr": {"chunks": 5627.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.310832, "cps": 43.960938, "mean_p": 0.451603, "boundary_entropy": 0.669638, "bpb": 3.100561}, "fi": {"chunks": 4195.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.223008, "cps": 32.773438, "mean_p": 0.436166, "boundary_entropy": 0.667579, "bpb": 3.351452}, "ru": {"chunks": 7609.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.40153, "cps": 59.445312, "mean_p": 0.427969, "boundary_entropy": 0.651208, "bpb": 2.17122}, "el": {"chunks": 6142.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.299274, "cps": 47.984375, "mean_p": 0.429758, "boundary_entropy": 0.666216, "bpb": 2.23198}, "hi": {"chunks": 6224.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.369926, "cps": 48.625, "mean_p": 0.427892, "boundary_entropy": 0.663442, "bpb": 1.656171}, "ta": {"chunks": 8982.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.480449, "cps": 70.171875, "mean_p": 0.41858, "boundary_entropy": 0.652894, "bpb": 1.397879}, "th": {"chunks": 4998.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.310724, "cps": 39.046875, "mean_p": 0.417089, "boundary_entropy": 0.664461, "bpb": 1.764239}, "zh": {"chunks": 3194.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.540806, "cps": 24.953125, "mean_p": 0.422643, "boundary_entropy": 0.655078, "bpb": 3.942878}, "my": {"chunks": 23320.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.230737, "cps": 182.1875, "mean_p": 0.489722, "boundary_entropy": 0.683317, "bpb": 1.670405}}, "bpb": {"en": 2.86954, "de": 3.17769, "vi": 2.53192, "tr": 3.10056, "fi": 3.35145, "ru": 2.17122, "el": 2.23198, "hi": 1.65617, "ta": 1.39788, "th": 1.76424, "zh": 3.94288, "my": 1.6704}, "gini_cpc": 0.263379, "gini_cps": 0.285353, "bpic": 4.472607, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0143315646537658, "vi": 1.0453367857692366, "tr": 1.1519202041959218, "fi": 0.8264491049991862, "ru": 1.4880393013983717, "el": 1.1090854271531976, "hi": 1.370915046411729, "ta": 1.7805067038212714, "th": 1.1515192924725384, "zh": 2.004183586581601, "my": 4.561011864624387}, "t": 85.6} {"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.87762, "de": 3.17799, "vi": 2.5376, "tr": 3.11225, "fi": 3.34352, "ru": 2.1541, "el": 2.22097, "hi": 1.64971, "ta": 1.41218, "th": 1.793, "zh": 3.95223, "my": 1.68373}, "bytes_per_chunk": {"en": 3.692, "de": 3.715, "vi": 4.93, "tr": 3.495, "fi": 4.587, "ru": 4.423, "el": 6.651, "hi": 7.297, "ta": 5.685, "th": 8.473, "zh": 4.403, "my": 2.3}, "bpb_mean": 2.49291, "frozen_inventory": false, "t": 90.6} {"event": "train", "step": 600, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.90372, "bpb": 2.74649, "chunk_loss": 0.033454, "F": 0.22635, "G": 0.43139, "chunks": 231.78, "bytes_per_chunk": 4.418, "boundary_entropy": 0.66509, "router_noise": 0.0, "grad_norm": 0.8332, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.345, "alpha": 0.03, "t": 91.2} {"event": "train", "step": 650, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.9427, "bpb": 2.80273, "chunk_loss": 0.035337, "F": 0.2576, "G": 0.43646, "chunks": 263.78, "bytes_per_chunk": 3.882, "boundary_entropy": 0.66668, "router_noise": 0.0, "grad_norm": 0.7685, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 100.1} {"event": "train", "step": 700, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.78194, "bpb": 2.5708, "chunk_loss": 0.034117, "F": 0.23749, "G": 0.4354, "chunks": 243.19, "bytes_per_chunk": 4.211, "boundary_entropy": 0.66526, "router_noise": 0.0, "grad_norm": 0.9415, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 108.7} {"event": "train", "step": 750, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.86252, "bpb": 2.68704, "chunk_loss": 0.034586, "F": 0.24408, "G": 0.43611, "chunks": 249.94, "bytes_per_chunk": 4.097, "boundary_entropy": 0.66657, "router_noise": 0.0, "grad_norm": 0.7466, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 117.2} {"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4828.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.277423, "cps": 37.71875, "mean_p": 0.442241, "boundary_entropy": 0.664711, "bpb": 2.66355}, "de": {"chunks": 5754.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.280232, "cps": 44.953125, "mean_p": 0.447745, "boundary_entropy": 0.669927, "bpb": 3.008263}, "vi": {"chunks": 5144.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.279034, "cps": 40.1875, "mean_p": 0.44528, "boundary_entropy": 0.672374, "bpb": 2.359905}, "tr": {"chunks": 5601.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.309396, "cps": 43.757812, "mean_p": 0.456435, "boundary_entropy": 0.672739, "bpb": 2.996157}, "fi": {"chunks": 4446.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.236351, "cps": 34.734375, "mean_p": 0.447292, "boundary_entropy": 0.67221, "bpb": 3.208088}, "ru": {"chunks": 6135.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.323747, "cps": 47.929688, "mean_p": 0.430708, "boundary_entropy": 0.658622, "bpb": 2.036752}, "el": {"chunks": 8512.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.414754, "cps": 66.5, "mean_p": 0.446153, "boundary_entropy": 0.672229, "bpb": 2.124456}, "hi": {"chunks": 10211.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.606895, "cps": 79.773438, "mean_p": 0.445928, "boundary_entropy": 0.66902, "bpb": 1.569797}, "ta": {"chunks": 11290.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.603905, "cps": 88.203125, "mean_p": 0.423039, "boundary_entropy": 0.649901, "bpb": 1.356767}, "th": {"chunks": 998.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.062045, "cps": 7.796875, "mean_p": 0.404602, "boundary_entropy": 0.661705, "bpb": 1.640402}, "zh": {"chunks": 2728.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.461903, "cps": 21.3125, "mean_p": 0.433396, "boundary_entropy": 0.666271, "bpb": 3.827874}, "my": {"chunks": 9170.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.483956, "cps": 71.640625, "mean_p": 0.472428, "boundary_entropy": 0.685712, "bpb": 1.49724}}, "bpb": {"en": 2.66355, "de": 3.00826, "vi": 2.3599, "tr": 2.99616, "fi": 3.20809, "ru": 2.03675, "el": 2.12446, "hi": 1.5698, "ta": 1.35677, "th": 1.6404, "zh": 3.82787, "my": 1.49724}, "gini_cpc": 0.231957, "gini_cps": 0.261942, "bpic": 5.129596, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0101231146047318, "vi": 1.005807053106944, "tr": 1.115249096340507, "fi": 0.8519506403844426, "ru": 1.1669767713841641, "el": 1.495022154907657, "hi": 2.1876108165468646, "ta": 2.1768340958388186, "th": 0.22364867770870317, "zh": 1.6649752501984278, "my": 1.7444672411775723}, "t": 128.8} {"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.67257, "de": 3.01332, "vi": 2.36568, "tr": 3.00655, "fi": 3.20671, "ru": 2.02081, "el": 2.11068, "hi": 1.56452, "ta": 1.36839, "th": 1.66676, "zh": 3.83762, "my": 1.50777}, "bytes_per_chunk": {"en": 3.595, "de": 3.592, "vi": 4.964, "tr": 3.563, "fi": 4.163, "ru": 5.602, "el": 4.426, "hi": 5.12, "ta": 4.765, "th": 41.198, "zh": 5.09, "my": 6.605}, "bpb_mean": 2.36178, "frozen_inventory": false, "t": 133.9} {"event": "train", "step": 800, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.8441, "bpb": 2.66048, "chunk_loss": 0.033188, "F": 0.21976, "G": 0.43556, "chunks": 225.03, "bytes_per_chunk": 4.55, "boundary_entropy": 0.66847, "router_noise": 0.0, "grad_norm": 0.811, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.341, "alpha": 0.03, "t": 134.2} {"event": "train", "step": 850, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.64442, "bpb": 2.3724, "chunk_loss": 0.035413, "F": 0.25815, "G": 0.43979, "chunks": 264.34, "bytes_per_chunk": 3.874, "boundary_entropy": 0.66866, "router_noise": 0.0, "grad_norm": 0.7676, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.165, "alpha": 0.03, "t": 142.5} {"event": "train", "step": 900, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.72557, "bpb": 2.48947, "chunk_loss": 0.034022, "F": 0.23334, "G": 0.43982, "chunks": 238.94, "bytes_per_chunk": 4.286, "boundary_entropy": 0.66957, "router_noise": 0.0, "grad_norm": 0.7671, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.171, "alpha": 0.03, "t": 151.1} {"event": "train", "step": 950, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.88159, "bpb": 2.71456, "chunk_loss": 0.03433, "F": 0.24026, "G": 0.43825, "chunks": 246.03, "bytes_per_chunk": 4.162, "boundary_entropy": 0.66898, "router_noise": 0.0, "grad_norm": 0.8157, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 159.5} {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4841.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.27817, "cps": 37.820312, "mean_p": 0.439367, "boundary_entropy": 0.661725, "bpb": 2.541487}, "de": {"chunks": 4770.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.232309, "cps": 37.265625, "mean_p": 0.440207, "boundary_entropy": 0.668549, "bpb": 2.882899}, "vi": {"chunks": 6834.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.370708, "cps": 53.390625, "mean_p": 0.447519, "boundary_entropy": 0.671012, "bpb": 2.270952}, "tr": {"chunks": 5638.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.31144, "cps": 44.046875, "mean_p": 0.451018, "boundary_entropy": 0.669016, "bpb": 2.855488}, "fi": {"chunks": 3832.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.203711, "cps": 29.9375, "mean_p": 0.43611, "boundary_entropy": 0.668679, "bpb": 3.107249}, "ru": {"chunks": 9490.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.500792, "cps": 74.140625, "mean_p": 0.437689, "boundary_entropy": 0.659218, "bpb": 1.931233}, "el": {"chunks": 8453.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.411879, "cps": 66.039062, "mean_p": 0.447874, "boundary_entropy": 0.674249, "bpb": 2.020058}, "hi": {"chunks": 2382.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.141575, "cps": 18.609375, "mean_p": 0.40803, "boundary_entropy": 0.655284, "bpb": 1.51524}, "ta": {"chunks": 7906.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.422894, "cps": 61.765625, "mean_p": 0.407987, "boundary_entropy": 0.647708, "bpb": 1.319647}, "th": {"chunks": 3357.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.208704, "cps": 26.226562, "mean_p": 0.427368, "boundary_entropy": 0.669399, "bpb": 1.564613}, "zh": {"chunks": 2628.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.444971, "cps": 20.53125, "mean_p": 0.447949, "boundary_entropy": 0.674024, "bpb": 3.693845}, "my": {"chunks": 19596.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.034199, "cps": 153.09375, "mean_p": 0.457377, "boundary_entropy": 0.680718, "bpb": 1.413303}}, "bpb": {"en": 2.54149, "de": 2.8829, "vi": 2.27095, "tr": 2.85549, "fi": 3.10725, "ru": 1.93123, "el": 2.02006, "hi": 1.51524, "ta": 1.31965, "th": 1.56461, "zh": 3.69384, "my": 1.4133}, "gini_cpc": 0.284268, "gini_cps": 0.328353, "bpic": 4.813689, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8351317777833019, "vi": 1.3326646260775323, "tr": 1.1196017217596346, "fi": 0.732322966831335, "ru": 1.8003047844952333, "el": 1.4806726735809017, "hi": 0.5089507067694744, "ta": 1.5202687837501054, "th": 0.7502729926361432, "zh": 1.599635213189518, "my": 3.717860516678639}, "t": 171.2} {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.54946, "de": 2.88782, "vi": 2.27596, "tr": 2.86791, "fi": 3.10547, "ru": 1.91344, "el": 2.00898, "hi": 1.51152, "ta": 1.32788, "th": 1.58869, "zh": 3.70543, "my": 1.42146}, "bytes_per_chunk": {"en": 3.544, "de": 4.365, "vi": 3.412, "tr": 3.501, "fi": 4.872, "ru": 3.718, "el": 4.341, "hi": 18.268, "ta": 7.284, "th": 13.966, "zh": 5.362, "my": 2.749}, "bpb_mean": 2.26367, "frozen_inventory": false, "t": 175.9} {"event": "train", "step": 1000, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.55568, "bpb": 2.24437, "chunk_loss": 0.032694, "F": 0.21045, "G": 0.43316, "chunks": 215.5, "bytes_per_chunk": 4.752, "boundary_entropy": 0.66686, "router_noise": 0.0, "grad_norm": 0.7411, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.342, "alpha": 0.03, "t": 176.6} {"event": "train", "step": 1050, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.76882, "bpb": 2.55187, "chunk_loss": 0.034058, "F": 0.23288, "G": 0.44223, "chunks": 238.47, "bytes_per_chunk": 4.294, "boundary_entropy": 0.67125, "router_noise": 0.0, "grad_norm": 0.8713, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.172, "alpha": 0.03, "t": 185.2} {"event": "train", "step": 1100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.5984, "bpb": 2.30601, "chunk_loss": 0.035975, "F": 0.26703, "G": 0.44105, "chunks": 273.44, "bytes_per_chunk": 3.745, "boundary_entropy": 0.66975, "router_noise": 0.0, "grad_norm": 0.8232, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.171, "alpha": 0.03, "t": 193.7} {"event": "train", "step": 1150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.61493, "bpb": 2.32986, "chunk_loss": 0.036331, "F": 0.27469, "G": 0.43841, "chunks": 281.28, "bytes_per_chunk": 3.64, "boundary_entropy": 0.66833, "router_noise": 0.0, "grad_norm": 0.8385, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 202.4} {"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4953.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.284606, "cps": 38.695312, "mean_p": 0.442106, "boundary_entropy": 0.662108, "bpb": 2.460986}, "de": {"chunks": 5926.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.288609, "cps": 46.296875, "mean_p": 0.449447, "boundary_entropy": 0.670378, "bpb": 2.787501}, "vi": {"chunks": 6516.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.353458, "cps": 50.90625, "mean_p": 0.446459, "boundary_entropy": 0.670954, "bpb": 2.21777}, "tr": {"chunks": 6389.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.352925, "cps": 49.914062, "mean_p": 0.462783, "boundary_entropy": 0.671956, "bpb": 2.780856}, "fi": {"chunks": 4823.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.256393, "cps": 37.679688, "mean_p": 0.44933, "boundary_entropy": 0.672246, "bpb": 3.031551}, "ru": {"chunks": 7472.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.394301, "cps": 58.375, "mean_p": 0.433307, "boundary_entropy": 0.659441, "bpb": 1.848449}, "el": {"chunks": 7094.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.345661, "cps": 55.421875, "mean_p": 0.444874, "boundary_entropy": 0.675679, "bpb": 1.943765}, "hi": {"chunks": 7315.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.43477, "cps": 57.148438, "mean_p": 0.4315, "boundary_entropy": 0.664757, "bpb": 1.471687}, "ta": {"chunks": 6026.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.322332, "cps": 47.078125, "mean_p": 0.410618, "boundary_entropy": 0.654534, "bpb": 1.204868}, "th": {"chunks": 10755.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.668635, "cps": 84.023438, "mean_p": 0.427228, "boundary_entropy": 0.666432, "bpb": 1.458618}, "zh": {"chunks": 2835.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.48002, "cps": 22.148438, "mean_p": 0.453982, "boundary_entropy": 0.676763, "bpb": 3.63533}, "my": {"chunks": 5281.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.27871, "cps": 41.257812, "mean_p": 0.45083, "boundary_entropy": 0.682506, "bpb": 1.380043}}, "bpb": {"en": 2.46099, "de": 2.7875, "vi": 2.21777, "tr": 2.78086, "fi": 3.03155, "ru": 1.84845, "el": 1.94376, "hi": 1.47169, "ta": 1.20487, "th": 1.45862, "zh": 3.63533, "my": 1.38004}, "gini_cpc": 0.149962, "gini_cps": 0.153316, "bpic": 5.090946, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0140632223915864, "vi": 1.2419202998010428, "tr": 1.2400469512015386, "fi": 0.9008680211821112, "ru": 1.3854263427138587, "el": 1.2145240837620157, "hi": 1.527618994476759, "ta": 1.1325553867763771, "th": 2.3493360435466717, "zh": 1.686612881027476, "my": 0.9792838303868908}, "t": 213.9} {"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.46793, "de": 2.78816, "vi": 2.22186, "tr": 2.79089, "fi": 3.03274, "ru": 1.83084, "el": 1.93428, "hi": 1.46822, "ta": 1.21513, "th": 1.48134, "zh": 3.64216, "my": 1.38813}, "bytes_per_chunk": {"en": 3.48, "de": 3.529, "vi": 3.796, "tr": 3.032, "fi": 3.86, "ru": 4.781, "el": 5.426, "hi": 6.599, "ta": 9.548, "th": 3.847, "zh": 5.366, "my": 10.162}, "bpb_mean": 2.18847, "frozen_inventory": false, "t": 218.7} {"event": "train", "step": 1200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.69979, "bpb": 2.45229, "chunk_loss": 0.034754, "F": 0.24689, "G": 0.44028, "chunks": 252.81, "bytes_per_chunk": 4.05, "boundary_entropy": 0.66962, "router_noise": 0.0, "grad_norm": 0.7955, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.334, "alpha": 0.03, "t": 219.1} {"event": "train", "step": 1250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.6012, "bpb": 2.31004, "chunk_loss": 0.034039, "F": 0.23441, "G": 0.44015, "chunks": 240.03, "bytes_per_chunk": 4.266, "boundary_entropy": 0.6703, "router_noise": 0.0, "grad_norm": 0.8553, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.161, "alpha": 0.03, "t": 227.2} {"event": "train", "step": 1300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.71091, "bpb": 2.46832, "chunk_loss": 0.034607, "F": 0.24371, "G": 0.43938, "chunks": 249.56, "bytes_per_chunk": 4.103, "boundary_entropy": 0.66937, "router_noise": 0.0, "grad_norm": 0.7163, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.171, "alpha": 0.03, "t": 235.7} {"event": "train", "step": 1350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.50946, "bpb": 2.1777, "chunk_loss": 0.035403, "F": 0.25876, "G": 0.43995, "chunks": 264.97, "bytes_per_chunk": 3.865, "boundary_entropy": 0.66863, "router_noise": 0.0, "grad_norm": 0.8155, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.159, "alpha": 0.03, "t": 243.7} {"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4843.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.278285, "cps": 37.835938, "mean_p": 0.44033, "boundary_entropy": 0.660869, "bpb": 2.384808}, "de": {"chunks": 5409.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.26343, "cps": 42.257812, "mean_p": 0.445602, "boundary_entropy": 0.668193, "bpb": 2.736342}, "vi": {"chunks": 6727.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.364904, "cps": 52.554688, "mean_p": 0.444582, "boundary_entropy": 0.668997, "bpb": 2.159138}, "tr": {"chunks": 5562.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.307242, "cps": 43.453125, "mean_p": 0.456974, "boundary_entropy": 0.670523, "bpb": 2.671654}, "fi": {"chunks": 4359.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.231726, "cps": 34.054688, "mean_p": 0.446222, "boundary_entropy": 0.671471, "bpb": 2.950627}, "ru": {"chunks": 9095.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.479947, "cps": 71.054688, "mean_p": 0.43743, "boundary_entropy": 0.65947, "bpb": 1.80796}, "el": {"chunks": 6387.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.311212, "cps": 49.898438, "mean_p": 0.445428, "boundary_entropy": 0.676201, "bpb": 1.858289}, "hi": {"chunks": 7355.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.437147, "cps": 57.460938, "mean_p": 0.441172, "boundary_entropy": 0.670793, "bpb": 1.432285}, "ta": {"chunks": 5470.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.292592, "cps": 42.734375, "mean_p": 0.421644, "boundary_entropy": 0.659271, "bpb": 1.168815}, "th": {"chunks": 3233.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.200995, "cps": 25.257812, "mean_p": 0.42195, "boundary_entropy": 0.667117, "bpb": 1.393427}, "zh": {"chunks": 3135.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.530816, "cps": 24.492188, "mean_p": 0.457349, "boundary_entropy": 0.677488, "bpb": 3.573694}, "my": {"chunks": 6582.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.347372, "cps": 51.421875, "mean_p": 0.460388, "boundary_entropy": 0.68448, "bpb": 1.349226}}, "bpb": {"en": 2.38481, "de": 2.73634, "vi": 2.15914, "tr": 2.67165, "fi": 2.95063, "ru": 1.80796, "el": 1.85829, "hi": 1.43228, "ta": 1.16881, "th": 1.39343, "zh": 3.57369, "my": 1.34923}, "gini_cpc": 0.158183, "gini_cps": 0.160114, "bpic": 5.630838, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9466168384190479, "vi": 1.311257353982588, "tr": 1.1040534120325922, "fi": 0.8326924687760674, "ru": 1.7246586074507342, "el": 1.1183190479552316, "hi": 1.570859183411115, "ta": 1.0514085587804758, "th": 0.7222612089873826, "zh": 1.907452596004903, "my": 1.2482574106226194}, "t": 255.0} {"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.39489, "de": 2.74123, "vi": 2.16288, "tr": 2.6851, "fi": 2.95291, "ru": 1.78807, "el": 1.84774, "hi": 1.42848, "ta": 1.17634, "th": 1.41404, "zh": 3.57877, "my": 1.35951}, "bytes_per_chunk": {"en": 3.504, "de": 3.75, "vi": 3.58, "tr": 3.602, "fi": 4.526, "ru": 3.92, "el": 5.704, "hi": 6.263, "ta": 9.199, "th": 13.736, "zh": 4.819, "my": 7.995}, "bpb_mean": 2.1275, "frozen_inventory": false, "t": 260.1} {"event": "train", "step": 1400, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.70058, "bpb": 2.45341, "chunk_loss": 0.033348, "F": 0.22217, "G": 0.4408, "chunks": 227.5, "bytes_per_chunk": 4.501, "boundary_entropy": 0.67054, "router_noise": 0.0, "grad_norm": 0.7083, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.338, "alpha": 0.03, "t": 260.6} {"event": "train", "step": 1450, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.7101, "bpb": 2.46715, "chunk_loss": 0.034769, "F": 0.2478, "G": 0.4426, "chunks": 253.75, "bytes_per_chunk": 4.035, "boundary_entropy": 0.67198, "router_noise": 0.0, "grad_norm": 0.7336, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.171, "alpha": 0.03, "t": 269.1} {"event": "train", "step": 1500, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.67953, "bpb": 2.42305, "chunk_loss": 0.035867, "F": 0.26575, "G": 0.44115, "chunks": 272.12, "bytes_per_chunk": 3.763, "boundary_entropy": 0.66888, "router_noise": 0.0, "grad_norm": 0.8448, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.162, "alpha": 0.03, "t": 277.2} {"event": "train", "step": 1550, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.56652, "bpb": 2.26001, "chunk_loss": 0.035916, "F": 0.26572, "G": 0.44283, "chunks": 272.09, "bytes_per_chunk": 3.763, "boundary_entropy": 0.66925, "router_noise": 0.0, "grad_norm": 0.8143, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.162, "alpha": 0.03, "t": 285.4} {"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4862.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.279377, "cps": 37.984375, "mean_p": 0.440437, "boundary_entropy": 0.660398, "bpb": 2.336735}, "de": {"chunks": 5367.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.261384, "cps": 41.929688, "mean_p": 0.44358, "boundary_entropy": 0.667814, "bpb": 2.65373}, "vi": {"chunks": 6909.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.374776, "cps": 53.976562, "mean_p": 0.445516, "boundary_entropy": 0.669448, "bpb": 2.11212}, "tr": {"chunks": 6316.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.348892, "cps": 49.34375, "mean_p": 0.461925, "boundary_entropy": 0.670272, "bpb": 2.620503}, "fi": {"chunks": 4781.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25416, "cps": 37.351562, "mean_p": 0.451042, "boundary_entropy": 0.671892, "bpb": 2.877122}, "ru": {"chunks": 7244.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.382269, "cps": 56.59375, "mean_p": 0.434141, "boundary_entropy": 0.660282, "bpb": 1.748642}, "el": {"chunks": 6776.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.330166, "cps": 52.9375, "mean_p": 0.445112, "boundary_entropy": 0.675301, "bpb": 1.812518}, "hi": {"chunks": 8204.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.487608, "cps": 64.09375, "mean_p": 0.441485, "boundary_entropy": 0.668405, "bpb": 1.404129}, "ta": {"chunks": 13375.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.715432, "cps": 104.492188, "mean_p": 0.433063, "boundary_entropy": 0.661749, "bpb": 1.12232}, "th": {"chunks": 16099.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.00087, "cps": 125.773438, "mean_p": 0.43358, "boundary_entropy": 0.666975, "bpb": 1.342969}, "zh": {"chunks": 3179.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.538266, "cps": 24.835938, "mean_p": 0.460378, "boundary_entropy": 0.678408, "bpb": 3.511492}, "my": {"chunks": 7180.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.378932, "cps": 56.09375, "mean_p": 0.457302, "boundary_entropy": 0.682635, "bpb": 1.303281}}, "bpb": {"en": 2.33674, "de": 2.65373, "vi": 2.11212, "tr": 2.6205, "fi": 2.87712, "ru": 1.74864, "el": 1.81252, "hi": 1.40413, "ta": 1.12232, "th": 1.34297, "zh": 3.51149, "my": 1.30328}, "gini_cpc": 0.233613, "gini_cps": 0.23532, "bpic": 4.250443, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9355959944725233, "vi": 1.3414707773440362, "tr": 1.2488225598239862, "fi": 0.9097373511796236, "ru": 1.3682907562715105, "el": 1.1817938289300405, "hi": 1.7453388041903033, "ta": 2.5608107654659897, "th": 3.58250661368473, "zh": 1.926665190549376, "my": 1.3563451971044267}, "t": 297.0} {"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.34583, "de": 2.66142, "vi": 2.11562, "tr": 2.62821, "fi": 2.88359, "ru": 1.73019, "el": 1.80398, "hi": 1.4015, "ta": 1.13062, "th": 1.36393, "zh": 3.52225, "my": 1.31296}, "bytes_per_chunk": {"en": 3.565, "de": 3.858, "vi": 3.467, "tr": 3.124, "fi": 3.946, "ru": 4.95, "el": 5.474, "hi": 5.478, "ta": 3.817, "th": 2.704, "zh": 4.825, "my": 8.132}, "bpb_mean": 2.07501, "frozen_inventory": false, "t": 302.2} {"event": "train", "step": 1600, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.78604, "bpb": 2.57671, "chunk_loss": 0.03479, "F": 0.24701, "G": 0.44328, "chunks": 252.94, "bytes_per_chunk": 4.048, "boundary_entropy": 0.67105, "router_noise": 0.0, "grad_norm": 0.8517, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.345, "alpha": 0.03, "t": 302.6} {"event": "train", "step": 1650, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.57277, "bpb": 2.26903, "chunk_loss": 0.034945, "F": 0.25012, "G": 0.44078, "chunks": 256.12, "bytes_per_chunk": 3.998, "boundary_entropy": 0.66949, "router_noise": 0.0, "grad_norm": 0.7613, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 310.9} {"event": "train", "step": 1700, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.55889, "bpb": 2.249, "chunk_loss": 0.034652, "F": 0.24667, "G": 0.43756, "chunks": 252.59, "bytes_per_chunk": 4.054, "boundary_entropy": 0.66824, "router_noise": 0.0, "grad_norm": 0.7274, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.162, "alpha": 0.03, "t": 319.0} {"event": "train", "step": 1750, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.50429, "bpb": 2.17023, "chunk_loss": 0.035608, "F": 0.26154, "G": 0.44141, "chunks": 267.81, "bytes_per_chunk": 3.824, "boundary_entropy": 0.66932, "router_noise": 0.0, "grad_norm": 0.7661, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 327.4} {"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4833.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.277711, "cps": 37.757812, "mean_p": 0.439764, "boundary_entropy": 0.660351, "bpb": 2.300663}, "de": {"chunks": 5259.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.256124, "cps": 41.085938, "mean_p": 0.443966, "boundary_entropy": 0.668422, "bpb": 2.605982}, "vi": {"chunks": 5759.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.312395, "cps": 44.992188, "mean_p": 0.439535, "boundary_entropy": 0.667812, "bpb": 2.096747}, "tr": {"chunks": 5446.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.300834, "cps": 42.546875, "mean_p": 0.457973, "boundary_entropy": 0.670181, "bpb": 2.579385}, "fi": {"chunks": 4161.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.2212, "cps": 32.507812, "mean_p": 0.445248, "boundary_entropy": 0.671462, "bpb": 2.818772}, "ru": {"chunks": 7574.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.399683, "cps": 59.171875, "mean_p": 0.434014, "boundary_entropy": 0.66058, "bpb": 1.709346}, "el": {"chunks": 5346.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.260488, "cps": 41.765625, "mean_p": 0.441204, "boundary_entropy": 0.674284, "bpb": 1.77892}, "hi": {"chunks": 7640.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.454086, "cps": 59.6875, "mean_p": 0.438011, "boundary_entropy": 0.66694, "bpb": 1.373313}, "ta": {"chunks": 8030.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.429527, "cps": 62.734375, "mean_p": 0.432234, "boundary_entropy": 0.663832, "bpb": 1.117951}, "th": {"chunks": 8586.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.533789, "cps": 67.078125, "mean_p": 0.421459, "boundary_entropy": 0.662807, "bpb": 1.308673}, "zh": {"chunks": 3022.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.511683, "cps": 23.609375, "mean_p": 0.457486, "boundary_entropy": 0.677579, "bpb": 3.502228}, "my": {"chunks": 7908.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.417353, "cps": 61.78125, "mean_p": 0.459229, "boundary_entropy": 0.682973, "bpb": 1.264166}}, "bpb": {"en": 2.30066, "de": 2.60598, "vi": 2.09675, "tr": 2.57939, "fi": 2.81877, "ru": 1.70935, "el": 1.77892, "hi": 1.37331, "ta": 1.11795, "th": 1.30867, "zh": 3.50223, "my": 1.26417}, "gini_cpc": 0.157902, "gini_cps": 0.154573, "bpic": 5.216968, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9222700143594075, "vi": 1.1248931227321253, "tr": 1.083264249715565, "fi": 0.7965135301099134, "ru": 1.4392074933381813, "el": 0.9379840086759181, "hi": 1.6351048650483353, "ta": 1.546669070573081, "th": 1.9221051571348442, "zh": 1.8425035887080927, "my": 1.5028326078350671}, "t": 339.1} {"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.31108, "de": 2.61093, "vi": 2.10146, "tr": 2.58579, "fi": 2.82705, "ru": 1.69031, "el": 1.76995, "hi": 1.37046, "ta": 1.12475, "th": 1.32659, "zh": 3.50952, "my": 1.27369}, "bytes_per_chunk": {"en": 3.476, "de": 3.889, "vi": 4.15, "tr": 3.64, "fi": 4.496, "ru": 4.519, "el": 6.782, "hi": 5.948, "ta": 6.477, "th": 5.066, "zh": 4.9, "my": 6.885}, "bpb_mean": 2.0418, "frozen_inventory": false, "t": 343.8} {"event": "train", "step": 1800, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.45102, "bpb": 2.09338, "chunk_loss": 0.03342, "F": 0.22495, "G": 0.44029, "chunks": 230.34, "bytes_per_chunk": 4.446, "boundary_entropy": 0.6698, "router_noise": 0.0, "grad_norm": 0.6717, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.335, "alpha": 0.03, "t": 344.2} {"event": "train", "step": 1850, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.43704, "bpb": 2.0732, "chunk_loss": 0.034822, "F": 0.24802, "G": 0.44024, "chunks": 253.97, "bytes_per_chunk": 4.032, "boundary_entropy": 0.66967, "router_noise": 0.0, "grad_norm": 0.7196, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.162, "alpha": 0.03, "t": 352.3} {"event": "train", "step": 1900, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.54356, "bpb": 2.22689, "chunk_loss": 0.03338, "F": 0.22461, "G": 0.43595, "chunks": 230.0, "bytes_per_chunk": 4.452, "boundary_entropy": 0.66657, "router_noise": 0.0, "grad_norm": 0.7068, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 360.7} {"event": "train", "step": 1950, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.60984, "bpb": 2.32251, "chunk_loss": 0.033971, "F": 0.23248, "G": 0.44078, "chunks": 238.06, "bytes_per_chunk": 4.301, "boundary_entropy": 0.66957, "router_noise": 0.0, "grad_norm": 0.7133, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 369.1} {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4918.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.282595, "cps": 38.421875, "mean_p": 0.438999, "boundary_entropy": 0.658871, "bpb": 2.264846}, "de": {"chunks": 5376.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.261822, "cps": 42.0, "mean_p": 0.44363, "boundary_entropy": 0.667169, "bpb": 2.558025}, "vi": {"chunks": 6259.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.339517, "cps": 48.898438, "mean_p": 0.440657, "boundary_entropy": 0.667447, "bpb": 2.073425}, "tr": {"chunks": 5921.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.327073, "cps": 46.257812, "mean_p": 0.459041, "boundary_entropy": 0.669521, "bpb": 2.513827}, "fi": {"chunks": 4665.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.247993, "cps": 36.445312, "mean_p": 0.450584, "boundary_entropy": 0.672216, "bpb": 2.79995}, "ru": {"chunks": 7056.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.372348, "cps": 55.125, "mean_p": 0.431796, "boundary_entropy": 0.659679, "bpb": 1.667978}, "el": {"chunks": 7550.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.36788, "cps": 58.984375, "mean_p": 0.442867, "boundary_entropy": 0.673302, "bpb": 1.750135}, "hi": {"chunks": 9113.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.541634, "cps": 71.195312, "mean_p": 0.439538, "boundary_entropy": 0.668161, "bpb": 1.340312}, "ta": {"chunks": 10140.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.542391, "cps": 79.21875, "mean_p": 0.438531, "boundary_entropy": 0.666987, "bpb": 1.106349}, "th": {"chunks": 15752.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.979297, "cps": 123.0625, "mean_p": 0.434068, "boundary_entropy": 0.66682, "bpb": 1.287921}, "zh": {"chunks": 2876.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.486962, "cps": 22.46875, "mean_p": 0.460498, "boundary_entropy": 0.678739, "bpb": 3.460603}, "my": {"chunks": 11190.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.590564, "cps": 87.421875, "mean_p": 0.461685, "boundary_entropy": 0.683201, "bpb": 1.244007}}, "bpb": {"en": 2.26485, "de": 2.55802, "vi": 2.07342, "tr": 2.51383, "fi": 2.79995, "ru": 1.66798, "el": 1.75014, "hi": 1.34031, "ta": 1.10635, "th": 1.28792, "zh": 3.4606, "my": 1.24401}, "gini_cpc": 0.223798, "gini_cps": 0.239756, "bpic": 4.225918, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9264936533691082, "vi": 1.2014270488410252, "tr": 1.157391012078842, "fi": 0.8775570518050436, "ru": 1.3176041486714571, "el": 1.3017923116125358, "hi": 1.9166459384122492, "ta": 1.9193230602637468, "th": 3.4653749657263795, "zh": 1.7231815454455437, "my": 2.0897883619414124}, "t": 380.3} {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.27406, "de": 2.56418, "vi": 2.07579, "tr": 2.52224, "fi": 2.80219, "ru": 1.64886, "el": 1.74261, "hi": 1.33688, "ta": 1.1145, "th": 1.30814, "zh": 3.46608, "my": 1.25281}, "bytes_per_chunk": {"en": 3.427, "de": 3.791, "vi": 3.77, "tr": 3.274, "fi": 4.088, "ru": 4.868, "el": 4.91, "hi": 4.866, "ta": 4.795, "th": 2.928, "zh": 5.292, "my": 5.195}, "bpb_mean": 2.00903, "frozen_inventory": false, "t": 385.6} {"event": "train", "step": 2000, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.40994, "bpb": 2.03412, "chunk_loss": 0.034557, "F": 0.24408, "G": 0.44027, "chunks": 249.94, "bytes_per_chunk": 4.097, "boundary_entropy": 0.6681, "router_noise": 0.0, "grad_norm": 1.0812, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.337, "alpha": 0.03, "t": 386.0} {"event": "train", "step": 2050, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.59309, "bpb": 2.29834, "chunk_loss": 0.033568, "F": 0.22833, "G": 0.43681, "chunks": 233.81, "bytes_per_chunk": 4.38, "boundary_entropy": 0.66752, "router_noise": 0.0, "grad_norm": 0.7835, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.163, "alpha": 0.03, "t": 394.1} {"event": "train", "step": 2100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.64103, "bpb": 2.36751, "chunk_loss": 0.035198, "F": 0.25363, "G": 0.44514, "chunks": 259.72, "bytes_per_chunk": 3.943, "boundary_entropy": 0.67135, "router_noise": 0.0, "grad_norm": 0.7886, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 402.6} {"event": "train", "step": 2150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.37341, "bpb": 1.98141, "chunk_loss": 0.033583, "F": 0.22794, "G": 0.43726, "chunks": 233.41, "bytes_per_chunk": 4.387, "boundary_entropy": 0.66685, "router_noise": 0.0, "grad_norm": 0.6901, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.163, "alpha": 0.03, "t": 410.7} {"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4573.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.262771, "cps": 35.726562, "mean_p": 0.435211, "boundary_entropy": 0.657413, "bpb": 2.226957}, "de": {"chunks": 5366.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.261335, "cps": 41.921875, "mean_p": 0.44322, "boundary_entropy": 0.666495, "bpb": 2.521178}, "vi": {"chunks": 5374.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.291511, "cps": 41.984375, "mean_p": 0.437468, "boundary_entropy": 0.666509, "bpb": 2.05368}, "tr": {"chunks": 5807.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.320776, "cps": 45.367188, "mean_p": 0.457102, "boundary_entropy": 0.668754, "bpb": 2.488759}, "fi": {"chunks": 4687.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.249163, "cps": 36.617188, "mean_p": 0.446682, "boundary_entropy": 0.669821, "bpb": 2.730502}, "ru": {"chunks": 7142.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.376887, "cps": 55.796875, "mean_p": 0.430762, "boundary_entropy": 0.659639, "bpb": 1.634328}, "el": {"chunks": 7085.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.345222, "cps": 55.351562, "mean_p": 0.44442, "boundary_entropy": 0.674181, "bpb": 1.72084}, "hi": {"chunks": 8816.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.523982, "cps": 68.875, "mean_p": 0.440917, "boundary_entropy": 0.667897, "bpb": 1.343143}, "ta": {"chunks": 9263.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.49548, "cps": 72.367188, "mean_p": 0.436396, "boundary_entropy": 0.666917, "bpb": 1.087717}, "th": {"chunks": 5827.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.362263, "cps": 45.523438, "mean_p": 0.422227, "boundary_entropy": 0.664657, "bpb": 1.261701}, "zh": {"chunks": 2956.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.500508, "cps": 23.09375, "mean_p": 0.457846, "boundary_entropy": 0.677591, "bpb": 3.449262}, "my": {"chunks": 11799.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.622704, "cps": 92.179688, "mean_p": 0.462118, "boundary_entropy": 0.683243, "bpb": 1.21935}}, "bpb": {"en": 2.22696, "de": 2.52118, "vi": 2.05368, "tr": 2.48876, "fi": 2.7305, "ru": 1.63433, "el": 1.72084, "hi": 1.34314, "ta": 1.08772, "th": 1.2617, "zh": 3.44926, "my": 1.21935}, "gini_cpc": 0.16878, "gini_cps": 0.193171, "bpic": 4.876816, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9945375383041719, "vi": 1.109372609634112, "tr": 1.2207428617006175, "fi": 0.9482131827454898, "ru": 1.4342787048218666, "el": 1.3137778280983123, "hi": 1.9940655355145147, "ta": 1.8855980195160413, "th": 1.378627291692851, "zh": 1.9047321218515696, "my": 2.3697620695973156}, "t": 421.4} {"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.23591, "de": 2.52628, "vi": 2.05709, "tr": 2.49495, "fi": 2.73893, "ru": 1.61582, "el": 1.71312, "hi": 1.33868, "ta": 1.09411, "th": 1.27897, "zh": 3.45192, "my": 1.2292}, "bytes_per_chunk": {"en": 3.772, "de": 3.803, "vi": 4.535, "tr": 3.369, "fi": 4.112, "ru": 4.848, "el": 5.416, "hi": 5.197, "ta": 5.344, "th": 7.922, "zh": 5.137, "my": 4.923}, "bpb_mean": 1.98125, "frozen_inventory": false, "t": 426.5} {"event": "train", "step": 2200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.28701, "bpb": 1.85676, "chunk_loss": 0.032958, "F": 0.21649, "G": 0.43692, "chunks": 221.69, "bytes_per_chunk": 4.619, "boundary_entropy": 0.6673, "router_noise": 0.0, "grad_norm": 0.7283, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.326, "alpha": 0.03, "t": 427.0} {"event": "train", "step": 2250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.52322, "bpb": 2.19755, "chunk_loss": 0.034096, "F": 0.23639, "G": 0.43847, "chunks": 242.06, "bytes_per_chunk": 4.23, "boundary_entropy": 0.66795, "router_noise": 0.0, "grad_norm": 0.6858, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 435.3} {"event": "train", "step": 2300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.47527, "bpb": 2.12836, "chunk_loss": 0.034034, "F": 0.23593, "G": 0.43853, "chunks": 241.59, "bytes_per_chunk": 4.239, "boundary_entropy": 0.66767, "router_noise": 0.0, "grad_norm": 0.7551, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.162, "alpha": 0.03, "t": 443.4} {"event": "train", "step": 2350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.4557, "bpb": 2.10013, "chunk_loss": 0.033642, "F": 0.22913, "G": 0.43934, "chunks": 234.62, "bytes_per_chunk": 4.364, "boundary_entropy": 0.66803, "router_noise": 0.0, "grad_norm": 0.6711, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.152, "alpha": 0.03, "t": 451.0} {"event": "probe", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4415.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.253692, "cps": 34.492188, "mean_p": 0.43503, "boundary_entropy": 0.657759, "bpb": 2.211682}, "de": {"chunks": 5313.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.258754, "cps": 41.507812, "mean_p": 0.4429, "boundary_entropy": 0.667132, "bpb": 2.468754}, "vi": {"chunks": 5364.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.290968, "cps": 41.90625, "mean_p": 0.44, "boundary_entropy": 0.667914, "bpb": 2.039664}, "tr": {"chunks": 5588.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.308678, "cps": 43.65625, "mean_p": 0.456621, "boundary_entropy": 0.669252, "bpb": 2.45845}, "fi": {"chunks": 5132.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.272819, "cps": 40.09375, "mean_p": 0.452424, "boundary_entropy": 0.672127, "bpb": 2.690278}, "ru": {"chunks": 7942.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.419103, "cps": 62.046875, "mean_p": 0.433667, "boundary_entropy": 0.660842, "bpb": 1.605954}, "el": {"chunks": 7009.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.341519, "cps": 54.757812, "mean_p": 0.443516, "boundary_entropy": 0.673741, "bpb": 1.691125}, "hi": {"chunks": 7544.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.44838, "cps": 58.9375, "mean_p": 0.438945, "boundary_entropy": 0.668811, "bpb": 1.315067}, "ta": {"chunks": 5627.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.30099, "cps": 43.960938, "mean_p": 0.42731, "boundary_entropy": 0.664761, "bpb": 1.069084}, "th": {"chunks": 3906.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.242835, "cps": 30.515625, "mean_p": 0.418201, "boundary_entropy": 0.664436, "bpb": 1.237208}, "zh": {"chunks": 3475.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.588385, "cps": 27.148438, "mean_p": 0.462502, "boundary_entropy": 0.678358, "bpb": 3.408902}, "my": {"chunks": 5801.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.306154, "cps": 45.320312, "mean_p": 0.457175, "boundary_entropy": 0.683552, "bpb": 1.206557}}, "bpb": {"en": 2.21168, "de": 2.46875, "vi": 2.03966, "tr": 2.45845, "fi": 2.69028, "ru": 1.60595, "el": 1.69113, "hi": 1.31507, "ta": 1.06908, "th": 1.23721, "zh": 3.4089, "my": 1.20656}, "gini_cpc": 0.146267, "gini_cps": 0.12981, "bpic": 5.718175, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.019954553173774, "vi": 1.146935616891524, "tr": 1.2167441259900174, "fi": 1.075395405196822, "ru": 1.6520153590093738, "el": 1.3461970270186603, "hi": 1.767421033589731, "ta": 1.1864374804825752, "th": 0.9572041789600899, "zh": 2.319288521299529, "my": 1.2067933315020114}, "t": 462.2} {"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.22256, "de": 2.47544, "vi": 2.04267, "tr": 2.46678, "fi": 2.7017, "ru": 1.58598, "el": 1.6807, "hi": 1.3141, "ta": 1.07703, "th": 1.25667, "zh": 3.41188, "my": 1.21369}, "bytes_per_chunk": {"en": 3.946, "de": 3.915, "vi": 4.528, "tr": 3.527, "fi": 3.74, "ru": 4.402, "el": 5.303, "hi": 6.611, "ta": 8.679, "th": 11.281, "zh": 4.409, "my": 8.651}, "bpb_mean": 1.9541, "frozen_inventory": false, "t": 467.2} {"event": "phase_transition", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 467.4} {"event": "train", "step": 2400, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.31801, "bpb": 1.90149, "chunk_loss": 0.041124, "F": 0.20844, "G": 0.43907, "chunks": 213.44, "bytes_per_chunk": 4.798, "boundary_entropy": 0.6696, "router_noise": 0.0, "grad_norm": 0.711, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.332, "n_target_mean": 10.8309, "alpha": 0.03, "t": 467.6} {"event": "train", "step": 2450, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.46989, "bpb": 2.1206, "chunk_loss": 0.036979, "F": 0.20255, "G": 0.43397, "chunks": 207.41, "bytes_per_chunk": 4.937, "boundary_entropy": 0.66667, "router_noise": 0.0, "grad_norm": 0.7384, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.161, "n_target_mean": 9.4427, "alpha": 0.03, "t": 475.7} {"event": "train", "step": 2500, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.46759, "bpb": 2.11729, "chunk_loss": 0.036863, "F": 0.19376, "G": 0.43173, "chunks": 198.41, "bytes_per_chunk": 5.161, "boundary_entropy": 0.66595, "router_noise": 0.0, "grad_norm": 0.7023, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.154, "n_target_mean": 9.5095, "alpha": 0.03, "t": 483.4} {"event": "train", "step": 2550, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.52731, "bpb": 2.20344, "chunk_loss": 0.037362, "F": 0.24231, "G": 0.43521, "chunks": 248.12, "bytes_per_chunk": 4.127, "boundary_entropy": 0.66462, "router_noise": 0.0, "grad_norm": 0.7736, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.157, "n_target_mean": 7.5354, "alpha": 0.03, "t": 491.3} {"event": "probe", "step": 2600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4456.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.256048, "cps": 34.8125, "mean_p": 0.433745, "boundary_entropy": 0.655428, "bpb": 2.181777}, "de": {"chunks": 5357.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.260897, "cps": 41.851562, "mean_p": 0.440578, "boundary_entropy": 0.664874, "bpb": 2.450248}, "vi": {"chunks": 5217.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.282994, "cps": 40.757812, "mean_p": 0.433247, "boundary_entropy": 0.664156, "bpb": 2.042219}, "tr": {"chunks": 5799.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.320334, "cps": 45.304688, "mean_p": 0.456771, "boundary_entropy": 0.667274, "bpb": 2.400591}, "fi": {"chunks": 5169.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.274786, "cps": 40.382812, "mean_p": 0.450377, "boundary_entropy": 0.670046, "bpb": 2.646709}, "ru": {"chunks": 5698.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.300686, "cps": 44.515625, "mean_p": 0.417873, "boundary_entropy": 0.654406, "bpb": 1.586663}, "el": {"chunks": 5123.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.249622, "cps": 40.023438, "mean_p": 0.430981, "boundary_entropy": 0.669507, "bpb": 1.663035}, "hi": {"chunks": 5382.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.319881, "cps": 42.046875, "mean_p": 0.419938, "boundary_entropy": 0.659771, "bpb": 1.313062}, "ta": {"chunks": 3637.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.194544, "cps": 28.414062, "mean_p": 0.404353, "boundary_entropy": 0.653412, "bpb": 1.068005}, "th": {"chunks": 8514.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.529313, "cps": 66.515625, "mean_p": 0.418317, "boundary_entropy": 0.660039, "bpb": 1.225205}, "zh": {"chunks": 3054.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.517101, "cps": 23.859375, "mean_p": 0.461287, "boundary_entropy": 0.678504, "bpb": 3.393552}, "my": {"chunks": 6067.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.320192, "cps": 47.398438, "mean_p": 0.457965, "boundary_entropy": 0.683369, "bpb": 1.188823}}, "bpb": {"en": 2.18178, "de": 2.45025, "vi": 2.04222, "tr": 2.40059, "fi": 2.64671, "ru": 1.58666, "el": 1.66303, "hi": 1.31306, "ta": 1.068, "th": 1.2252, "zh": 3.39355, "my": 1.18882}, "gini_cpc": 0.152215, "gini_cps": 0.124705, "bpic": 6.046366, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0189389812659588, "vi": 1.1052400982831134, "tr": 1.2510696695579289, "fi": 1.0731825113404845, "ru": 1.174335442888069, "el": 0.9749053389414392, "hi": 1.2493023547394904, "ta": 0.759795591905016, "th": 2.06724294577338, "zh": 2.01954961968506, "my": 1.2505168757241307}, "t": 502.8} {"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.19072, "de": 2.45505, "vi": 2.04519, "tr": 2.40932, "fi": 2.65607, "ru": 1.56483, "el": 1.6541, "hi": 1.31102, "ta": 1.07476, "th": 1.24407, "zh": 3.39517, "my": 1.19811}, "bytes_per_chunk": {"en": 3.874, "de": 3.825, "vi": 4.701, "tr": 3.363, "fi": 3.654, "ru": 6.159, "el": 7.206, "hi": 8.779, "ta": 13.494, "th": 5.341, "zh": 5.495, "my": 8.742}, "bpb_mean": 1.9332, "frozen_inventory": false, "t": 507.7} {"event": "train", "step": 2600, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.49391, "bpb": 2.15526, "chunk_loss": 0.038468, "F": 0.21564, "G": 0.43287, "chunks": 220.81, "bytes_per_chunk": 4.637, "boundary_entropy": 0.66501, "router_noise": 0.0, "grad_norm": 0.7323, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.338, "n_target_mean": 9.2882, "alpha": 0.03, "t": 508.2} {"event": "train", "step": 2650, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.44137, "bpb": 2.07945, "chunk_loss": 0.036366, "F": 0.21625, "G": 0.43105, "chunks": 221.44, "bytes_per_chunk": 4.624, "boundary_entropy": 0.66444, "router_noise": 0.0, "grad_norm": 0.746, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.151, "n_target_mean": 8.0661, "alpha": 0.03, "t": 515.7} {"event": "train", "step": 2700, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.49101, "bpb": 2.15107, "chunk_loss": 0.038294, "F": 0.22974, "G": 0.43052, "chunks": 235.25, "bytes_per_chunk": 4.353, "boundary_entropy": 0.66331, "router_noise": 0.0, "grad_norm": 0.7906, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.16, "n_target_mean": 8.2992, "alpha": 0.03, "t": 523.8} {"event": "train", "step": 2750, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.52938, "bpb": 2.20642, "chunk_loss": 0.036853, "F": 0.2218, "G": 0.43245, "chunks": 227.12, "bytes_per_chunk": 4.509, "boundary_entropy": 0.66474, "router_noise": 0.0, "grad_norm": 0.7674, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.161, "n_target_mean": 8.0733, "alpha": 0.03, "t": 531.8} {"event": "probe", "step": 2800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4595.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.264035, "cps": 35.898438, "mean_p": 0.433024, "boundary_entropy": 0.655653, "bpb": 2.166889}, "de": {"chunks": 5264.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.256368, "cps": 41.125, "mean_p": 0.438536, "boundary_entropy": 0.664355, "bpb": 2.409957}, "vi": {"chunks": 5298.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.287388, "cps": 41.390625, "mean_p": 0.432102, "boundary_entropy": 0.662782, "bpb": 2.018518}, "tr": {"chunks": 5775.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.319008, "cps": 45.117188, "mean_p": 0.454314, "boundary_entropy": 0.667419, "bpb": 2.379331}, "fi": {"chunks": 4895.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.26022, "cps": 38.242188, "mean_p": 0.44741, "boundary_entropy": 0.669463, "bpb": 2.619908}, "ru": {"chunks": 5493.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.289868, "cps": 42.914062, "mean_p": 0.410367, "boundary_entropy": 0.651726, "bpb": 1.569568}, "el": {"chunks": 5326.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.259514, "cps": 41.609375, "mean_p": 0.431283, "boundary_entropy": 0.670174, "bpb": 1.640619}, "hi": {"chunks": 5338.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.317266, "cps": 41.703125, "mean_p": 0.419476, "boundary_entropy": 0.660817, "bpb": 1.280311}, "ta": {"chunks": 4316.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.230864, "cps": 33.71875, "mean_p": 0.40838, "boundary_entropy": 0.656383, "bpb": 1.054222}, "th": {"chunks": 4623.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.287411, "cps": 36.117188, "mean_p": 0.404798, "boundary_entropy": 0.652374, "bpb": 1.205109}, "zh": {"chunks": 2916.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.493735, "cps": 22.78125, "mean_p": 0.459935, "boundary_entropy": 0.678065, "bpb": 3.336265}, "my": {"chunks": 3732.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.19696, "cps": 29.15625, "mean_p": 0.444017, "boundary_entropy": 0.679926, "bpb": 1.172008}}, "bpb": {"en": 2.16689, "de": 2.40996, "vi": 2.01852, "tr": 2.37933, "fi": 2.61991, "ru": 1.56957, "el": 1.64062, "hi": 1.28031, "ta": 1.05422, "th": 1.20511, "zh": 3.33626, "my": 1.17201}, "gini_cpc": 0.114094, "gini_cps": 0.087574, "bpic": 6.666221, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9709616521656506, "vi": 1.088447325293439, "tr": 1.2082033668182255, "fi": 0.9855517131093104, "ru": 1.097839845421058, "el": 0.9828764320102232, "hi": 1.2016060353734193, "ta": 0.8743686362113778, "th": 1.0885325814337121, "zh": 1.8699615705906867, "my": 0.7459622727845959}, "t": 542.9} {"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.17721, "de": 2.41627, "vi": 2.02111, "tr": 2.38458, "fi": 2.62986, "ru": 1.54943, "el": 1.6321, "hi": 1.27833, "ta": 1.06081, "th": 1.22288, "zh": 3.33925, "my": 1.18113}, "bytes_per_chunk": {"en": 3.778, "de": 3.918, "vi": 4.601, "tr": 3.487, "fi": 3.938, "ru": 6.346, "el": 6.955, "hi": 8.475, "ta": 11.144, "th": 9.623, "zh": 5.161, "my": 13.925}, "bpb_mean": 1.90775, "frozen_inventory": false, "t": 547.7} {"event": "train", "step": 2800, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.40411, "bpb": 2.0257, "chunk_loss": 0.037214, "F": 0.21503, "G": 0.42705, "chunks": 220.19, "bytes_per_chunk": 4.651, "boundary_entropy": 0.66188, "router_noise": 0.0, "grad_norm": 0.7319, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.324, "n_target_mean": 8.7824, "alpha": 0.03, "t": 548.0} {"event": "train", "step": 2850, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.32139, "bpb": 1.90636, "chunk_loss": 0.037242, "F": 0.20023, "G": 0.42875, "chunks": 205.03, "bytes_per_chunk": 4.994, "boundary_entropy": 0.66439, "router_noise": 0.0, "grad_norm": 0.716, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.156, "n_target_mean": 9.986, "alpha": 0.03, "t": 555.8} {"event": "train", "step": 2900, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.411, "bpb": 2.03564, "chunk_loss": 0.035339, "F": 0.19711, "G": 0.424, "chunks": 201.84, "bytes_per_chunk": 5.073, "boundary_entropy": 0.66061, "router_noise": 0.0, "grad_norm": 0.7597, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.162, "n_target_mean": 8.5507, "alpha": 0.03, "t": 563.9} {"event": "train", "step": 2950, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.30656, "bpb": 1.88496, "chunk_loss": 0.035118, "F": 0.18301, "G": 0.42068, "chunks": 187.41, "bytes_per_chunk": 5.464, "boundary_entropy": 0.65933, "router_noise": 0.0, "grad_norm": 0.6674, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.161, "n_target_mean": 9.6732, "alpha": 0.03, "t": 572.0} {"event": "probe", "step": 3000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4193.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.240935, "cps": 32.757812, "mean_p": 0.428747, "boundary_entropy": 0.653287, "bpb": 2.151176}, "de": {"chunks": 5113.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.249014, "cps": 39.945312, "mean_p": 0.434899, "boundary_entropy": 0.662264, "bpb": 2.405898}, "vi": {"chunks": 5067.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.274858, "cps": 39.585938, "mean_p": 0.426201, "boundary_entropy": 0.659765, "bpb": 2.003592}, "tr": {"chunks": 5636.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.31133, "cps": 44.03125, "mean_p": 0.45359, "boundary_entropy": 0.665748, "bpb": 2.353012}, "fi": {"chunks": 4318.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.229547, "cps": 33.734375, "mean_p": 0.441432, "boundary_entropy": 0.667706, "bpb": 2.607592}, "ru": {"chunks": 5700.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.300792, "cps": 44.53125, "mean_p": 0.412081, "boundary_entropy": 0.651692, "bpb": 1.54087}, "el": {"chunks": 4859.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.236759, "cps": 37.960938, "mean_p": 0.429295, "boundary_entropy": 0.669631, "bpb": 1.608757}, "hi": {"chunks": 3491.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.207489, "cps": 27.273438, "mean_p": 0.414692, "boundary_entropy": 0.659561, "bpb": 1.27009}, "ta": {"chunks": 3877.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.207382, "cps": 30.289062, "mean_p": 0.403963, "boundary_entropy": 0.655558, "bpb": 1.040837}, "th": {"chunks": 2916.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.181287, "cps": 22.78125, "mean_p": 0.396251, "boundary_entropy": 0.648799, "bpb": 1.182177}, "zh": {"chunks": 3022.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.511683, "cps": 23.609375, "mean_p": 0.459837, "boundary_entropy": 0.677871, "bpb": 3.328544}, "my": {"chunks": 2616.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.138062, "cps": 20.4375, "mean_p": 0.429889, "boundary_entropy": 0.675701, "bpb": 1.157735}}, "bpb": {"en": 2.15118, "de": 2.4059, "vi": 2.00359, "tr": 2.35301, "fi": 2.60759, "ru": 1.54087, "el": 1.60876, "hi": 1.27009, "ta": 1.04084, "th": 1.18218, "zh": 3.32854, "my": 1.15773}, "gini_cpc": 0.170854, "gini_cps": 0.13776, "bpic": 7.553555, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0335289435192705, "vi": 1.140793453073888, "tr": 1.292170109775028, "fi": 0.9527303765754627, "ru": 1.2484320204487063, "el": 0.9826646013798552, "hi": 0.8611801952521875, "ta": 0.8607352503028678, "th": 0.7524293230846947, "zh": 2.1237347589378035, "my": 0.5730250680172494}, "t": 582.7} {"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.15926, "de": 2.41107, "vi": 2.00644, "tr": 2.36011, "fi": 2.62042, "ru": 1.51798, "el": 1.60138, "hi": 1.26874, "ta": 1.04727, "th": 1.20106, "zh": 3.32712, "my": 1.16689}, "bytes_per_chunk": {"en": 4.07, "de": 4.01, "vi": 4.645, "tr": 3.503, "fi": 4.522, "ru": 5.996, "el": 7.549, "hi": 13.087, "ta": 12.318, "th": 15.017, "zh": 5.287, "my": 19.385}, "bpb_mean": 1.89065, "frozen_inventory": false, "t": 586.9} {"event": "train", "step": 3000, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.51485, "bpb": 2.18546, "chunk_loss": 0.034895, "F": 0.1915, "G": 0.43105, "chunks": 196.09, "bytes_per_chunk": 5.222, "boundary_entropy": 0.66591, "router_noise": 0.0, "grad_norm": 0.6929, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.314, "n_target_mean": 9.0106, "alpha": 0.03, "t": 587.7} {"event": "train", "step": 3050, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.47511, "bpb": 2.12813, "chunk_loss": 0.037577, "F": 0.21182, "G": 0.43021, "chunks": 216.91, "bytes_per_chunk": 4.721, "boundary_entropy": 0.66358, "router_noise": 0.0, "grad_norm": 0.8053, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.16, "n_target_mean": 9.0983, "alpha": 0.03, "t": 595.7} {"event": "train", "step": 3100, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.41402, "bpb": 2.04, "chunk_loss": 0.037437, "F": 0.19235, "G": 0.42933, "chunks": 196.97, "bytes_per_chunk": 5.199, "boundary_entropy": 0.66543, "router_noise": 0.0, "grad_norm": 0.7293, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.162, "n_target_mean": 9.8865, "alpha": 0.03, "t": 603.8} {"event": "train", "step": 3150, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.32456, "bpb": 1.91093, "chunk_loss": 0.036844, "F": 0.19037, "G": 0.42405, "chunks": 194.94, "bytes_per_chunk": 5.253, "boundary_entropy": 0.6615, "router_noise": 0.0, "grad_norm": 0.7508, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.161, "n_target_mean": 9.6709, "alpha": 0.03, "t": 611.9} {"event": "probe", "step": 3200, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4273.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.245532, "cps": 33.382812, "mean_p": 0.4279, "boundary_entropy": 0.652611, "bpb": 2.126801}, "de": {"chunks": 5265.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.256417, "cps": 41.132812, "mean_p": 0.43583, "boundary_entropy": 0.662066, "bpb": 2.371016}, "vi": {"chunks": 4393.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.238297, "cps": 34.320312, "mean_p": 0.420086, "boundary_entropy": 0.657422, "bpb": 1.994849}, "tr": {"chunks": 5526.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.305253, "cps": 43.171875, "mean_p": 0.451194, "boundary_entropy": 0.664801, "bpb": 2.321676}, "fi": {"chunks": 4684.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.249003, "cps": 36.59375, "mean_p": 0.444449, "boundary_entropy": 0.667648, "bpb": 2.553334}, "ru": {"chunks": 4954.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.261425, "cps": 38.703125, "mean_p": 0.404804, "boundary_entropy": 0.648398, "bpb": 1.532964}, "el": {"chunks": 4182.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.203771, "cps": 32.671875, "mean_p": 0.423994, "boundary_entropy": 0.667774, "bpb": 1.609501}, "hi": {"chunks": 3825.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.22734, "cps": 29.882812, "mean_p": 0.411341, "boundary_entropy": 0.657234, "bpb": 1.261119}, "ta": {"chunks": 3814.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.204012, "cps": 29.796875, "mean_p": 0.397496, "boundary_entropy": 0.653171, "bpb": 1.035737}, "th": {"chunks": 3485.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.216661, "cps": 27.226562, "mean_p": 0.401917, "boundary_entropy": 0.650937, "bpb": 1.17572}, "zh": {"chunks": 3016.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.510667, "cps": 23.5625, "mean_p": 0.46009, "boundary_entropy": 0.677893, "bpb": 3.288747}, "my": {"chunks": 2906.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.153367, "cps": 22.703125, "mean_p": 0.434173, "boundary_entropy": 0.676918, "bpb": 1.150308}}, "bpb": {"en": 2.1268, "de": 2.37102, "vi": 1.99485, "tr": 2.32168, "fi": 2.55333, "ru": 1.53296, "el": 1.6095, "hi": 1.26112, "ta": 1.03574, "th": 1.17572, "zh": 3.28875, "my": 1.15031}, "gini_cpc": 0.149658, "gini_cps": 0.108448, "bpic": 7.626354, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0443286576554163, "vi": 0.9705307247410605, "tr": 1.2432302150581018, "fi": 1.0141360716500967, "ru": 1.0647263821985875, "el": 0.8299165221683266, "hi": 0.92590748293737, "ta": 0.830895575926689, "th": 0.8824151271609454, "zh": 2.0798361473839826, "my": 0.6246308951726728}, "t": 622.6} {"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.13862, "de": 2.37487, "vi": 1.99723, "tr": 2.32937, "fi": 2.56613, "ru": 1.51096, "el": 1.59972, "hi": 1.2565, "ta": 1.04227, "th": 1.19403, "zh": 3.28832, "my": 1.15889}, "bytes_per_chunk": {"en": 4.007, "de": 3.969, "vi": 5.412, "tr": 3.546, "fi": 4.01, "ru": 7.047, "el": 8.751, "hi": 11.878, "ta": 12.692, "th": 13.24, "zh": 5.098, "my": 19.03}, "bpb_mean": 1.87141, "frozen_inventory": false, "t": 627.4} {"event": "train", "step": 3200, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.39766, "bpb": 2.0164, "chunk_loss": 0.035467, "F": 0.19055, "G": 0.4236, "chunks": 195.12, "bytes_per_chunk": 5.248, "boundary_entropy": 0.66172, "router_noise": 0.0, "grad_norm": 0.7296, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.321, "n_target_mean": 9.2811, "alpha": 0.03, "t": 627.9} {"event": "train", "step": 3250, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.37352, "bpb": 1.98158, "chunk_loss": 0.03556, "F": 0.20477, "G": 0.42781, "chunks": 209.69, "bytes_per_chunk": 4.883, "boundary_entropy": 0.66175, "router_noise": 0.0, "grad_norm": 0.7077, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.161, "n_target_mean": 9.2351, "alpha": 0.03, "t": 636.0} {"event": "train", "step": 3300, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.2837, "bpb": 1.85199, "chunk_loss": 0.037419, "F": 0.18011, "G": 0.42063, "chunks": 184.44, "bytes_per_chunk": 5.552, "boundary_entropy": 0.66002, "router_noise": 0.0, "grad_norm": 0.7717, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.156, "n_target_mean": 10.7221, "alpha": 0.03, "t": 643.8} {"event": "train", "step": 3350, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.36115, "bpb": 1.96372, "chunk_loss": 0.034979, "F": 0.1785, "G": 0.42448, "chunks": 182.78, "bytes_per_chunk": 5.602, "boundary_entropy": 0.66286, "router_noise": 0.0, "grad_norm": 0.7078, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.157, "n_target_mean": 10.0553, "alpha": 0.03, "t": 651.6} {"event": "probe", "step": 3400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4202.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.241453, "cps": 32.828125, "mean_p": 0.427438, "boundary_entropy": 0.652684, "bpb": 2.119011}, "de": {"chunks": 5033.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.245118, "cps": 39.320312, "mean_p": 0.434329, "boundary_entropy": 0.662131, "bpb": 2.363952}, "vi": {"chunks": 4508.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.244535, "cps": 35.21875, "mean_p": 0.423232, "boundary_entropy": 0.658467, "bpb": 1.985157}, "tr": {"chunks": 5261.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.290615, "cps": 41.101562, "mean_p": 0.449206, "boundary_entropy": 0.664226, "bpb": 2.318623}, "fi": {"chunks": 4374.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.232524, "cps": 34.171875, "mean_p": 0.44254, "boundary_entropy": 0.667512, "bpb": 2.543975}, "ru": {"chunks": 5572.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.294037, "cps": 43.53125, "mean_p": 0.407076, "boundary_entropy": 0.648282, "bpb": 1.512656}, "el": {"chunks": 4554.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.221897, "cps": 35.578125, "mean_p": 0.425689, "boundary_entropy": 0.668399, "bpb": 1.571958}, "hi": {"chunks": 3631.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.21581, "cps": 28.367188, "mean_p": 0.40909, "boundary_entropy": 0.656471, "bpb": 1.261788}, "ta": {"chunks": 4152.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.222091, "cps": 32.4375, "mean_p": 0.4008, "boundary_entropy": 0.654311, "bpb": 1.027953}, "th": {"chunks": 3449.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.214423, "cps": 26.945312, "mean_p": 0.401204, "boundary_entropy": 0.650384, "bpb": 1.157722}, "zh": {"chunks": 3156.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.534372, "cps": 24.65625, "mean_p": 0.461034, "boundary_entropy": 0.678249, "bpb": 3.264582}, "my": {"chunks": 2147.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.11331, "cps": 16.773438, "mean_p": 0.427924, "boundary_entropy": 0.675895, "bpb": 1.156158}}, "bpb": {"en": 2.11901, "de": 2.36395, "vi": 1.98516, "tr": 2.31862, "fi": 2.54398, "ru": 1.51266, "el": 1.57196, "hi": 1.26179, "ta": 1.02795, "th": 1.15772, "zh": 3.26458, "my": 1.15616}, "gini_cpc": 0.164599, "gini_cps": 0.122509, "bpic": 7.669638, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0151789298767104, "vi": 1.012765357578959, "tr": 1.2036100974162787, "fi": 0.9630192477281778, "ru": 1.217783163941282, "el": 0.9190100342024642, "hi": 0.8937977316398545, "ta": 0.9198138560084036, "th": 0.8880556899870201, "zh": 2.213153944560288, "my": 0.4692850730931529}, "t": 662.5} {"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 2.12799, "de": 2.36817, "vi": 1.98637, "tr": 2.32536, "fi": 2.55745, "ru": 1.49059, "el": 1.56464, "hi": 1.25997, "ta": 1.03524, "th": 1.17391, "zh": 3.26286, "my": 1.16521}, "bytes_per_chunk": {"en": 4.141, "de": 4.14, "vi": 5.164, "tr": 3.803, "fi": 4.422, "ru": 6.129, "el": 7.993, "hi": 12.263, "ta": 11.613, "th": 13.334, "zh": 4.8, "my": 23.429}, "bpb_mean": 1.85981, "frozen_inventory": false, "t": 667.3} {"event": "train", "step": 3400, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.30846, "bpb": 1.88771, "chunk_loss": 0.036093, "F": 0.18063, "G": 0.42047, "chunks": 184.97, "bytes_per_chunk": 5.536, "boundary_entropy": 0.65976, "router_noise": 0.0, "grad_norm": 0.7163, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.321, "n_target_mean": 10.5619, "alpha": 0.03, "t": 667.7} {"event": "train", "step": 3450, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.21638, "bpb": 1.75487, "chunk_loss": 0.036352, "F": 0.18066, "G": 0.4186, "chunks": 185.0, "bytes_per_chunk": 5.535, "boundary_entropy": 0.65761, "router_noise": 0.0, "grad_norm": 0.6818, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.156, "n_target_mean": 10.9543, "alpha": 0.03, "t": 675.5} {"event": "train", "step": 3500, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.32629, "bpb": 1.91344, "chunk_loss": 0.037032, "F": 0.19891, "G": 0.42562, "chunks": 203.69, "bytes_per_chunk": 5.027, "boundary_entropy": 0.66099, "router_noise": 0.0, "grad_norm": 0.7505, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.159, "n_target_mean": 10.0046, "alpha": 0.03, "t": 683.4} {"event": "train", "step": 3550, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.37052, "bpb": 1.97724, "chunk_loss": 0.035197, "F": 0.18143, "G": 0.42429, "chunks": 185.78, "bytes_per_chunk": 5.512, "boundary_entropy": 0.66256, "router_noise": 0.0, "grad_norm": 0.741, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.155, "n_target_mean": 9.7442, "alpha": 0.03, "t": 691.2} {"event": "probe", "step": 3600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4362.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.250646, "cps": 34.078125, "mean_p": 0.427544, "boundary_entropy": 0.651531, "bpb": 2.100276}, "de": {"chunks": 5042.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.245556, "cps": 39.390625, "mean_p": 0.434439, "boundary_entropy": 0.661552, "bpb": 2.325901}, "vi": {"chunks": 4118.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.223379, "cps": 32.171875, "mean_p": 0.417927, "boundary_entropy": 0.656725, "bpb": 1.967578}, "tr": {"chunks": 5611.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.309949, "cps": 43.835938, "mean_p": 0.451281, "boundary_entropy": 0.664133, "bpb": 2.282312}, "fi": {"chunks": 4727.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251289, "cps": 36.929688, "mean_p": 0.445962, "boundary_entropy": 0.667932, "bpb": 2.504782}, "ru": {"chunks": 5041.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.266016, "cps": 39.382812, "mean_p": 0.40233, "boundary_entropy": 0.64689, "bpb": 1.503556}, "el": {"chunks": 4800.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.233884, "cps": 37.5, "mean_p": 0.425233, "boundary_entropy": 0.667902, "bpb": 1.561889}, "hi": {"chunks": 4095.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.243388, "cps": 31.992188, "mean_p": 0.409172, "boundary_entropy": 0.656707, "bpb": 1.241812}, "ta": {"chunks": 4234.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.226478, "cps": 33.078125, "mean_p": 0.40176, "boundary_entropy": 0.654254, "bpb": 1.024371}, "th": {"chunks": 2295.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.14268, "cps": 17.929688, "mean_p": 0.394968, "boundary_entropy": 0.649148, "bpb": 1.15384}, "zh": {"chunks": 2813.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.476295, "cps": 21.976562, "mean_p": 0.458359, "boundary_entropy": 0.677601, "bpb": 3.245043}, "my": {"chunks": 2615.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.138009, "cps": 20.429688, "mean_p": 0.432642, "boundary_entropy": 0.67713, "bpb": 1.141341}}, "bpb": {"en": 2.10028, "de": 2.3259, "vi": 1.96758, "tr": 2.28231, "fi": 2.50478, "ru": 1.50356, "el": 1.56189, "hi": 1.24181, "ta": 1.02437, "th": 1.15384, "zh": 3.24504, "my": 1.14134}, "gini_cpc": 0.157977, "gini_cps": 0.132982, "bpic": 7.713726, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9796904918582768, "vi": 0.8912133004582442, "tr": 1.2365969648992343, "fi": 1.0025641659373263, "ru": 1.0613190071606668, "el": 0.9331229074958073, "hi": 0.9710403845721493, "ta": 0.9035742441928695, "th": 0.5692461258299808, "zh": 1.900267533867361, "my": 0.5506133996205935}, "t": 702.3} {"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 2.10918, "de": 2.33308, "vi": 1.97079, "tr": 2.28864, "fi": 2.51742, "ru": 1.48205, "el": 1.5555, "hi": 1.23782, "ta": 1.0329, "th": 1.16885, "zh": 3.24566, "my": 1.15082}, "bytes_per_chunk": {"en": 4.003, "de": 4.125, "vi": 5.71, "tr": 3.544, "fi": 3.97, "ru": 6.956, "el": 7.778, "hi": 10.923, "ta": 11.505, "th": 18.777, "zh": 5.413, "my": 19.394}, "bpb_mean": 1.84106, "frozen_inventory": false, "t": 706.9} {"event": "train", "step": 3600, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.3179, "bpb": 1.90133, "chunk_loss": 0.034779, "F": 0.17075, "G": 0.41815, "chunks": 174.84, "bytes_per_chunk": 5.857, "boundary_entropy": 0.65882, "router_noise": 0.0, "grad_norm": 0.7463, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.326, "n_target_mean": 10.3328, "alpha": 0.03, "t": 707.5} {"event": "train", "step": 3650, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.30712, "bpb": 1.88578, "chunk_loss": 0.037058, "F": 0.2063, "G": 0.41996, "chunks": 211.25, "bytes_per_chunk": 4.847, "boundary_entropy": 0.65724, "router_noise": 0.0, "grad_norm": 0.7035, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.161, "n_target_mean": 9.3376, "alpha": 0.03, "t": 715.5} {"event": "train", "step": 3700, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.52669, "bpb": 2.20255, "chunk_loss": 0.036462, "F": 0.22137, "G": 0.43073, "chunks": 226.69, "bytes_per_chunk": 4.517, "boundary_entropy": 0.66333, "router_noise": 0.0, "grad_norm": 0.7906, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.154, "n_target_mean": 8.3595, "alpha": 0.03, "t": 723.2} {"event": "train", "step": 3750, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.41948, "bpb": 2.04787, "chunk_loss": 0.036838, "F": 0.22882, "G": 0.42422, "chunks": 234.31, "bytes_per_chunk": 4.37, "boundary_entropy": 0.65824, "router_noise": 0.0, "grad_norm": 0.6843, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.16, "n_target_mean": 8.2469, "alpha": 0.03, "t": 731.2} {"event": "probe", "step": 3800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4415.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.253692, "cps": 34.492188, "mean_p": 0.426518, "boundary_entropy": 0.650274, "bpb": 2.084828}, "de": {"chunks": 5208.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.25364, "cps": 40.6875, "mean_p": 0.432979, "boundary_entropy": 0.660189, "bpb": 2.306821}, "vi": {"chunks": 5309.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.287985, "cps": 41.476562, "mean_p": 0.422674, "boundary_entropy": 0.656852, "bpb": 1.951627}, "tr": {"chunks": 5582.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.308347, "cps": 43.609375, "mean_p": 0.449193, "boundary_entropy": 0.662947, "bpb": 2.257113}, "fi": {"chunks": 4794.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254851, "cps": 37.453125, "mean_p": 0.444342, "boundary_entropy": 0.666781, "bpb": 2.474211}, "ru": {"chunks": 4999.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.263799, "cps": 39.054688, "mean_p": 0.400668, "boundary_entropy": 0.645404, "bpb": 1.482126}, "el": {"chunks": 4813.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.234517, "cps": 37.601562, "mean_p": 0.423341, "boundary_entropy": 0.666777, "bpb": 1.536175}, "hi": {"chunks": 4165.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.247548, "cps": 32.539062, "mean_p": 0.407111, "boundary_entropy": 0.654752, "bpb": 1.224632}, "ta": {"chunks": 3912.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.209254, "cps": 30.5625, "mean_p": 0.399316, "boundary_entropy": 0.651588, "bpb": 1.010716}, "th": {"chunks": 2699.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.167796, "cps": 21.085938, "mean_p": 0.394901, "boundary_entropy": 0.647412, "bpb": 1.14053}, "zh": {"chunks": 2996.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.507281, "cps": 23.40625, "mean_p": 0.461073, "boundary_entropy": 0.678282, "bpb": 3.242092}, "my": {"chunks": 3661.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.193213, "cps": 28.601562, "mean_p": 0.441057, "boundary_entropy": 0.678934, "bpb": 1.126576}}, "bpb": {"en": 2.08483, "de": 2.30682, "vi": 1.95163, "tr": 2.25711, "fi": 2.47421, "ru": 1.48213, "el": 1.53617, "hi": 1.22463, "ta": 1.01072, "th": 1.14053, "zh": 3.24209, "my": 1.12658}, "gini_cpc": 0.143782, "gini_cps": 0.11277, "bpic": 7.302742, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9997973485655964, "vi": 1.1351754642201901, "tr": 1.2154376720251032, "fi": 1.0045685059457452, "ru": 1.0398419516101562, "el": 0.9244180754802128, "hi": 0.9757832191014356, "ta": 0.8248344452901785, "th": 0.6614168149035542, "zh": 1.9995937869966585, "my": 0.7616049623562944}, "t": 742.2} {"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 2.09278, "de": 2.31498, "vi": 1.95362, "tr": 2.26512, "fi": 2.4874, "ru": 1.45952, "el": 1.52977, "hi": 1.22143, "ta": 1.01791, "th": 1.15761, "zh": 3.23965, "my": 1.13458}, "bytes_per_chunk": {"en": 3.927, "de": 4.015, "vi": 4.501, "tr": 3.514, "fi": 3.913, "ru": 6.974, "el": 7.682, "hi": 10.587, "ta": 12.329, "th": 16.378, "zh": 5.114, "my": 15.071}, "bpb_mean": 1.82286, "frozen_inventory": false, "t": 747.1} {"event": "train", "step": 3800, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.37371, "bpb": 1.98184, "chunk_loss": 0.036404, "F": 0.19861, "G": 0.42098, "chunks": 203.38, "bytes_per_chunk": 5.035, "boundary_entropy": 0.65792, "router_noise": 0.0, "grad_norm": 0.7085, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.325, "n_target_mean": 9.6644, "alpha": 0.03, "t": 747.5} {"event": "train", "step": 3850, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.52979, "bpb": 2.20702, "chunk_loss": 0.036117, "F": 0.20895, "G": 0.42651, "chunks": 213.97, "bytes_per_chunk": 4.786, "boundary_entropy": 0.66143, "router_noise": 0.0, "grad_norm": 0.7397, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.158, "n_target_mean": 8.3671, "alpha": 0.03, "t": 755.4} {"event": "train", "step": 3900, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.46922, "bpb": 2.11963, "chunk_loss": 0.035188, "F": 0.21005, "G": 0.42443, "chunks": 215.09, "bytes_per_chunk": 4.761, "boundary_entropy": 0.66001, "router_noise": 0.0, "grad_norm": 0.714, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.159, "n_target_mean": 8.3124, "alpha": 0.03, "t": 763.3} {"event": "train", "step": 3950, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.42129, "bpb": 2.05049, "chunk_loss": 0.035816, "F": 0.20648, "G": 0.42188, "chunks": 211.44, "bytes_per_chunk": 4.843, "boundary_entropy": 0.65816, "router_noise": 0.0, "grad_norm": 0.7231, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.159, "n_target_mean": 8.5064, "alpha": 0.03, "t": 771.3} {"event": "probe", "step": 4000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 4309.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.247601, "cps": 33.664062, "mean_p": 0.425515, "boundary_entropy": 0.650302, "bpb": 2.078337}, "de": {"chunks": 5276.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.256952, "cps": 41.21875, "mean_p": 0.434034, "boundary_entropy": 0.660759, "bpb": 2.291653}, "vi": {"chunks": 4747.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.257499, "cps": 37.085938, "mean_p": 0.419548, "boundary_entropy": 0.655894, "bpb": 1.945171}, "tr": {"chunks": 5395.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.298017, "cps": 42.148438, "mean_p": 0.447756, "boundary_entropy": 0.661887, "bpb": 2.240156}, "fi": {"chunks": 4709.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.250332, "cps": 36.789062, "mean_p": 0.443515, "boundary_entropy": 0.666426, "bpb": 2.455523}, "ru": {"chunks": 4884.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.257731, "cps": 38.15625, "mean_p": 0.39928, "boundary_entropy": 0.644929, "bpb": 1.47604}, "el": {"chunks": 4500.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.219266, "cps": 35.15625, "mean_p": 0.421372, "boundary_entropy": 0.666457, "bpb": 1.53084}, "hi": {"chunks": 3971.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.236018, "cps": 31.023438, "mean_p": 0.40558, "boundary_entropy": 0.655317, "bpb": 1.222051}, "ta": {"chunks": 3582.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.191602, "cps": 27.984375, "mean_p": 0.394179, "boundary_entropy": 0.650674, "bpb": 1.002484}, "th": {"chunks": 5538.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.344296, "cps": 43.265625, "mean_p": 0.397531, "boundary_entropy": 0.646837, "bpb": 1.125213}, "zh": {"chunks": 3205.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.542668, "cps": 25.039062, "mean_p": 0.462248, "boundary_entropy": 0.67838, "bpb": 3.194982}, "my": {"chunks": 2574.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.135845, "cps": 20.109375, "mean_p": 0.430164, "boundary_entropy": 0.67603, "bpb": 1.119427}}, "bpb": {"en": 2.07834, "de": 2.29165, "vi": 1.94517, "tr": 2.24016, "fi": 2.45552, "ru": 1.47604, "el": 1.53084, "hi": 1.22205, "ta": 1.00248, "th": 1.12521, "zh": 3.19498, "my": 1.11943}, "gini_cpc": 0.168463, "gini_cps": 0.111122, "bpic": 7.283754, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0377673569798835, "vi": 1.0399769551533766, "tr": 1.2036175835941079, "fi": 1.01103090940196, "ru": 1.0409121241605745, "el": 0.8855626571920787, "hi": 0.953218451170138, "ta": 0.7738338765011079, "th": 1.3905272761545495, "zh": 2.1917056001594406, "my": 0.5486467268604778}, "t": 782.5} {"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 2.08598, "de": 2.2976, "vi": 1.94571, "tr": 2.24742, "fi": 2.46925, "ru": 1.45272, "el": 1.52456, "hi": 1.21895, "ta": 1.01039, "th": 1.14272, "zh": 3.19367, "my": 1.12867}, "bytes_per_chunk": {"en": 4.071, "de": 3.948, "vi": 4.981, "tr": 3.633, "fi": 4.009, "ru": 7.047, "el": 7.987, "hi": 11.074, "ta": 13.359, "th": 8.83, "zh": 4.81, "my": 19.734}, "bpb_mean": 1.8098, "frozen_inventory": false, "t": 787.4} {"event": "train", "step": 4000, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.30965, "bpb": 1.88942, "chunk_loss": 0.036241, "F": 0.18918, "G": 0.41994, "chunks": 193.72, "bytes_per_chunk": 5.286, "boundary_entropy": 0.65839, "router_noise": 0.0, "grad_norm": 0.7149, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.335, "n_target_mean": 10.021, "alpha": 0.03, "t": 788.0} {"event": "train", "step": 4050, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.46362, "bpb": 2.11155, "chunk_loss": 0.034854, "F": 0.19986, "G": 0.42553, "chunks": 204.66, "bytes_per_chunk": 5.004, "boundary_entropy": 0.66133, "router_noise": 0.0, "grad_norm": 0.6956, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.154, "n_target_mean": 8.3995, "alpha": 0.03, "t": 795.7} {"event": "train", "step": 4100, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.32855, "bpb": 1.91669, "chunk_loss": 0.03471, "F": 0.17981, "G": 0.41792, "chunks": 184.12, "bytes_per_chunk": 5.561, "boundary_entropy": 0.65742, "router_noise": 0.0, "grad_norm": 0.77, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.156, "n_target_mean": 9.8635, "alpha": 0.03, "t": 803.6} {"event": "train", "step": 4150, "arch": "hnet", "phase": 2, "objective": "parity_a", "lm_loss": 1.4761, "bpb": 2.12956, "chunk_loss": 0.035316, "F": 0.20782, "G": 0.42568, "chunks": 212.81, "bytes_per_chunk": 4.812, "boundary_entropy": 0.6613, "router_noise": 0.0, "grad_norm": 0.7382, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.16, "n_target_mean": 8.0388, "alpha": 0.03, "t": 811.6} {"event": "probe", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4148.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.23835, "cps": 32.40625, "mean_p": 0.423241, "boundary_entropy": 0.649077, "bpb": 2.065375}, "de": {"chunks": 4980.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.242536, "cps": 38.90625, "mean_p": 0.429913, "boundary_entropy": 0.658807, "bpb": 2.281451}, "vi": {"chunks": 4185.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227014, "cps": 32.695312, "mean_p": 0.415388, "boundary_entropy": 0.654944, "bpb": 1.944914}, "tr": {"chunks": 5454.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.301276, "cps": 42.609375, "mean_p": 0.448212, "boundary_entropy": 0.66203, "bpb": 2.226007}, "fi": {"chunks": 4724.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25113, "cps": 36.90625, "mean_p": 0.443329, "boundary_entropy": 0.665906, "bpb": 2.437983}, "ru": {"chunks": 4894.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.258259, "cps": 38.234375, "mean_p": 0.396868, "boundary_entropy": 0.643403, "bpb": 1.455777}, "el": {"chunks": 4470.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.217804, "cps": 34.921875, "mean_p": 0.419128, "boundary_entropy": 0.665198, "bpb": 1.51624}, "hi": {"chunks": 3897.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.23162, "cps": 30.445312, "mean_p": 0.401724, "boundary_entropy": 0.652027, "bpb": 1.209711}, "ta": {"chunks": 3768.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.201551, "cps": 29.4375, "mean_p": 0.392141, "boundary_entropy": 0.648497, "bpb": 0.994416}, "th": {"chunks": 3739.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.232453, "cps": 29.210938, "mean_p": 0.393787, "boundary_entropy": 0.644864, "bpb": 1.122084}, "zh": {"chunks": 3043.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.515239, "cps": 23.773438, "mean_p": 0.460393, "boundary_entropy": 0.677994, "bpb": 3.185907}, "my": {"chunks": 3133.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.165347, "cps": 24.476562, "mean_p": 0.435143, "boundary_entropy": 0.676928, "bpb": 1.111318}}, "bpb": {"en": 2.06538, "de": 2.28145, "vi": 1.94491, "tr": 2.22601, "fi": 2.43798, "ru": 1.45578, "el": 1.51624, "hi": 1.20971, "ta": 0.99442, "th": 1.12208, "zh": 3.18591, "my": 1.11132}, "gini_cpc": 0.14027, "gini_cps": 0.095445, "bpic": 7.609418, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0175653454530307, "vi": 0.9524401458485193, "tr": 1.2640083842531626, "fi": 1.0536184766837886, "ru": 1.083527961467904, "el": 0.9138018868310179, "hi": 0.9717637745181623, "ta": 0.8456113374545404, "th": 0.9752585633997786, "zh": 2.1616923328873665, "my": 0.693717086238559}, "t": 822.6} {"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 2.07437, "de": 2.2884, "vi": 1.94637, "tr": 2.23294, "fi": 2.45086, "ru": 1.43321, "el": 1.51069, "hi": 1.20579, "ta": 1.00319, "th": 1.13919, "zh": 3.18374, "my": 1.11977}, "bytes_per_chunk": {"en": 4.126, "de": 4.147, "vi": 5.763, "tr": 3.601, "fi": 4.064, "ru": 7.141, "el": 8.258, "hi": 11.186, "ta": 12.925, "th": 12.449, "zh": 5.21, "my": 17.247}, "bpb_mean": 1.79904, "frozen_inventory": false, "t": 827.1} {"event": "phase_transition", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 131584, "t": 827.3} {"event": "train", "step": 4200, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.29387, "bpb": 1.86666, "chunk_loss": 0.0, "F": 0.17944, "G": 0.41711, "chunks": 183.75, "bytes_per_chunk": 5.573, "boundary_entropy": 0.65697, "router_noise": 0.0, "grad_norm": 0.6812, "lr": 9.12e-05, "router_trainable": false, "sec_per_step": 0.318, "alpha": 0.0, "t": 827.5} {"event": "train", "step": 4250, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.30611, "bpb": 1.88432, "chunk_loss": 0.0, "F": 0.1965, "G": 0.42212, "chunks": 201.22, "bytes_per_chunk": 5.089, "boundary_entropy": 0.65828, "router_noise": 0.0, "grad_norm": 0.7288, "lr": 8.81e-05, "router_trainable": false, "sec_per_step": 0.155, "alpha": 0.0, "t": 835.2} {"event": "train", "step": 4300, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.34755, "bpb": 1.9441, "chunk_loss": 0.0, "F": 0.21802, "G": 0.42192, "chunks": 223.25, "bytes_per_chunk": 4.587, "boundary_entropy": 0.65708, "router_noise": 0.0, "grad_norm": 0.7261, "lr": 8.5e-05, "router_trainable": false, "sec_per_step": 0.161, "alpha": 0.0, "t": 843.3} {"event": "train", "step": 4350, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.47817, "bpb": 2.13254, "chunk_loss": 0.0, "F": 0.21204, "G": 0.43118, "chunks": 217.12, "bytes_per_chunk": 4.716, "boundary_entropy": 0.66432, "router_noise": 0.0, "grad_norm": 0.7578, "lr": 8.21e-05, "router_trainable": false, "sec_per_step": 0.16, "alpha": 0.0, "t": 851.3} {"event": "probe", "step": 4400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4587.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.263575, "cps": 35.835938, "mean_p": 0.429682, "boundary_entropy": 0.651704, "bpb": 2.04803}, "de": {"chunks": 5509.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.2683, "cps": 43.039062, "mean_p": 0.43675, "boundary_entropy": 0.661147, "bpb": 2.261301}, "vi": {"chunks": 5209.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.28256, "cps": 40.695312, "mean_p": 0.423421, "boundary_entropy": 0.657558, "bpb": 1.927826}, "tr": {"chunks": 5812.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.321052, "cps": 45.40625, "mean_p": 0.454072, "boundary_entropy": 0.663814, "bpb": 2.204025}, "fi": {"chunks": 5036.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.267716, "cps": 39.34375, "mean_p": 0.44705, "boundary_entropy": 0.667587, "bpb": 2.418278}, "ru": {"chunks": 5608.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.295937, "cps": 43.8125, "mean_p": 0.405996, "boundary_entropy": 0.64821, "bpb": 1.439857}, "el": {"chunks": 4541.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.221264, "cps": 35.476562, "mean_p": 0.425314, "boundary_entropy": 0.668299, "bpb": 1.506566}, "hi": {"chunks": 4368.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.259614, "cps": 34.125, "mean_p": 0.411683, "boundary_entropy": 0.658288, "bpb": 1.199687}, "ta": {"chunks": 4121.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.220433, "cps": 32.195312, "mean_p": 0.401512, "boundary_entropy": 0.654364, "bpb": 0.98918}, "th": {"chunks": 4001.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.248741, "cps": 31.257812, "mean_p": 0.399792, "boundary_entropy": 0.649548, "bpb": 1.113556}, "zh": {"chunks": 3328.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.563495, "cps": 26.0, "mean_p": 0.463401, "boundary_entropy": 0.678559, "bpb": 3.170101}, "my": {"chunks": 3628.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.191471, "cps": 28.34375, "mean_p": 0.438187, "boundary_entropy": 0.67797, "bpb": 1.105911}}, "bpb": {"en": 2.04803, "de": 2.2613, "vi": 1.92783, "tr": 2.20402, "fi": 2.41828, "ru": 1.43986, "el": 1.50657, "hi": 1.19969, "ta": 0.98918, "th": 1.11356, "zh": 3.1701, "my": 1.10591}, "gini_cpc": 0.140021, "gini_cps": 0.094458, "bpic": 6.884211, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0179249170495759, "vi": 1.0720291523272247, "tr": 1.21806491572622, "fi": 1.0157088068122062, "ru": 1.12277871442352, "el": 0.8394716553238121, "hi": 0.9849698498664259, "ta": 0.8363200815262595, "th": 0.9437193636731613, "zh": 2.1378895036850487, "my": 0.7264392182494434}, "t": 862.2} {"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 2.05574, "de": 2.26717, "vi": 1.9292, "tr": 2.21144, "fi": 2.43036, "ru": 1.41631, "el": 1.49913, "hi": 1.19734, "ta": 0.99743, "th": 1.13063, "zh": 3.16745, "my": 1.11474}, "bytes_per_chunk": {"en": 3.757, "de": 3.764, "vi": 4.559, "tr": 3.365, "fi": 3.786, "ru": 6.093, "el": 8.031, "hi": 10.248, "ta": 11.632, "th": 11.761, "zh": 4.666, "my": 14.473}, "bpb_mean": 1.78474, "frozen_inventory": false, "t": 866.8} {"event": "train", "step": 4400, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.43824, "bpb": 2.07494, "chunk_loss": 0.0, "F": 0.23624, "G": 0.43126, "chunks": 241.91, "bytes_per_chunk": 4.233, "boundary_entropy": 0.6619, "router_noise": 0.0, "grad_norm": 0.7868, "lr": 7.92e-05, "router_trainable": false, "sec_per_step": 0.321, "alpha": 0.0, "t": 867.3} {"event": "train", "step": 4450, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.3693, "bpb": 1.97548, "chunk_loss": 0.0, "F": 0.23541, "G": 0.43265, "chunks": 241.06, "bytes_per_chunk": 4.248, "boundary_entropy": 0.6628, "router_noise": 0.0, "grad_norm": 0.7547, "lr": 7.63e-05, "router_trainable": false, "sec_per_step": 0.161, "alpha": 0.0, "t": 875.4} {"event": "train", "step": 4500, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.30873, "bpb": 1.88809, "chunk_loss": 0.0, "F": 0.22147, "G": 0.42512, "chunks": 226.78, "bytes_per_chunk": 4.515, "boundary_entropy": 0.65946, "router_noise": 0.0, "grad_norm": 0.731, "lr": 7.36e-05, "router_trainable": false, "sec_per_step": 0.161, "alpha": 0.0, "t": 883.5} {"event": "train", "step": 4550, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.35523, "bpb": 1.95518, "chunk_loss": 0.0, "F": 0.21298, "G": 0.43015, "chunks": 218.09, "bytes_per_chunk": 4.695, "boundary_entropy": 0.66351, "router_noise": 0.0, "grad_norm": 0.7408, "lr": 7.09e-05, "router_trainable": false, "sec_per_step": 0.159, "alpha": 0.0, "t": 891.4} {"event": "probe", "step": 4600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4769.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.274033, "cps": 37.257812, "mean_p": 0.432302, "boundary_entropy": 0.652348, "bpb": 2.038689}, "de": {"chunks": 5570.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.271271, "cps": 43.515625, "mean_p": 0.439151, "boundary_entropy": 0.662202, "bpb": 2.24093}, "vi": {"chunks": 5085.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.275834, "cps": 39.726562, "mean_p": 0.424879, "boundary_entropy": 0.658399, "bpb": 1.917887}, "tr": {"chunks": 5915.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.326741, "cps": 46.210938, "mean_p": 0.456162, "boundary_entropy": 0.664136, "bpb": 2.19388}, "fi": {"chunks": 5152.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.273882, "cps": 40.25, "mean_p": 0.449026, "boundary_entropy": 0.667731, "bpb": 2.40583}, "ru": {"chunks": 6032.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.318311, "cps": 47.125, "mean_p": 0.410015, "boundary_entropy": 0.650037, "bpb": 1.428786}, "el": {"chunks": 5123.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.249622, "cps": 40.023438, "mean_p": 0.429022, "boundary_entropy": 0.669005, "bpb": 1.486221}, "hi": {"chunks": 5099.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.303061, "cps": 39.835938, "mean_p": 0.414941, "boundary_entropy": 0.658507, "bpb": 1.195869}, "ta": {"chunks": 4104.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.219524, "cps": 32.0625, "mean_p": 0.404028, "boundary_entropy": 0.65495, "bpb": 0.984501}, "th": {"chunks": 5494.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.34156, "cps": 42.921875, "mean_p": 0.402446, "boundary_entropy": 0.650113, "bpb": 1.103375}, "zh": {"chunks": 3292.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.557399, "cps": 25.71875, "mean_p": 0.46319, "boundary_entropy": 0.678515, "bpb": 3.156145}, "my": {"chunks": 5749.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.303409, "cps": 44.914062, "mean_p": 0.445391, "boundary_entropy": 0.679542, "bpb": 1.102014}}, "bpb": {"en": 2.03869, "de": 2.24093, "vi": 1.91789, "tr": 2.19388, "fi": 2.40583, "ru": 1.42879, "el": 1.48622, "hi": 1.19587, "ta": 0.9845, "th": 1.10337, "zh": 3.15615, "my": 1.10201}, "gini_cpc": 0.118223, "gini_cps": 0.077298, "bpic": 6.252134, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9899188309063969, "vi": 1.0065714615735164, "tr": 1.1923424204547386, "fi": 0.9994493012173862, "ru": 1.1615794397745773, "el": 0.9109201489459114, "hi": 1.1059277023446061, "ta": 0.801085148573931, "th": 1.2464199384260317, "zh": 2.0340572939141537, "my": 1.107199115939163}, "t": 902.2} {"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 2.04723, "de": 2.24813, "vi": 1.91947, "tr": 2.1998, "fi": 2.41659, "ru": 1.40588, "el": 1.48055, "hi": 1.19285, "ta": 0.99177, "th": 1.12057, "zh": 3.1526, "my": 1.11006}, "bytes_per_chunk": {"en": 3.584, "de": 3.719, "vi": 4.598, "tr": 3.303, "fi": 3.664, "ru": 5.686, "el": 7.149, "hi": 8.692, "ta": 11.658, "th": 8.583, "zh": 4.692, "my": 9.757}, "bpb_mean": 1.77379, "frozen_inventory": false, "t": 907.2} {"event": "train", "step": 4600, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.404, "bpb": 2.02554, "chunk_loss": 0.0, "F": 0.22375, "G": 0.43402, "chunks": 229.12, "bytes_per_chunk": 4.469, "boundary_entropy": 0.66484, "router_noise": 0.0, "grad_norm": 0.7236, "lr": 6.82e-05, "router_trainable": false, "sec_per_step": 0.324, "alpha": 0.0, "t": 907.6} {"event": "train", "step": 4650, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.50703, "bpb": 2.17418, "chunk_loss": 0.0, "F": 0.23062, "G": 0.43618, "chunks": 236.16, "bytes_per_chunk": 4.336, "boundary_entropy": 0.66582, "router_noise": 0.0, "grad_norm": 0.7777, "lr": 6.57e-05, "router_trainable": false, "sec_per_step": 0.163, "alpha": 0.0, "t": 915.7} {"event": "train", "step": 4700, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.44655, "bpb": 2.08693, "chunk_loss": 0.0, "F": 0.21747, "G": 0.43221, "chunks": 222.69, "bytes_per_chunk": 4.598, "boundary_entropy": 0.66507, "router_noise": 0.0, "grad_norm": 0.7969, "lr": 6.32e-05, "router_trainable": false, "sec_per_step": 0.162, "alpha": 0.0, "t": 923.8} {"event": "train", "step": 4750, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.44959, "bpb": 2.09132, "chunk_loss": 0.0, "F": 0.23856, "G": 0.43248, "chunks": 244.28, "bytes_per_chunk": 4.192, "boundary_entropy": 0.66251, "router_noise": 0.0, "grad_norm": 0.7449, "lr": 6.08e-05, "router_trainable": false, "sec_per_step": 0.151, "alpha": 0.0, "t": 931.4} {"event": "probe", "step": 4800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4869.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.279779, "cps": 38.039062, "mean_p": 0.433969, "boundary_entropy": 0.653123, "bpb": 2.027226}, "de": {"chunks": 5705.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.277845, "cps": 44.570312, "mean_p": 0.440105, "boundary_entropy": 0.662388, "bpb": 2.237945}, "vi": {"chunks": 5004.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.27144, "cps": 39.09375, "mean_p": 0.424693, "boundary_entropy": 0.658834, "bpb": 1.914668}, "tr": {"chunks": 5951.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.32873, "cps": 46.492188, "mean_p": 0.456352, "boundary_entropy": 0.664344, "bpb": 2.181503}, "fi": {"chunks": 5382.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.286109, "cps": 42.046875, "mean_p": 0.450104, "boundary_entropy": 0.667748, "bpb": 2.385064}, "ru": {"chunks": 5988.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.315989, "cps": 46.78125, "mean_p": 0.409773, "boundary_entropy": 0.649705, "bpb": 1.421564}, "el": {"chunks": 4977.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.242508, "cps": 38.882812, "mean_p": 0.427762, "boundary_entropy": 0.669024, "bpb": 1.477926}, "hi": {"chunks": 4330.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.257355, "cps": 33.828125, "mean_p": 0.409934, "boundary_entropy": 0.657172, "bpb": 1.187546}, "ta": {"chunks": 3936.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.210538, "cps": 30.75, "mean_p": 0.402314, "boundary_entropy": 0.653743, "bpb": 0.980989}, "th": {"chunks": 3956.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.245943, "cps": 30.90625, "mean_p": 0.399914, "boundary_entropy": 0.649345, "bpb": 1.099273}, "zh": {"chunks": 3360.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.568913, "cps": 26.25, "mean_p": 0.464013, "boundary_entropy": 0.678728, "bpb": 3.145536}, "my": {"chunks": 4439.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.234273, "cps": 34.679688, "mean_p": 0.441891, "boundary_entropy": 0.678935, "bpb": 1.100564}}, "bpb": {"en": 2.02723, "de": 2.23794, "vi": 1.91467, "tr": 2.1815, "fi": 2.38506, "ru": 1.42156, "el": 1.47793, "hi": 1.18755, "ta": 0.98099, "th": 1.09927, "zh": 3.14554, "my": 1.10056}, "gini_cpc": 0.13215, "gini_cps": 0.095499, "bpic": 6.628685, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9930876641508721, "vi": 0.9701938218258988, "tr": 1.1749617927047082, "fi": 1.02262441126839, "ru": 1.129423768161179, "el": 0.866784509296319, "hi": 0.9198503312798285, "ta": 0.7525129288673763, "th": 0.879062114371989, "zh": 2.0334344659054704, "my": 0.8373482453713512}, "t": 942.6} {"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 2.03549, "de": 2.24348, "vi": 1.9144, "tr": 2.18874, "fi": 2.39673, "ru": 1.39811, "el": 1.47158, "hi": 1.18471, "ta": 0.98893, "th": 1.11555, "zh": 3.14165, "my": 1.10807}, "bytes_per_chunk": {"en": 3.513, "de": 3.675, "vi": 4.678, "tr": 3.264, "fi": 3.525, "ru": 5.747, "el": 7.331, "hi": 10.128, "ta": 12.219, "th": 11.583, "zh": 4.555, "my": 12.31}, "bpb_mean": 1.76562, "frozen_inventory": false, "t": 947.4} {"event": "train", "step": 4800, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.34301, "bpb": 1.93755, "chunk_loss": 0.0, "F": 0.23083, "G": 0.43055, "chunks": 236.38, "bytes_per_chunk": 4.332, "boundary_entropy": 0.66183, "router_noise": 0.0, "grad_norm": 0.7018, "lr": 5.85e-05, "router_trainable": false, "sec_per_step": 0.328, "alpha": 0.0, "t": 947.8} {"event": "train", "step": 4850, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.27529, "bpb": 1.83985, "chunk_loss": 0.0, "F": 0.23663, "G": 0.43164, "chunks": 242.31, "bytes_per_chunk": 4.226, "boundary_entropy": 0.66126, "router_noise": 0.0, "grad_norm": 0.6668, "lr": 5.62e-05, "router_trainable": false, "sec_per_step": 0.161, "alpha": 0.0, "t": 955.9} {"event": "train", "step": 4900, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.3755, "bpb": 1.98443, "chunk_loss": 0.0, "F": 0.22278, "G": 0.4326, "chunks": 228.12, "bytes_per_chunk": 4.489, "boundary_entropy": 0.66426, "router_noise": 0.0, "grad_norm": 0.7874, "lr": 5.41e-05, "router_trainable": false, "sec_per_step": 0.148, "alpha": 0.0, "t": 963.3} {"event": "train", "step": 4950, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.25238, "bpb": 1.8068, "chunk_loss": 0.0, "F": 0.20142, "G": 0.42215, "chunks": 206.25, "bytes_per_chunk": 4.965, "boundary_entropy": 0.6585, "router_noise": 0.0, "grad_norm": 0.754, "lr": 5.2e-05, "router_trainable": false, "sec_per_step": 0.16, "alpha": 0.0, "t": 971.3} {"event": "probe", "step": 5000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4765.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.273803, "cps": 37.226562, "mean_p": 0.433452, "boundary_entropy": 0.65321, "bpb": 2.021936}, "de": {"chunks": 5687.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.276969, "cps": 44.429688, "mean_p": 0.43993, "boundary_entropy": 0.662577, "bpb": 2.223269}, "vi": {"chunks": 5481.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.297315, "cps": 42.820312, "mean_p": 0.427016, "boundary_entropy": 0.65934, "bpb": 1.904299}, "tr": {"chunks": 5949.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.32862, "cps": 46.476562, "mean_p": 0.456251, "boundary_entropy": 0.664437, "bpb": 2.170676}, "fi": {"chunks": 5027.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.267237, "cps": 39.273438, "mean_p": 0.448285, "boundary_entropy": 0.667797, "bpb": 2.373983}, "ru": {"chunks": 5895.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.311082, "cps": 46.054688, "mean_p": 0.411353, "boundary_entropy": 0.651584, "bpb": 1.414798}, "el": {"chunks": 5199.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.253326, "cps": 40.617188, "mean_p": 0.42998, "boundary_entropy": 0.669456, "bpb": 1.470725}, "hi": {"chunks": 4879.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.289985, "cps": 38.117188, "mean_p": 0.415397, "boundary_entropy": 0.659693, "bpb": 1.180346}, "ta": {"chunks": 4382.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.234394, "cps": 34.234375, "mean_p": 0.407263, "boundary_entropy": 0.65634, "bpb": 0.975889}, "th": {"chunks": 5894.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.366428, "cps": 46.046875, "mean_p": 0.403486, "boundary_entropy": 0.6506, "bpb": 1.100056}, "zh": {"chunks": 3224.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.545886, "cps": 25.1875, "mean_p": 0.462854, "boundary_entropy": 0.678425, "bpb": 3.137969}, "my": {"chunks": 3690.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.194744, "cps": 28.828125, "mean_p": 0.442304, "boundary_entropy": 0.679511, "bpb": 1.09283}}, "bpb": {"en": 2.02194, "de": 2.22327, "vi": 1.9043, "tr": 2.17068, "fi": 2.37398, "ru": 1.4148, "el": 1.47073, "hi": 1.18035, "ta": 0.97589, "th": 1.10006, "zh": 3.13797, "my": 1.09283}, "gini_cpc": 0.134639, "gini_cps": 0.092933, "bpic": 6.388684, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0115609050289327, "vi": 1.0858701014397598, "tr": 1.200202792009974, "fi": 0.9760188762318581, "ru": 1.1361503597859282, "el": 0.9252097217060337, "hi": 1.0590999814459383, "ta": 0.8560677153859084, "th": 1.3382901499477624, "zh": 1.993713757578062, "my": 0.7112531541008793}, "t": 982.4} {"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 2.03042, "de": 2.23136, "vi": 1.90451, "tr": 2.1772, "fi": 2.38625, "ru": 1.3929, "el": 1.46441, "hi": 1.17674, "ta": 0.98385, "th": 1.1159, "zh": 3.13378, "my": 1.10102}, "bytes_per_chunk": {"en": 3.591, "de": 3.667, "vi": 4.255, "tr": 3.267, "fi": 3.782, "ru": 5.839, "el": 7.086, "hi": 9.233, "ta": 10.939, "th": 8.042, "zh": 4.747, "my": 14.547}, "bpb_mean": 1.7582, "frozen_inventory": false, "t": 987.3} {"event": "train", "step": 5000, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.36434, "bpb": 1.96833, "chunk_loss": 0.0, "F": 0.20389, "G": 0.43123, "chunks": 208.78, "bytes_per_chunk": 4.905, "boundary_entropy": 0.6651, "router_noise": 0.0, "grad_norm": 0.7292, "lr": 5e-05, "router_trainable": false, "sec_per_step": 0.329, "alpha": 0.0, "t": 987.7} {"event": "train", "step": 5050, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.37623, "bpb": 1.98549, "chunk_loss": 0.0, "F": 0.21231, "G": 0.42741, "chunks": 217.41, "bytes_per_chunk": 4.71, "boundary_entropy": 0.66116, "router_noise": 0.0, "grad_norm": 0.8407, "lr": 4.81e-05, "router_trainable": false, "sec_per_step": 0.15, "alpha": 0.0, "t": 995.2} {"event": "train", "step": 5100, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.47517, "bpb": 2.12822, "chunk_loss": 0.0, "F": 0.23608, "G": 0.43636, "chunks": 241.75, "bytes_per_chunk": 4.236, "boundary_entropy": 0.66622, "router_noise": 0.0, "grad_norm": 0.7328, "lr": 4.63e-05, "router_trainable": false, "sec_per_step": 0.162, "alpha": 0.0, "t": 1003.4} {"event": "train", "step": 5150, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.32735, "bpb": 1.91497, "chunk_loss": 0.0, "F": 0.23404, "G": 0.43147, "chunks": 239.66, "bytes_per_chunk": 4.273, "boundary_entropy": 0.66183, "router_noise": 0.0, "grad_norm": 0.7228, "lr": 4.45e-05, "router_trainable": false, "sec_per_step": 0.163, "alpha": 0.0, "t": 1011.5} {"event": "probe", "step": 5200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4826.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.277309, "cps": 37.703125, "mean_p": 0.432932, "boundary_entropy": 0.652433, "bpb": 2.014715}, "de": {"chunks": 5712.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.278186, "cps": 44.625, "mean_p": 0.439945, "boundary_entropy": 0.662114, "bpb": 2.216592}, "vi": {"chunks": 5705.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.309466, "cps": 44.570312, "mean_p": 0.42754, "boundary_entropy": 0.659094, "bpb": 1.897918}, "tr": {"chunks": 6030.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.333094, "cps": 47.109375, "mean_p": 0.45782, "boundary_entropy": 0.664315, "bpb": 2.158545}, "fi": {"chunks": 5330.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.283345, "cps": 41.640625, "mean_p": 0.449894, "boundary_entropy": 0.667635, "bpb": 2.360697}, "ru": {"chunks": 6095.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.321636, "cps": 47.617188, "mean_p": 0.411136, "boundary_entropy": 0.650428, "bpb": 1.406053}, "el": {"chunks": 5408.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.263509, "cps": 42.25, "mean_p": 0.430463, "boundary_entropy": 0.669665, "bpb": 1.464286}, "hi": {"chunks": 5019.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.298306, "cps": 39.210938, "mean_p": 0.414812, "boundary_entropy": 0.658884, "bpb": 1.178818}, "ta": {"chunks": 4022.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.215138, "cps": 31.421875, "mean_p": 0.405005, "boundary_entropy": 0.656004, "bpb": 0.9724}, "th": {"chunks": 4655.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.2894, "cps": 36.367188, "mean_p": 0.401616, "boundary_entropy": 0.650205, "bpb": 1.090871}, "zh": {"chunks": 3294.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.557738, "cps": 25.734375, "mean_p": 0.463269, "boundary_entropy": 0.678568, "bpb": 3.130825}, "my": {"chunks": 3918.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.206776, "cps": 30.609375, "mean_p": 0.4416, "boundary_entropy": 0.679178, "bpb": 1.088218}}, "bpb": {"en": 2.01472, "de": 2.21659, "vi": 1.89792, "tr": 2.15855, "fi": 2.3607, "ru": 1.40605, "el": 1.46429, "hi": 1.17882, "ta": 0.9724, "th": 1.09087, "zh": 3.13083, "my": 1.08822}, "gini_cpc": 0.127417, "gini_cps": 0.09575, "bpic": 6.394858, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.003165514706507, "vi": 1.1159617504451915, "tr": 1.2011674733467301, "fi": 1.0217676024047353, "ru": 1.1598485883959686, "el": 0.9502385394418348, "hi": 1.0757192128570539, "ta": 0.7758064739591987, "th": 1.043603249469477, "zh": 2.011254157375622, "my": 0.7456548692471405}, "t": 1022.4} {"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 2.02291, "de": 2.22312, "vi": 1.89706, "tr": 2.16691, "fi": 2.37194, "ru": 1.38419, "el": 1.45751, "hi": 1.17538, "ta": 0.98053, "th": 1.10731, "zh": 3.12723, "my": 1.09647}, "bytes_per_chunk": {"en": 3.531, "de": 3.644, "vi": 4.151, "tr": 3.234, "fi": 3.531, "ru": 5.637, "el": 6.8, "hi": 8.864, "ta": 12.087, "th": 10.005, "zh": 4.656, "my": 13.845}, "bpb_mean": 1.75088, "frozen_inventory": false, "t": 1026.7} {"event": "train", "step": 5200, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.24826, "bpb": 1.80085, "chunk_loss": 0.0, "F": 0.20633, "G": 0.42871, "chunks": 211.28, "bytes_per_chunk": 4.847, "boundary_entropy": 0.66255, "router_noise": 0.0, "grad_norm": 0.7459, "lr": 4.29e-05, "router_trainable": false, "sec_per_step": 0.311, "alpha": 0.0, "t": 1027.1} {"event": "train", "step": 5250, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.28935, "bpb": 1.86014, "chunk_loss": 0.0, "F": 0.21408, "G": 0.42874, "chunks": 219.22, "bytes_per_chunk": 4.671, "boundary_entropy": 0.66228, "router_noise": 0.0, "grad_norm": 0.706, "lr": 4.14e-05, "router_trainable": false, "sec_per_step": 0.123, "alpha": 0.0, "t": 1033.2} {"event": "train", "step": 5300, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.22444, "bpb": 1.76649, "chunk_loss": 0.0, "F": 0.21436, "G": 0.42566, "chunks": 219.5, "bytes_per_chunk": 4.665, "boundary_entropy": 0.66043, "router_noise": 0.0, "grad_norm": 0.7014, "lr": 3.99e-05, "router_trainable": false, "sec_per_step": 0.123, "alpha": 0.0, "t": 1039.4} {"event": "train", "step": 5350, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.44692, "bpb": 2.08747, "chunk_loss": 0.0, "F": 0.21396, "G": 0.43702, "chunks": 219.09, "bytes_per_chunk": 4.674, "boundary_entropy": 0.66813, "router_noise": 0.0, "grad_norm": 0.7789, "lr": 3.86e-05, "router_trainable": false, "sec_per_step": 0.132, "alpha": 0.0, "t": 1046.0} {"event": "probe", "step": 5400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4788.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.275125, "cps": 37.40625, "mean_p": 0.43412, "boundary_entropy": 0.653346, "bpb": 2.012357}, "de": {"chunks": 5734.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.279258, "cps": 44.796875, "mean_p": 0.441315, "boundary_entropy": 0.662814, "bpb": 2.208951}, "vi": {"chunks": 5338.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.289558, "cps": 41.703125, "mean_p": 0.427875, "boundary_entropy": 0.660033, "bpb": 1.895162}, "tr": {"chunks": 6070.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.335304, "cps": 47.421875, "mean_p": 0.458569, "boundary_entropy": 0.664574, "bpb": 2.150263}, "fi": {"chunks": 5267.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.279996, "cps": 41.148438, "mean_p": 0.450369, "boundary_entropy": 0.668284, "bpb": 2.351413}, "ru": {"chunks": 5928.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.312823, "cps": 46.3125, "mean_p": 0.410816, "boundary_entropy": 0.650516, "bpb": 1.398031}, "el": {"chunks": 5540.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.269941, "cps": 43.28125, "mean_p": 0.430398, "boundary_entropy": 0.66921, "bpb": 1.455884}, "hi": {"chunks": 4234.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.251649, "cps": 33.078125, "mean_p": 0.412607, "boundary_entropy": 0.658778, "bpb": 1.171869}, "ta": {"chunks": 4344.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.232362, "cps": 33.9375, "mean_p": 0.40531, "boundary_entropy": 0.655815, "bpb": 0.966689}, "th": {"chunks": 4160.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.258626, "cps": 32.5, "mean_p": 0.402793, "boundary_entropy": 0.652179, "bpb": 1.086624}, "zh": {"chunks": 3403.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.576194, "cps": 26.585938, "mean_p": 0.464794, "boundary_entropy": 0.678971, "bpb": 3.12026}, "my": {"chunks": 4418.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.233164, "cps": 34.515625, "mean_p": 0.44151, "boundary_entropy": 0.67876, "bpb": 1.086242}}, "bpb": {"en": 2.01236, "de": 2.20895, "vi": 1.89516, "tr": 2.15026, "fi": 2.35141, "ru": 1.39803, "el": 1.45588, "hi": 1.17187, "ta": 0.96669, "th": 1.08662, "zh": 3.12026, "my": 1.08624}, "gini_cpc": 0.123321, "gini_cps": 0.091416, "bpic": 6.48016, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0150215430457505, "vi": 1.0524595323404797, "tr": 1.2187317296168112, "fi": 1.0177038404314438, "ru": 1.1370222389747457, "el": 0.9811578842345351, "hi": 0.9146727889560455, "ta": 0.8445674228363099, "th": 0.9400311471872096, "zh": 2.094298033434416, "my": 0.8474855732322151}, "t": 1054.9} {"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 2.02051, "de": 2.21597, "vi": 1.89406, "tr": 2.15799, "fi": 2.36301, "ru": 1.37529, "el": 1.44983, "hi": 1.16767, "ta": 0.97478, "th": 1.10347, "zh": 3.11662, "my": 1.09453}, "bytes_per_chunk": {"en": 3.538, "de": 3.624, "vi": 4.4, "tr": 3.185, "fi": 3.595, "ru": 5.779, "el": 6.674, "hi": 10.586, "ta": 11.091, "th": 11.291, "zh": 4.422, "my": 12.447}, "bpb_mean": 1.74448, "frozen_inventory": false, "t": 1058.3} {"event": "train", "step": 5400, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.34585, "bpb": 1.94165, "chunk_loss": 0.0, "F": 0.22519, "G": 0.43079, "chunks": 230.59, "bytes_per_chunk": 4.441, "boundary_entropy": 0.66312, "router_noise": 0.0, "grad_norm": 0.8131, "lr": 3.73e-05, "router_trainable": false, "sec_per_step": 0.255, "alpha": 0.0, "t": 1058.7} {"event": "train", "step": 5450, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.20477, "bpb": 1.73812, "chunk_loss": 0.0, "F": 0.20584, "G": 0.42704, "chunks": 210.78, "bytes_per_chunk": 4.858, "boundary_entropy": 0.66179, "router_noise": 0.0, "grad_norm": 0.7281, "lr": 3.62e-05, "router_trainable": false, "sec_per_step": 0.133, "alpha": 0.0, "t": 1065.4} {"event": "train", "step": 5500, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.31094, "bpb": 1.89129, "chunk_loss": 0.0, "F": 0.22977, "G": 0.43177, "chunks": 235.28, "bytes_per_chunk": 4.352, "boundary_entropy": 0.66266, "router_noise": 0.0, "grad_norm": 0.7787, "lr": 3.51e-05, "router_trainable": false, "sec_per_step": 0.128, "alpha": 0.0, "t": 1071.8} {"event": "train", "step": 5550, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.37949, "bpb": 1.99018, "chunk_loss": 0.0, "F": 0.22516, "G": 0.43453, "chunks": 230.56, "bytes_per_chunk": 4.441, "boundary_entropy": 0.66525, "router_noise": 0.0, "grad_norm": 0.8013, "lr": 3.41e-05, "router_trainable": false, "sec_per_step": 0.131, "alpha": 0.0, "t": 1078.3} {"event": "probe", "step": 5600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4858.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.279147, "cps": 37.953125, "mean_p": 0.43447, "boundary_entropy": 0.653132, "bpb": 2.005094}, "de": {"chunks": 5748.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.27994, "cps": 44.90625, "mean_p": 0.440835, "boundary_entropy": 0.662358, "bpb": 2.202516}, "vi": {"chunks": 5512.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.298996, "cps": 43.0625, "mean_p": 0.427762, "boundary_entropy": 0.659551, "bpb": 1.890846}, "tr": {"chunks": 6121.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.338121, "cps": 47.820312, "mean_p": 0.458711, "boundary_entropy": 0.664213, "bpb": 2.141544}, "fi": {"chunks": 5218.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.277391, "cps": 40.765625, "mean_p": 0.449961, "boundary_entropy": 0.667807, "bpb": 2.347726}, "ru": {"chunks": 5863.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.309393, "cps": 45.804688, "mean_p": 0.410867, "boundary_entropy": 0.650751, "bpb": 1.393205}, "el": {"chunks": 5586.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.272182, "cps": 43.640625, "mean_p": 0.431802, "boundary_entropy": 0.669897, "bpb": 1.449337}, "hi": {"chunks": 4384.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.260565, "cps": 34.25, "mean_p": 0.414891, "boundary_entropy": 0.660004, "bpb": 1.168245}, "ta": {"chunks": 3977.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.212731, "cps": 31.070312, "mean_p": 0.405873, "boundary_entropy": 0.657278, "bpb": 0.965399}, "th": {"chunks": 5624.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.349643, "cps": 43.9375, "mean_p": 0.405534, "boundary_entropy": 0.652379, "bpb": 1.082605}, "zh": {"chunks": 3206.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.542838, "cps": 25.046875, "mean_p": 0.463364, "boundary_entropy": 0.678668, "bpb": 3.118554}, "my": {"chunks": 4376.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.230948, "cps": 34.1875, "mean_p": 0.443602, "boundary_entropy": 0.679476, "bpb": 1.082149}}, "bpb": {"en": 2.00509, "de": 2.20252, "vi": 1.89085, "tr": 2.14154, "fi": 2.34773, "ru": 1.39321, "el": 1.44934, "hi": 1.16824, "ta": 0.9654, "th": 1.0826, "zh": 3.11855, "my": 1.08215}, "gini_cpc": 0.125368, "gini_cps": 0.09277, "bpic": 6.34632, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0028384155103063, "vi": 1.0711065538762197, "tr": 1.2112629626540976, "fi": 0.9937080340180067, "ru": 1.1083509289141433, "el": 0.9750495718516977, "hi": 0.9334307039077713, "ta": 0.7620732741387912, "th": 1.2525378441234318, "zh": 1.9446286734244762, "my": 0.8273333800908435}, "t": 1086.0} {"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 2.01342, "de": 2.20994, "vi": 1.88978, "tr": 2.14886, "fi": 2.3585, "ru": 1.37142, "el": 1.44341, "hi": 1.16489, "ta": 0.97401, "th": 1.09909, "zh": 3.11327, "my": 1.09006}, "bytes_per_chunk": {"en": 3.49, "de": 3.616, "vi": 4.267, "tr": 3.179, "fi": 3.63, "ru": 5.841, "el": 6.536, "hi": 10.087, "ta": 11.966, "th": 8.331, "zh": 4.76, "my": 12.583}, "bpb_mean": 1.73972, "frozen_inventory": false, "t": 1088.0} {"event": "train", "step": 5600, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.33011, "bpb": 1.91895, "chunk_loss": 0.0, "F": 0.22348, "G": 0.4309, "chunks": 228.84, "bytes_per_chunk": 4.475, "boundary_entropy": 0.66197, "router_noise": 0.0, "grad_norm": 0.7599, "lr": 3.33e-05, "router_trainable": false, "sec_per_step": 0.198, "alpha": 0.0, "t": 1088.3} {"event": "train", "step": 5650, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.30839, "bpb": 1.88761, "chunk_loss": 0.0, "F": 0.22009, "G": 0.43026, "chunks": 225.38, "bytes_per_chunk": 4.544, "boundary_entropy": 0.6628, "router_noise": 0.0, "grad_norm": 0.7969, "lr": 3.25e-05, "router_trainable": false, "sec_per_step": 0.083, "alpha": 0.0, "t": 1092.4} {"event": "train", "step": 5700, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.28169, "bpb": 1.84909, "chunk_loss": 0.0, "F": 0.24176, "G": 0.42879, "chunks": 247.56, "bytes_per_chunk": 4.136, "boundary_entropy": 0.65965, "router_noise": 0.0, "grad_norm": 0.7417, "lr": 3.18e-05, "router_trainable": false, "sec_per_step": 0.094, "alpha": 0.0, "t": 1097.1} {"event": "train", "step": 5750, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.19631, "bpb": 1.72591, "chunk_loss": 0.0, "F": 0.21521, "G": 0.42845, "chunks": 220.38, "bytes_per_chunk": 4.647, "boundary_entropy": 0.66167, "router_noise": 0.0, "grad_norm": 0.7423, "lr": 3.13e-05, "router_trainable": false, "sec_per_step": 0.092, "alpha": 0.0, "t": 1101.7} {"event": "probe", "step": 5800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4907.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.281963, "cps": 38.335938, "mean_p": 0.435174, "boundary_entropy": 0.653475, "bpb": 1.998621}, "de": {"chunks": 5800.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.282472, "cps": 45.3125, "mean_p": 0.442148, "boundary_entropy": 0.663002, "bpb": 2.196676}, "vi": {"chunks": 5413.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.293626, "cps": 42.289062, "mean_p": 0.428009, "boundary_entropy": 0.660147, "bpb": 1.885548}, "tr": {"chunks": 6143.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.339336, "cps": 47.992188, "mean_p": 0.458995, "boundary_entropy": 0.664639, "bpb": 2.137769}, "fi": {"chunks": 5341.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.28393, "cps": 41.726562, "mean_p": 0.44976, "boundary_entropy": 0.667807, "bpb": 2.336056}, "ru": {"chunks": 5771.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.304538, "cps": 45.085938, "mean_p": 0.41036, "boundary_entropy": 0.650073, "bpb": 1.390708}, "el": {"chunks": 5316.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.259026, "cps": 41.53125, "mean_p": 0.430139, "boundary_entropy": 0.669465, "bpb": 1.445996}, "hi": {"chunks": 4589.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.272749, "cps": 35.851562, "mean_p": 0.414575, "boundary_entropy": 0.659718, "bpb": 1.165011}, "ta": {"chunks": 4194.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.224338, "cps": 32.765625, "mean_p": 0.406562, "boundary_entropy": 0.657376, "bpb": 0.962343}, "th": {"chunks": 5520.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.343177, "cps": 43.125, "mean_p": 0.405256, "boundary_entropy": 0.652539, "bpb": 1.080849}, "zh": {"chunks": 3280.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.555367, "cps": 25.625, "mean_p": 0.46426, "boundary_entropy": 0.678846, "bpb": 3.110109}, "my": {"chunks": 5680.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.299768, "cps": 44.375, "mean_p": 0.444558, "boundary_entropy": 0.679145, "bpb": 1.079223}}, "bpb": {"en": 1.99862, "de": 2.19668, "vi": 1.88555, "tr": 2.13777, "fi": 2.33606, "ru": 1.39071, "el": 1.446, "hi": 1.16501, "ta": 0.96234, "th": 1.08085, "zh": 3.11011, "my": 1.07922}, "gini_cpc": 0.112274, "gini_cps": 0.079051, "bpic": 6.194612, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0018060465643783, "vi": 1.041364928765353, "tr": 1.203477641092784, "fi": 1.0069751581748956, "ru": 1.0800650731575645, "el": 0.9186544631475664, "hi": 0.9673219596228332, "ta": 0.7956297592563212, "th": 1.2170994739012275, "zh": 1.9696472919844497, "my": 1.0631462749896654}, "t": 1107.7} {"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 2.00774, "de": 2.20481, "vi": 1.88516, "tr": 2.14434, "fi": 2.34739, "ru": 1.3684, "el": 1.43938, "hi": 1.16053, "ta": 0.97102, "th": 1.09671, "zh": 3.10541, "my": 1.08715}, "bytes_per_chunk": {"en": 3.45, "de": 3.572, "vi": 4.31, "tr": 3.162, "fi": 3.559, "ru": 5.912, "el": 6.782, "hi": 9.572, "ta": 11.472, "th": 8.599, "zh": 4.686, "my": 9.896}, "bpb_mean": 1.73484, "frozen_inventory": false, "t": 1109.7} {"event": "train", "step": 5800, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.2833, "bpb": 1.8514, "chunk_loss": 0.0, "F": 0.22296, "G": 0.42917, "chunks": 228.31, "bytes_per_chunk": 4.485, "boundary_entropy": 0.66157, "router_noise": 0.0, "grad_norm": 0.7724, "lr": 3.08e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1110.1} {"event": "train", "step": 5850, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.27429, "bpb": 1.83841, "chunk_loss": 0.0, "F": 0.21793, "G": 0.43262, "chunks": 223.16, "bytes_per_chunk": 4.589, "boundary_entropy": 0.66413, "router_noise": 0.0, "grad_norm": 0.7488, "lr": 3.05e-05, "router_trainable": false, "sec_per_step": 0.089, "alpha": 0.0, "t": 1114.5} {"event": "train", "step": 5900, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.34697, "bpb": 1.94326, "chunk_loss": 0.0, "F": 0.22137, "G": 0.43444, "chunks": 226.69, "bytes_per_chunk": 4.517, "boundary_entropy": 0.66503, "router_noise": 0.0, "grad_norm": 0.8003, "lr": 3.02e-05, "router_trainable": false, "sec_per_step": 0.097, "alpha": 0.0, "t": 1119.4} {"event": "train", "step": 5950, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.39454, "bpb": 2.01189, "chunk_loss": 0.0, "F": 0.23083, "G": 0.43678, "chunks": 236.38, "bytes_per_chunk": 4.332, "boundary_entropy": 0.66552, "router_noise": 0.0, "grad_norm": 0.7676, "lr": 3.01e-05, "router_trainable": false, "sec_per_step": 0.091, "alpha": 0.0, "t": 1123.9} {"event": "train", "step": 5999, "arch": "hnet", "phase": 3, "objective": "none", "lm_loss": 1.24531, "bpb": 1.7966, "chunk_loss": 0.0, "F": 0.21881, "G": 0.43048, "chunks": 224.06, "bytes_per_chunk": 4.57, "boundary_entropy": 0.66284, "router_noise": 0.0, "grad_norm": 0.7627, "lr": 3e-05, "router_trainable": false, "sec_per_step": 0.095, "alpha": 0.0, "t": 1128.7} {"event": "probe", "step": 6000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 4901.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.281618, "cps": 38.289062, "mean_p": 0.435568, "boundary_entropy": 0.653855, "bpb": 1.995904}, "de": {"chunks": 5810.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.282959, "cps": 45.390625, "mean_p": 0.441346, "boundary_entropy": 0.662686, "bpb": 2.195214}, "vi": {"chunks": 5282.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.28652, "cps": 41.265625, "mean_p": 0.427889, "boundary_entropy": 0.660036, "bpb": 1.883723}, "tr": {"chunks": 6071.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.335359, "cps": 47.429688, "mean_p": 0.458042, "boundary_entropy": 0.664234, "bpb": 2.134688}, "fi": {"chunks": 5306.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.282069, "cps": 41.453125, "mean_p": 0.45126, "boundary_entropy": 0.668375, "bpb": 2.334769}, "ru": {"chunks": 5860.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.309235, "cps": 45.78125, "mean_p": 0.412499, "boundary_entropy": 0.651727, "bpb": 1.388367}, "el": {"chunks": 5119.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.249427, "cps": 39.992188, "mean_p": 0.430534, "boundary_entropy": 0.669875, "bpb": 1.442486}, "hi": {"chunks": 5500.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.326895, "cps": 42.96875, "mean_p": 0.4195, "boundary_entropy": 0.661648, "bpb": 1.163224}, "ta": {"chunks": 4250.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.227334, "cps": 33.203125, "mean_p": 0.409597, "boundary_entropy": 0.658695, "bpb": 0.960352}, "th": {"chunks": 5742.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.356979, "cps": 44.859375, "mean_p": 0.40536, "boundary_entropy": 0.652336, "bpb": 1.079212}, "zh": {"chunks": 3307.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.559939, "cps": 25.835938, "mean_p": 0.464913, "boundary_entropy": 0.678986, "bpb": 3.104764}, "my": {"chunks": 3902.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.205932, "cps": 30.484375, "mean_p": 0.444494, "boundary_entropy": 0.679977, "bpb": 1.077151}}, "bpb": {"en": 1.9959, "de": 2.19521, "vi": 1.88372, "tr": 2.13469, "fi": 2.33477, "ru": 1.38837, "el": 1.44249, "hi": 1.16322, "ta": 0.96035, "th": 1.07921, "zh": 3.10476, "my": 1.07715}, "gini_cpc": 0.134391, "gini_cps": 0.087794, "bpic": 6.286339, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.004761863924852, "vi": 1.0174068857935537, "tr": 1.1908281684923774, "fi": 1.0016010701757105, "ru": 1.098064419570827, "el": 0.8856939988888605, "hi": 1.1607722959168052, "ta": 0.80724038012012, "th": 1.2675979862895528, "zh": 1.9882920222021492, "my": 0.7312456686729092}, "t": 1130.8} {"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 2.00466, "de": 2.20313, "vi": 1.88281, "tr": 2.14183, "fi": 2.34728, "ru": 1.36565, "el": 1.43639, "hi": 1.15878, "ta": 0.96952, "th": 1.0945, "zh": 3.09989, "my": 1.08503}, "bytes_per_chunk": {"en": 3.476, "de": 3.597, "vi": 4.424, "tr": 3.209, "fi": 3.564, "ru": 5.814, "el": 7.127, "hi": 8.107, "ta": 11.286, "th": 8.251, "zh": 4.602, "my": 13.582}, "bpb_mean": 1.73246, "frozen_inventory": false, "t": 1133.3} {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s2/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s2/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s2/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s2/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/parity_B_ste_s2/probe/reps/th0.70"], "t": 1139.7} {"event": "final", "run": "parity_B_ste_s2", "arch": "hnet", "objective": "parity", "condition": "B", "step": 6000, "params": 22516992, "seed": 2, "per_lang": {"en": {"cpc": 0.2816181118198012, "cps": 38.2890625, "bpb": 2.0135295581715775, "bytes_per_chunk": 3.4952999373324976}, "de": {"cpc": 0.2829591389470608, "cps": 45.390625, "bpb": 2.1950287400328916, "bytes_per_chunk": 3.5935981356454985}, "vi": {"cpc": 0.2865202061296447, "cps": 41.265625, "bpb": 1.8864606495338812, "bytes_per_chunk": 4.391095328229953}, "tr": {"cpc": 0.33535878031265537, "cps": 47.4296875, "bpb": 2.1344755189754245, "bytes_per_chunk": 3.20739988498856}, "fi": {"cpc": 0.28206900217957576, "cps": 41.453125, "bpb": 2.3628800021588425, "bytes_per_chunk": 3.546536247471775}, "ru": {"cpc": 0.3092348284960422, "cps": 45.78125, "bpb": 1.3645900121596202, "bytes_per_chunk": 5.797273239934983}, "el": {"cpc": 0.24942747161720996, "cps": 39.9921875, "bpb": 1.431534441739535, "bytes_per_chunk": 7.108122398622541}, "hi": {"cpc": 0.32689450222882616, "cps": 42.96875, "bpb": 1.1660458178773176, "bytes_per_chunk": 8.080140554202755}, "ta": {"cpc": 0.22733351163412677, "cps": 33.203125, "bpb": 0.9700471441848934, "bytes_per_chunk": 11.362735961509287}, "th": {"cpc": 0.35697855144544605, "cps": 44.859375, "bpb": 1.0921100516892268, "bytes_per_chunk": 8.269526813880127}, "zh": {"cpc": 0.5599390450389434, "cps": 25.8359375, "bpb": 3.106413392053783, "bytes_per_chunk": 4.593698524515474}, "my": {"cpc": 0.20593202448807263, "cps": 30.484375, "bpb": 1.0940867342869187, "bytes_per_chunk": 13.50631150497192}}, "bpb_mean": 1.734766838571992, "gini_cpc": 0.1343910771829675, "gini_cps": 0.08779415779415789, "bpic": 6.286339066339067, "mean_boundary_entropy": 0.662835955619812, "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": 1143.6260423660278, "t": 1143.6}