| {"event": "start", "run": "exp32_local_s1", "condition": "A", "params": 82470528, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.184, "vi": 1.39, "tr": 1.12, "fi": 1.104, "ru": 1.971, "el": 2.174, "hi": 2.552, "ta": 3.192, "th": 2.762, "zh": 0.915, "my": 3.524}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "global", "p3_objective": "global", "freeze_in_p3": false, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 2.1} |
| {"event": "probe_schedule", "n": 21, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 11000], "t": 2.1} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14139.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.812446, "cps": 110.460938, "mean_p": 0.520947, "boundary_entropy": 0.686709, "bpb": 8.053723}, "de": {"chunks": 18668.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.909171, "cps": 145.84375, "mean_p": 0.525359, "boundary_entropy": 0.687245, "bpb": 8.19356}, "vi": {"chunks": 21341.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.157635, "cps": 166.726562, "mean_p": 0.523718, "boundary_entropy": 0.68791, "bpb": 8.0595}, "tr": {"chunks": 17844.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.985693, "cps": 139.40625, "mean_p": 0.529758, "boundary_entropy": 0.686399, "bpb": 8.194766}, "fi": {"chunks": 17557.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.933337, "cps": 137.164062, "mean_p": 0.526424, "boundary_entropy": 0.686837, "bpb": 8.037401}, "ru": {"chunks": 34128.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.80095, "cps": 266.625, "mean_p": 0.540375, "boundary_entropy": 0.687168, "bpb": 8.258426}, "el": {"chunks": 36914.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.798665, "cps": 288.390625, "mean_p": 0.53644, "boundary_entropy": 0.688054, "bpb": 8.180766}, "hi": {"chunks": 33657.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.000416, "cps": 262.945312, "mean_p": 0.509672, "boundary_entropy": 0.690736, "bpb": 8.396576}, "ta": {"chunks": 50614.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.707355, "cps": 395.421875, "mean_p": 0.521149, "boundary_entropy": 0.690436, "bpb": 8.104079}, "th": {"chunks": 45393.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.82207, "cps": 354.632812, "mean_p": 0.531267, "boundary_entropy": 0.689124, "bpb": 8.02752}, "zh": {"chunks": 4875.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.825432, "cps": 38.085938, "mean_p": 0.492214, "boundary_entropy": 0.686444, "bpb": 8.097877}, "my": {"chunks": 30094.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.588242, "cps": 235.109375, "mean_p": 0.506988, "boundary_entropy": 0.690851, "bpb": 8.217318}}, "bpb": {"en": 8.05372, "de": 8.19356, "vi": 8.0595, "tr": 8.19477, "fi": 8.0374, "ru": 8.25843, "el": 8.18077, "hi": 8.39658, "ta": 8.10408, "th": 8.02752, "zh": 8.09788, "my": 8.21732}, "gini_cpc": 0.244844, "gini_cps": 0.27034, "bpic": 1.180051, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1190533992007132, "vi": 1.4248759281827625, "tr": 1.2132410293977671, "fi": 1.1487984777338243, "ru": 2.216700654508078, "el": 2.2138882150706616, "hi": 2.462213768688232, "ta": 3.332350057232739, "th": 3.473547534610352, "zh": 1.0159833788983057, "my": 1.9548883851595404}, "t": 10.2} |
| {"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.63249, "bpb": 8.12596, "chunk_loss": 0.084119, "F": 0.86429, "G": 0.52219, "chunks": 885.03, "bytes_per_chunk": 1.157, "boundary_entropy": 0.69071, "router_noise": 0.0, "grad_norm": 9.604, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.005, "alpha": 0.03, "t": 11.0} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14074.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.808711, "cps": 109.953125, "mean_p": 0.520666, "boundary_entropy": 0.68673, "bpb": 8.017048}, "de": {"chunks": 18634.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.907515, "cps": 145.578125, "mean_p": 0.525093, "boundary_entropy": 0.68727, "bpb": 8.160104}, "vi": {"chunks": 21323.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.156659, "cps": 166.585938, "mean_p": 0.523605, "boundary_entropy": 0.687919, "bpb": 8.044163}, "tr": {"chunks": 17807.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.983649, "cps": 139.117188, "mean_p": 0.529553, "boundary_entropy": 0.686421, "bpb": 8.169354}, "fi": {"chunks": 17517.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.93121, "cps": 136.851562, "mean_p": 0.52614, "boundary_entropy": 0.686866, "bpb": 8.006126}, "ru": {"chunks": 34126.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.800844, "cps": 266.609375, "mean_p": 0.540288, "boundary_entropy": 0.687181, "bpb": 8.224495}, "el": {"chunks": 36912.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.798567, "cps": 288.375, "mean_p": 0.536349, "boundary_entropy": 0.688067, "bpb": 8.168318}, "hi": {"chunks": 33524.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.992511, "cps": 261.90625, "mean_p": 0.509549, "boundary_entropy": 0.69074, "bpb": 8.349052}, "ta": {"chunks": 50597.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.706446, "cps": 395.289062, "mean_p": 0.520975, "boundary_entropy": 0.69045, "bpb": 8.070066}, "th": {"chunks": 45385.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.821573, "cps": 354.570312, "mean_p": 0.531156, "boundary_entropy": 0.689138, "bpb": 7.982322}, "zh": {"chunks": 4858.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.822553, "cps": 37.953125, "mean_p": 0.492169, "boundary_entropy": 0.686442, "bpb": 8.092523}, "my": {"chunks": 30055.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.586183, "cps": 234.804688, "mean_p": 0.506977, "boundary_entropy": 0.690852, "bpb": 8.203248}}, "bpb": {"en": 8.01705, "de": 8.1601, "vi": 8.04416, "tr": 8.16935, "fi": 8.00613, "ru": 8.22449, "el": 8.16832, "hi": 8.34905, "ta": 8.07007, "th": 7.98232, "zh": 8.09252, "my": 8.20325}, "gini_cpc": 0.24529, "gini_cps": 0.270846, "bpic": 1.181548, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1221741429333723, "vi": 1.430249282156515, "tr": 1.2163170112795654, "fi": 1.1514747527039255, "ru": 2.2268078602996675, "el": 2.223992434418055, "hi": 2.4638106750964597, "ta": 3.346615910103867, "th": 3.488974923567772, "zh": 1.0171163634566476, "my": 1.961371813043208}, "t": 18.7} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 13930.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.800437, "cps": 108.828125, "mean_p": 0.520056, "boundary_entropy": 0.686775, "bpb": 7.943438}, "de": {"chunks": 18539.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.902888, "cps": 144.835938, "mean_p": 0.5245, "boundary_entropy": 0.687325, "bpb": 8.090808}, "vi": {"chunks": 21261.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.153295, "cps": 166.101562, "mean_p": 0.523354, "boundary_entropy": 0.687941, "bpb": 8.011837}, "tr": {"chunks": 17750.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.9805, "cps": 138.671875, "mean_p": 0.529108, "boundary_entropy": 0.686469, "bpb": 8.116938}, "fi": {"chunks": 17433.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.926745, "cps": 136.195312, "mean_p": 0.525569, "boundary_entropy": 0.686922, "bpb": 7.944074}, "ru": {"chunks": 34116.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.800317, "cps": 266.53125, "mean_p": 0.5401, "boundary_entropy": 0.68721, "bpb": 8.156733}, "el": {"chunks": 36885.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.797252, "cps": 288.164062, "mean_p": 0.536141, "boundary_entropy": 0.688095, "bpb": 8.137098}, "hi": {"chunks": 33262.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.976939, "cps": 259.859375, "mean_p": 0.509337, "boundary_entropy": 0.690746, "bpb": 8.264668}, "ta": {"chunks": 50567.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.704841, "cps": 395.054688, "mean_p": 0.520678, "boundary_entropy": 0.690474, "bpb": 8.007497}, "th": {"chunks": 45373.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.820827, "cps": 354.476562, "mean_p": 0.530947, "boundary_entropy": 0.689163, "bpb": 7.906248}, "zh": {"chunks": 4830.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.817812, "cps": 37.734375, "mean_p": 0.492079, "boundary_entropy": 0.686438, "bpb": 8.083402}, "my": {"chunks": 30009.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.583756, "cps": 234.445312, "mean_p": 0.506943, "boundary_entropy": 0.690852, "bpb": 8.172437}}, "bpb": {"en": 7.94344, "de": 8.09081, "vi": 8.01184, "tr": 8.11694, "fi": 7.94407, "ru": 8.15673, "el": 8.1371, "hi": 8.26467, "ta": 8.0075, "th": 7.90625, "zh": 8.0834, "my": 8.17244}, "gini_cpc": 0.246248, "gini_cps": 0.271901, "bpic": 1.184674, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1279942895838817, "vi": 1.4408326766927744, "tr": 1.2249569038998704, "fi": 1.1577992142566917, "ru": 2.2491679960299042, "el": 2.2453391374774343, "hi": 2.469825612887191, "ta": 3.379206432191523, "th": 3.5241098203839485, "zh": 1.0217077599307067, "my": 1.9786143351335526}, "t": 27.1} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 13378.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.768718, "cps": 104.515625, "mean_p": 0.517866, "boundary_entropy": 0.686922, "bpb": 7.68911}, "de": {"chunks": 18122.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.882579, "cps": 141.578125, "mean_p": 0.522388, "boundary_entropy": 0.687509, "bpb": 7.847162}, "vi": {"chunks": 21035.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.141036, "cps": 164.335938, "mean_p": 0.522443, "boundary_entropy": 0.688019, "bpb": 7.899189}, "tr": {"chunks": 17536.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.968679, "cps": 137.0, "mean_p": 0.527581, "boundary_entropy": 0.686627, "bpb": 7.931916}, "fi": {"chunks": 17142.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.911275, "cps": 133.921875, "mean_p": 0.523587, "boundary_entropy": 0.687105, "bpb": 7.734126}, "ru": {"chunks": 34092.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.79905, "cps": 266.34375, "mean_p": 0.539348, "boundary_entropy": 0.687325, "bpb": 7.911607}, "el": {"chunks": 36843.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.795205, "cps": 287.835938, "mean_p": 0.53546, "boundary_entropy": 0.688188, "bpb": 8.042989}, "hi": {"chunks": 32504.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.931887, "cps": 253.9375, "mean_p": 0.508797, "boundary_entropy": 0.690763, "bpb": 8.039317}, "ta": {"chunks": 50493.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.700883, "cps": 394.476562, "mean_p": 0.519756, "boundary_entropy": 0.690545, "bpb": 7.824514}, "th": {"chunks": 45302.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.816413, "cps": 353.921875, "mean_p": 0.530283, "boundary_entropy": 0.689241, "bpb": 7.673422}, "zh": {"chunks": 4736.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.801896, "cps": 37.0, "mean_p": 0.491745, "boundary_entropy": 0.686422, "bpb": 8.049757}, "my": {"chunks": 29944.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.580325, "cps": 233.9375, "mean_p": 0.506909, "boundary_entropy": 0.690854, "bpb": 8.085544}}, "bpb": {"en": 7.68911, "de": 7.84716, "vi": 7.89919, "tr": 7.93192, "fi": 7.73413, "ru": 7.91161, "el": 8.04299, "hi": 8.03932, "ta": 7.82451, "th": 7.67342, "zh": 8.04976, "my": 8.08554}, "gini_cpc": 0.250113, "gini_cps": 0.275951, "bpic": 1.195107, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1481183217150808, "vi": 1.4843362814311682, "tr": 1.2601229289051759, "fi": 1.1854480754053158, "ru": 2.340325119293638, "el": 2.3353236071527, "hi": 2.5131283247184344, "ta": 3.5134892880180626, "th": 3.66377874716529, "zh": 1.0431606100599227, "my": 2.0557929227146063}, "t": 36.6} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 11233.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.645463, "cps": 87.757812, "mean_p": 0.510808, "boundary_entropy": 0.687265, "bpb": 7.066313}, "de": {"chunks": 16269.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.792334, "cps": 127.101562, "mean_p": 0.515362, "boundary_entropy": 0.687993, "bpb": 7.200115}, "vi": {"chunks": 20145.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.092758, "cps": 157.382812, "mean_p": 0.519353, "boundary_entropy": 0.688256, "bpb": 7.55741}, "tr": {"chunks": 16597.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.916809, "cps": 129.664062, "mean_p": 0.522586, "boundary_entropy": 0.687084, "bpb": 7.403334}, "fi": {"chunks": 15627.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.830737, "cps": 122.085938, "mean_p": 0.517335, "boundary_entropy": 0.687577, "bpb": 7.198402}, "ru": {"chunks": 33945.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.791293, "cps": 265.195312, "mean_p": 0.53622, "boundary_entropy": 0.687784, "bpb": 7.173025}, "el": {"chunks": 36656.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.786094, "cps": 286.375, "mean_p": 0.532536, "boundary_entropy": 0.688565, "bpb": 7.593601}, "hi": {"chunks": 29941.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.779554, "cps": 233.914062, "mean_p": 0.506905, "boundary_entropy": 0.690812, "bpb": 7.233695}, "ta": {"chunks": 49806.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.664135, "cps": 389.109375, "mean_p": 0.515727, "boundary_entropy": 0.690815, "bpb": 7.092899}, "th": {"chunks": 44844.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.787939, "cps": 350.34375, "mean_p": 0.527349, "boundary_entropy": 0.689567, "bpb": 6.802021}, "zh": {"chunks": 4358.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.737894, "cps": 34.046875, "mean_p": 0.490318, "boundary_entropy": 0.686347, "bpb": 7.906512}, "my": {"chunks": 29099.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.535729, "cps": 227.335938, "mean_p": 0.506482, "boundary_entropy": 0.690861, "bpb": 7.673017}}, "bpb": {"en": 7.06631, "de": 7.20011, "vi": 7.55741, "tr": 7.40333, "fi": 7.1984, "ru": 7.17302, "el": 7.5936, "hi": 7.23369, "ta": 7.0929, "th": 6.80202, "zh": 7.90651, "my": 7.67302}, "gini_cpc": 0.267385, "gini_cps": 0.293605, "bpic": 1.243942, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2275432803995716, "vi": 1.69298258639744, "tr": 1.4203893387315003, "fi": 1.2870401346897595, "ru": 2.7752043016864887, "el": 2.7671492752425784, "hi": 2.7570179246592077, "ta": 4.127475994333073, "th": 4.319282800596327, "zh": 1.1431998125835574, "my": 2.3792662807557146}, "t": 46.0} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 7313.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.420215, "cps": 57.132812, "mean_p": 0.500679, "boundary_entropy": 0.687489, "bpb": 6.25924}, "de": {"chunks": 11437.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.557006, "cps": 89.351562, "mean_p": 0.504443, "boundary_entropy": 0.688403, "bpb": 6.387667}, "vi": {"chunks": 17692.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.959696, "cps": 138.21875, "mean_p": 0.512504, "boundary_entropy": 0.68867, "bpb": 7.001968}, "tr": {"chunks": 14176.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.783075, "cps": 110.75, "mean_p": 0.513303, "boundary_entropy": 0.68775, "bpb": 6.696437}, "fi": {"chunks": 11639.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.618734, "cps": 90.929688, "mean_p": 0.506513, "boundary_entropy": 0.688048, "bpb": 6.462795}, "ru": {"chunks": 29346.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.548602, "cps": 229.265625, "mean_p": 0.514197, "boundary_entropy": 0.689946, "bpb": 5.759787}, "el": {"chunks": 35087.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.709643, "cps": 274.117188, "mean_p": 0.52142, "boundary_entropy": 0.689714, "bpb": 6.450143}, "hi": {"chunks": 20911.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.242853, "cps": 163.367188, "mean_p": 0.500782, "boundary_entropy": 0.690879, "bpb": 5.78185}, "ta": {"chunks": 24762.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.324525, "cps": 193.453125, "mean_p": 0.502558, "boundary_entropy": 0.691251, "bpb": 5.964513}, "th": {"chunks": 39827.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.476034, "cps": 311.148438, "mean_p": 0.514341, "boundary_entropy": 0.690628, "bpb": 5.531964}, "zh": {"chunks": 3271.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.553844, "cps": 25.554688, "mean_p": 0.485631, "boundary_entropy": 0.686044, "bpb": 7.51115}, "my": {"chunks": 24745.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.305943, "cps": 193.320312, "mean_p": 0.504152, "boundary_entropy": 0.690888, "bpb": 6.385357}}, "bpb": {"en": 6.25924, "de": 6.38767, "vi": 7.00197, "tr": 6.69644, "fi": 6.4628, "ru": 5.75979, "el": 6.45014, "hi": 5.78185, "ta": 5.96451, "th": 5.53196, "zh": 7.51115, "my": 6.38536}, "gini_cpc": 0.279861, "gini_cps": 0.304595, "bpic": 1.597716, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.3255260303477086, "vi": 2.283822438211787, "tr": 1.8635098810392141, "fi": 1.4724221144206153, "ru": 3.685260953223504, "el": 4.068496422809896, "hi": 2.9576601907206306, "ta": 3.1520187810494176, "th": 5.892303055850609, "zh": 1.3180007222871617, "my": 3.107796897904579}, "t": 56.8} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 1360.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.078147, "cps": 10.625, "mean_p": 0.485581, "boundary_entropy": 0.687085, "bpb": 5.795209}, "de": {"chunks": 2004.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.097599, "cps": 15.65625, "mean_p": 0.487145, "boundary_entropy": 0.688059, "bpb": 5.936048}, "vi": {"chunks": 3582.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.194304, "cps": 27.984375, "mean_p": 0.486358, "boundary_entropy": 0.688453, "bpb": 6.612103}, "tr": {"chunks": 4791.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.264652, "cps": 37.429688, "mean_p": 0.493392, "boundary_entropy": 0.688032, "bpb": 6.302473}, "fi": {"chunks": 3013.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.160172, "cps": 23.539062, "mean_p": 0.490261, "boundary_entropy": 0.687927, "bpb": 5.99539}, "ru": {"chunks": 418.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.022058, "cps": 3.265625, "mean_p": 0.477107, "boundary_entropy": 0.689146, "bpb": 5.356508}, "el": {"chunks": 7189.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.35029, "cps": 56.164062, "mean_p": 0.494495, "boundary_entropy": 0.690498, "bpb": 5.590452}, "hi": {"chunks": 252.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.014978, "cps": 1.96875, "mean_p": 0.472563, "boundary_entropy": 0.689255, "bpb": 5.427786}, "ta": {"chunks": 736.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.039369, "cps": 5.75, "mean_p": 0.479596, "boundary_entropy": 0.690336, "bpb": 5.335613}, "th": {"chunks": 1073.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.066708, "cps": 8.382812, "mean_p": 0.482789, "boundary_entropy": 0.690374, "bpb": 5.04145}, "zh": {"chunks": 3440.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.582459, "cps": 26.875, "mean_p": 0.487072, "boundary_entropy": 0.686202, "bpb": 6.987966}, "my": {"chunks": 4074.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.215009, "cps": 31.828125, "mean_p": 0.479989, "boundary_entropy": 0.69016, "bpb": 4.922773}}, "bpb": {"en": 5.79521, "de": 5.93605, "vi": 6.6121, "tr": 6.30247, "fi": 5.99539, "ru": 5.35651, "el": 5.59045, "hi": 5.42779, "ta": 5.33561, "th": 5.04145, "zh": 6.98797, "my": 4.92277}, "gini_cpc": 0.476485, "gini_cps": 0.414485, "bpic": 12.018696, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2489082137506051, "vi": 2.486380845259178, "tr": 3.386576038745609, "fi": 2.0496157989536785, "ru": 0.2822619121527239, "el": 4.482423127853685, "hi": 0.19165964513591469, "ta": 0.503776096156569, "th": 0.8536185978898865, "zh": 7.453327622955718, "my": 2.751331119845025}, "t": 68.3} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 1492.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.085732, "cps": 11.65625, "mean_p": 0.488496, "boundary_entropy": 0.687339, "bpb": 5.395132}, "de": {"chunks": 1627.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.079238, "cps": 12.710938, "mean_p": 0.487479, "boundary_entropy": 0.688136, "bpb": 5.569456}, "vi": {"chunks": 4710.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.255492, "cps": 36.796875, "mean_p": 0.490453, "boundary_entropy": 0.688787, "bpb": 6.02921}, "tr": {"chunks": 2942.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.162515, "cps": 22.984375, "mean_p": 0.489717, "boundary_entropy": 0.687917, "bpb": 5.911938}, "fi": {"chunks": 1760.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.093562, "cps": 13.75, "mean_p": 0.487375, "boundary_entropy": 0.687816, "bpb": 5.577003}, "ru": {"chunks": 2960.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.156201, "cps": 23.125, "mean_p": 0.491258, "boundary_entropy": 0.69021, "bpb": 4.821545}, "el": {"chunks": 5957.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.29026, "cps": 46.539062, "mean_p": 0.496044, "boundary_entropy": 0.690621, "bpb": 5.056178}, "hi": {"chunks": 19834.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.178841, "cps": 154.953125, "mean_p": 0.501014, "boundary_entropy": 0.690986, "bpb": 4.431196}, "ta": {"chunks": 9564.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.511581, "cps": 74.71875, "mean_p": 0.494511, "boundary_entropy": 0.691208, "bpb": 4.41933}, "th": {"chunks": 9204.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.57221, "cps": 71.90625, "mean_p": 0.493762, "boundary_entropy": 0.690934, "bpb": 4.445493}, "zh": {"chunks": 3608.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.610904, "cps": 28.1875, "mean_p": 0.491909, "boundary_entropy": 0.686853, "bpb": 6.667395}, "my": {"chunks": 844.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.044543, "cps": 6.59375, "mean_p": 0.482698, "boundary_entropy": 0.690591, "bpb": 4.392868}}, "bpb": {"en": 5.39513, "de": 5.56946, "vi": 6.02921, "tr": 5.91194, "fi": 5.577, "ru": 4.82154, "el": 5.05618, "hi": 4.4312, "ta": 4.41933, "th": 4.44549, "zh": 6.6674, "my": 4.39287}, "gini_cpc": 0.47951, "gini_cps": 0.467058, "bpic": 5.949909, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9242520932432183, "vi": 2.9801152662314006, "tr": 1.8956031164535259, "fi": 1.091329968789296, "ru": 1.8219556190624402, "el": 3.3856499390796624, "hi": 13.7502480589892, "ta": 5.967183523859443, "th": 6.674378656950633, "zh": 7.125713932938648, "my": 0.519558396310829}, "t": 84.1} |
| {"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.30811, "bpb": 4.77259, "chunk_loss": 0.033786, "F": 0.22348, "G": 0.47714, "chunks": 228.84, "bytes_per_chunk": 4.475, "boundary_entropy": 0.68918, "router_noise": 0.0, "grad_norm": 2.4893, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 0.836, "alpha": 0.03, "t": 94.6} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3164.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.181808, "cps": 24.71875, "mean_p": 0.43417, "boundary_entropy": 0.667979, "bpb": 4.333644}, "de": {"chunks": 4037.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.19661, "cps": 31.539062, "mean_p": 0.43695, "boundary_entropy": 0.67015, "bpb": 4.580305}, "vi": {"chunks": 6237.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.338324, "cps": 48.726562, "mean_p": 0.476953, "boundary_entropy": 0.685448, "bpb": 4.760798}, "tr": {"chunks": 3944.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.217864, "cps": 30.8125, "mean_p": 0.448988, "boundary_entropy": 0.675252, "bpb": 4.778883}, "fi": {"chunks": 2661.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.14146, "cps": 20.789062, "mean_p": 0.442552, "boundary_entropy": 0.675211, "bpb": 4.605934}, "ru": {"chunks": 12493.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.659261, "cps": 97.601562, "mean_p": 0.402684, "boundary_entropy": 0.609261, "bpb": 3.211979}, "el": {"chunks": 14266.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.695123, "cps": 111.453125, "mean_p": 0.481204, "boundary_entropy": 0.686665, "bpb": 4.149384}, "hi": {"chunks": 9940.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.590788, "cps": 77.65625, "mean_p": 0.489844, "boundary_entropy": 0.689742, "bpb": 2.806752}, "ta": {"chunks": 12716.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.680182, "cps": 99.34375, "mean_p": 0.476756, "boundary_entropy": 0.688074, "bpb": 2.914983}, "th": {"chunks": 8404.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.522474, "cps": 65.65625, "mean_p": 0.487937, "boundary_entropy": 0.690268, "bpb": 2.805669}, "zh": {"chunks": 3860.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.653573, "cps": 30.15625, "mean_p": 0.491003, "boundary_entropy": 0.686682, "bpb": 6.220086}, "my": {"chunks": 5569.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.29391, "cps": 43.507812, "mean_p": 0.471522, "boundary_entropy": 0.687998, "bpb": 3.095336}}, "bpb": {"en": 4.33364, "de": 4.58031, "vi": 4.7608, "tr": 4.77888, "fi": 4.60593, "ru": 3.21198, "el": 4.14938, "hi": 2.80675, "ta": 2.91498, "th": 2.80567, "zh": 6.22009, "my": 3.09534}, "gini_cpc": 0.275032, "gini_cps": 0.303394, "bpic": 4.39657, "worst_lang": "el", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0814189800107785, "vi": 1.8608880520172528, "tr": 1.1983232902495784, "fi": 0.7780735220892153, "ru": 3.6261450386772034, "el": 3.823393697293647, "hi": 3.249518073873424, "ta": 3.7412152427567555, "th": 2.873774083399387, "zh": 3.594856074283026, "my": 1.61659595280663}, "t": 106.0} |
| {"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.36244, "bpb": 3.40829, "chunk_loss": 0.031628, "F": 0.19948, "G": 0.36492, "chunks": 204.27, "bytes_per_chunk": 5.013, "boundary_entropy": 0.60106, "router_noise": 0.0, "grad_norm": 2.0566, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.318, "alpha": 0.03, "t": 126.3} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3048.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.175142, "cps": 23.8125, "mean_p": 0.328172, "boundary_entropy": 0.576587, "bpb": 3.53413}, "de": {"chunks": 3685.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.179467, "cps": 28.789062, "mean_p": 0.326855, "boundary_entropy": 0.567445, "bpb": 3.750434}, "vi": {"chunks": 5357.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.290589, "cps": 41.851562, "mean_p": 0.375556, "boundary_entropy": 0.570193, "bpb": 3.060669}, "tr": {"chunks": 4003.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.221124, "cps": 31.273438, "mean_p": 0.337955, "boundary_entropy": 0.556045, "bpb": 3.749533}, "fi": {"chunks": 3256.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.17309, "cps": 25.4375, "mean_p": 0.340861, "boundary_entropy": 0.582461, "bpb": 3.801811}, "ru": {"chunks": 7714.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.407071, "cps": 60.265625, "mean_p": 0.245534, "boundary_entropy": 0.421172, "bpb": 2.586781}, "el": {"chunks": 5907.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.287823, "cps": 46.148438, "mean_p": 0.274611, "boundary_entropy": 0.47433, "bpb": 2.707105}, "hi": {"chunks": 7617.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.452719, "cps": 59.507812, "mean_p": 0.372104, "boundary_entropy": 0.607849, "bpb": 2.041344}, "ta": {"chunks": 9189.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.491522, "cps": 71.789062, "mean_p": 0.382673, "boundary_entropy": 0.582069, "bpb": 1.862829}, "th": {"chunks": 6605.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.410631, "cps": 51.601562, "mean_p": 0.425593, "boundary_entropy": 0.668122, "bpb": 2.075413}, "zh": {"chunks": 2956.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.500508, "cps": 23.09375, "mean_p": 0.410548, "boundary_entropy": 0.647493, "bpb": 4.683374}, "my": {"chunks": 16018.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.845366, "cps": 125.140625, "mean_p": 0.456371, "boundary_entropy": 0.677354, "bpb": 2.159535}}, "bpb": {"en": 3.53413, "de": 3.75043, "vi": 3.06067, "tr": 3.74953, "fi": 3.80181, "ru": 2.58678, "el": 2.7071, "hi": 2.04134, "ta": 1.86283, "th": 2.07541, "zh": 4.68337, "my": 2.15953}, "gini_cpc": 0.265885, "gini_cps": 0.285507, "bpic": 5.092973, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0246941163657812, "vi": 1.6591576810628534, "tr": 1.2625372377049786, "fi": 0.9882837860407527, "ru": 2.3242325431616564, "el": 1.6433697299860615, "hi": 2.5848660364334104, "ta": 2.806415301156358, "th": 2.3445576403897603, "zh": 2.8577230948908228, "my": 4.826741837170607}, "t": 143.5} |
| {"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.11542, "bpb": 3.05191, "chunk_loss": 0.030259, "F": 0.16942, "G": 0.3103, "chunks": 173.48, "bytes_per_chunk": 5.903, "boundary_entropy": 0.52627, "router_noise": 0.0, "grad_norm": 1.1012, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.285, "alpha": 0.03, "t": 154.9} |
| {"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 1.97706, "bpb": 2.85229, "chunk_loss": 0.030373, "F": 0.17603, "G": 0.30185, "chunks": 180.25, "bytes_per_chunk": 5.681, "boundary_entropy": 0.51222, "router_noise": 0.0, "grad_norm": 1.4163, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.246, "alpha": 0.03, "t": 179.5} |
| {"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.75949, "bpb": 2.5384, "chunk_loss": 0.030087, "F": 0.1656, "G": 0.28019, "chunks": 169.58, "bytes_per_chunk": 6.039, "boundary_entropy": 0.47198, "router_noise": 0.0, "grad_norm": 1.1286, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 203.7} |
| {"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.75656, "bpb": 2.53418, "chunk_loss": 0.030712, "F": 0.19101, "G": 0.29216, "chunks": 195.59, "bytes_per_chunk": 5.235, "boundary_entropy": 0.45596, "router_noise": 0.0, "grad_norm": 0.8859, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 227.1} |
| {"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.71907, "bpb": 2.4801, "chunk_loss": 0.030456, "F": 0.18156, "G": 0.26524, "chunks": 185.92, "bytes_per_chunk": 5.508, "boundary_entropy": 0.40159, "router_noise": 0.0, "grad_norm": 0.7619, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 250.6} |
| {"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.58739, "bpb": 2.29012, "chunk_loss": 0.030314, "F": 0.17615, "G": 0.26338, "chunks": 180.38, "bytes_per_chunk": 5.677, "boundary_entropy": 0.39317, "router_noise": 0.0, "grad_norm": 0.7558, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 274.7} |
| {"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.53037, "bpb": 2.20785, "chunk_loss": 0.030161, "F": 0.17206, "G": 0.26154, "chunks": 176.19, "bytes_per_chunk": 5.812, "boundary_entropy": 0.39883, "router_noise": 0.0, "grad_norm": 0.7276, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.223, "alpha": 0.03, "t": 297.0} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2746.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.157789, "cps": 21.453125, "mean_p": 0.274602, "boundary_entropy": 0.441449, "bpb": 2.315191}, "de": {"chunks": 3576.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174159, "cps": 27.9375, "mean_p": 0.249395, "boundary_entropy": 0.395395, "bpb": 2.598539}, "vi": {"chunks": 4249.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.230485, "cps": 33.195312, "mean_p": 0.252985, "boundary_entropy": 0.325453, "bpb": 2.084197}, "tr": {"chunks": 3356.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185384, "cps": 26.21875, "mean_p": 0.247337, "boundary_entropy": 0.380619, "bpb": 2.59926}, "fi": {"chunks": 3611.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.191962, "cps": 28.210938, "mean_p": 0.252905, "boundary_entropy": 0.37307, "bpb": 2.819229}, "ru": {"chunks": 5438.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.286966, "cps": 42.484375, "mean_p": 0.222899, "boundary_entropy": 0.349723, "bpb": 1.651608}, "el": {"chunks": 6300.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.306973, "cps": 49.21875, "mean_p": 0.2458, "boundary_entropy": 0.390596, "bpb": 1.735688}, "hi": {"chunks": 8087.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.480654, "cps": 63.179688, "mean_p": 0.242933, "boundary_entropy": 0.363339, "bpb": 1.338992}, "ta": {"chunks": 8213.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.439315, "cps": 64.164062, "mean_p": 0.239454, "boundary_entropy": 0.391726, "bpb": 1.110418}, "th": {"chunks": 9731.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.604974, "cps": 76.023438, "mean_p": 0.297235, "boundary_entropy": 0.45631, "bpb": 1.314781}, "zh": {"chunks": 4151.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.702845, "cps": 32.429688, "mean_p": 0.273908, "boundary_entropy": 0.161195, "bpb": 3.413381}, "my": {"chunks": 8483.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.447699, "cps": 66.273438, "mean_p": 0.322714, "boundary_entropy": 0.528066, "bpb": 1.251052}}, "bpb": {"en": 2.31519, "de": 2.59854, "vi": 2.0842, "tr": 2.59926, "fi": 2.81923, "ru": 1.65161, "el": 1.73569, "hi": 1.33899, "ta": 1.11042, "th": 1.31478, "zh": 3.41338, "my": 1.25105}, "gini_cpc": 0.274498, "gini_cps": 0.227468, "bpic": 5.648739, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1037448501442386, "vi": 1.4607206754465552, "tr": 1.174883996121731, "fi": 1.2165758532677355, "ru": 1.8186686319462995, "el": 1.9454644157964067, "hi": 3.046182771319948, "ta": 2.784196868230324, "th": 3.8340696196756983, "zh": 4.454335018963025, "my": 2.837328877699707}, "t": 324.2} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.32755, "de": 2.59232, "vi": 2.08876, "tr": 2.59319, "fi": 2.82797, "ru": 1.65135, "el": 1.73033, "hi": 1.34746, "ta": 1.11068, "th": 1.31408, "zh": 3.419, "my": 1.26309}, "bytes_per_chunk": {"en": 5.627, "de": 5.543, "vi": 5.561, "tr": 5.981, "fi": 5.219, "ru": 6.209, "el": 5.95, "hi": 5.329, "ta": 6.036, "th": 4.728, "zh": 3.82, "my": 6.509}, "bpb_mean": 2.02215, "t": 330.7} |
| {"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.51467, "bpb": 2.18521, "chunk_loss": 0.030795, "F": 0.19392, "G": 0.26618, "chunks": 198.58, "bytes_per_chunk": 5.157, "boundary_entropy": 0.36546, "router_noise": 0.0, "grad_norm": 0.626, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.34, "alpha": 0.03, "t": 331.0} |
| {"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.43118, "bpb": 2.06476, "chunk_loss": 0.030442, "F": 0.18524, "G": 0.26039, "chunks": 189.69, "bytes_per_chunk": 5.398, "boundary_entropy": 0.36342, "router_noise": 0.0, "grad_norm": 0.6703, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 354.5} |
| {"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.49413, "bpb": 2.15557, "chunk_loss": 0.030494, "F": 0.18706, "G": 0.26553, "chunks": 191.55, "bytes_per_chunk": 5.346, "boundary_entropy": 0.3663, "router_noise": 0.0, "grad_norm": 0.7029, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 377.1} |
| {"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.36749, "bpb": 1.97287, "chunk_loss": 0.030343, "F": 0.17995, "G": 0.25722, "chunks": 184.27, "bytes_per_chunk": 5.557, "boundary_entropy": 0.36595, "router_noise": 0.0, "grad_norm": 0.6565, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 400.6} |
| {"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.3343, "bpb": 1.92499, "chunk_loss": 0.03036, "F": 0.17886, "G": 0.261, "chunks": 183.16, "bytes_per_chunk": 5.591, "boundary_entropy": 0.38016, "router_noise": 0.0, "grad_norm": 0.5995, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.223, "alpha": 0.03, "t": 422.9} |
| {"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.40441, "bpb": 2.02613, "chunk_loss": 0.030446, "F": 0.18336, "G": 0.25287, "chunks": 187.77, "bytes_per_chunk": 5.454, "boundary_entropy": 0.33916, "router_noise": 0.0, "grad_norm": 0.5936, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 446.1} |
| {"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.44886, "bpb": 2.09026, "chunk_loss": 0.030234, "F": 0.17523, "G": 0.25614, "chunks": 179.44, "bytes_per_chunk": 5.707, "boundary_entropy": 0.34565, "router_noise": 0.0, "grad_norm": 0.5948, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.212, "alpha": 0.03, "t": 467.3} |
| {"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.47517, "bpb": 2.12822, "chunk_loss": 0.030394, "F": 0.18207, "G": 0.25769, "chunks": 186.44, "bytes_per_chunk": 5.492, "boundary_entropy": 0.339, "router_noise": 0.0, "grad_norm": 0.6035, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.219, "alpha": 0.03, "t": 489.3} |
| {"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.28435, "bpb": 1.85292, "chunk_loss": 0.030125, "F": 0.17075, "G": 0.25251, "chunks": 174.84, "bytes_per_chunk": 5.857, "boundary_entropy": 0.35893, "router_noise": 0.0, "grad_norm": 0.5349, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.22, "alpha": 0.03, "t": 511.2} |
| {"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.27086, "bpb": 1.83347, "chunk_loss": 0.030305, "F": 0.17892, "G": 0.25908, "chunks": 183.22, "bytes_per_chunk": 5.589, "boundary_entropy": 0.35809, "router_noise": 0.0, "grad_norm": 0.5118, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 534.3} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2645.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.151985, "cps": 20.664062, "mean_p": 0.246756, "boundary_entropy": 0.379901, "bpb": 2.091317}, "de": {"chunks": 3058.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.148931, "cps": 23.890625, "mean_p": 0.215877, "boundary_entropy": 0.314796, "bpb": 2.303843}, "vi": {"chunks": 4280.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.232167, "cps": 33.4375, "mean_p": 0.231595, "boundary_entropy": 0.25859, "bpb": 1.92935}, "tr": {"chunks": 2996.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.165497, "cps": 23.40625, "mean_p": 0.239948, "boundary_entropy": 0.358995, "bpb": 2.282291}, "fi": {"chunks": 3582.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.19042, "cps": 27.984375, "mean_p": 0.239223, "boundary_entropy": 0.315897, "bpb": 2.474585}, "ru": {"chunks": 5324.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.28095, "cps": 41.59375, "mean_p": 0.216149, "boundary_entropy": 0.303467, "bpb": 1.406721}, "el": {"chunks": 6413.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.312479, "cps": 50.101562, "mean_p": 0.241421, "boundary_entropy": 0.365504, "bpb": 1.462878}, "hi": {"chunks": 7509.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.4463, "cps": 58.664062, "mean_p": 0.247882, "boundary_entropy": 0.355766, "bpb": 1.191793}, "ta": {"chunks": 9556.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.511153, "cps": 74.65625, "mean_p": 0.277184, "boundary_entropy": 0.404224, "bpb": 0.996058}, "th": {"chunks": 6295.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.391358, "cps": 49.179688, "mean_p": 0.266685, "boundary_entropy": 0.445018, "bpb": 1.126838}, "zh": {"chunks": 3972.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.672536, "cps": 31.03125, "mean_p": 0.257051, "boundary_entropy": 0.141011, "bpb": 3.111172}, "my": {"chunks": 9533.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.503114, "cps": 74.476562, "mean_p": 0.355364, "boundary_entropy": 0.570949, "bpb": 1.151672}}, "bpb": {"en": 2.09132, "de": 2.30384, "vi": 1.92935, "tr": 2.28229, "fi": 2.47458, "ru": 1.40672, "el": 1.46288, "hi": 1.19179, "ta": 0.99606, "th": 1.12684, "zh": 3.11117, "my": 1.15167}, "gini_cpc": 0.273362, "gini_cps": 0.240652, "bpic": 5.889554, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9799039712641102, "vi": 1.5275627902255051, "tr": 1.0889042714724095, "fi": 1.2528876891904237, "ru": 1.848533290771157, "el": 2.05597977721892, "hi": 2.9364693708446508, "ta": 3.3631722845741336, "th": 2.574975569884234, "zh": 4.42500984228656, "my": 3.3102794714656327}, "t": 560.6} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.10052, "de": 2.29617, "vi": 1.93356, "tr": 2.27544, "fi": 2.48852, "ru": 1.40465, "el": 1.45921, "hi": 1.19871, "ta": 0.99565, "th": 1.12268, "zh": 3.12076, "my": 1.16005}, "bytes_per_chunk": {"en": 5.871, "de": 6.27, "vi": 5.418, "tr": 6.49, "fi": 5.173, "ru": 6.356, "el": 5.762, "hi": 5.711, "ta": 5.223, "th": 6.981, "zh": 3.97, "my": 5.647}, "bpb_mean": 1.79633, "t": 567.1} |
| {"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.36795, "bpb": 1.97354, "chunk_loss": 0.030467, "F": 0.18356, "G": 0.25508, "chunks": 187.97, "bytes_per_chunk": 5.448, "boundary_entropy": 0.33011, "router_noise": 0.0, "grad_norm": 0.5255, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.331, "alpha": 0.03, "t": 567.4} |
| {"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.38328, "bpb": 1.99565, "chunk_loss": 0.030379, "F": 0.18271, "G": 0.25136, "chunks": 187.09, "bytes_per_chunk": 5.473, "boundary_entropy": 0.31743, "router_noise": 0.0, "grad_norm": 0.5253, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 590.5} |
| {"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.25227, "bpb": 1.80664, "chunk_loss": 0.030072, "F": 0.16777, "G": 0.2477, "chunks": 171.8, "bytes_per_chunk": 5.961, "boundary_entropy": 0.34582, "router_noise": 0.0, "grad_norm": 0.4875, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.221, "alpha": 0.03, "t": 612.6} |
| {"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.39369, "bpb": 2.01067, "chunk_loss": 0.030254, "F": 0.17549, "G": 0.24935, "chunks": 179.7, "bytes_per_chunk": 5.698, "boundary_entropy": 0.32605, "router_noise": 0.0, "grad_norm": 0.4814, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 635.0} |
| {"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.29571, "bpb": 1.86931, "chunk_loss": 0.030241, "F": 0.17914, "G": 0.24894, "chunks": 183.44, "bytes_per_chunk": 5.582, "boundary_entropy": 0.32795, "router_noise": 0.0, "grad_norm": 0.5029, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 658.2} |
| {"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.26768, "bpb": 1.82888, "chunk_loss": 0.030488, "F": 0.18106, "G": 0.25135, "chunks": 185.41, "bytes_per_chunk": 5.523, "boundary_entropy": 0.32485, "router_noise": 0.0, "grad_norm": 0.4688, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.23, "alpha": 0.03, "t": 681.1} |
| {"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.36321, "bpb": 1.96669, "chunk_loss": 0.030637, "F": 0.19232, "G": 0.25719, "chunks": 196.94, "bytes_per_chunk": 5.2, "boundary_entropy": 0.30853, "router_noise": 0.0, "grad_norm": 0.4542, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 704.0} |
| {"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.22975, "bpb": 1.77415, "chunk_loss": 0.030095, "F": 0.16934, "G": 0.24693, "chunks": 173.41, "bytes_per_chunk": 5.905, "boundary_entropy": 0.33102, "router_noise": 0.0, "grad_norm": 0.4608, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 727.0} |
| {"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.29324, "bpb": 1.86575, "chunk_loss": 0.030383, "F": 0.18164, "G": 0.24521, "chunks": 186.0, "bytes_per_chunk": 5.505, "boundary_entropy": 0.29977, "router_noise": 0.0, "grad_norm": 0.436, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 750.6} |
| {"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.36453, "bpb": 1.96861, "chunk_loss": 0.030409, "F": 0.18286, "G": 0.24939, "chunks": 187.25, "bytes_per_chunk": 5.469, "boundary_entropy": 0.30723, "router_noise": 0.0, "grad_norm": 0.481, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 773.2} |
| {"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2835.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.162903, "cps": 22.148438, "mean_p": 0.23845, "boundary_entropy": 0.335982, "bpb": 1.991957}, "de": {"chunks": 3430.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.167048, "cps": 26.796875, "mean_p": 0.212812, "boundary_entropy": 0.261634, "bpb": 2.156328}, "vi": {"chunks": 4257.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.230919, "cps": 33.257812, "mean_p": 0.220653, "boundary_entropy": 0.220123, "bpb": 1.818417}, "tr": {"chunks": 3296.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.182069, "cps": 25.75, "mean_p": 0.240767, "boundary_entropy": 0.328592, "bpb": 2.15587}, "fi": {"chunks": 3429.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.182287, "cps": 26.789062, "mean_p": 0.221422, "boundary_entropy": 0.2654, "bpb": 2.330073}, "ru": {"chunks": 5914.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.312084, "cps": 46.203125, "mean_p": 0.226108, "boundary_entropy": 0.279843, "bpb": 1.302194}, "el": {"chunks": 6035.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.29406, "cps": 47.148438, "mean_p": 0.223097, "boundary_entropy": 0.331723, "bpb": 1.344343}, "hi": {"chunks": 6783.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.40315, "cps": 52.992188, "mean_p": 0.316236, "boundary_entropy": 0.471907, "bpb": 1.102348}, "ta": {"chunks": 8397.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.449158, "cps": 65.601562, "mean_p": 0.29188, "boundary_entropy": 0.429768, "bpb": 0.9272}, "th": {"chunks": 7135.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.443581, "cps": 55.742188, "mean_p": 0.318768, "boundary_entropy": 0.500804, "bpb": 1.055341}, "zh": {"chunks": 3946.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.668134, "cps": 30.828125, "mean_p": 0.251419, "boundary_entropy": 0.129761, "bpb": 2.942619}, "my": {"chunks": 8130.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.429069, "cps": 63.515625, "mean_p": 0.359315, "boundary_entropy": 0.575353, "bpb": 1.058114}}, "bpb": {"en": 1.99196, "de": 2.15633, "vi": 1.81842, "tr": 2.15587, "fi": 2.33007, "ru": 1.30219, "el": 1.34434, "hi": 1.10235, "ta": 0.9272, "th": 1.05534, "zh": 2.94262, "my": 1.05811}, "gini_cpc": 0.248351, "gini_cps": 0.203552, "bpic": 6.035526, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0254459397789648, "vi": 1.4175277358027563, "tr": 1.1176548538974658, "fi": 1.1189911677817692, "ru": 1.9157690943865089, "el": 1.8051258531207857, "hi": 2.4747868581806176, "ta": 2.757209337867268, "th": 2.7229769758212408, "zh": 4.101424253337562, "my": 2.633893596844832}, "t": 799.3} |
| {"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 2.00104, "de": 2.14903, "vi": 1.82409, "tr": 2.14936, "fi": 2.34598, "ru": 1.30132, "el": 1.34126, "hi": 1.10971, "ta": 0.92803, "th": 1.0521, "zh": 2.95447, "my": 1.06624}, "bytes_per_chunk": {"en": 5.614, "de": 5.744, "vi": 5.453, "tr": 5.84, "fi": 5.397, "ru": 5.749, "el": 6.049, "hi": 6.152, "ta": 5.985, "th": 6.224, "zh": 3.978, "my": 6.685}, "bpb_mean": 1.68522, "t": 806.6} |
| {"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.39076, "bpb": 2.00644, "chunk_loss": 0.030445, "F": 0.18762, "G": 0.25402, "chunks": 192.12, "bytes_per_chunk": 5.33, "boundary_entropy": 0.29758, "router_noise": 0.0, "grad_norm": 0.4731, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.336, "alpha": 0.03, "t": 806.8} |
| {"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.27526, "bpb": 1.83982, "chunk_loss": 0.030207, "F": 0.17471, "G": 0.24481, "chunks": 178.91, "bytes_per_chunk": 5.724, "boundary_entropy": 0.30019, "router_noise": 0.0, "grad_norm": 0.4503, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 829.4} |
| {"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.30955, "bpb": 1.88927, "chunk_loss": 0.030381, "F": 0.18297, "G": 0.24623, "chunks": 187.36, "bytes_per_chunk": 5.465, "boundary_entropy": 0.28975, "router_noise": 0.0, "grad_norm": 0.4503, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 852.2} |
| {"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.176, "bpb": 1.69661, "chunk_loss": 0.030407, "F": 0.1787, "G": 0.25027, "chunks": 182.98, "bytes_per_chunk": 5.596, "boundary_entropy": 0.30222, "router_noise": 0.0, "grad_norm": 0.4248, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 875.0} |
| {"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.23388, "bpb": 1.78011, "chunk_loss": 0.030282, "F": 0.17862, "G": 0.25138, "chunks": 182.91, "bytes_per_chunk": 5.598, "boundary_entropy": 0.30375, "router_noise": 0.0, "grad_norm": 0.4339, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 897.9} |
| {"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.2132, "bpb": 1.75027, "chunk_loss": 0.030518, "F": 0.1864, "G": 0.2553, "chunks": 190.88, "bytes_per_chunk": 5.365, "boundary_entropy": 0.30572, "router_noise": 0.0, "grad_norm": 0.4387, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 920.6} |
| {"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.27748, "bpb": 1.84301, "chunk_loss": 0.030284, "F": 0.17836, "G": 0.24828, "chunks": 182.64, "bytes_per_chunk": 5.607, "boundary_entropy": 0.30683, "router_noise": 0.0, "grad_norm": 0.4437, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 944.2} |
| {"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.19158, "bpb": 1.71908, "chunk_loss": 0.030242, "F": 0.17485, "G": 0.24031, "chunks": 179.05, "bytes_per_chunk": 5.719, "boundary_entropy": 0.29651, "router_noise": 0.0, "grad_norm": 0.4511, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 967.0} |
| {"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.32737, "bpb": 1.915, "chunk_loss": 0.030678, "F": 0.19421, "G": 0.25133, "chunks": 198.88, "bytes_per_chunk": 5.149, "boundary_entropy": 0.26105, "router_noise": 0.0, "grad_norm": 0.4188, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 989.8} |
| {"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.19724, "bpb": 1.72725, "chunk_loss": 0.030277, "F": 0.17809, "G": 0.24815, "chunks": 182.36, "bytes_per_chunk": 5.615, "boundary_entropy": 0.3054, "router_noise": 0.0, "grad_norm": 0.4254, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 1014.2} |
| {"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2639.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.151641, "cps": 20.617188, "mean_p": 0.210135, "boundary_entropy": 0.280267, "bpb": 1.9303}, "de": {"chunks": 3447.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.167876, "cps": 26.929688, "mean_p": 0.2011, "boundary_entropy": 0.213886, "bpb": 2.085702}, "vi": {"chunks": 4185.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227014, "cps": 32.695312, "mean_p": 0.20819, "boundary_entropy": 0.185181, "bpb": 1.754444}, "tr": {"chunks": 3119.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.172292, "cps": 24.367188, "mean_p": 0.214948, "boundary_entropy": 0.284535, "bpb": 2.070583}, "fi": {"chunks": 3448.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.183297, "cps": 26.9375, "mean_p": 0.213194, "boundary_entropy": 0.228886, "bpb": 2.210402}, "ru": {"chunks": 5891.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.310871, "cps": 46.023438, "mean_p": 0.220847, "boundary_entropy": 0.265616, "bpb": 1.250932}, "el": {"chunks": 6002.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.292452, "cps": 46.890625, "mean_p": 0.223178, "boundary_entropy": 0.322275, "bpb": 1.277412}, "hi": {"chunks": 7337.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.436077, "cps": 57.320312, "mean_p": 0.351274, "boundary_entropy": 0.493331, "bpb": 1.053886}, "ta": {"chunks": 8348.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.446537, "cps": 65.21875, "mean_p": 0.313492, "boundary_entropy": 0.458177, "bpb": 0.885951}, "th": {"chunks": 7772.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.483183, "cps": 60.71875, "mean_p": 0.359284, "boundary_entropy": 0.520439, "bpb": 1.004679}, "zh": {"chunks": 3958.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.670166, "cps": 30.921875, "mean_p": 0.246375, "boundary_entropy": 0.113438, "bpb": 2.833377}, "my": {"chunks": 7036.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.371332, "cps": 54.96875, "mean_p": 0.352565, "boundary_entropy": 0.569572, "bpb": 1.02126}}, "bpb": {"en": 1.9303, "de": 2.0857, "vi": 1.75444, "tr": 2.07058, "fi": 2.2104, "ru": 1.25093, "el": 1.27741, "hi": 1.05389, "ta": 0.88595, "th": 1.00468, "zh": 2.83338, "my": 1.02126}, "gini_cpc": 0.259335, "gini_cps": 0.207976, "bpic": 6.074214, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1070662376281424, "vi": 1.4970525672526178, "tr": 1.13618631310117, "fi": 1.208760065951108, "ru": 2.050050400877441, "el": 1.9285898461558404, "hi": 2.8757304508270334, "ta": 2.9447043701565008, "th": 3.1863718500958167, "zh": 4.419438321758581, "my": 2.448765425136328}, "t": 1040.4} |
| {"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.93773, "de": 2.0785, "vi": 1.76074, "tr": 2.06544, "fi": 2.2276, "ru": 1.24956, "el": 1.27597, "hi": 1.06142, "ta": 0.88712, "th": 1.00279, "zh": 2.84217, "my": 1.02977}, "bytes_per_chunk": {"en": 5.983, "de": 5.744, "vi": 5.564, "tr": 6.131, "fi": 5.364, "ru": 5.757, "el": 6.156, "hi": 5.629, "ta": 5.933, "th": 5.766, "zh": 3.976, "my": 7.53}, "bpb_mean": 1.61823, "t": 1047.4} |
| {"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.20706, "bpb": 1.74142, "chunk_loss": 0.030219, "F": 0.17406, "G": 0.24595, "chunks": 178.23, "bytes_per_chunk": 5.745, "boundary_entropy": 0.29809, "router_noise": 0.0, "grad_norm": 0.4174, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.335, "alpha": 0.03, "t": 1047.7} |
| {"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.26669, "bpb": 1.82744, "chunk_loss": 0.03032, "F": 0.18623, "G": 0.25452, "chunks": 190.7, "bytes_per_chunk": 5.37, "boundary_entropy": 0.28236, "router_noise": 0.0, "grad_norm": 0.4284, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 1072.0} |
| {"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.0872, "bpb": 1.56849, "chunk_loss": 0.030314, "F": 0.17764, "G": 0.26101, "chunks": 181.91, "bytes_per_chunk": 5.629, "boundary_entropy": 0.32813, "router_noise": 0.0, "grad_norm": 0.4566, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 1095.6} |
| {"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.2487, "bpb": 1.80149, "chunk_loss": 0.03031, "F": 0.17743, "G": 0.23828, "chunks": 181.69, "bytes_per_chunk": 5.636, "boundary_entropy": 0.27787, "router_noise": 0.0, "grad_norm": 0.4081, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 1118.8} |
| {"event": "phase_transition", "step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1142.0} |
| {"event": "train", "step": 4400, "phase": 2, "objective": "global", "lm_loss": 1.26754, "bpb": 1.82868, "chunk_loss": 0.030212, "F": 0.17677, "G": 0.23771, "chunks": 181.02, "bytes_per_chunk": 5.657, "boundary_entropy": 0.26874, "router_noise": 0.0, "grad_norm": 0.4233, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 1142.3} |
| {"event": "train", "step": 4500, "phase": 2, "objective": "global", "lm_loss": 1.26136, "bpb": 1.81976, "chunk_loss": 0.030344, "F": 0.18184, "G": 0.23942, "chunks": 186.2, "bytes_per_chunk": 5.499, "boundary_entropy": 0.26952, "router_noise": 0.0, "grad_norm": 0.4204, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 1165.6} |
| {"event": "train", "step": 4600, "phase": 2, "objective": "global", "lm_loss": 1.18816, "bpb": 1.71415, "chunk_loss": 0.030317, "F": 0.18208, "G": 0.24671, "chunks": 186.45, "bytes_per_chunk": 5.492, "boundary_entropy": 0.27458, "router_noise": 0.0, "grad_norm": 0.4108, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.229, "alpha": 0.03, "t": 1188.5} |
| {"event": "train", "step": 4700, "phase": 2, "objective": "global", "lm_loss": 1.20412, "bpb": 1.73718, "chunk_loss": 0.030379, "F": 0.18469, "G": 0.2417, "chunks": 189.12, "bytes_per_chunk": 5.414, "boundary_entropy": 0.25898, "router_noise": 0.0, "grad_norm": 0.4124, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 1212.1} |
| {"event": "train", "step": 4800, "phase": 2, "objective": "global", "lm_loss": 1.21589, "bpb": 1.75416, "chunk_loss": 0.030239, "F": 0.17979, "G": 0.23578, "chunks": 184.11, "bytes_per_chunk": 5.562, "boundary_entropy": 0.25451, "router_noise": 0.0, "grad_norm": 0.4046, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 1235.5} |
| {"event": "train", "step": 4900, "phase": 2, "objective": "global", "lm_loss": 1.2027, "bpb": 1.73513, "chunk_loss": 0.030221, "F": 0.17897, "G": 0.22841, "chunks": 183.27, "bytes_per_chunk": 5.588, "boundary_entropy": 0.24344, "router_noise": 0.0, "grad_norm": 0.4189, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 1258.1} |
| {"event": "probe", "step": 5000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2593.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.148997, "cps": 20.257812, "mean_p": 0.192499, "boundary_entropy": 0.230741, "bpb": 1.885057}, "de": {"chunks": 3628.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176691, "cps": 28.34375, "mean_p": 0.198225, "boundary_entropy": 0.182505, "bpb": 2.018872}, "vi": {"chunks": 4208.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.228261, "cps": 32.875, "mean_p": 0.205532, "boundary_entropy": 0.171698, "bpb": 1.690972}, "tr": {"chunks": 3145.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.173728, "cps": 24.570312, "mean_p": 0.206823, "boundary_entropy": 0.264704, "bpb": 2.009813}, "fi": {"chunks": 3303.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.175589, "cps": 25.804688, "mean_p": 0.198291, "boundary_entropy": 0.197913, "bpb": 2.136361}, "ru": {"chunks": 5728.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.302269, "cps": 44.75, "mean_p": 0.208328, "boundary_entropy": 0.235023, "bpb": 1.208642}, "el": {"chunks": 6200.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.3021, "cps": 48.4375, "mean_p": 0.210061, "boundary_entropy": 0.272095, "bpb": 1.226557}, "hi": {"chunks": 6142.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.365052, "cps": 47.984375, "mean_p": 0.321077, "boundary_entropy": 0.468649, "bpb": 1.012831}, "ta": {"chunks": 7985.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.42712, "cps": 62.382812, "mean_p": 0.329881, "boundary_entropy": 0.463048, "bpb": 0.857783}, "th": {"chunks": 7112.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.442151, "cps": 55.5625, "mean_p": 0.353797, "boundary_entropy": 0.512127, "bpb": 0.975436}, "zh": {"chunks": 3958.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.670166, "cps": 30.921875, "mean_p": 0.246513, "boundary_entropy": 0.100891, "bpb": 2.753145}, "my": {"chunks": 7389.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.389962, "cps": 57.726562, "mean_p": 0.365171, "boundary_entropy": 0.576957, "bpb": 0.977239}}, "bpb": {"en": 1.88506, "de": 2.01887, "vi": 1.69097, "tr": 2.00981, "fi": 2.13636, "ru": 1.20864, "el": 1.22656, "hi": 1.01283, "ta": 0.85778, "th": 0.97544, "zh": 2.75314, "my": 0.97724}, "gini_cpc": 0.249562, "gini_cps": 0.196966, "bpic": 6.251421, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1858683403156254, "vi": 1.5319838696973795, "tr": 1.1659816128214695, "fi": 1.1784693552727794, "ru": 2.0286886451955595, "el": 2.0275540846734486, "hi": 2.4500578709739163, "ta": 2.8666261243805846, "th": 2.9675106492324512, "zh": 4.497839464373658, "my": 2.6172420779186525}, "t": 1285.7} |
| {"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.89187, "de": 2.0129, "vi": 1.70023, "tr": 2.00183, "fi": 2.15319, "ru": 1.20718, "el": 1.22438, "hi": 1.02105, "ta": 0.85957, "th": 0.97318, "zh": 2.76144, "my": 0.98619}, "bytes_per_chunk": {"en": 6.085, "de": 5.528, "vi": 5.52, "tr": 6.146, "fi": 5.649, "ru": 5.847, "el": 5.949, "hi": 6.817, "ta": 6.264, "th": 6.249, "zh": 3.981, "my": 7.27}, "bpb_mean": 1.56608, "t": 1292.9} |
| {"event": "train", "step": 5000, "phase": 2, "objective": "global", "lm_loss": 1.15338, "bpb": 1.66398, "chunk_loss": 0.030213, "F": 0.17656, "G": 0.2371, "chunks": 180.8, "bytes_per_chunk": 5.664, "boundary_entropy": 0.25714, "router_noise": 0.0, "grad_norm": 0.4243, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.349, "alpha": 0.03, "t": 1293.1} |
| {"event": "train", "step": 5100, "phase": 2, "objective": "global", "lm_loss": 1.13917, "bpb": 1.64348, "chunk_loss": 0.030286, "F": 0.18236, "G": 0.25226, "chunks": 186.73, "bytes_per_chunk": 5.484, "boundary_entropy": 0.27919, "router_noise": 0.0, "grad_norm": 0.4061, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 1316.9} |
| {"event": "train", "step": 5200, "phase": 2, "objective": "global", "lm_loss": 1.27032, "bpb": 1.83268, "chunk_loss": 0.030363, "F": 0.18213, "G": 0.22972, "chunks": 186.5, "bytes_per_chunk": 5.491, "boundary_entropy": 0.23169, "router_noise": 0.0, "grad_norm": 0.4326, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 1339.5} |
| {"event": "train", "step": 5300, "phase": 2, "objective": "global", "lm_loss": 1.16738, "bpb": 1.68418, "chunk_loss": 0.030379, "F": 0.18561, "G": 0.23633, "chunks": 190.06, "bytes_per_chunk": 5.388, "boundary_entropy": 0.23966, "router_noise": 0.0, "grad_norm": 0.4151, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 1362.1} |
| {"event": "train", "step": 5400, "phase": 2, "objective": "global", "lm_loss": 1.11641, "bpb": 1.61064, "chunk_loss": 0.030289, "F": 0.17592, "G": 0.23316, "chunks": 180.14, "bytes_per_chunk": 5.684, "boundary_entropy": 0.25216, "router_noise": 0.0, "grad_norm": 0.4116, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 1384.8} |
| {"event": "train", "step": 5500, "phase": 2, "objective": "global", "lm_loss": 1.1302, "bpb": 1.63054, "chunk_loss": 0.03024, "F": 0.17775, "G": 0.23363, "chunks": 182.02, "bytes_per_chunk": 5.626, "boundary_entropy": 0.25263, "router_noise": 0.0, "grad_norm": 0.4093, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 1407.6} |
| {"event": "train", "step": 5600, "phase": 2, "objective": "global", "lm_loss": 1.16091, "bpb": 1.67483, "chunk_loss": 0.030233, "F": 0.17639, "G": 0.22826, "chunks": 180.62, "bytes_per_chunk": 5.669, "boundary_entropy": 0.23764, "router_noise": 0.0, "grad_norm": 0.4177, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 1430.3} |
| {"event": "train", "step": 5700, "phase": 2, "objective": "global", "lm_loss": 1.24338, "bpb": 1.79382, "chunk_loss": 0.030457, "F": 0.18889, "G": 0.23062, "chunks": 193.42, "bytes_per_chunk": 5.294, "boundary_entropy": 0.2154, "router_noise": 0.0, "grad_norm": 0.4146, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 1453.1} |
| {"event": "train", "step": 5800, "phase": 2, "objective": "global", "lm_loss": 1.24786, "bpb": 1.80028, "chunk_loss": 0.030459, "F": 0.18675, "G": 0.22769, "chunks": 191.23, "bytes_per_chunk": 5.355, "boundary_entropy": 0.20429, "router_noise": 0.0, "grad_norm": 0.4337, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 1475.7} |
| {"event": "train", "step": 5900, "phase": 2, "objective": "global", "lm_loss": 1.19444, "bpb": 1.72321, "chunk_loss": 0.030287, "F": 0.18413, "G": 0.23829, "chunks": 188.55, "bytes_per_chunk": 5.431, "boundary_entropy": 0.24155, "router_noise": 0.0, "grad_norm": 0.4331, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.222, "alpha": 0.03, "t": 1497.9} |
| {"event": "probe", "step": 6000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2679.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.153939, "cps": 20.929688, "mean_p": 0.190737, "boundary_entropy": 0.203989, "bpb": 1.84384}, "de": {"chunks": 3851.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.187552, "cps": 30.085938, "mean_p": 0.20298, "boundary_entropy": 0.143784, "bpb": 1.964257}, "vi": {"chunks": 4269.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.23157, "cps": 33.351562, "mean_p": 0.201417, "boundary_entropy": 0.152201, "bpb": 1.644266}, "tr": {"chunks": 3290.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181738, "cps": 25.703125, "mean_p": 0.209236, "boundary_entropy": 0.247231, "bpb": 1.952506}, "fi": {"chunks": 3371.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.179204, "cps": 26.335938, "mean_p": 0.197279, "boundary_entropy": 0.176276, "bpb": 2.075946}, "ru": {"chunks": 5620.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.29657, "cps": 43.90625, "mean_p": 0.195629, "boundary_entropy": 0.203708, "bpb": 1.172652}, "el": {"chunks": 6084.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.296448, "cps": 47.53125, "mean_p": 0.212639, "boundary_entropy": 0.272245, "bpb": 1.191743}, "hi": {"chunks": 6374.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.378841, "cps": 49.796875, "mean_p": 0.333516, "boundary_entropy": 0.472902, "bpb": 0.988686}, "ta": {"chunks": 8063.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.431292, "cps": 62.992188, "mean_p": 0.326651, "boundary_entropy": 0.458856, "bpb": 0.83513}, "th": {"chunks": 7143.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.444078, "cps": 55.804688, "mean_p": 0.350523, "boundary_entropy": 0.509488, "bpb": 0.948479}, "zh": {"chunks": 3787.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.641212, "cps": 29.585938, "mean_p": 0.242214, "boundary_entropy": 0.086876, "bpb": 2.702771}, "my": {"chunks": 8136.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.429386, "cps": 63.5625, "mean_p": 0.369373, "boundary_entropy": 0.570953, "bpb": 0.941679}}, "bpb": {"en": 1.84384, "de": 1.96426, "vi": 1.64427, "tr": 1.95251, "fi": 2.07595, "ru": 1.17265, "el": 1.19174, "hi": 0.98869, "ta": 0.83513, "th": 0.94848, "zh": 2.70277, "my": 0.94168}, "gini_cpc": 0.241051, "gini_cps": 0.199963, "bpic": 6.124132, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2183512635737985, "vi": 1.504299875079385, "tr": 1.1805836194640609, "fi": 1.164121407365529, "ru": 1.9265421173773145, "el": 1.9257493804645696, "hi": 2.4609817484180243, "ta": 2.8017062367640864, "th": 2.88476866144687, "zh": 4.165366971694282, "my": 2.7893240438084685}, "t": 1525.1} |
| {"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.8491, "de": 1.95951, "vi": 1.6539, "tr": 1.94554, "fi": 2.09165, "ru": 1.17038, "el": 1.19048, "hi": 0.9969, "ta": 0.83783, "th": 0.94527, "zh": 2.71226, "my": 0.9505}, "bytes_per_chunk": {"en": 5.814, "de": 5.231, "vi": 5.49, "tr": 5.824, "fi": 5.514, "ru": 5.986, "el": 6.027, "hi": 6.499, "ta": 6.193, "th": 6.222, "zh": 4.134, "my": 6.557}, "bpb_mean": 1.52528, "t": 1531.8} |
| {"event": "train", "step": 6000, "phase": 2, "objective": "global", "lm_loss": 1.19523, "bpb": 1.72435, "chunk_loss": 0.030272, "F": 0.18515, "G": 0.2458, "chunks": 189.59, "bytes_per_chunk": 5.401, "boundary_entropy": 0.24154, "router_noise": 0.0, "grad_norm": 0.3997, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.341, "alpha": 0.03, "t": 1532.0} |
| {"event": "train", "step": 6100, "phase": 2, "objective": "global", "lm_loss": 1.15539, "bpb": 1.66688, "chunk_loss": 0.030452, "F": 0.18561, "G": 0.23258, "chunks": 190.06, "bytes_per_chunk": 5.388, "boundary_entropy": 0.22144, "router_noise": 0.0, "grad_norm": 0.4258, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 1555.5} |
| {"event": "train", "step": 6200, "phase": 2, "objective": "global", "lm_loss": 1.09503, "bpb": 1.57979, "chunk_loss": 0.030176, "F": 0.177, "G": 0.23185, "chunks": 181.25, "bytes_per_chunk": 5.65, "boundary_entropy": 0.23487, "router_noise": 0.0, "grad_norm": 0.4233, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 1578.4} |
| {"event": "train", "step": 6300, "phase": 2, "objective": "global", "lm_loss": 1.15867, "bpb": 1.67161, "chunk_loss": 0.031103, "F": 0.19507, "G": 0.2437, "chunks": 199.75, "bytes_per_chunk": 5.126, "boundary_entropy": 0.21532, "router_noise": 0.0, "grad_norm": 0.4159, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 1602.1} |
| {"event": "train", "step": 6400, "phase": 2, "objective": "global", "lm_loss": 1.1149, "bpb": 1.60846, "chunk_loss": 0.030327, "F": 0.18239, "G": 0.23822, "chunks": 186.77, "bytes_per_chunk": 5.483, "boundary_entropy": 0.23324, "router_noise": 0.0, "grad_norm": 0.3909, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 1624.9} |
| {"event": "train", "step": 6500, "phase": 2, "objective": "global", "lm_loss": 1.10692, "bpb": 1.59694, "chunk_loss": 0.030251, "F": 0.17928, "G": 0.23658, "chunks": 183.58, "bytes_per_chunk": 5.578, "boundary_entropy": 0.24113, "router_noise": 0.0, "grad_norm": 0.4122, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 1647.7} |
| {"event": "train", "step": 6600, "phase": 2, "objective": "global", "lm_loss": 1.15399, "bpb": 1.66486, "chunk_loss": 0.030265, "F": 0.18263, "G": 0.22589, "chunks": 187.02, "bytes_per_chunk": 5.475, "boundary_entropy": 0.2059, "router_noise": 0.0, "grad_norm": 0.3943, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 1670.7} |
| {"event": "train", "step": 6700, "phase": 2, "objective": "global", "lm_loss": 1.22212, "bpb": 1.76315, "chunk_loss": 0.031153, "F": 0.19653, "G": 0.23385, "chunks": 201.25, "bytes_per_chunk": 5.088, "boundary_entropy": 0.19086, "router_noise": 0.0, "grad_norm": 0.4625, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.23, "alpha": 0.03, "t": 1693.7} |
| {"event": "train", "step": 6800, "phase": 2, "objective": "global", "lm_loss": 1.09182, "bpb": 1.57516, "chunk_loss": 0.030307, "F": 0.18019, "G": 0.23098, "chunks": 184.52, "bytes_per_chunk": 5.55, "boundary_entropy": 0.22187, "router_noise": 0.0, "grad_norm": 0.42, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.221, "alpha": 0.03, "t": 1715.8} |
| {"event": "train", "step": 6900, "phase": 2, "objective": "global", "lm_loss": 1.15227, "bpb": 1.66238, "chunk_loss": 0.030371, "F": 0.18575, "G": 0.23844, "chunks": 190.2, "bytes_per_chunk": 5.384, "boundary_entropy": 0.2204, "router_noise": 0.0, "grad_norm": 0.4104, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 1739.9} |
| {"event": "probe", "step": 7000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2738.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.157329, "cps": 21.390625, "mean_p": 0.185851, "boundary_entropy": 0.173144, "bpb": 1.81215}, "de": {"chunks": 3856.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.187795, "cps": 30.125, "mean_p": 0.199439, "boundary_entropy": 0.119983, "bpb": 1.924649}, "vi": {"chunks": 4367.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.236886, "cps": 34.117188, "mean_p": 0.204591, "boundary_entropy": 0.143557, "bpb": 1.597606}, "tr": {"chunks": 3205.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177042, "cps": 25.039062, "mean_p": 0.192326, "boundary_entropy": 0.212895, "bpb": 1.910876}, "fi": {"chunks": 3416.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.181596, "cps": 26.6875, "mean_p": 0.193877, "boundary_entropy": 0.151597, "bpb": 2.027686}, "ru": {"chunks": 5605.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.295778, "cps": 43.789062, "mean_p": 0.193229, "boundary_entropy": 0.185893, "bpb": 1.146598}, "el": {"chunks": 6216.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.30288, "cps": 48.5625, "mean_p": 0.211068, "boundary_entropy": 0.254471, "bpb": 1.157135}, "hi": {"chunks": 6971.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.414324, "cps": 54.460938, "mean_p": 0.340326, "boundary_entropy": 0.467391, "bpb": 0.956446}, "ta": {"chunks": 8321.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.445092, "cps": 65.007812, "mean_p": 0.334571, "boundary_entropy": 0.454213, "bpb": 0.815788}, "th": {"chunks": 7152.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.444638, "cps": 55.875, "mean_p": 0.34538, "boundary_entropy": 0.497966, "bpb": 0.925902}, "zh": {"chunks": 3780.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.640027, "cps": 29.53125, "mean_p": 0.23829, "boundary_entropy": 0.081155, "bpb": 2.654453}, "my": {"chunks": 9516.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.502217, "cps": 74.34375, "mean_p": 0.399983, "boundary_entropy": 0.582955, "bpb": 0.905913}}, "bpb": {"en": 1.81215, "de": 1.92465, "vi": 1.59761, "tr": 1.91088, "fi": 2.02769, "ru": 1.1466, "el": 1.15713, "hi": 0.95645, "ta": 0.81579, "th": 0.9259, "zh": 2.65445, "my": 0.90591}, "gini_cpc": 0.248903, "gini_cps": 0.218784, "bpic": 5.891362, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1936453058496546, "vi": 1.5056732209965997, "tr": 1.1252995853166647, "fi": 1.1542414201410447, "ru": 1.8799966657094234, "el": 1.925133436316012, "hi": 2.633484012907329, "ta": 2.8290506889359017, "th": 2.826162418288707, "zh": 4.068075772938442, "my": 3.1921385552061827}, "t": 1767.4} |
| {"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.81485, "de": 1.92011, "vi": 1.60791, "tr": 1.9031, "fi": 2.04441, "ru": 1.14453, "el": 1.15587, "hi": 0.96459, "ta": 0.81964, "th": 0.9219, "zh": 2.66346, "my": 0.91692}, "bytes_per_chunk": {"en": 5.701, "de": 5.213, "vi": 5.426, "tr": 6.003, "fi": 5.487, "ru": 5.985, "el": 5.929, "hi": 5.97, "ta": 6.008, "th": 6.206, "zh": 4.167, "my": 5.7}, "bpb_mean": 1.48977, "t": 1774.0} |
| {"event": "train", "step": 7000, "phase": 2, "objective": "global", "lm_loss": 1.16168, "bpb": 1.67595, "chunk_loss": 0.030292, "F": 0.18671, "G": 0.23167, "chunks": 191.19, "bytes_per_chunk": 5.356, "boundary_entropy": 0.20195, "router_noise": 0.0, "grad_norm": 0.4656, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.343, "alpha": 0.03, "t": 1774.2} |
| {"event": "train", "step": 7100, "phase": 2, "objective": "global", "lm_loss": 1.06196, "bpb": 1.53209, "chunk_loss": 0.030275, "F": 0.17955, "G": 0.23373, "chunks": 183.86, "bytes_per_chunk": 5.569, "boundary_entropy": 0.22637, "router_noise": 0.0, "grad_norm": 0.4105, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 1798.3} |
| {"event": "train", "step": 7200, "phase": 2, "objective": "global", "lm_loss": 1.11618, "bpb": 1.61031, "chunk_loss": 0.030197, "F": 0.17804, "G": 0.22437, "chunks": 182.31, "bytes_per_chunk": 5.617, "boundary_entropy": 0.20721, "router_noise": 0.0, "grad_norm": 0.4286, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 1821.0} |
| {"event": "train", "step": 7300, "phase": 2, "objective": "global", "lm_loss": 1.1625, "bpb": 1.67713, "chunk_loss": 0.030306, "F": 0.18185, "G": 0.22094, "chunks": 186.22, "bytes_per_chunk": 5.499, "boundary_entropy": 0.19963, "router_noise": 0.0, "grad_norm": 0.4273, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 1843.5} |
| {"event": "train", "step": 7400, "phase": 2, "objective": "global", "lm_loss": 1.11179, "bpb": 1.60398, "chunk_loss": 0.030186, "F": 0.1797, "G": 0.23107, "chunks": 184.02, "bytes_per_chunk": 5.565, "boundary_entropy": 0.21779, "router_noise": 0.0, "grad_norm": 0.4254, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1867.4} |
| {"event": "train", "step": 7500, "phase": 2, "objective": "global", "lm_loss": 1.15727, "bpb": 1.66958, "chunk_loss": 0.030345, "F": 0.18253, "G": 0.22548, "chunks": 186.91, "bytes_per_chunk": 5.479, "boundary_entropy": 0.19741, "router_noise": 0.0, "grad_norm": 0.4187, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 1890.1} |
| {"event": "train", "step": 7600, "phase": 2, "objective": "global", "lm_loss": 1.10093, "bpb": 1.58831, "chunk_loss": 0.030152, "F": 0.17737, "G": 0.22327, "chunks": 181.62, "bytes_per_chunk": 5.638, "boundary_entropy": 0.20004, "router_noise": 0.0, "grad_norm": 0.4224, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 1914.4} |
| {"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1936.9} |
| {"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.12363, "bpb": 1.62105, "chunk_loss": 0.030254, "F": 0.18196, "G": 0.23566, "chunks": 186.33, "bytes_per_chunk": 5.496, "boundary_entropy": 0.22026, "router_noise": 0.0, "grad_norm": 0.455, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 1937.1} |
| {"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.14006, "bpb": 1.64476, "chunk_loss": 0.0303, "F": 0.18408, "G": 0.21879, "chunks": 188.5, "bytes_per_chunk": 5.432, "boundary_entropy": 0.18119, "router_noise": 0.0, "grad_norm": 0.4122, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 1961.4} |
| {"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.04546, "bpb": 1.50828, "chunk_loss": 0.03027, "F": 0.18039, "G": 0.2386, "chunks": 184.72, "bytes_per_chunk": 5.544, "boundary_entropy": 0.22426, "router_noise": 0.0, "grad_norm": 0.4155, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1983.8} |
| {"event": "probe", "step": 8000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2835.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.162903, "cps": 22.148438, "mean_p": 0.185799, "boundary_entropy": 0.155883, "bpb": 1.780718}, "de": {"chunks": 3879.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.188915, "cps": 30.304688, "mean_p": 0.19711, "boundary_entropy": 0.105816, "bpb": 1.888216}, "vi": {"chunks": 4347.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.235801, "cps": 33.960938, "mean_p": 0.197368, "boundary_entropy": 0.123913, "bpb": 1.565918}, "tr": {"chunks": 3319.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18334, "cps": 25.929688, "mean_p": 0.195201, "boundary_entropy": 0.206895, "bpb": 1.866071}, "fi": {"chunks": 3386.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.180001, "cps": 26.453125, "mean_p": 0.187722, "boundary_entropy": 0.13153, "bpb": 1.985933}, "ru": {"chunks": 5632.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.297203, "cps": 44.0, "mean_p": 0.191822, "boundary_entropy": 0.178335, "bpb": 1.121896}, "el": {"chunks": 6187.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.301467, "cps": 48.335938, "mean_p": 0.203695, "boundary_entropy": 0.235805, "bpb": 1.131401}, "hi": {"chunks": 6983.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.415037, "cps": 54.554688, "mean_p": 0.339192, "boundary_entropy": 0.462503, "bpb": 0.936882}, "ta": {"chunks": 8474.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.453276, "cps": 66.203125, "mean_p": 0.336178, "boundary_entropy": 0.455276, "bpb": 0.793234}, "th": {"chunks": 7413.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.460864, "cps": 57.914062, "mean_p": 0.348542, "boundary_entropy": 0.492274, "bpb": 0.904686}, "zh": {"chunks": 3771.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.638503, "cps": 29.460938, "mean_p": 0.236619, "boundary_entropy": 0.073115, "bpb": 2.612786}, "my": {"chunks": 8484.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.447752, "cps": 66.28125, "mean_p": 0.376112, "boundary_entropy": 0.56126, "bpb": 0.884142}}, "bpb": {"en": 1.78072, "de": 1.88822, "vi": 1.56592, "tr": 1.86607, "fi": 1.98593, "ru": 1.1219, "el": 1.1314, "hi": 0.93688, "ta": 0.79323, "th": 0.90469, "zh": 2.61279, "my": 0.88414}, "gini_cpc": 0.241381, "gini_cps": 0.207325, "bpic": 5.930783, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1596806998258322, "vi": 1.4474966097097912, "tr": 1.1254540230842502, "fi": 1.1049589075850308, "ru": 1.824418589716743, "el": 1.8505904976401495, "hi": 2.547757132636776, "ta": 2.7824927866008373, "th": 2.829071944185404, "zh": 3.919531388579814, "my": 2.748579738700068}, "t": 2011.2} |
| {"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.78221, "de": 1.88481, "vi": 1.57722, "tr": 1.8604, "fi": 2.00286, "ru": 1.12039, "el": 1.13029, "hi": 0.94413, "ta": 0.7975, "th": 0.90161, "zh": 2.6242, "my": 0.89461}, "bytes_per_chunk": {"en": 5.627, "de": 5.198, "vi": 5.442, "tr": 5.733, "fi": 5.509, "ru": 5.967, "el": 5.975, "hi": 5.903, "ta": 5.854, "th": 6.103, "zh": 4.163, "my": 6.174}, "bpb_mean": 1.46002, "t": 2017.9} |
| {"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.05945, "bpb": 1.52846, "chunk_loss": 0.030284, "F": 0.18481, "G": 0.24112, "chunks": 189.25, "bytes_per_chunk": 5.411, "boundary_entropy": 0.22234, "router_noise": 0.0, "grad_norm": 0.4214, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.343, "alpha": 0.03, "t": 2018.1} |
| {"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.13875, "bpb": 1.64287, "chunk_loss": 0.030217, "F": 0.17923, "G": 0.2194, "chunks": 183.53, "bytes_per_chunk": 5.579, "boundary_entropy": 0.18905, "router_noise": 0.0, "grad_norm": 0.4345, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 2041.9} |
| {"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.1355, "bpb": 1.63818, "chunk_loss": 0.030324, "F": 0.1835, "G": 0.22064, "chunks": 187.91, "bytes_per_chunk": 5.45, "boundary_entropy": 0.18322, "router_noise": 0.0, "grad_norm": 0.4386, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.212, "alpha": 0.03, "t": 2063.2} |
| {"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.07868, "bpb": 1.55621, "chunk_loss": 0.030249, "F": 0.18306, "G": 0.22848, "chunks": 187.45, "bytes_per_chunk": 5.463, "boundary_entropy": 0.20584, "router_noise": 0.0, "grad_norm": 0.4144, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 2085.8} |
| {"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.21429, "bpb": 1.75185, "chunk_loss": 0.03046, "F": 0.19257, "G": 0.22324, "chunks": 197.19, "bytes_per_chunk": 5.193, "boundary_entropy": 0.1641, "router_noise": 0.0, "grad_norm": 0.4506, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 2108.5} |
| {"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.11411, "bpb": 1.60732, "chunk_loss": 0.030265, "F": 0.18201, "G": 0.23019, "chunks": 186.38, "bytes_per_chunk": 5.494, "boundary_entropy": 0.20203, "router_noise": 0.0, "grad_norm": 0.4267, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.213, "alpha": 0.03, "t": 2129.8} |
| {"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.12314, "bpb": 1.62036, "chunk_loss": 0.030324, "F": 0.18434, "G": 0.21604, "chunks": 188.77, "bytes_per_chunk": 5.425, "boundary_entropy": 0.17107, "router_noise": 0.0, "grad_norm": 0.4405, "lr": 6.33e-05, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 2152.2} |
| {"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 1.0935, "bpb": 1.57758, "chunk_loss": 0.030259, "F": 0.18068, "G": 0.22589, "chunks": 185.02, "bytes_per_chunk": 5.535, "boundary_entropy": 0.20045, "router_noise": 0.0, "grad_norm": 0.4561, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 2174.8} |
| {"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.15033, "bpb": 1.65957, "chunk_loss": 0.030553, "F": 0.1886, "G": 0.22493, "chunks": 193.12, "bytes_per_chunk": 5.302, "boundary_entropy": 0.18188, "router_noise": 0.0, "grad_norm": 0.4351, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.212, "alpha": 0.03, "t": 2196.0} |
| {"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.07961, "bpb": 1.55755, "chunk_loss": 0.030413, "F": 0.18909, "G": 0.24378, "chunks": 193.62, "bytes_per_chunk": 5.289, "boundary_entropy": 0.21474, "router_noise": 0.0, "grad_norm": 0.4357, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 2218.5} |
| {"event": "probe", "step": 9000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2756.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.158364, "cps": 21.53125, "mean_p": 0.1784, "boundary_entropy": 0.137747, "bpb": 1.758645}, "de": {"chunks": 3911.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.190474, "cps": 30.554688, "mean_p": 0.198202, "boundary_entropy": 0.093207, "bpb": 1.862445}, "vi": {"chunks": 4374.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.237266, "cps": 34.171875, "mean_p": 0.197356, "boundary_entropy": 0.116012, "bpb": 1.534269}, "tr": {"chunks": 3396.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.187593, "cps": 26.53125, "mean_p": 0.194779, "boundary_entropy": 0.193427, "bpb": 1.837094}, "fi": {"chunks": 3422.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.181915, "cps": 26.734375, "mean_p": 0.186887, "boundary_entropy": 0.118936, "bpb": 1.952335}, "ru": {"chunks": 5644.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.297836, "cps": 44.09375, "mean_p": 0.187454, "boundary_entropy": 0.161544, "bpb": 1.104828}, "el": {"chunks": 6427.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.313161, "cps": 50.210938, "mean_p": 0.205923, "boundary_entropy": 0.222939, "bpb": 1.109577}, "hi": {"chunks": 6937.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.412303, "cps": 54.195312, "mean_p": 0.333496, "boundary_entropy": 0.453257, "bpb": 0.91903}, "ta": {"chunks": 8377.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.448088, "cps": 65.445312, "mean_p": 0.341388, "boundary_entropy": 0.451632, "bpb": 0.777358}, "th": {"chunks": 7419.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.461237, "cps": 57.960938, "mean_p": 0.3403, "boundary_entropy": 0.479346, "bpb": 0.891755}, "zh": {"chunks": 3767.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.637826, "cps": 29.429688, "mean_p": 0.235479, "boundary_entropy": 0.066742, "bpb": 2.58132}, "my": {"chunks": 8942.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.471923, "cps": 69.859375, "mean_p": 0.382576, "boundary_entropy": 0.566958, "bpb": 0.864254}}, "bpb": {"en": 1.75865, "de": 1.86245, "vi": 1.53427, "tr": 1.83709, "fi": 1.95233, "ru": 1.10483, "el": 1.10958, "hi": 0.91903, "ta": 0.77736, "th": 0.89176, "zh": 2.58132, "my": 0.86425}, "gini_cpc": 0.242376, "gini_cps": 0.211, "bpic": 5.870724, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2027637092670465, "vi": 1.4982370884561653, "tr": 1.1845735661771575, "fi": 1.148716948263681, "ru": 1.8807137413788484, "el": 1.9774811926468283, "hi": 2.6035236587685495, "ta": 2.8294886286663425, "th": 2.9125219883664117, "zh": 4.027606977136237, "my": 2.9799995358161504}, "t": 2245.1} |
| {"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.75991, "de": 1.85976, "vi": 1.54562, "tr": 1.83183, "fi": 1.96948, "ru": 1.103, "el": 1.10938, "hi": 0.92803, "ta": 0.78233, "th": 0.88794, "zh": 2.59197, "my": 0.87326}, "bytes_per_chunk": {"en": 5.708, "de": 5.144, "vi": 5.417, "tr": 5.607, "fi": 5.479, "ru": 5.965, "el": 5.783, "hi": 5.98, "ta": 5.968, "th": 6.014, "zh": 4.173, "my": 5.882}, "bpb_mean": 1.43687, "t": 2252.4} |
| {"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.13975, "bpb": 1.64431, "chunk_loss": 0.030367, "F": 0.185, "G": 0.22045, "chunks": 189.44, "bytes_per_chunk": 5.405, "boundary_entropy": 0.17124, "router_noise": 0.0, "grad_norm": 0.4765, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.342, "alpha": 0.03, "t": 2252.8} |
| {"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.12428, "bpb": 1.622, "chunk_loss": 0.030337, "F": 0.18568, "G": 0.22447, "chunks": 190.14, "bytes_per_chunk": 5.385, "boundary_entropy": 0.17533, "router_noise": 0.0, "grad_norm": 0.4598, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 2276.4} |
| {"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 1.05092, "bpb": 1.51616, "chunk_loss": 0.030248, "F": 0.18169, "G": 0.226, "chunks": 186.05, "bytes_per_chunk": 5.504, "boundary_entropy": 0.18988, "router_noise": 0.0, "grad_norm": 0.4391, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 2300.6} |
| {"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 1.05978, "bpb": 1.52893, "chunk_loss": 0.030373, "F": 0.18539, "G": 0.23068, "chunks": 189.84, "bytes_per_chunk": 5.394, "boundary_entropy": 0.18985, "router_noise": 0.0, "grad_norm": 0.4371, "lr": 4.71e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 2324.7} |
| {"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.03925, "bpb": 1.49932, "chunk_loss": 0.030374, "F": 0.18817, "G": 0.24554, "chunks": 192.69, "bytes_per_chunk": 5.314, "boundary_entropy": 0.21911, "router_noise": 0.0, "grad_norm": 0.4289, "lr": 4.52e-05, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 2347.5} |
| {"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.12632, "bpb": 1.62493, "chunk_loss": 0.030234, "F": 0.18169, "G": 0.21703, "chunks": 186.05, "bytes_per_chunk": 5.504, "boundary_entropy": 0.17138, "router_noise": 0.0, "grad_norm": 0.4816, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 2371.6} |
| {"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 1.07463, "bpb": 1.55036, "chunk_loss": 0.030399, "F": 0.18681, "G": 0.22322, "chunks": 191.3, "bytes_per_chunk": 5.353, "boundary_entropy": 0.17636, "router_noise": 0.0, "grad_norm": 0.424, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.23, "alpha": 0.03, "t": 2394.5} |
| {"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 1.09665, "bpb": 1.58213, "chunk_loss": 0.030161, "F": 0.1788, "G": 0.21567, "chunks": 183.09, "bytes_per_chunk": 5.593, "boundary_entropy": 0.17495, "router_noise": 0.0, "grad_norm": 0.4467, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 2418.4} |
| {"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 1.01908, "bpb": 1.47022, "chunk_loss": 0.030234, "F": 0.18124, "G": 0.22693, "chunks": 185.59, "bytes_per_chunk": 5.517, "boundary_entropy": 0.18966, "router_noise": 0.0, "grad_norm": 0.4295, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 2441.0} |
| {"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.09334, "bpb": 1.57736, "chunk_loss": 0.030301, "F": 0.18393, "G": 0.21972, "chunks": 188.34, "bytes_per_chunk": 5.437, "boundary_entropy": 0.1719, "router_noise": 0.0, "grad_norm": 0.4352, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 2464.7} |
| {"event": "probe", "step": 10000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2856.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.16411, "cps": 22.3125, "mean_p": 0.179999, "boundary_entropy": 0.127824, "bpb": 1.742013}, "de": {"chunks": 3906.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.19023, "cps": 30.515625, "mean_p": 0.195724, "boundary_entropy": 0.085126, "bpb": 1.843103}, "vi": {"chunks": 4400.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.238676, "cps": 34.375, "mean_p": 0.196935, "boundary_entropy": 0.112514, "bpb": 1.515511}, "tr": {"chunks": 3351.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185107, "cps": 26.179688, "mean_p": 0.190062, "boundary_entropy": 0.185473, "bpb": 1.813518}, "fi": {"chunks": 3446.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.183191, "cps": 26.921875, "mean_p": 0.187442, "boundary_entropy": 0.11264, "bpb": 1.928447}, "ru": {"chunks": 5779.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.30496, "cps": 45.148438, "mean_p": 0.189582, "boundary_entropy": 0.162118, "bpb": 1.091117}, "el": {"chunks": 6322.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.308045, "cps": 49.390625, "mean_p": 0.203807, "boundary_entropy": 0.220476, "bpb": 1.096387}, "hi": {"chunks": 7151.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.425022, "cps": 55.867188, "mean_p": 0.331638, "boundary_entropy": 0.445673, "bpb": 0.907718}, "ta": {"chunks": 8359.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.447125, "cps": 65.304688, "mean_p": 0.334387, "boundary_entropy": 0.444348, "bpb": 0.766665}, "th": {"chunks": 7700.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.478707, "cps": 60.15625, "mean_p": 0.351494, "boundary_entropy": 0.483103, "bpb": 0.877919}, "zh": {"chunks": 3760.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.636641, "cps": 29.375, "mean_p": 0.234291, "boundary_entropy": 0.063629, "bpb": 2.55675}, "my": {"chunks": 8911.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.470287, "cps": 69.617188, "mean_p": 0.379719, "boundary_entropy": 0.561664, "bpb": 0.851669}}, "bpb": {"en": 1.74201, "de": 1.8431, "vi": 1.51551, "tr": 1.81352, "fi": 1.92845, "ru": 1.09112, "el": 1.09639, "hi": 0.90772, "ta": 0.76667, "th": 0.87792, "zh": 2.55675, "my": 0.85167}, "gini_cpc": 0.241816, "gini_cps": 0.211094, "bpic": 5.820066, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1591663061757114, "vi": 1.454371821959052, "tr": 1.1279498569576425, "fi": 1.116270137846218, "ru": 1.858272488414891, "el": 1.8770660873617098, "hi": 2.5898679560977436, "ta": 2.724549940329167, "th": 2.91699427672841, "zh": 3.879362107186254, "my": 2.8656885252517723}, "t": 2492.1} |
| {"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.74192, "de": 1.84112, "vi": 1.52774, "tr": 1.80679, "fi": 1.94567, "ru": 1.08988, "el": 1.09624, "hi": 0.91599, "ta": 0.77178, "th": 0.87495, "zh": 2.56674, "my": 0.86111}, "bytes_per_chunk": {"en": 5.586, "de": 5.162, "vi": 5.392, "tr": 5.691, "fi": 5.459, "ru": 5.856, "el": 5.906, "hi": 5.772, "ta": 5.997, "th": 5.861, "zh": 4.188, "my": 5.915}, "bpb_mean": 1.41999, "t": 2498.5} |
| {"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.10369, "bpb": 1.59229, "chunk_loss": 0.030288, "F": 0.18303, "G": 0.21996, "chunks": 187.42, "bytes_per_chunk": 5.464, "boundary_entropy": 0.17332, "router_noise": 0.0, "grad_norm": 0.4344, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.341, "alpha": 0.03, "t": 2498.7} |
| {"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 1.15826, "bpb": 1.67102, "chunk_loss": 0.030463, "F": 0.19069, "G": 0.22412, "chunks": 195.27, "bytes_per_chunk": 5.244, "boundary_entropy": 0.1639, "router_noise": 0.0, "grad_norm": 0.4517, "lr": 3.49e-05, "router_trainable": true, "sec_per_step": 0.229, "alpha": 0.03, "t": 2521.6} |
| {"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.0642, "bpb": 1.53531, "chunk_loss": 0.03041, "F": 0.18716, "G": 0.22174, "chunks": 191.66, "bytes_per_chunk": 5.343, "boundary_entropy": 0.165, "router_noise": 0.0, "grad_norm": 0.4684, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 2545.4} |
| {"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 1.06527, "bpb": 1.53686, "chunk_loss": 0.030229, "F": 0.17957, "G": 0.21369, "chunks": 183.88, "bytes_per_chunk": 5.569, "boundary_entropy": 0.16493, "router_noise": 0.0, "grad_norm": 0.4516, "lr": 3.3e-05, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 2569.1} |
| {"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.06111, "bpb": 1.53086, "chunk_loss": 0.030407, "F": 0.18645, "G": 0.22667, "chunks": 190.92, "bytes_per_chunk": 5.363, "boundary_entropy": 0.17772, "router_noise": 0.0, "grad_norm": 0.4525, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 2593.1} |
| {"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.02938, "bpb": 1.48507, "chunk_loss": 0.030307, "F": 0.18161, "G": 0.23678, "chunks": 185.97, "bytes_per_chunk": 5.506, "boundary_entropy": 0.21269, "router_noise": 0.0, "grad_norm": 0.4544, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 2616.8} |
| {"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 1.11353, "bpb": 1.60648, "chunk_loss": 0.030343, "F": 0.18657, "G": 0.22609, "chunks": 191.05, "bytes_per_chunk": 5.36, "boundary_entropy": 0.17665, "router_noise": 0.0, "grad_norm": 0.4403, "lr": 3.1e-05, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 2640.0} |
| {"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.09533, "bpb": 1.58022, "chunk_loss": 0.030484, "F": 0.18851, "G": 0.22858, "chunks": 193.03, "bytes_per_chunk": 5.305, "boundary_entropy": 0.18259, "router_noise": 0.0, "grad_norm": 0.4619, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 2662.9} |
| {"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.06811, "bpb": 1.54095, "chunk_loss": 0.030233, "F": 0.1801, "G": 0.22427, "chunks": 184.42, "bytes_per_chunk": 5.552, "boundary_entropy": 0.19051, "router_noise": 0.0, "grad_norm": 0.4662, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 2686.6} |
| {"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.13337, "bpb": 1.63511, "chunk_loss": 0.030476, "F": 0.18726, "G": 0.22291, "chunks": 191.75, "bytes_per_chunk": 5.34, "boundary_entropy": 0.1643, "router_noise": 0.0, "grad_norm": 0.4609, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 2710.5} |
| {"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2808.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.161351, "cps": 21.9375, "mean_p": 0.177803, "boundary_entropy": 0.123881, "bpb": 1.73157}, "de": {"chunks": 3883.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18911, "cps": 30.335938, "mean_p": 0.193506, "boundary_entropy": 0.081432, "bpb": 1.831054}, "vi": {"chunks": 4379.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.237537, "cps": 34.210938, "mean_p": 0.195739, "boundary_entropy": 0.106918, "bpb": 1.503171}, "tr": {"chunks": 3242.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179086, "cps": 25.328125, "mean_p": 0.184298, "boundary_entropy": 0.180344, "bpb": 1.80063}, "fi": {"chunks": 3403.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.180905, "cps": 26.585938, "mean_p": 0.185069, "boundary_entropy": 0.109966, "bpb": 1.914367}, "ru": {"chunks": 5729.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.302322, "cps": 44.757812, "mean_p": 0.189163, "boundary_entropy": 0.158817, "bpb": 1.084048}, "el": {"chunks": 6430.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.313307, "cps": 50.234375, "mean_p": 0.204967, "boundary_entropy": 0.212139, "bpb": 1.086471}, "hi": {"chunks": 6839.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.406478, "cps": 53.429688, "mean_p": 0.321523, "boundary_entropy": 0.440885, "bpb": 0.899474}, "ta": {"chunks": 8033.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.429687, "cps": 62.757812, "mean_p": 0.322283, "boundary_entropy": 0.437149, "bpb": 0.76276}, "th": {"chunks": 7543.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.468946, "cps": 58.929688, "mean_p": 0.33746, "boundary_entropy": 0.470847, "bpb": 0.871889}, "zh": {"chunks": 3769.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.638165, "cps": 29.445312, "mean_p": 0.234839, "boundary_entropy": 0.060697, "bpb": 2.545574}, "my": {"chunks": 8554.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.451446, "cps": 66.828125, "mean_p": 0.371021, "boundary_entropy": 0.555963, "bpb": 0.844456}}, "bpb": {"en": 1.73157, "de": 1.83105, "vi": 1.50317, "tr": 1.80063, "fi": 1.91437, "ru": 1.08405, "el": 1.08647, "hi": 0.89947, "ta": 0.76276, "th": 0.87189, "zh": 2.54557, "my": 0.84446}, "gini_cpc": 0.241435, "gini_cps": 0.205903, "bpic": 5.939779, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1720388297179316, "vi": 1.4721729036415774, "tr": 1.109914374110916, "fi": 1.121184492888549, "ru": 1.8736851927023432, "el": 1.9417671272297787, "hi": 2.519211020942076, "ta": 2.6630499607966844, "th": 2.9063643597448383, "zh": 3.955120426568954, "my": 2.7979044988623833}, "t": 2737.3} |
| {"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.73144, "de": 1.8293, "vi": 1.51538, "tr": 1.79427, "fi": 1.93166, "ru": 1.08283, "el": 1.08724, "hi": 0.90779, "ta": 0.7679, "th": 0.86862, "zh": 2.55525, "my": 0.85443}, "bytes_per_chunk": {"en": 5.592, "de": 5.191, "vi": 5.4, "tr": 5.839, "fi": 5.49, "ru": 5.89, "el": 5.786, "hi": 6.022, "ta": 6.174, "th": 6.007, "zh": 4.172, "my": 6.118}, "bpb_mean": 1.41051, "t": 2743.8} |
| {"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2808.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.161351, "cps": 21.9375, "mean_p": 0.177803, "boundary_entropy": 0.123881, "bpb": 1.73157}, "de": {"chunks": 3883.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18911, "cps": 30.335938, "mean_p": 0.193506, "boundary_entropy": 0.081432, "bpb": 1.831054}, "vi": {"chunks": 4379.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.237537, "cps": 34.210938, "mean_p": 0.195739, "boundary_entropy": 0.106918, "bpb": 1.503171}, "tr": {"chunks": 3242.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179086, "cps": 25.328125, "mean_p": 0.184298, "boundary_entropy": 0.180344, "bpb": 1.80063}, "fi": {"chunks": 3403.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.180905, "cps": 26.585938, "mean_p": 0.185069, "boundary_entropy": 0.109966, "bpb": 1.914367}, "ru": {"chunks": 5729.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.302322, "cps": 44.757812, "mean_p": 0.189163, "boundary_entropy": 0.158817, "bpb": 1.084048}, "el": {"chunks": 6430.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.313307, "cps": 50.234375, "mean_p": 0.204967, "boundary_entropy": 0.212139, "bpb": 1.086471}, "hi": {"chunks": 6839.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.406478, "cps": 53.429688, "mean_p": 0.321523, "boundary_entropy": 0.440885, "bpb": 0.899474}, "ta": {"chunks": 8033.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.429687, "cps": 62.757812, "mean_p": 0.322283, "boundary_entropy": 0.437149, "bpb": 0.76276}, "th": {"chunks": 7543.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.468946, "cps": 58.929688, "mean_p": 0.33746, "boundary_entropy": 0.470847, "bpb": 0.871889}, "zh": {"chunks": 3769.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.638165, "cps": 29.445312, "mean_p": 0.234839, "boundary_entropy": 0.060697, "bpb": 2.545574}, "my": {"chunks": 8554.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.451446, "cps": 66.828125, "mean_p": 0.371021, "boundary_entropy": 0.555963, "bpb": 0.844456}}, "bpb": {"en": 1.73157, "de": 1.83105, "vi": 1.50317, "tr": 1.80063, "fi": 1.91437, "ru": 1.08405, "el": 1.08647, "hi": 0.89947, "ta": 0.76276, "th": 0.87189, "zh": 2.54557, "my": 0.84446}, "gini_cpc": 0.241435, "gini_cps": 0.205903, "bpic": 5.939779, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1720388297179316, "vi": 1.4721729036415774, "tr": 1.109914374110916, "fi": 1.121184492888549, "ru": 1.8736851927023432, "el": 1.9417671272297787, "hi": 2.519211020942076, "ta": 2.6630499607966844, "th": 2.9063643597448383, "zh": 3.955120426568954, "my": 2.7979044988623833}, "t": 2747.6} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s1/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s1/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s1/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s1/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_s1/th0.70"], "t": 2751.5} |
| {"event": "final", "run": "exp32_local_s1", "condition": "A", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 0.1613514911222203, "cps": 21.9375, "bpb": 1.7265047131991618, "bytes_per_chunk": 5.603822187069122}, "de": {"cpc": 0.18911021282813031, "cps": 30.3359375, "bpb": 1.830438612383034, "bytes_per_chunk": 5.152834221971056}, "vi": {"cpc": 0.23753729319229727, "cps": 34.2109375, "bpb": 1.5175986125724954, "bytes_per_chunk": 5.3968244164800945}, "tr": {"cpc": 0.1790863392807822, "cps": 25.328125, "bpb": 1.7923800878723977, "bytes_per_chunk": 5.838136385907086}, "fi": {"cpc": 0.1809047897506778, "cps": 26.5859375, "bpb": 1.9390279421898589, "bytes_per_chunk": 5.51278600269179}, "ru": {"cpc": 0.30232189973614776, "cps": 44.7578125, "bpb": 1.0836855245323749, "bytes_per_chunk": 5.890181495441549}, "el": {"cpc": 0.3133070213906349, "cps": 50.234375, "bpb": 1.084785278342233, "bytes_per_chunk": 5.798234942823649}, "hi": {"cpc": 0.4064784546805349, "cps": 53.4296875, "bpb": 0.9085640183086692, "bytes_per_chunk": 6.025294634802246}, "ta": {"cpc": 0.4296870821075154, "cps": 62.7578125, "bpb": 0.7655357592313872, "bytes_per_chunk": 6.1524235330012385}, "th": {"cpc": 0.4689462231893068, "cps": 58.9296875, "bpb": 0.870829783209608, "bytes_per_chunk": 6.016547893643635}, "zh": {"cpc": 0.6381645783948527, "cps": 29.4453125, "bpb": 2.5609352228924527, "bytes_per_chunk": 4.183160061595909}, "my": {"cpc": 0.4514460629090141, "cps": 66.828125, "bpb": 0.852660729803785, "bytes_per_chunk": 6.11849829032898}}, "gini_cpc": 0.2414345530143711, "gini_cps": 0.2059034441486618, "bpic": 5.939778988423203, "seed": 1, "mean_boundary_entropy": 0.18706578016281128, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}], "t": 2766.1} |
|
|
