Text Generation
PyTorch
English
ScalarPotentialLMSARFMassLNMultiXi
research
semsimula
conservative-language-model
scalar-potential
lagrangian-mechanics
energy-based-model
physics-informed
splm
multi-channel-xi
tinystories
non-transformer
attention-free
constant-memory-inference
riemannian-geometry
riemannian-geodesics
damped-riemannian-geometry
Eval Results (legacy)
Upload training_log_8k.jsonl
Browse files- training_log_8k.jsonl +160 -0
training_log_8k.jsonl
ADDED
|
@@ -0,0 +1,160 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 50, "train_loss": 10.710833740234374, "lr": 6.25e-05, "grad_norm": 0.9682838320732117, "mass_mean": 1.4175937175750732, "mass_std": 0.18117541074752808, "mass_min": 1.1736551523208618, "mass_max": 2.138187885284424, "gamma": 0.30000001192092896, "xi_alphas": [0.24970565736293793, 0.49961602687835693, 0.749688446521759, 0.9499717950820923], "elapsed_sec": 12.864405870437622}
|
| 2 |
+
{"step": 100, "train_loss": 9.42455171585083, "lr": 0.000125, "grad_norm": 1.8334816694259644, "mass_mean": 1.4169409275054932, "mass_std": 0.1844644844532013, "mass_min": 1.1704672574996948, "mass_max": 2.1302497386932373, "gamma": 0.30000001192092896, "xi_alphas": [0.24942590296268463, 0.4995148181915283, 0.7500290274620056, 0.9501466155052185], "elapsed_sec": 24.92237401008606}
|
| 3 |
+
{"step": 150, "train_loss": 7.091764478683472, "lr": 0.0001875, "grad_norm": 1.124351978302002, "mass_mean": 1.4127107858657837, "mass_std": 0.1808931529521942, "mass_min": 1.169831395149231, "mass_max": 2.1279470920562744, "gamma": 0.30000001192092896, "xi_alphas": [0.24780765175819397, 0.49737486243247986, 0.7487191557884216, 0.9503628611564636], "elapsed_sec": 36.976197481155396}
|
| 4 |
+
{"step": 200, "train_loss": 5.776824855804444, "lr": 0.00025, "grad_norm": 1.5183931589126587, "mass_mean": 1.4047632217407227, "mass_std": 0.17364487051963806, "mass_min": 1.1675240993499756, "mass_max": 2.0633134841918945, "gamma": 0.30000001192092896, "xi_alphas": [0.24563640356063843, 0.4945879876613617, 0.7466389536857605, 0.9499326348304749], "elapsed_sec": 49.0284960269928}
|
| 5 |
+
{"step": 250, "train_loss": 5.218332147598266, "lr": 0.0003125, "grad_norm": 0.7288146018981934, "mass_mean": 1.4060076475143433, "mass_std": 0.17503400146961212, "mass_min": 1.1687674522399902, "mass_max": 2.1314537525177, "gamma": 0.30000001192092896, "xi_alphas": [0.24284406006336212, 0.4909031391143799, 0.7446586489677429, 0.9493913054466248], "elapsed_sec": 61.08041715621948}
|
| 6 |
+
{"step": 300, "train_loss": 4.688025569915771, "lr": 0.000375, "grad_norm": 0.6014610528945923, "mass_mean": 1.409529447555542, "mass_std": 0.1808263063430786, "mass_min": 1.1675745248794556, "mass_max": 2.1328794956207275, "gamma": 0.30000001192092896, "xi_alphas": [0.23976817727088928, 0.4874734580516815, 0.7437582015991211, 0.9490185976028442], "elapsed_sec": 73.1288697719574}
|
| 7 |
+
{"step": 350, "train_loss": 4.25768497467041, "lr": 0.00043750000000000006, "grad_norm": 0.8546426296234131, "mass_mean": 1.4106091260910034, "mass_std": 0.1819382905960083, "mass_min": 1.1673364639282227, "mass_max": 2.134350538253784, "gamma": 0.30000001192092896, "xi_alphas": [0.23659947514533997, 0.48521801829338074, 0.743736982345581, 0.9492318034172058], "elapsed_sec": 85.17746305465698}
|
| 8 |
+
{"step": 400, "train_loss": 3.9475585508346556, "lr": 0.0005, "grad_norm": 0.9556416273117065, "mass_mean": 1.411723017692566, "mass_std": 0.1832975596189499, "mass_min": 1.1673237085342407, "mass_max": 2.134643793106079, "gamma": 0.30000001192092896, "xi_alphas": [0.23344793915748596, 0.484361469745636, 0.744271457195282, 0.9491086006164551], "elapsed_sec": 97.2241849899292}
|
| 9 |
+
{"step": 450, "train_loss": 3.775391206741333, "lr": 0.0004999487185984905, "grad_norm": 0.730624258518219, "mass_mean": 1.4089395999908447, "mass_std": 0.18093320727348328, "mass_min": 1.167720079421997, "mass_max": 2.1362781524658203, "gamma": 0.30000001192092896, "xi_alphas": [0.23135249316692352, 0.48539999127388, 0.7448858618736267, 0.9489767551422119], "elapsed_sec": 112.55407667160034}
|
| 10 |
+
{"step": 500, "train_loss": 3.6441031980514524, "lr": 0.0004997906888557206, "grad_norm": 0.7579625844955444, "mass_mean": 1.413572072982788, "mass_std": 0.1821599304676056, "mass_min": 1.167608380317688, "mass_max": 2.101710081100464, "gamma": 0.30000001192092896, "xi_alphas": [0.2295292168855667, 0.48657989501953125, 0.7453075051307678, 0.9491447806358337], "elapsed_sec": 124.60110402107239}
|
| 11 |
+
{"step": 550, "train_loss": 3.5518261003494263, "lr": 0.0004995259569176092, "grad_norm": 0.7679300308227539, "mass_mean": 1.4087003469467163, "mass_std": 0.18013456463813782, "mass_min": 1.1676177978515625, "mass_max": 2.138892412185669, "gamma": 0.30000001192092896, "xi_alphas": [0.2281777411699295, 0.4883781671524048, 0.7461714744567871, 0.9495478868484497], "elapsed_sec": 136.6480040550232}
|
| 12 |
+
{"step": 600, "train_loss": 3.4652013874053953, "lr": 0.0004991546358687517, "grad_norm": 0.7766846418380737, "mass_mean": 1.4134891033172607, "mass_std": 0.1824030727148056, "mass_min": 1.1675206422805786, "mass_max": 2.1043317317962646, "gamma": 0.30000001192092896, "xi_alphas": [0.22761499881744385, 0.490164190530777, 0.7468090057373047, 0.9499388337135315], "elapsed_sec": 148.69352412223816}
|
| 13 |
+
{"step": 650, "train_loss": 3.394323019981384, "lr": 0.0004986768843250328, "grad_norm": 0.7676606774330139, "mass_mean": 1.4106507301330566, "mass_std": 0.17959433794021606, "mass_min": 1.1676863431930542, "mass_max": 2.1059317588806152, "gamma": 0.30000001192092896, "xi_alphas": [0.22702081501483917, 0.4918900430202484, 0.7473559379577637, 0.9502513408660889], "elapsed_sec": 160.74370408058167}
|
| 14 |
+
{"step": 700, "train_loss": 3.3144438123703, "lr": 0.0004980929063658709, "grad_norm": 0.7813747525215149, "mass_mean": 1.413272500038147, "mass_std": 0.18151448667049408, "mass_min": 1.1677641868591309, "mass_max": 2.143267869949341, "gamma": 0.30000001192092896, "xi_alphas": [0.22686295211315155, 0.4940818250179291, 0.7478107213973999, 0.9505071640014648], "elapsed_sec": 172.79203844070435}
|
| 15 |
+
{"step": 750, "train_loss": 3.2680360651016236, "lr": 0.0004974029514470429, "grad_norm": 0.7106603384017944, "mass_mean": 1.4154741764068604, "mass_std": 0.18199731409549713, "mass_min": 1.1677521467208862, "mass_max": 2.1444735527038574, "gamma": 0.30000001192092896, "xi_alphas": [0.2268262505531311, 0.4966575801372528, 0.7482381463050842, 0.9506319165229797], "elapsed_sec": 184.83882188796997}
|
| 16 |
+
{"step": 800, "train_loss": 3.2345056533813477, "lr": 0.0004966073142941239, "grad_norm": 0.6092336177825928, "mass_mean": 1.411515712738037, "mass_std": 0.18295760452747345, "mass_min": 1.1676464080810547, "mass_max": 2.1453261375427246, "gamma": 0.30000001192092896, "xi_alphas": [0.2265288531780243, 0.49871763586997986, 0.748778223991394, 0.9508890509605408], "elapsed_sec": 196.88716983795166}
|
| 17 |
+
{"step": 850, "train_loss": 3.1987895917892457, "lr": 0.0004957063347765917, "grad_norm": 0.6847230195999146, "mass_mean": 1.4192955493927002, "mass_std": 0.19048024713993073, "mass_min": 1.167596459388733, "mass_max": 2.146206855773926, "gamma": 0.30000001192092896, "xi_alphas": [0.22652092576026917, 0.5010342001914978, 0.7493801712989807, 0.9512712955474854], "elapsed_sec": 212.22000789642334}
|
| 18 |
+
{"step": 900, "train_loss": 3.148645339012146, "lr": 0.0004947003977626449, "grad_norm": 0.8420872688293457, "mass_mean": 1.4177525043487549, "mass_std": 0.1901637762784958, "mass_min": 1.1678310632705688, "mass_max": 2.147797107696533, "gamma": 0.30000001192092896, "xi_alphas": [0.2267524003982544, 0.503413200378418, 0.7503959536552429, 0.9515497088432312], "elapsed_sec": 224.26658296585083}
|
| 19 |
+
{"step": 950, "train_loss": 3.112411117553711, "lr": 0.0004935899329548007, "grad_norm": 0.7710803151130676, "mass_mean": 1.411266803741455, "mass_std": 0.18331919610500336, "mass_min": 1.1676422357559204, "mass_max": 2.147867441177368, "gamma": 0.30000001192092896, "xi_alphas": [0.22700271010398865, 0.5053178668022156, 0.7508543729782104, 0.9517343044281006], "elapsed_sec": 236.31347346305847}
|
| 20 |
+
{"step": 1000, "train_loss": 3.067617697715759, "lr": 0.0004923754147063402, "grad_norm": 0.635733962059021, "mass_mean": 1.4136836528778076, "mass_std": 0.18131142854690552, "mass_min": 1.167313814163208, "mass_max": 2.147975206375122, "gamma": 0.30000001192092896, "xi_alphas": [0.22722472250461578, 0.5075888633728027, 0.7514410614967346, 0.9519443511962891], "elapsed_sec": 248.3612174987793}
|
| 21 |
+
{"step": 1050, "train_loss": 3.0509614372253417, "lr": 0.0004910573618186806, "grad_norm": 0.732552170753479, "mass_mean": 1.4145317077636719, "mass_std": 0.18478912115097046, "mass_min": 1.1675443649291992, "mass_max": 2.149434804916382, "gamma": 0.30000001192092896, "xi_alphas": [0.22760677337646484, 0.5095449686050415, 0.7519340515136719, 0.952024519443512], "elapsed_sec": 260.40899109840393}
|
| 22 |
+
{"step": 1100, "train_loss": 3.0499465322494506, "lr": 0.0004896363373197609, "grad_norm": 0.7469215989112854, "mass_mean": 1.4152402877807617, "mass_std": 0.18534304201602936, "mass_min": 1.167373538017273, "mass_max": 2.1498334407806396, "gamma": 0.30000001192092896, "xi_alphas": [0.2279195487499237, 0.511699378490448, 0.7526265978813171, 0.9523733258247375], "elapsed_sec": 272.4507088661194}
|
| 23 |
+
{"step": 1150, "train_loss": 3.0049140214920045, "lr": 0.0004881129482235347, "grad_norm": 0.6452631950378418, "mass_mean": 1.422040581703186, "mass_std": 0.18992768228054047, "mass_min": 1.1670337915420532, "mass_max": 2.150092601776123, "gamma": 0.30000001192092896, "xi_alphas": [0.22801782190799713, 0.5136008858680725, 0.752997636795044, 0.9524611234664917], "elapsed_sec": 284.4893341064453}
|
| 24 |
+
{"step": 1200, "train_loss": 2.994425916671753, "lr": 0.0004864878452706749, "grad_norm": 0.6671186685562134, "mass_mean": 1.4191575050354004, "mass_std": 0.18851907551288605, "mass_min": 1.1669771671295166, "mass_max": 2.1507468223571777, "gamma": 0.30000001192092896, "xi_alphas": [0.22864319384098053, 0.5155860781669617, 0.7532868385314941, 0.9524447917938232], "elapsed_sec": 296.52639842033386}
|
| 25 |
+
{"step": 1250, "train_loss": 2.967375249862671, "lr": 0.0004847617226505988, "grad_norm": 0.7919197678565979, "mass_mean": 1.4160499572753906, "mass_std": 0.18703578412532806, "mass_min": 1.1667309999465942, "mass_max": 2.1509621143341064, "gamma": 0.30000001192092896, "xi_alphas": [0.22882376611232758, 0.5176374316215515, 0.7537884712219238, 0.9526219367980957], "elapsed_sec": 311.85505056381226}
|
| 26 |
+
{"step": 1300, "train_loss": 2.9395450496673585, "lr": 0.00048293531770493347, "grad_norm": 0.6504496335983276, "mass_mean": 1.4131519794464111, "mass_std": 0.18456992506980896, "mass_min": 1.1667211055755615, "mass_max": 2.151838541030884, "gamma": 0.30000001192092896, "xi_alphas": [0.22887234389781952, 0.5190824866294861, 0.754238486289978, 0.9526829719543457], "elapsed_sec": 323.8959755897522}
|
| 27 |
+
{"step": 1350, "train_loss": 2.928374466896057, "lr": 0.00048100941061254875, "grad_norm": 0.6478497982025146, "mass_mean": 1.4159281253814697, "mass_std": 0.18342596292495728, "mass_min": 1.1665433645248413, "mass_max": 2.1523597240448, "gamma": 0.30000001192092896, "xi_alphas": [0.22950932383537292, 0.5211201906204224, 0.7546830177307129, 0.9527530670166016], "elapsed_sec": 335.9385898113251}
|
| 28 |
+
{"step": 1400, "train_loss": 2.9023393535614015, "lr": 0.0004789848240562905, "grad_norm": 0.66156005859375, "mass_mean": 1.4113184213638306, "mass_std": 0.18156257271766663, "mass_min": 1.1664448976516724, "mass_max": 2.1534855365753174, "gamma": 0.30000001192092896, "xi_alphas": [0.23009569942951202, 0.5225571990013123, 0.7550778985023499, 0.9529501795768738], "elapsed_sec": 347.980318069458}
|
| 29 |
+
{"step": 1450, "train_loss": 2.8722211694717408, "lr": 0.00047686242287155823, "grad_norm": 0.6736901998519897, "mass_mean": 1.4156627655029297, "mass_std": 0.18407133221626282, "mass_min": 1.1661463975906372, "mass_max": 2.1537697315216064, "gamma": 0.30000001192092896, "xi_alphas": [0.23100239038467407, 0.524471640586853, 0.7557423114776611, 0.9530144929885864], "elapsed_sec": 360.01903796195984}
|
| 30 |
+
{"step": 1500, "train_loss": 2.8971552085876464, "lr": 0.00047464311367687663, "grad_norm": 0.6980342864990234, "mass_mean": 1.4166027307510376, "mass_std": 0.18655115365982056, "mass_min": 1.1659709215164185, "mass_max": 2.1546969413757324, "gamma": 0.30000001192092896, "xi_alphas": [0.23152029514312744, 0.525783360004425, 0.7560976147651672, 0.9531355500221252], "elapsed_sec": 372.0605046749115}
|
| 31 |
+
{"step": 1550, "train_loss": 2.877853956222534, "lr": 0.0004723278444866184, "grad_norm": 0.6732420921325684, "mass_mean": 1.4157471656799316, "mass_std": 0.1864490509033203, "mass_min": 1.166011095046997, "mass_max": 2.155735969543457, "gamma": 0.30000001192092896, "xi_alphas": [0.2319328635931015, 0.527226984500885, 0.7568075656890869, 0.9534279704093933], "elapsed_sec": 384.09823656082153}
|
| 32 |
+
{"step": 1600, "train_loss": 2.8506073474884035, "lr": 0.0004699176043060446, "grad_norm": 0.6752021312713623, "mass_mean": 1.4160867929458618, "mass_std": 0.19048257172107697, "mass_min": 1.1657923460006714, "mass_max": 2.1563880443573, "gamma": 0.30000001192092896, "xi_alphas": [0.2323075830936432, 0.528435468673706, 0.7573753595352173, 0.9536482691764832], "elapsed_sec": 396.1383550167084}
|
| 33 |
+
{"step": 1650, "train_loss": 2.8509015369415285, "lr": 0.00046741342270883437, "grad_norm": 0.6233871579170227, "mass_mean": 1.4135499000549316, "mass_std": 0.18643790483474731, "mass_min": 1.165041446685791, "mass_max": 2.15571665763855, "gamma": 0.30000001192092896, "xi_alphas": [0.2327430248260498, 0.5300411581993103, 0.757632315158844, 0.9538801312446594], "elapsed_sec": 411.4735703468323}
|
| 34 |
+
{"step": 1700, "train_loss": 2.8131636095047, "lr": 0.000464816369397286, "grad_norm": 0.6583446264266968, "mass_mean": 1.4164799451828003, "mass_std": 0.1889234483242035, "mass_min": 1.164900302886963, "mass_max": 2.156386613845825, "gamma": 0.30000001192092896, "xi_alphas": [0.23355993628501892, 0.5321005582809448, 0.7583050727844238, 0.9540621042251587], "elapsed_sec": 423.52097368240356}
|
| 35 |
+
{"step": 1750, "train_loss": 2.8037508010864256, "lr": 0.00046212755374537594, "grad_norm": 0.6464589238166809, "mass_mean": 1.4140607118606567, "mass_std": 0.18550755083560944, "mass_min": 1.164468765258789, "mass_max": 2.156606912612915, "gamma": 0.30000001192092896, "xi_alphas": [0.23415061831474304, 0.5337773561477661, 0.7585653066635132, 0.9540983438491821], "elapsed_sec": 435.5689260959625}
|
| 36 |
+
{"step": 1800, "train_loss": 2.8055628728866577, "lr": 0.0004593481243248713, "grad_norm": 0.5985241532325745, "mass_mean": 1.4083009958267212, "mass_std": 0.1799345463514328, "mass_min": 1.1640698909759521, "mass_max": 2.1567342281341553, "gamma": 0.30000001192092896, "xi_alphas": [0.23445068299770355, 0.5352583527565002, 0.759380578994751, 0.9540552496910095], "elapsed_sec": 447.61445665359497}
|
| 37 |
+
{"step": 1850, "train_loss": 2.776411271095276, "lr": 0.00045647926841469876, "grad_norm": 0.6420720219612122, "mass_mean": 1.4139044284820557, "mass_std": 0.1903570294380188, "mass_min": 1.1638624668121338, "mass_max": 2.1573755741119385, "gamma": 0.30000001192092896, "xi_alphas": [0.23513588309288025, 0.5368234515190125, 0.759910523891449, 0.9542032480239868], "elapsed_sec": 459.66126704216003}
|
| 38 |
+
{"step": 1900, "train_loss": 2.7826163005828857, "lr": 0.000453522211493779, "grad_norm": 0.6480479836463928, "mass_mean": 1.4153497219085693, "mass_std": 0.1923118531703949, "mass_min": 1.163523554801941, "mass_max": 2.157683849334717, "gamma": 0.30000001192092896, "xi_alphas": [0.23567213118076324, 0.5379819869995117, 0.760291576385498, 0.9543280005455017], "elapsed_sec": 471.7068750858307}
|
| 39 |
+
{"step": 1950, "train_loss": 2.7631636571884157, "lr": 0.00045047821671754276, "grad_norm": 0.5833965539932251, "mass_mean": 1.4153172969818115, "mass_std": 0.18921034038066864, "mass_min": 1.163025975227356, "mass_max": 2.157550573348999, "gamma": 0.30000001192092896, "xi_alphas": [0.23623435199260712, 0.5389108061790466, 0.7607093453407288, 0.9544969797134399], "elapsed_sec": 483.75455474853516}
|
| 40 |
+
{"step": 2000, "train_loss": 2.736636338233948, "lr": 0.0004473485843783537, "grad_norm": 0.630669891834259, "mass_mean": 1.4197648763656616, "mass_std": 0.19357115030288696, "mass_min": 1.1628143787384033, "mass_max": 2.1577706336975098, "gamma": 0.30000001192092896, "xi_alphas": [0.2368144989013672, 0.5404422879219055, 0.7612862586975098, 0.9546716213226318], "elapsed_sec": 495.80084466934204}
|
| 41 |
+
{"step": 2050, "train_loss": 2.74223162651062, "lr": 0.00044413465135006704, "grad_norm": 0.6576671600341797, "mass_mean": 1.4111905097961426, "mass_std": 0.18680506944656372, "mass_min": 1.1624459028244019, "mass_max": 2.1578078269958496, "gamma": 0.30000001192092896, "xi_alphas": [0.23667456209659576, 0.5416415929794312, 0.7619953155517578, 0.9548693895339966], "elapsed_sec": 511.13305020332336}
|
| 42 |
+
{"step": 2100, "train_loss": 2.7253036642074586, "lr": 0.0004408377905169619, "grad_norm": 0.6857367753982544, "mass_mean": 1.4191486835479736, "mass_std": 0.192742720246315, "mass_min": 1.1620959043502808, "mass_max": 2.1577606201171875, "gamma": 0.30000001192092896, "xi_alphas": [0.23705145716667175, 0.5430116653442383, 0.7625554800033569, 0.9550303220748901], "elapsed_sec": 523.1798477172852}
|
| 43 |
+
{"step": 2150, "train_loss": 2.744077067375183, "lr": 0.0004374594101872913, "grad_norm": 0.6678242087364197, "mass_mean": 1.4114651679992676, "mass_std": 0.19012418389320374, "mass_min": 1.1615310907363892, "mass_max": 2.1573376655578613, "gamma": 0.30000001192092896, "xi_alphas": [0.23757828772068024, 0.5439454913139343, 0.7629664540290833, 0.9551858305931091], "elapsed_sec": 535.2264339923859}
|
| 44 |
+
{"step": 2200, "train_loss": 2.7188238525390624, "lr": 0.00043400095349169996, "grad_norm": 0.6841303706169128, "mass_mean": 1.4092342853546143, "mass_std": 0.18959562480449677, "mass_min": 1.1610773801803589, "mass_max": 2.157228469848633, "gamma": 0.30000001192092896, "xi_alphas": [0.23721042275428772, 0.5447936058044434, 0.7634981274604797, 0.9552717208862305], "elapsed_sec": 547.2729926109314}
|
| 45 |
+
{"step": 2250, "train_loss": 2.7248430585861207, "lr": 0.0004304638977667675, "grad_norm": 0.6921964883804321, "mass_mean": 1.414386510848999, "mass_std": 0.1892378032207489, "mass_min": 1.1607245206832886, "mass_max": 2.157384157180786, "gamma": 0.30000001192092896, "xi_alphas": [0.23740527033805847, 0.5461157560348511, 0.7638437747955322, 0.9552867412567139], "elapsed_sec": 559.3212027549744}
|
| 46 |
+
{"step": 2300, "train_loss": 2.7070662546157838, "lr": 0.0004268497539239402, "grad_norm": 0.6516231298446655, "mass_mean": 1.4225709438323975, "mass_std": 0.19885550439357758, "mass_min": 1.16044020652771, "mass_max": 2.157514810562134, "gamma": 0.30000001192092896, "xi_alphas": [0.23779581487178802, 0.5465686917304993, 0.7640372514724731, 0.9553521275520325], "elapsed_sec": 571.3708555698395}
|
| 47 |
+
{"step": 2350, "train_loss": 2.6858531427383423, "lr": 0.0004231600658041194, "grad_norm": 0.7233913540840149, "mass_mean": 1.4170650243759155, "mass_std": 0.1940237283706665, "mass_min": 1.1601238250732422, "mass_max": 2.1577441692352295, "gamma": 0.30000001192092896, "xi_alphas": [0.23807653784751892, 0.547373354434967, 0.7647578120231628, 0.9554088115692139], "elapsed_sec": 583.4177281856537}
|
| 48 |
+
{"step": 2400, "train_loss": 2.6901263189315796, "lr": 0.0004193964095181856, "grad_norm": 0.65864098072052, "mass_mean": 1.4095525741577148, "mass_std": 0.18953651189804077, "mass_min": 1.1594982147216797, "mass_max": 2.157188892364502, "gamma": 0.30000001192092896, "xi_alphas": [0.23810990154743195, 0.5488593578338623, 0.7651115655899048, 0.9555315971374512], "elapsed_sec": 595.4619357585907}
|
| 49 |
+
{"step": 2450, "train_loss": 2.6644471406936647, "lr": 0.00041556039277373605, "grad_norm": 0.6763158440589905, "mass_mean": 1.406935214996338, "mass_std": 0.18825151026248932, "mass_min": 1.158647894859314, "mass_max": 2.119032621383667, "gamma": 0.30000001192092896, "xi_alphas": [0.23826761543750763, 0.549780011177063, 0.7658705711364746, 0.9556534290313721], "elapsed_sec": 610.7835988998413}
|
| 50 |
+
{"step": 2500, "train_loss": 2.687615084648132, "lr": 0.0004116536541883265, "grad_norm": 0.6825906038284302, "mass_mean": 1.4096286296844482, "mass_std": 0.1881411373615265, "mass_min": 1.1584457159042358, "mass_max": 2.1562447547912598, "gamma": 0.30000001192092896, "xi_alphas": [0.23813672363758087, 0.5507972240447998, 0.7664183378219604, 0.9556832313537598], "elapsed_sec": 622.8267521858215}
|
| 51 |
+
{"step": 2550, "train_loss": 2.6573412132263186, "lr": 0.0004076778625895085, "grad_norm": 0.6510889530181885, "mass_mean": 1.4139782190322876, "mass_std": 0.19365887343883514, "mass_min": 1.1584030389785767, "mass_max": 2.1570706367492676, "gamma": 0.30000001192092896, "xi_alphas": [0.23864905536174774, 0.5516353249549866, 0.7669681310653687, 0.9557558298110962], "elapsed_sec": 634.8734786510468}
|
| 52 |
+
{"step": 2600, "train_loss": 2.664740591049194, "lr": 0.0004036347163019624, "grad_norm": 0.6903866529464722, "mass_mean": 1.4147069454193115, "mass_std": 0.1914646327495575, "mass_min": 1.1578150987625122, "mass_max": 2.1566357612609863, "gamma": 0.30000001192092896, "xi_alphas": [0.23944421112537384, 0.5528238415718079, 0.7675026059150696, 0.9559381604194641], "elapsed_sec": 646.9176943302155}
|
| 53 |
+
{"step": 2650, "train_loss": 2.6360889673233032, "lr": 0.00039952594242203014, "grad_norm": 0.6416949033737183, "mass_mean": 1.4032799005508423, "mass_std": 0.18518002331256866, "mass_min": 1.1574313640594482, "mass_max": 2.1565396785736084, "gamma": 0.30000001192092896, "xi_alphas": [0.23919489979743958, 0.5533547401428223, 0.7677310109138489, 0.95602947473526], "elapsed_sec": 658.9633233547211}
|
| 54 |
+
{"step": 2700, "train_loss": 2.640004105567932, "lr": 0.0003953532960799577, "grad_norm": 0.6376296281814575, "mass_mean": 1.4067085981369019, "mass_std": 0.18512345850467682, "mass_min": 1.156988501548767, "mass_max": 2.156214714050293, "gamma": 0.30000001192092896, "xi_alphas": [0.23926541209220886, 0.554172933101654, 0.7684805989265442, 0.9562491774559021], "elapsed_sec": 671.0097932815552}
|
| 55 |
+
{"step": 2750, "train_loss": 2.6255293130874633, "lr": 0.000391118559690162, "grad_norm": 0.6398818492889404, "mass_mean": 1.4020700454711914, "mass_std": 0.18420979380607605, "mass_min": 1.1566271781921387, "mass_max": 2.156520366668701, "gamma": 0.30000001192092896, "xi_alphas": [0.23940058052539825, 0.5546916723251343, 0.7686693072319031, 0.9563309550285339], "elapsed_sec": 683.053297996521}
|
| 56 |
+
{"step": 2800, "train_loss": 2.6130700969696044, "lr": 0.000386823542189844, "grad_norm": 0.6864411234855652, "mass_mean": 1.4076392650604248, "mass_std": 0.19044040143489838, "mass_min": 1.1563822031021118, "mass_max": 2.156726837158203, "gamma": 0.30000001192092896, "xi_alphas": [0.23912616074085236, 0.5550875663757324, 0.7689499258995056, 0.9563734531402588], "elapsed_sec": 695.0977432727814}
|
| 57 |
+
{"step": 2850, "train_loss": 2.625071477890015, "lr": 0.0003824700782662709, "grad_norm": 0.7078414559364319, "mass_mean": 1.4109859466552734, "mass_std": 0.1928011029958725, "mass_min": 1.1558068990707397, "mass_max": 2.15604829788208, "gamma": 0.30000001192092896, "xi_alphas": [0.2396467924118042, 0.5558207035064697, 0.7693738341331482, 0.9565103054046631], "elapsed_sec": 710.4257435798645}
|
| 58 |
+
{"step": 2900, "train_loss": 2.6103227090835572, "lr": 0.00037806002757305945, "grad_norm": 0.6718555092811584, "mass_mean": 1.4034377336502075, "mass_std": 0.1838555634021759, "mass_min": 1.155241847038269, "mass_max": 2.155397415161133, "gamma": 0.30000001192092896, "xi_alphas": [0.23994481563568115, 0.5561223030090332, 0.7699478268623352, 0.9565250277519226], "elapsed_sec": 722.4689931869507}
|
| 59 |
+
{"step": 2950, "train_loss": 2.6036356687545776, "lr": 0.0003735952739357946, "grad_norm": 0.6451295018196106, "mass_mean": 1.404793381690979, "mass_std": 0.1874324381351471, "mass_min": 1.1550490856170654, "mass_max": 2.155693769454956, "gamma": 0.30000001192092896, "xi_alphas": [0.2400139421224594, 0.5567435622215271, 0.7701931595802307, 0.9565309286117554], "elapsed_sec": 734.5149445533752}
|
| 60 |
+
{"step": 3000, "train_loss": 2.6073888111114503, "lr": 0.0003690777245473223, "grad_norm": 0.7790660262107849, "mass_mean": 1.4056974649429321, "mass_std": 0.18714071810245514, "mass_min": 1.154615879058838, "mass_max": 2.1555347442626953, "gamma": 0.30000001192092896, "xi_alphas": [0.24036622047424316, 0.5570517778396606, 0.770297110080719, 0.9565351009368896], "elapsed_sec": 746.5540733337402}
|
| 61 |
+
{"step": 3050, "train_loss": 2.576490387916565, "lr": 0.00036450930915306063, "grad_norm": 0.6355403661727905, "mass_mean": 1.4066424369812012, "mass_std": 0.1906476616859436, "mass_min": 1.1542150974273682, "mass_max": 2.155407667160034, "gamma": 0.30000001192092896, "xi_alphas": [0.24078918993473053, 0.5581237077713013, 0.770884096622467, 0.9565325975418091], "elapsed_sec": 758.5972454547882}
|
| 62 |
+
{"step": 3100, "train_loss": 2.568551721572876, "lr": 0.00035989197922667754, "grad_norm": 0.6463014483451843, "mass_mean": 1.4093470573425293, "mass_std": 0.1876067817211151, "mass_min": 1.1539198160171509, "mass_max": 2.155468225479126, "gamma": 0.30000001192092896, "xi_alphas": [0.2406838983297348, 0.5587382912635803, 0.7713608145713806, 0.9565272331237793], "elapsed_sec": 770.6398487091064}
|
| 63 |
+
{"step": 3150, "train_loss": 2.5737042236328125, "lr": 0.0003552277071364868, "grad_norm": 0.6288552284240723, "mass_mean": 1.408245325088501, "mass_std": 0.18951405584812164, "mass_min": 1.153349757194519, "mass_max": 2.1548633575439453, "gamma": 0.30000001192092896, "xi_alphas": [0.24053913354873657, 0.5587605834007263, 0.7716845870018005, 0.956638514995575], "elapsed_sec": 782.6869616508484}
|
| 64 |
+
{"step": 3200, "train_loss": 2.5607348442077638, "lr": 0.00035051848530291826, "grad_norm": 0.6213096380233765, "mass_mean": 1.4074952602386475, "mass_std": 0.18658043444156647, "mass_min": 1.1531420946121216, "mass_max": 2.1550047397613525, "gamma": 0.30000001192092896, "xi_alphas": [0.24105821549892426, 0.5595161318778992, 0.772022545337677, 0.9567014575004578], "elapsed_sec": 794.7322280406952}
|
| 65 |
+
{"step": 3250, "train_loss": 2.5672537326812743, "lr": 0.00034576632534742306, "grad_norm": 0.6409585475921631, "mass_mean": 1.4038612842559814, "mass_std": 0.186325341463089, "mass_min": 1.1528925895690918, "mass_max": 2.1552083492279053, "gamma": 0.30000001192092896, "xi_alphas": [0.24109232425689697, 0.5597496032714844, 0.7722247242927551, 0.9568249583244324], "elapsed_sec": 810.0534636974335}
|
| 66 |
+
{"step": 3300, "train_loss": 2.559220404624939, "lr": 0.00034097325723317644, "grad_norm": 0.7230061888694763, "mass_mean": 1.401374101638794, "mass_std": 0.18336988985538483, "mass_min": 1.1522572040557861, "mass_max": 2.1543047428131104, "gamma": 0.30000001192092896, "xi_alphas": [0.24122236669063568, 0.5602351427078247, 0.7726011276245117, 0.9568972587585449], "elapsed_sec": 822.0951797962189}
|
| 67 |
+
{"step": 3350, "train_loss": 2.562384467124939, "lr": 0.00033614132839794537, "grad_norm": 0.6589606404304504, "mass_mean": 1.4000264406204224, "mass_std": 0.18472260236740112, "mass_min": 1.1518094539642334, "mass_max": 2.1540894508361816, "gamma": 0.30000001192092896, "xi_alphas": [0.24144603312015533, 0.5603081583976746, 0.7726443409919739, 0.9569592475891113], "elapsed_sec": 834.1409585475922}
|
| 68 |
+
{"step": 3400, "train_loss": 2.539666619300842, "lr": 0.00033127260287949196, "grad_norm": 0.6301694512367249, "mass_mean": 1.4076941013336182, "mass_std": 0.19050860404968262, "mass_min": 1.1511560678482056, "mass_max": 2.1533219814300537, "gamma": 0.30000001192092896, "xi_alphas": [0.24159690737724304, 0.5609506964683533, 0.7730424404144287, 0.9570215940475464], "elapsed_sec": 846.1821830272675}
|
| 69 |
+
{"step": 3450, "train_loss": 2.5386516857147217, "lr": 0.00032636916043388614, "grad_norm": 0.6629562973976135, "mass_mean": 1.4008352756500244, "mass_std": 0.18694178760051727, "mass_min": 1.1507160663604736, "mass_max": 2.115689992904663, "gamma": 0.30000001192092896, "xi_alphas": [0.24132023751735687, 0.5607804656028748, 0.7731350660324097, 0.9569639563560486], "elapsed_sec": 858.2250123023987}
|
| 70 |
+
{"step": 3500, "train_loss": 2.5363344144821167, "lr": 0.00032143309564710345, "grad_norm": 0.6463574767112732, "mass_mean": 1.4022088050842285, "mass_std": 0.1897350698709488, "mass_min": 1.15045964717865, "mass_max": 2.1530184745788574, "gamma": 0.30000001192092896, "xi_alphas": [0.2407192587852478, 0.5611729621887207, 0.7733687162399292, 0.9569782018661499], "elapsed_sec": 870.2672526836395}
|
| 71 |
+
{"step": 3550, "train_loss": 2.5501129817962647, "lr": 0.0003164665170402881, "grad_norm": 0.7643730044364929, "mass_mean": 1.415751576423645, "mass_std": 0.19814376533031464, "mass_min": 1.150088906288147, "mass_max": 2.1526482105255127, "gamma": 0.30000001192092896, "xi_alphas": [0.24120232462882996, 0.5616384148597717, 0.773857057094574, 0.9570930004119873], "elapsed_sec": 882.3122043609619}
|
| 72 |
+
{"step": 3600, "train_loss": 2.515745897293091, "lr": 0.0003114715461690636, "grad_norm": 0.6643994450569153, "mass_mean": 1.3996036052703857, "mass_std": 0.18874584138393402, "mass_min": 1.1499533653259277, "mass_max": 2.1530354022979736, "gamma": 0.30000001192092896, "xi_alphas": [0.2412603795528412, 0.5619657635688782, 0.7743445634841919, 0.9571900963783264], "elapsed_sec": 894.3558006286621}
|
| 73 |
+
{"step": 3650, "train_loss": 2.511483564376831, "lr": 0.000306450316717276, "grad_norm": 0.6936405301094055, "mass_mean": 1.4053699970245361, "mass_std": 0.19398246705532074, "mass_min": 1.1493945121765137, "mass_max": 2.1521968841552734, "gamma": 0.30000001192092896, "xi_alphas": [0.24142219126224518, 0.5620796084403992, 0.7745555639266968, 0.9572893381118774], "elapsed_sec": 909.6830959320068}
|
| 74 |
+
{"step": 3700, "train_loss": 2.5248458671569822, "lr": 0.0003014049735855547, "grad_norm": 0.6516889929771423, "mass_mean": 1.403290033340454, "mass_std": 0.18986517190933228, "mass_min": 1.148711085319519, "mass_max": 2.15148663520813, "gamma": 0.30000001192092896, "xi_alphas": [0.24129901826381683, 0.5621029138565063, 0.7746524214744568, 0.9572323560714722], "elapsed_sec": 921.7296793460846}
|
| 75 |
+
{"step": 3750, "train_loss": 2.5206124544143678, "lr": 0.000296337671975084, "grad_norm": 0.6467770934104919, "mass_mean": 1.4015765190124512, "mass_std": 0.19187438488006592, "mass_min": 1.1482418775558472, "mass_max": 2.1510608196258545, "gamma": 0.30000001192092896, "xi_alphas": [0.241408571600914, 0.5626148581504822, 0.7749279141426086, 0.9572584629058838], "elapsed_sec": 933.77055311203}
|
| 76 |
+
{"step": 3800, "train_loss": 2.488644251823425, "lr": 0.0002912505764669728, "grad_norm": 0.6456804275512695, "mass_mean": 1.3974355459213257, "mass_std": 0.18688328564167023, "mass_min": 1.1476325988769531, "mass_max": 2.15027117729187, "gamma": 0.30000001192092896, "xi_alphas": [0.24145521223545074, 0.5626316070556641, 0.7749617695808411, 0.957253098487854], "elapsed_sec": 945.8144164085388}
|
| 77 |
+
{"step": 3850, "train_loss": 2.4984075021743775, "lr": 0.0002861458600976186, "grad_norm": 0.6627992391586304, "mass_mean": 1.396712303161621, "mass_std": 0.18726781010627747, "mass_min": 1.1473060846328735, "mass_max": 2.14998459815979, "gamma": 0.30000001192092896, "xi_alphas": [0.24132561683654785, 0.5623385906219482, 0.7752498984336853, 0.9573481678962708], "elapsed_sec": 957.8567774295807}
|
| 78 |
+
{"step": 3900, "train_loss": 2.494299063682556, "lr": 0.0002810257034304591, "grad_norm": 0.6645398736000061, "mass_mean": 1.396019458770752, "mass_std": 0.18605796992778778, "mass_min": 1.1469473838806152, "mass_max": 2.1496753692626953, "gamma": 0.30000001192092896, "xi_alphas": [0.24123315513134003, 0.5625745058059692, 0.7755181193351746, 0.9573421478271484], "elapsed_sec": 969.9020338058472}
|
| 79 |
+
{"step": 3950, "train_loss": 2.494398760795593, "lr": 0.0002758922936245088, "grad_norm": 0.6396831274032593, "mass_mean": 1.396317481994629, "mass_std": 0.1866597831249237, "mass_min": 1.146051049232483, "mass_max": 2.1483397483825684, "gamma": 0.30000001192092896, "xi_alphas": [0.2412148118019104, 0.5628124475479126, 0.7755597233772278, 0.9573342800140381], "elapsed_sec": 981.9503629207611}
|
| 80 |
+
{"step": 4000, "train_loss": 2.4778757524490356, "lr": 0.0002707478235000782, "grad_norm": 0.6739593744277954, "mass_mean": 1.3977783918380737, "mass_std": 0.18995489180088043, "mass_min": 1.145607352256775, "mass_max": 2.1477503776550293, "gamma": 0.30000001192092896, "xi_alphas": [0.24108600616455078, 0.5633942484855652, 0.7757781147956848, 0.9572867155075073], "elapsed_sec": 993.99791264534}
|
| 81 |
+
{"step": 4050, "train_loss": 2.4763082456588745, "lr": 0.0002655944906020745, "grad_norm": 0.6727248430252075, "mass_mean": 1.3980237245559692, "mass_std": 0.19149015843868256, "mass_min": 1.1452833414077759, "mass_max": 2.147456407546997, "gamma": 0.30000001192092896, "xi_alphas": [0.24106530845165253, 0.5635172724723816, 0.7759854793548584, 0.9573198556900024], "elapsed_sec": 1009.3344736099243}
|
| 82 |
+
{"step": 4100, "train_loss": 2.4800765323638916, "lr": 0.000260434496261284, "grad_norm": 0.6763479709625244, "mass_mean": 1.3951085805892944, "mass_std": 0.18948222696781158, "mass_min": 1.1446902751922607, "mass_max": 2.146531581878662, "gamma": 0.30000001192092896, "xi_alphas": [0.24051888287067413, 0.5634520053863525, 0.7764024138450623, 0.9573954343795776], "elapsed_sec": 1021.3831565380096}
|
| 83 |
+
{"step": 4150, "train_loss": 2.4630517578125, "lr": 0.0002552700446540379, "grad_norm": 0.6647515296936035, "mass_mean": 1.395348072052002, "mass_std": 0.1914258897304535, "mass_min": 1.1441714763641357, "mass_max": 2.1458587646484375, "gamma": 0.30000001192092896, "xi_alphas": [0.2404259592294693, 0.5641335844993591, 0.7769373655319214, 0.9573417901992798], "elapsed_sec": 1033.4273390769958}
|
| 84 |
+
{"step": 4200, "train_loss": 2.468280243873596, "lr": 0.0002501033418606619, "grad_norm": 0.6232271790504456, "mass_mean": 1.3931090831756592, "mass_std": 0.18417075276374817, "mass_min": 1.144016146659851, "mass_max": 2.146254777908325, "gamma": 0.30000001192092896, "xi_alphas": [0.24073855578899384, 0.5645049810409546, 0.7770383358001709, 0.9574196934700012], "elapsed_sec": 1045.476084947586}
|
| 85 |
+
{"step": 4250, "train_loss": 2.4729485082626343, "lr": 0.0002449365949231128, "grad_norm": 0.6835941672325134, "mass_mean": 1.3966901302337646, "mass_std": 0.1906009465456009, "mass_min": 1.1434906721115112, "mass_max": 2.1455347537994385, "gamma": 0.30000001192092896, "xi_alphas": [0.24080516397953033, 0.5644053220748901, 0.7771746516227722, 0.9574581980705261], "elapsed_sec": 1057.5267443656921}
|
| 86 |
+
{"step": 4300, "train_loss": 2.4704685497283934, "lr": 0.0002397720109022044, "grad_norm": 0.6170490384101868, "mass_mean": 1.3977785110473633, "mass_std": 0.19374310970306396, "mass_min": 1.1431689262390137, "mass_max": 2.145350694656372, "gamma": 0.30000001192092896, "xi_alphas": [0.24057690799236298, 0.5644688606262207, 0.7772825360298157, 0.957489013671875], "elapsed_sec": 1069.5726068019867}
|
| 87 |
+
{"step": 4350, "train_loss": 2.466846241950989, "lr": 0.00023461179593482493, "grad_norm": 0.7269933223724365, "mass_mean": 1.3986365795135498, "mass_std": 0.19222834706306458, "mass_min": 1.1424542665481567, "mass_max": 2.1443369388580322, "gamma": 0.30000001192092896, "xi_alphas": [0.24038977921009064, 0.5643216967582703, 0.7774296998977661, 0.9575079083442688], "elapsed_sec": 1081.6182458400726}
|
| 88 |
+
{"step": 4400, "train_loss": 2.4461480808258056, "lr": 0.00022945815429154993, "grad_norm": 0.6415592432022095, "mass_mean": 1.396254062652588, "mass_std": 0.18896795809268951, "mass_min": 1.1420530080795288, "mass_max": 2.1438210010528564, "gamma": 0.30000001192092896, "xi_alphas": [0.24034462869167328, 0.5646272301673889, 0.7776577472686768, 0.957511842250824], "elapsed_sec": 1093.6585302352905}
|
| 89 |
+
{"step": 4450, "train_loss": 2.475518488883972, "lr": 0.00022431328743505147, "grad_norm": 0.6400965452194214, "mass_mean": 1.3909931182861328, "mass_std": 0.18930862843990326, "mass_min": 1.1416815519332886, "mass_max": 2.1434526443481445, "gamma": 0.30000001192092896, "xi_alphas": [0.24019421637058258, 0.5643366575241089, 0.7776191234588623, 0.9575992822647095], "elapsed_sec": 1108.9801154136658}
|
| 90 |
+
{"step": 4500, "train_loss": 2.4503334760665894, "lr": 0.00021917939307970722, "grad_norm": 0.6399940252304077, "mass_mean": 1.389721393585205, "mass_std": 0.18672248721122742, "mass_min": 1.1407772302627563, "mass_max": 2.141695737838745, "gamma": 0.30000001192092896, "xi_alphas": [0.23980650305747986, 0.564163327217102, 0.777787983417511, 0.9576596021652222], "elapsed_sec": 1121.0251026153564}
|
| 91 |
+
{"step": 4550, "train_loss": 2.4326301431655883, "lr": 0.00021405866425281076, "grad_norm": 0.6305260062217712, "mass_mean": 1.3923313617706299, "mass_std": 0.19089335203170776, "mass_min": 1.1405469179153442, "mass_max": 2.141573667526245, "gamma": 0.30000001192092896, "xi_alphas": [0.23985466361045837, 0.5647352337837219, 0.7781181931495667, 0.9576584100723267], "elapsed_sec": 1133.0762507915497}
|
| 92 |
+
{"step": 4600, "train_loss": 2.437916884422302, "lr": 0.0002089532883577843, "grad_norm": 0.6196517944335938, "mass_mean": 1.384275197982788, "mass_std": 0.18106651306152344, "mass_min": 1.1401664018630981, "mass_max": 2.140972137451172, "gamma": 0.30000001192092896, "xi_alphas": [0.2395007610321045, 0.5645055770874023, 0.7780175805091858, 0.9576199650764465], "elapsed_sec": 1145.1250443458557}
|
| 93 |
+
{"step": 4650, "train_loss": 2.444408268928528, "lr": 0.00020386544623979264, "grad_norm": 0.6178796887397766, "mass_mean": 1.3925650119781494, "mass_std": 0.18865108489990234, "mass_min": 1.1401846408843994, "mass_max": 2.1414225101470947, "gamma": 0.30000001192092896, "xi_alphas": [0.23941121995449066, 0.5645480751991272, 0.7782779335975647, 0.9576817154884338], "elapsed_sec": 1157.171705007553}
|
| 94 |
+
{"step": 4700, "train_loss": 2.440186710357666, "lr": 0.00019879731125415983, "grad_norm": 0.6793596148490906, "mass_mean": 1.3950936794281006, "mass_std": 0.19212906062602997, "mass_min": 1.1391915082931519, "mass_max": 2.139622926712036, "gamma": 0.30000001192092896, "xi_alphas": [0.23927095532417297, 0.5644919872283936, 0.7783718705177307, 0.9577459096908569], "elapsed_sec": 1169.2204475402832}
|
| 95 |
+
{"step": 4750, "train_loss": 2.4123098373413088, "lr": 0.00019375104833798496, "grad_norm": 0.6201149821281433, "mass_mean": 1.3907593488693237, "mass_std": 0.19020381569862366, "mass_min": 1.1385819911956787, "mass_max": 2.1385576725006104, "gamma": 0.30000001192092896, "xi_alphas": [0.23912487924098969, 0.564548134803772, 0.7784080505371094, 0.9577162861824036], "elapsed_sec": 1181.2677783966064}
|
| 96 |
+
{"step": 4800, "train_loss": 2.443775372505188, "lr": 0.000188728813085353, "grad_norm": 0.6537901163101196, "mass_mean": 1.3885341882705688, "mass_std": 0.19039493799209595, "mass_min": 1.1381911039352417, "mass_max": 2.1380460262298584, "gamma": 0.30000001192092896, "xi_alphas": [0.2392953336238861, 0.564193844795227, 0.7782662510871887, 0.9577362537384033], "elapsed_sec": 1193.3160290718079}
|
| 97 |
+
{"step": 4850, "train_loss": 2.407232894897461, "lr": 0.00018373275082653808, "grad_norm": 0.6475687026977539, "mass_mean": 1.3962621688842773, "mass_std": 0.1914663463830948, "mass_min": 1.1379152536392212, "mass_max": 2.1377274990081787, "gamma": 0.30000001192092896, "xi_alphas": [0.2391398847103119, 0.5643151998519897, 0.7783728837966919, 0.9577777981758118], "elapsed_sec": 1208.6519660949707}
|
| 98 |
+
{"step": 4900, "train_loss": 2.4115958404541016, "lr": 0.00017876499571159, "grad_norm": 0.6412304639816284, "mass_mean": 1.394254446029663, "mass_std": 0.1917795091867447, "mass_min": 1.1374778747558594, "mass_max": 2.1370835304260254, "gamma": 0.30000001192092896, "xi_alphas": [0.2391776740550995, 0.5643866658210754, 0.7786238789558411, 0.9578651189804077], "elapsed_sec": 1220.6909210681915}
|
| 99 |
+
{"step": 4950, "train_loss": 2.4023128366470337, "lr": 0.0001738276697986973, "grad_norm": 0.6962430477142334, "mass_mean": 1.3879177570343018, "mass_std": 0.18473735451698303, "mass_min": 1.136900782585144, "mass_max": 2.1360840797424316, "gamma": 0.30000001192092896, "xi_alphas": [0.23917484283447266, 0.5644170045852661, 0.7787288427352905, 0.9578987956047058], "elapsed_sec": 1232.733253479004}
|
| 100 |
+
{"step": 5000, "train_loss": 2.4180736541748047, "lr": 0.00016892288214771488, "grad_norm": 0.6635139584541321, "mass_mean": 1.3891189098358154, "mass_std": 0.1910502016544342, "mass_min": 1.136370062828064, "mass_max": 2.1353118419647217, "gamma": 0.30000001192092896, "xi_alphas": [0.23895123600959778, 0.5644741654396057, 0.7788195610046387, 0.957885205745697], "elapsed_sec": 1244.7709484100342}
|
| 101 |
+
{"step": 5050, "train_loss": 2.397653789520264, "lr": 0.0001640527279192449, "grad_norm": 0.6332620978355408, "mass_mean": 1.3858160972595215, "mass_std": 0.18563680350780487, "mass_min": 1.1360609531402588, "mass_max": 2.1347270011901855, "gamma": 0.30000001192092896, "xi_alphas": [0.2389550507068634, 0.56452876329422, 0.779059648513794, 0.957894504070282], "elapsed_sec": 1256.8089780807495}
|
| 102 |
+
{"step": 5100, "train_loss": 2.3993259048461915, "lr": 0.00015921928747965412, "grad_norm": 0.6533944010734558, "mass_mean": 1.382611870765686, "mass_std": 0.18443764746189117, "mass_min": 1.1356964111328125, "mass_max": 2.1344072818756104, "gamma": 0.30000001192092896, "xi_alphas": [0.23867443203926086, 0.5644546747207642, 0.7790777683258057, 0.9579062461853027], "elapsed_sec": 1268.8495185375214}
|
| 103 |
+
{"step": 5150, "train_loss": 2.3896528339385985, "lr": 0.00015442462551241122, "grad_norm": 0.6678941249847412, "mass_mean": 1.389876365661621, "mass_std": 0.19036439061164856, "mass_min": 1.1352901458740234, "mass_max": 2.13373064994812, "gamma": 0.30000001192092896, "xi_alphas": [0.23854109644889832, 0.56443190574646, 0.7792742848396301, 0.9579315185546875], "elapsed_sec": 1280.889086008072}
|
| 104 |
+
{"step": 5200, "train_loss": 2.4020469093322756, "lr": 0.00014967079013612392, "grad_norm": 0.6776208281517029, "mass_mean": 1.385870099067688, "mass_std": 0.18924948573112488, "mass_min": 1.134748101234436, "mass_max": 2.1328132152557373, "gamma": 0.30000001192092896, "xi_alphas": [0.23836879432201385, 0.5645069479942322, 0.7793180346488953, 0.9579556584358215], "elapsed_sec": 1292.9292585849762}
|
| 105 |
+
{"step": 5250, "train_loss": 2.3775475454330444, "lr": 0.00014495981202965072, "grad_norm": 0.7160329818725586, "mass_mean": 1.383440613746643, "mass_std": 0.185940682888031, "mass_min": 1.1344016790390015, "mass_max": 2.132385730743408, "gamma": 0.30000001192092896, "xi_alphas": [0.23818416893482208, 0.5643402338027954, 0.779400646686554, 0.9579341411590576], "elapsed_sec": 1308.2507524490356}
|
| 106 |
+
{"step": 5300, "train_loss": 2.4065380764007567, "lr": 0.00014029370356466353, "grad_norm": 0.6683937907218933, "mass_mean": 1.3877664804458618, "mass_std": 0.1921834796667099, "mass_min": 1.1339818239212036, "mass_max": 2.131589889526367, "gamma": 0.30000001192092896, "xi_alphas": [0.23805606365203857, 0.5643436312675476, 0.7794761061668396, 0.9579334855079651], "elapsed_sec": 1320.2935075759888}
|
| 107 |
+
{"step": 5350, "train_loss": 2.3760496377944946, "lr": 0.00013567445794603018, "grad_norm": 0.6501907706260681, "mass_mean": 1.3779785633087158, "mass_std": 0.18100374937057495, "mass_min": 1.1336642503738403, "mass_max": 2.1311094760894775, "gamma": 0.30000001192092896, "xi_alphas": [0.23787571489810944, 0.5642873048782349, 0.779639482498169, 0.9579082131385803], "elapsed_sec": 1332.3367869853973}
|
| 108 |
+
{"step": 5400, "train_loss": 2.40710373878479, "lr": 0.00013110404836038365, "grad_norm": 0.7069498896598816, "mass_mean": 1.3875031471252441, "mass_std": 0.18947520852088928, "mass_min": 1.1333590745925903, "mass_max": 2.1306445598602295, "gamma": 0.30000001192092896, "xi_alphas": [0.237660214304924, 0.5641263723373413, 0.7797678112983704, 0.9579697847366333], "elapsed_sec": 1344.376229763031}
|
| 109 |
+
{"step": 5450, "train_loss": 2.4052069854736327, "lr": 0.00012658442713324347, "grad_norm": 0.6855981349945068, "mass_mean": 1.3883607387542725, "mass_std": 0.19207797944545746, "mass_min": 1.1326249837875366, "mass_max": 2.1290292739868164, "gamma": 0.30000001192092896, "xi_alphas": [0.23740261793136597, 0.5640828013420105, 0.7798908352851868, 0.9580326676368713], "elapsed_sec": 1356.4208097457886}
|
| 110 |
+
{"step": 5500, "train_loss": 2.390788493156433, "lr": 0.00012211752489504763, "grad_norm": 0.6341361403465271, "mass_mean": 1.3813470602035522, "mass_std": 0.18444472551345825, "mass_min": 1.1324735879898071, "mass_max": 2.1287970542907715, "gamma": 0.30000001192092896, "xi_alphas": [0.23721691966056824, 0.5640236735343933, 0.7800944447517395, 0.958082377910614], "elapsed_sec": 1368.4639592170715}
|
| 111 |
+
{"step": 5550, "train_loss": 2.3690747213363648, "lr": 0.00011770524975645239, "grad_norm": 0.6778855919837952, "mass_mean": 1.3822286128997803, "mass_std": 0.18623344600200653, "mass_min": 1.132056474685669, "mass_max": 2.128100633621216, "gamma": 0.30000001192092896, "xi_alphas": [0.23707181215286255, 0.5639477968215942, 0.7801977396011353, 0.9580932855606079], "elapsed_sec": 1380.510015487671}
|
| 112 |
+
{"step": 5600, "train_loss": 2.361921715736389, "lr": 0.00011334948649325089, "grad_norm": 0.667397677898407, "mass_mean": 1.3805723190307617, "mass_std": 0.1830095499753952, "mass_min": 1.1315438747406006, "mass_max": 2.127124786376953, "gamma": 0.30000001192092896, "xi_alphas": [0.23689059913158417, 0.5638990998268127, 0.7803668975830078, 0.9581379294395447], "elapsed_sec": 1392.5573360919952}
|
| 113 |
+
{"step": 5650, "train_loss": 2.3781905460357664, "lr": 0.0001090520957412606, "grad_norm": 0.7096372842788696, "mass_mean": 1.3824913501739502, "mass_std": 0.1903201788663864, "mass_min": 1.1312694549560547, "mass_max": 2.1268258094787598, "gamma": 0.30000001192092896, "xi_alphas": [0.23674756288528442, 0.5637490749359131, 0.7802444696426392, 0.9581693410873413], "elapsed_sec": 1407.8847250938416}
|
| 114 |
+
{"step": 5700, "train_loss": 2.37968701839447, "lr": 0.00010481491320152237, "grad_norm": 0.6635118126869202, "mass_mean": 1.3823018074035645, "mass_std": 0.18975257873535156, "mass_min": 1.1308743953704834, "mass_max": 2.126042604446411, "gamma": 0.30000001192092896, "xi_alphas": [0.23662270605564117, 0.5636773705482483, 0.7801873087882996, 0.9581543803215027], "elapsed_sec": 1419.9325969219208}
|
| 115 |
+
{"step": 5750, "train_loss": 2.3720221757888793, "lr": 0.00010063974885614999, "grad_norm": 0.6909535527229309, "mass_mean": 1.3796719312667847, "mass_std": 0.18610915541648865, "mass_min": 1.1303696632385254, "mass_max": 2.125086545944214, "gamma": 0.30000001192092896, "xi_alphas": [0.23653852939605713, 0.56365567445755, 0.7802801728248596, 0.9581860899925232], "elapsed_sec": 1431.9776635169983}
|
| 116 |
+
{"step": 5800, "train_loss": 2.3468648481369017, "lr": 9.652838619516696e-05, "grad_norm": 0.6918841600418091, "mass_mean": 1.3766560554504395, "mass_std": 0.18268625438213348, "mass_min": 1.1301286220550537, "mass_max": 2.124842882156372, "gamma": 0.30000001192092896, "xi_alphas": [0.23627911508083344, 0.5636196732521057, 0.7803155183792114, 0.9581575393676758], "elapsed_sec": 1444.0196261405945}
|
| 117 |
+
{"step": 5850, "train_loss": 2.343415479660034, "lr": 9.248258145465977e-05, "grad_norm": 0.7008745670318604, "mass_mean": 1.3802812099456787, "mass_std": 0.1921987533569336, "mass_min": 1.1296162605285645, "mass_max": 2.1238303184509277, "gamma": 0.30000001192092896, "xi_alphas": [0.23615436255931854, 0.5636155605316162, 0.780341625213623, 0.9581623673439026], "elapsed_sec": 1456.071626663208}
|
| 118 |
+
{"step": 5900, "train_loss": 2.349937090873718, "lr": 8.850406286657198e-05, "grad_norm": 0.6796526908874512, "mass_mean": 1.3823322057724, "mass_std": 0.18922695517539978, "mass_min": 1.1292656660079956, "mass_max": 2.1232216358184814, "gamma": 0.30000001192092896, "xi_alphas": [0.23611891269683838, 0.5636048913002014, 0.7803387641906738, 0.958184003829956], "elapsed_sec": 1468.118893623352}
|
| 119 |
+
{"step": 5950, "train_loss": 2.355285768508911, "lr": 8.459452992046224e-05, "grad_norm": 0.7244963049888611, "mass_mean": 1.3811211585998535, "mass_std": 0.18681849539279938, "mass_min": 1.1289262771606445, "mass_max": 2.1225998401641846, "gamma": 0.30000001192092896, "xi_alphas": [0.2358267456293106, 0.5632041096687317, 0.7804699540138245, 0.9582046270370483], "elapsed_sec": 1480.1682589054108}
|
| 120 |
+
{"step": 6000, "train_loss": 2.344728960990906, "lr": 8.075565263753929e-05, "grad_norm": 0.6774550080299377, "mass_mean": 1.3736581802368164, "mass_std": 0.1870976835489273, "mass_min": 1.1286828517913818, "mass_max": 2.1222355365753174, "gamma": 0.30000001192092896, "xi_alphas": [0.23585011065006256, 0.5631469488143921, 0.7805365324020386, 0.9582127928733826], "elapsed_sec": 1492.218517780304}
|
| 121 |
+
{"step": 6050, "train_loss": 2.358675799369812, "lr": 7.698907085728466e-05, "grad_norm": 0.7155231237411499, "mass_mean": 1.3758316040039062, "mass_std": 0.18912334740161896, "mass_min": 1.1283760070800781, "mass_max": 2.1216845512390137, "gamma": 0.30000001192092896, "xi_alphas": [0.2358061671257019, 0.5630794763565063, 0.780606746673584, 0.9582396149635315], "elapsed_sec": 1507.5566048622131}
|
| 122 |
+
{"step": 6100, "train_loss": 2.3418849658966066, "lr": 7.32963935369687e-05, "grad_norm": 0.6918966770172119, "mass_mean": 1.3799128532409668, "mass_std": 0.1885310411453247, "mass_min": 1.128076434135437, "mass_max": 2.121129035949707, "gamma": 0.30000001192092896, "xi_alphas": [0.23560062050819397, 0.5629249811172485, 0.7806670069694519, 0.9582622647285461], "elapsed_sec": 1519.6022644042969}
|
| 123 |
+
{"step": 6150, "train_loss": 2.3346440029144286, "lr": 6.967919806435793e-05, "grad_norm": 0.6389872431755066, "mass_mean": 1.3715214729309082, "mass_std": 0.18190789222717285, "mass_min": 1.1277892589569092, "mass_max": 2.1205852031707764, "gamma": 0.30000001192092896, "xi_alphas": [0.23539814352989197, 0.5628481507301331, 0.7806963920593262, 0.9582615494728088], "elapsed_sec": 1531.647228717804}
|
| 124 |
+
{"step": 6200, "train_loss": 2.3370265674591066, "lr": 6.61390295839077e-05, "grad_norm": 0.6738641858100891, "mass_mean": 1.3747421503067017, "mass_std": 0.1865377575159073, "mass_min": 1.1273871660232544, "mass_max": 2.1198208332061768, "gamma": 0.30000001192092896, "xi_alphas": [0.2353152334690094, 0.5627930760383606, 0.7807725667953491, 0.9582776427268982], "elapsed_sec": 1543.6933453083038}
|
| 125 |
+
{"step": 6250, "train_loss": 2.324606761932373, "lr": 6.267740033672855e-05, "grad_norm": 0.6728448867797852, "mass_mean": 1.3753405809402466, "mass_std": 0.18568404018878937, "mass_min": 1.1269946098327637, "mass_max": 2.119009017944336, "gamma": 0.30000001192092896, "xi_alphas": [0.23521795868873596, 0.5626643896102905, 0.7807649970054626, 0.9582740068435669], "elapsed_sec": 1555.7375416755676}
|
| 126 |
+
{"step": 6300, "train_loss": 2.344477906227112, "lr": 5.929578901460789e-05, "grad_norm": 0.6921743154525757, "mass_mean": 1.3860300779342651, "mass_std": 0.19181770086288452, "mass_min": 1.12679123878479, "mass_max": 2.1187150478363037, "gamma": 0.30000001192092896, "xi_alphas": [0.23509474098682404, 0.5625805258750916, 0.7808221578598022, 0.9582804441452026], "elapsed_sec": 1567.7812609672546}
|
| 127 |
+
{"step": 6350, "train_loss": 2.3353663444519044, "lr": 5.599564012836189e-05, "grad_norm": 0.6630412340164185, "mass_mean": 1.3717715740203857, "mass_std": 0.1866508275270462, "mass_min": 1.1264839172363281, "mass_max": 2.118100643157959, "gamma": 0.30000001192092896, "xi_alphas": [0.23496679961681366, 0.5623671412467957, 0.7808255553245544, 0.9582775235176086], "elapsed_sec": 1579.8267362117767}
|
| 128 |
+
{"step": 6400, "train_loss": 2.3531339979171753, "lr": 5.2778363390789426e-05, "grad_norm": 0.690943717956543, "mass_mean": 1.376152753829956, "mass_std": 0.1872827410697937, "mass_min": 1.1262866258621216, "mass_max": 2.117755889892578, "gamma": 0.30000001192092896, "xi_alphas": [0.2348826378583908, 0.5622847676277161, 0.780893087387085, 0.9582808017730713], "elapsed_sec": 1591.87286901474}
|
| 129 |
+
{"step": 6450, "train_loss": 2.3315902614593504, "lr": 4.964533311448991e-05, "grad_norm": 0.6437000632286072, "mass_mean": 1.3688137531280518, "mass_std": 0.1825641542673111, "mass_min": 1.1260675191879272, "mass_max": 2.117387056350708, "gamma": 0.30000001192092896, "xi_alphas": [0.23476316034793854, 0.562218964099884, 0.7809376120567322, 0.9583011269569397], "elapsed_sec": 1607.2040598392487}
|
| 130 |
+
{"step": 6500, "train_loss": 2.3336797094345094, "lr": 4.659788762480327e-05, "grad_norm": 0.6693686246871948, "mass_mean": 1.3799389600753784, "mass_std": 0.19223082065582275, "mass_min": 1.1258286237716675, "mass_max": 2.1169137954711914, "gamma": 0.30000001192092896, "xi_alphas": [0.23463991284370422, 0.5621395111083984, 0.78093022108078, 0.9583077430725098], "elapsed_sec": 1619.2512834072113}
|
| 131 |
+
{"step": 6550, "train_loss": 2.3454887008666994, "lr": 4.3637328688122065e-05, "grad_norm": 0.6522312164306641, "mass_mean": 1.3786200284957886, "mass_std": 0.19034676253795624, "mass_min": 1.1256448030471802, "mass_max": 2.1165480613708496, "gamma": 0.30000001192092896, "xi_alphas": [0.23461288213729858, 0.5621427893638611, 0.780890703201294, 0.9583122134208679], "elapsed_sec": 1631.2992639541626}
|
| 132 |
+
{"step": 6600, "train_loss": 2.339022326469421, "lr": 4.0764920955820656e-05, "grad_norm": 0.6821866631507874, "mass_mean": 1.3805702924728394, "mass_std": 0.18970640003681183, "mass_min": 1.125601887702942, "mass_max": 2.116474151611328, "gamma": 0.30000001192092896, "xi_alphas": [0.23450012505054474, 0.562017023563385, 0.7808924317359924, 0.9583160281181335], "elapsed_sec": 1643.3468978404999}
|
| 133 |
+
{"step": 6650, "train_loss": 2.320140995979309, "lr": 3.798189142403882e-05, "grad_norm": 0.704889714717865, "mass_mean": 1.374986171722412, "mass_std": 0.18897446990013123, "mass_min": 1.1253076791763306, "mass_max": 2.1158435344696045, "gamma": 0.30000001192092896, "xi_alphas": [0.23439854383468628, 0.5618451237678528, 0.780893087387085, 0.9583301544189453], "elapsed_sec": 1655.3933911323547}
|
| 134 |
+
{"step": 6700, "train_loss": 2.328099737167358, "lr": 3.5289428909549865e-05, "grad_norm": 0.6907082200050354, "mass_mean": 1.3780606985092163, "mass_std": 0.1903715282678604, "mass_min": 1.1250368356704712, "mass_max": 2.1153151988983154, "gamma": 0.30000001192092896, "xi_alphas": [0.23431965708732605, 0.5618900060653687, 0.78095543384552, 0.9583415389060974], "elapsed_sec": 1667.4379878044128}
|
| 135 |
+
{"step": 6750, "train_loss": 2.3497699737548827, "lr": 3.268868354193818e-05, "grad_norm": 0.6545105576515198, "mass_mean": 1.3755924701690674, "mass_std": 0.18945588171482086, "mass_min": 1.1247426271438599, "mass_max": 2.1147234439849854, "gamma": 0.30000001192092896, "xi_alphas": [0.2342512309551239, 0.5618326663970947, 0.780936062335968, 0.9583420157432556], "elapsed_sec": 1679.4858787059784}
|
| 136 |
+
{"step": 6800, "train_loss": 2.3191409587860106, "lr": 3.018076627230268e-05, "grad_norm": 0.6702688932418823, "mass_mean": 1.3742754459381104, "mass_std": 0.18841096758842468, "mass_min": 1.1246343851089478, "mass_max": 2.114577293395996, "gamma": 0.30000001192092896, "xi_alphas": [0.23419073224067688, 0.5617204904556274, 0.7809297442436218, 0.9583470225334167], "elapsed_sec": 1691.5315244197845}
|
| 137 |
+
{"step": 6850, "train_loss": 2.3447596788406373, "lr": 2.776674839869539e-05, "grad_norm": 0.6560508012771606, "mass_mean": 1.3745648860931396, "mass_std": 0.18535466492176056, "mass_min": 1.1243743896484375, "mass_max": 2.113996744155884, "gamma": 0.30000001192092896, "xi_alphas": [0.2340872585773468, 0.5616840720176697, 0.7809404730796814, 0.9583523869514465], "elapsed_sec": 1706.8522708415985}
|
| 138 |
+
{"step": 6900, "train_loss": 2.358266496658325, "lr": 2.544766110849911e-05, "grad_norm": 0.6503906846046448, "mass_mean": 1.376518964767456, "mass_std": 0.18984109163284302, "mass_min": 1.1243185997009277, "mass_max": 2.1139070987701416, "gamma": 0.30000001192092896, "xi_alphas": [0.23398461937904358, 0.5616374015808105, 0.7809559106826782, 0.9583579897880554], "elapsed_sec": 1718.8923690319061}
|
| 139 |
+
{"step": 6950, "train_loss": 2.348797764778137, "lr": 2.322449503793897e-05, "grad_norm": 0.6981320381164551, "mass_mean": 1.3712810277938843, "mass_std": 0.1854468435049057, "mass_min": 1.124185562133789, "mass_max": 2.113689661026001, "gamma": 0.30000001192092896, "xi_alphas": [0.23390062153339386, 0.5615124106407166, 0.7809725403785706, 0.9583633542060852], "elapsed_sec": 1730.935523033142}
|
| 140 |
+
{"step": 7000, "train_loss": 2.340227174758911, "lr": 2.1098199848915534e-05, "grad_norm": 0.6868717670440674, "mass_mean": 1.3793973922729492, "mass_std": 0.19127365946769714, "mass_min": 1.124101996421814, "mass_max": 2.1135637760162354, "gamma": 0.30000001192092896, "xi_alphas": [0.23385022580623627, 0.5614722371101379, 0.7809935808181763, 0.9583613872528076], "elapsed_sec": 1742.9804904460907}
|
| 141 |
+
{"step": 7050, "train_loss": 2.343977680206299, "lr": 1.9069683823341373e-05, "grad_norm": 0.6665303707122803, "mass_mean": 1.3753364086151123, "mass_std": 0.18912743031978607, "mass_min": 1.1240041255950928, "mass_max": 2.113372802734375, "gamma": 0.30000001192092896, "xi_alphas": [0.2337760329246521, 0.5613991022109985, 0.7809893488883972, 0.958368718624115], "elapsed_sec": 1755.0235164165497}
|
| 142 |
+
{"step": 7100, "train_loss": 2.339254312515259, "lr": 1.7139813475153476e-05, "grad_norm": 0.6851592659950256, "mass_mean": 1.3732528686523438, "mass_std": 0.1899390071630478, "mass_min": 1.1238943338394165, "mass_max": 2.1131415367126465, "gamma": 0.30000001192092896, "xi_alphas": [0.23377655446529388, 0.5613913536071777, 0.7810121178627014, 0.9583763480186462], "elapsed_sec": 1767.065702199936}
|
| 143 |
+
{"step": 7150, "train_loss": 2.314516520500183, "lr": 1.5309413180167487e-05, "grad_norm": 0.6556789875030518, "mass_mean": 1.372882604598999, "mass_std": 0.18678215146064758, "mass_min": 1.123813271522522, "mass_max": 2.113006114959717, "gamma": 0.30000001192092896, "xi_alphas": [0.23373064398765564, 0.5613178014755249, 0.7810277938842773, 0.958378255367279], "elapsed_sec": 1779.106957912445}
|
| 144 |
+
{"step": 7200, "train_loss": 2.3479841232299803, "lr": 1.3579264823932152e-05, "grad_norm": 0.684283971786499, "mass_mean": 1.3762527704238892, "mass_std": 0.1877031773328781, "mass_min": 1.123703122138977, "mass_max": 2.1127634048461914, "gamma": 0.30000001192092896, "xi_alphas": [0.23367439210414886, 0.5612765550613403, 0.781048059463501, 0.9583815336227417], "elapsed_sec": 1791.1483354568481}
|
| 145 |
+
{"step": 7250, "train_loss": 2.337676224708557, "lr": 1.1950107467734089e-05, "grad_norm": 0.6963022351264954, "mass_mean": 1.3663733005523682, "mass_std": 0.18131524324417114, "mass_min": 1.1235991716384888, "mass_max": 2.1125142574310303, "gamma": 0.30000001192092896, "xi_alphas": [0.2336505502462387, 0.5612300038337708, 0.7810488343238831, 0.958381175994873], "elapsed_sec": 1806.4671308994293}
|
| 146 |
+
{"step": 7300, "train_loss": 2.321146106719971, "lr": 1.0422637032895421e-05, "grad_norm": 0.6750226020812988, "mass_mean": 1.3718879222869873, "mass_std": 0.1866150051355362, "mass_min": 1.123534083366394, "mass_max": 2.112391948699951, "gamma": 0.30000001192092896, "xi_alphas": [0.23363682627677917, 0.5612058639526367, 0.7810466289520264, 0.9583843350410461], "elapsed_sec": 1818.505514383316}
|
| 147 |
+
{"step": 7350, "train_loss": 2.3112298393249513, "lr": 8.997506003499855e-06, "grad_norm": 0.6703722476959229, "mass_mean": 1.3756892681121826, "mass_std": 0.18888036906719208, "mass_min": 1.1235158443450928, "mass_max": 2.112380027770996, "gamma": 0.30000001192092896, "xi_alphas": [0.23361819982528687, 0.5611794590950012, 0.781049907207489, 0.9583821892738342], "elapsed_sec": 1830.548709154129}
|
| 148 |
+
{"step": 7400, "train_loss": 2.3276697969436646, "lr": 7.675323147673053e-06, "grad_norm": 0.6711066961288452, "mass_mean": 1.3779394626617432, "mass_std": 0.1895599067211151, "mass_min": 1.1234731674194336, "mass_max": 2.112313985824585, "gamma": 0.30000001192092896, "xi_alphas": [0.23358480632305145, 0.5611801743507385, 0.781059741973877, 0.9583827257156372], "elapsed_sec": 1842.5973794460297}
|
| 149 |
+
{"step": 7450, "train_loss": 2.303600902557373, "lr": 6.456653257537665e-06, "grad_norm": 0.6984114050865173, "mass_mean": 1.3807638883590698, "mass_std": 0.19214557111263275, "mass_min": 1.1234261989593506, "mass_max": 2.1122117042541504, "gamma": 0.30000001192092896, "xi_alphas": [0.23355892300605774, 0.5611635446548462, 0.7810595631599426, 0.9583845734596252], "elapsed_sec": 1854.6425378322601}
|
| 150 |
+
{"step": 7500, "train_loss": 2.306391882896423, "lr": 5.342016907952857e-06, "grad_norm": 0.6655080914497375, "mass_mean": 1.378909707069397, "mass_std": 0.19310471415519714, "mass_min": 1.1234036684036255, "mass_max": 2.112196445465088, "gamma": 0.30000001192092896, "xi_alphas": [0.23354105651378632, 0.5611318945884705, 0.7810553908348083, 0.9583882093429565], "elapsed_sec": 1866.685940504074}
|
| 151 |
+
{"step": 7550, "train_loss": 2.3259192895889282, "lr": 4.331890234142266e-06, "grad_norm": 0.6717228293418884, "mass_mean": 1.3710167407989502, "mass_std": 0.1883702278137207, "mass_min": 1.1233782768249512, "mass_max": 2.1121442317962646, "gamma": 0.30000001192092896, "xi_alphas": [0.23353320360183716, 0.561122715473175, 0.7810526490211487, 0.9583870768547058], "elapsed_sec": 1878.7329201698303}
|
| 152 |
+
{"step": 7600, "train_loss": 2.329669575691223, "lr": 3.4267047283051135e-06, "grad_norm": 0.6910347938537598, "mass_mean": 1.3747506141662598, "mass_std": 0.18841147422790527, "mass_min": 1.123335838317871, "mass_max": 2.112053871154785, "gamma": 0.30000001192092896, "xi_alphas": [0.2335200160741806, 0.5611228346824646, 0.7810591459274292, 0.9583878517150879], "elapsed_sec": 1890.779738664627}
|
| 153 |
+
{"step": 7650, "train_loss": 2.3131145429611206, "lr": 2.626847055297088e-06, "grad_norm": 0.7139986157417297, "mass_mean": 1.379054069519043, "mass_std": 0.19413639605045319, "mass_min": 1.123329520225525, "mass_max": 2.1120574474334717, "gamma": 0.30000001192092896, "xi_alphas": [0.23351261019706726, 0.5611178278923035, 0.781061053276062, 0.9583887457847595], "elapsed_sec": 1906.109662771225}
|
| 154 |
+
{"step": 7700, "train_loss": 2.332669205665588, "lr": 1.932658887460165e-06, "grad_norm": 0.6930757761001587, "mass_mean": 1.3747150897979736, "mass_std": 0.1905948370695114, "mass_min": 1.1233031749725342, "mass_max": 2.1119964122772217, "gamma": 0.30000001192092896, "xi_alphas": [0.23350340127944946, 0.5611103773117065, 0.781065046787262, 0.9583883881568909], "elapsed_sec": 1918.1558866500854}
|
| 155 |
+
{"step": 7750, "train_loss": 2.317634630203247, "lr": 1.3444367586716854e-06, "grad_norm": 0.6677200794219971, "mass_mean": 1.3750543594360352, "mass_std": 0.1905270665884018, "mass_min": 1.123300552368164, "mass_max": 2.1119933128356934, "gamma": 0.30000001192092896, "xi_alphas": [0.23349760472774506, 0.5611069202423096, 0.781067430973053, 0.9583892822265625], "elapsed_sec": 1930.202341079712}
|
| 156 |
+
{"step": 7800, "train_loss": 2.333816046714783, "lr": 8.624319376750456e-07, "grad_norm": 0.6659238934516907, "mass_mean": 1.3708887100219727, "mass_std": 0.18568763136863708, "mass_min": 1.1232837438583374, "mass_max": 2.1119544506073, "gamma": 0.30000001192092896, "xi_alphas": [0.23349544405937195, 0.5611060261726379, 0.7810678482055664, 0.958389401435852], "elapsed_sec": 1942.2508549690247}
|
| 157 |
+
{"step": 7850, "train_loss": 2.3559260082244875, "lr": 4.868503207459963e-07, "grad_norm": 0.6850001811981201, "mass_mean": 1.3745372295379639, "mass_std": 0.1897892951965332, "mass_min": 1.1232821941375732, "mass_max": 2.1119537353515625, "gamma": 0.30000001192092896, "xi_alphas": [0.23349300026893616, 0.5611035227775574, 0.7810688614845276, 0.9583897590637207], "elapsed_sec": 1954.296730041504}
|
| 158 |
+
{"step": 7900, "train_loss": 2.309685173034668, "lr": 2.178523437408597e-07, "grad_norm": 0.6655879020690918, "mass_mean": 1.3703397512435913, "mass_std": 0.1873304694890976, "mass_min": 1.1232856512069702, "mass_max": 2.1119625568389893, "gamma": 0.30000001192092896, "xi_alphas": [0.23349131643772125, 0.5611025094985962, 0.781069278717041, 0.9583898186683655], "elapsed_sec": 1966.3441739082336}
|
| 159 |
+
{"step": 7950, "train_loss": 2.3399563217163086, "lr": 5.555291356351688e-08, "grad_norm": 0.6699886918067932, "mass_mean": 1.3743369579315186, "mass_std": 0.18664635717868805, "mass_min": 1.123284935951233, "mass_max": 2.111961841583252, "gamma": 0.30000001192092896, "xi_alphas": [0.23349103331565857, 0.561102569103241, 0.781069278717041, 0.9583898186683655], "elapsed_sec": 1978.3909268379211}
|
| 160 |
+
{"step": 8000, "train_loss": 2.320168867111206, "lr": 2.1359081253669033e-11, "grad_norm": 0.6800988912582397, "mass_mean": 1.3744497299194336, "mass_std": 0.1896265596151352, "mass_min": 1.123284935951233, "mass_max": 2.111961841583252, "gamma": 0.30000001192092896, "xi_alphas": [0.23349103331565857, 0.5611025094985962, 0.781069278717041, 0.9583898186683655], "elapsed_sec": 1990.4420058727264}
|