Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -3219,3 +3219,47 @@
 {"timestamp": 1787650303.7365785, "elapsed_hours": 6.401897132330471, "session_id": 2, "event": "train_step", "step": 122500, "epoch": 1, "loss": 1.7083448004722595, "loss_raw": 1.0824265480041504, "perplexity": 5.519817512828083, "muon_lr": 0.0017454355344592978, "adamw_lr": 5.236306603377893e-05, "tokens_per_sec": 699074.3477736248, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 16056320000, "tokens_processed_billions": 16.05632}
 {"timestamp": 1787650303.7385075, "elapsed_hours": 6.401897667977545, "session_id": 2, "event": "lambda_stats", "step": 122500, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1892361111111112, "lambda_std": 0.20446532211957086, "lambda_all": [2.015625, 1.25, 1.1875, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
 {"timestamp": 1787650329.416899, "elapsed_hours": 6.409030554691951, "session_id": 2, "event": "validation", "step": 122500, "epoch": 1, "val_loss": 1.7506743031740188, "val_perplexity": 5.758484331414582, "is_best": false, "best_loss": 1.6233880519866943, "best_perplexity": 5.070239483980759, "train_loss": 1.0824265480041504, "train_loss_avg": 1.7083448004722595}
+{"timestamp": 1787650337.6606615, "elapsed_hours": 6.411320488585366, "session_id": 2, "event": "train_step", "step": 122510, "epoch": 1, "loss": 1.6962854409217833, "loss_raw": 1.7312554121017456, "perplexity": 5.453651808657196, "muon_lr": 0.0017450143479496361, "adamw_lr": 5.235043043848908e-05, "tokens_per_sec": 698100.3077871846, "vram_gb": 2.785593032836914, "grad_norm": 0.09423828125, "tokens_processed": 16057630720, "tokens_processed_billions": 16.05763072}
+{"timestamp": 1787650345.1924222, "elapsed_hours": 6.4134126442670825, "session_id": 2, "event": "train_step", "step": 122520, "epoch": 1, "loss": 1.6763129127025604, "loss_raw": 1.3838354349136353, "perplexity": 5.345809124401892, "muon_lr": 0.0017441722997946635, "adamw_lr": 5.2325168993839896e-05, "tokens_per_sec": 697928.75978256, "vram_gb": 2.363645076751709, "grad_norm": 0.091796875, "tokens_processed": 16058941440, "tokens_processed_billions": 16.05894144}
+{"timestamp": 1787650352.6430643, "elapsed_hours": 6.415482267075115, "session_id": 2, "event": "train_step", "step": 122530, "epoch": 1, "loss": 1.695100018978119, "loss_raw": 1.5432242155075073, "perplexity": 5.44719076042089, "muon_lr": 0.001743751438173634, "adamw_lr": 5.231254314520902e-05, "tokens_per_sec": 697759.7852638473, "vram_gb": 2.785593032836914, "grad_norm": 0.11328125, "tokens_processed": 16060252160, "tokens_processed_billions": 16.06025216}
+{"timestamp": 1787650360.0943828, "elapsed_hours": 6.417552077902688, "session_id": 2, "event": "train_step", "step": 122540, "epoch": 1, "loss": 1.6680652260780335, "loss_raw": 1.0803476572036743, "perplexity": 5.301899889700914, "muon_lr": 0.0017433306848730948, "adamw_lr": 5.2299920546192843e-05, "tokens_per_sec": 697590.8991703263, "vram_gb": 2.785593032836914, "grad_norm": 0.09423828125, "tokens_processed": 16061562880, "tokens_processed_billions": 16.06156288}
+{"timestamp": 1787650367.5375443, "elapsed_hours": 6.419619622694121, "session_id": 2, "event": "train_step", "step": 122550, "epoch": 1, "loss": 1.6802010750770568, "loss_raw": 1.5728596448898315, "perplexity": 5.366634959177489, "muon_lr": 0.0017429100399051775, "adamw_lr": 5.228730119715532e-05, "tokens_per_sec": 697422.370502952, "vram_gb": 2.785593032836914, "grad_norm": 0.0966796875, "tokens_processed": 16062873600, "tokens_processed_billions": 16.0628736}
+{"timestamp": 1787650375.0837104, "elapsed_hours": 6.421715779900551, "session_id": 2, "event": "train_step", "step": 122560, "epoch": 1, "loss": 1.7016386699676513, "loss_raw": 1.6928342580795288, "perplexity": 5.482924738367831, "muon_lr": 0.0017420690750157262, "adamw_lr": 5.226207225047178e-05, "tokens_per_sec": 697250.8272007816, "vram_gb": 2.363645076751709, "grad_norm": 0.09375, "tokens_processed": 16064184320, "tokens_processed_billions": 16.06418432}
+{"timestamp": 1787650382.5542233, "elapsed_hours": 6.423790922363599, "session_id": 2, "event": "train_step", "step": 122570, "epoch": 1, "loss": 1.6871020460128785, "loss_raw": 1.0012205839157104, "perplexity": 5.403798034218427, "muon_lr": 0.001741648755118445, "adamw_lr": 5.224946265355334e-05, "tokens_per_sec": 697081.6906903051, "vram_gb": 2.785593032836914, "grad_norm": 0.08935546875, "tokens_processed": 16065495040, "tokens_processed_billions": 16.06549504}
+{"timestamp": 1787650389.9929626, "elapsed_hours": 6.425857238835759, "session_id": 2, "event": "train_step", "step": 122580, "epoch": 1, "loss": 1.6607141757011414, "loss_raw": 1.6712682247161865, "perplexity": 5.263068258023785, "muon_lr": 0.001741228543602287, "adamw_lr": 5.223685630806861e-05, "tokens_per_sec": 696913.621647102, "vram_gb": 2.785593032836914, "grad_norm": 0.091796875, "tokens_processed": 16066805760, "tokens_processed_billions": 16.06680576}
+{"timestamp": 1787650397.4679525, "elapsed_hours": 6.427933624916607, "session_id": 2, "event": "train_step", "step": 122590, "epoch": 1, "loss": 1.6473762345314027, "loss_raw": 2.4913673400878906, "perplexity": 5.193335840451492, "muon_lr": 0.0017408084404793725, "adamw_lr": 5.222425321438117e-05, "tokens_per_sec": 696744.5664796198, "vram_gb": 2.785593032836914, "grad_norm": 0.0908203125, "tokens_processed": 16068116480, "tokens_processed_billions": 16.06811648}
+{"timestamp": 1787650405.000789, "elapsed_hours": 6.430026079482502, "session_id": 2, "event": "train_step", "step": 122600, "epoch": 1, "loss": 1.67447998046875, "loss_raw": 2.270676612854004, "perplexity": 5.336019593056532, "muon_lr": 0.001739968559461725, "adamw_lr": 5.219905678385174e-05, "tokens_per_sec": 696573.8725071611, "vram_gb": 2.363645076751709, "grad_norm": 0.103515625, "tokens_processed": 16069427200, "tokens_processed_billions": 16.0694272}
+{"timestamp": 1787650405.002621, "elapsed_hours": 6.430026588439941, "session_id": 2, "event": "lambda_stats", "step": 122600, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1892361111111112, "lambda_std": 0.20446532211957086, "lambda_all": [2.015625, 1.25, 1.1875, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787650412.4514349, "elapsed_hours": 6.432095703350173, "session_id": 2, "event": "train_step", "step": 122610, "epoch": 1, "loss": 1.672255139350891, "loss_raw": 2.004153251647949, "perplexity": 5.32416099390129, "muon_lr": 0.0017395487815912137, "adamw_lr": 5.2186463447736396e-05, "tokens_per_sec": 696405.7738622106, "vram_gb": 2.785593032836914, "grad_norm": 0.09765625, "tokens_processed": 16070737920, "tokens_processed_billions": 16.07073792}
+{"timestamp": 1787650419.9209094, "elapsed_hours": 6.434170557459195, "session_id": 2, "event": "train_step", "step": 122620, "epoch": 1, "loss": 1.7604724335670472, "loss_raw": 1.7327791452407837, "perplexity": 5.815184033685275, "muon_lr": 0.0017391291121623834, "adamw_lr": 5.217387336487149e-05, "tokens_per_sec": 696237.2129276649, "vram_gb": 2.785593032836914, "grad_norm": 0.09814453125, "tokens_processed": 16072048640, "tokens_processed_billions": 16.07204864}
+{"timestamp": 1787650427.3632553, "elapsed_hours": 6.436237875686752, "session_id": 2, "event": "train_step", "step": 122630, "epoch": 1, "loss": 1.7653036952018737, "loss_raw": 1.6931116580963135, "perplexity": 5.843346684996191, "muon_lr": 0.001738709551187339, "adamw_lr": 5.216128653562016e-05, "tokens_per_sec": 696069.5800914866, "vram_gb": 2.785593032836914, "grad_norm": 0.09423828125, "tokens_processed": 16073359360, "tokens_processed_billions": 16.07335936}
+{"timestamp": 1787650434.949871, "elapsed_hours": 6.438345269097222, "session_id": 2, "event": "train_step", "step": 122640, "epoch": 1, "loss": 1.7663790225982665, "loss_raw": 1.218908429145813, "perplexity": 5.849633575400123, "muon_lr": 0.0017378707546469954, "adamw_lr": 5.213612263940985e-05, "tokens_per_sec": 695897.7067227343, "vram_gb": 2.363645076751709, "grad_norm": 0.09228515625, "tokens_processed": 16074670080, "tokens_processed_billions": 16.07467008}
+{"timestamp": 1787650442.4124506, "elapsed_hours": 6.440418207777871, "session_id": 2, "event": "train_step", "step": 122650, "epoch": 1, "loss": 1.7061746549606323, "loss_raw": 1.3284748792648315, "perplexity": 5.507851694107629, "muon_lr": 0.0017374515191058862, "adamw_lr": 5.212354557317658e-05, "tokens_per_sec": 695729.6846156984, "vram_gb": 2.785593032836914, "grad_norm": 0.09423828125, "tokens_processed": 16075980800, "tokens_processed_billions": 16.0759808}
+{"timestamp": 1787650449.8534975, "elapsed_hours": 6.4424851653973265, "session_id": 2, "event": "train_step", "step": 122660, "epoch": 1, "loss": 1.6710890293121339, "loss_raw": 1.3726239204406738, "perplexity": 5.317956054842335, "muon_lr": 0.0017370323920669378, "adamw_lr": 5.211097176200813e-05, "tokens_per_sec": 695562.417593638, "vram_gb": 2.785593032836914, "grad_norm": 0.08740234375, "tokens_processed": 16077291520, "tokens_processed_billions": 16.07729152}
+{"timestamp": 1787650457.3547401, "elapsed_hours": 6.4445688437753255, "session_id": 2, "event": "train_step", "step": 122670, "epoch": 1, "loss": 1.619570686817169, "loss_raw": 1.5210777521133423, "perplexity": 5.050921423877354, "muon_lr": 0.0017366133735422398, "adamw_lr": 5.2098401206267185e-05, "tokens_per_sec": 695393.4460551895, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 16078602240, "tokens_processed_billions": 16.07860224}
+{"timestamp": 1787650464.8920686, "elapsed_hours": 6.446662546065118, "session_id": 2, "event": "train_step", "step": 122680, "epoch": 1, "loss": 1.6374526619911194, "loss_raw": 1.9616146087646484, "perplexity": 5.142054264570373, "muon_lr": 0.0017357756620839178, "adamw_lr": 5.207326986251753e-05, "tokens_per_sec": 695223.499364834, "vram_gb": 2.363645076751709, "grad_norm": 0.09521484375, "tokens_processed": 16079912960, "tokens_processed_billions": 16.07991296}
+{"timestamp": 1787650472.3319163, "elapsed_hours": 6.448729170428382, "session_id": 2, "event": "train_step", "step": 122690, "epoch": 1, "loss": 1.6940782690048217, "loss_raw": 1.6832534074783325, "perplexity": 5.441627935798793, "muon_lr": 0.0017353569691744533, "adamw_lr": 5.20607090752336e-05, "tokens_per_sec": 695056.5942281866, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 16081223680, "tokens_processed_billions": 16.08122368}
+{"timestamp": 1787650479.8004744, "elapsed_hours": 6.450803769959344, "session_id": 2, "event": "train_step", "step": 122700, "epoch": 1, "loss": 1.6896642684936523, "loss_raw": 0.9320913553237915, "perplexity": 5.417661520106402, "muon_lr": 0.001734938384827551, "adamw_lr": 5.204815154482652e-05, "tokens_per_sec": 694888.9343880086, "vram_gb": 2.785593032836914, "grad_norm": 0.091796875, "tokens_processed": 16082534400, "tokens_processed_billions": 16.0825344}
+{"timestamp": 1787650479.8022306, "elapsed_hours": 6.450804257790248, "session_id": 2, "event": "lambda_stats", "step": 122700, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1892361111111112, "lambda_std": 0.20446532211957086, "lambda_all": [2.015625, 1.25, 1.1875, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787650487.2505038, "elapsed_hours": 6.452873222563002, "session_id": 2, "event": "train_step", "step": 122710, "epoch": 1, "loss": 1.7031325244903563, "loss_raw": 1.8330963850021362, "perplexity": 5.491121551184507, "muon_lr": 0.001734519909055284, "adamw_lr": 5.203559727165852e-05, "tokens_per_sec": 694721.9370194931, "vram_gb": 2.785593032836914, "grad_norm": 0.10009765625, "tokens_processed": 16083845120, "tokens_processed_billions": 16.08384512}
+{"timestamp": 1787650494.8096912, "elapsed_hours": 6.4549729967779585, "session_id": 2, "event": "train_step", "step": 122720, "epoch": 1, "loss": 1.6551755344867707, "loss_raw": 0.8305220603942871, "perplexity": 5.233998588802916, "muon_lr": 0.0017336832832829199, "adamw_lr": 5.201049849848759e-05, "tokens_per_sec": 694551.7728008319, "vram_gb": 2.363645076751709, "grad_norm": 0.08837890625, "tokens_processed": 16085155840, "tokens_processed_billions": 16.08515584}
+{"timestamp": 1787650502.283445, "elapsed_hours": 6.457049039536052, "session_id": 2, "event": "train_step", "step": 122730, "epoch": 1, "loss": 1.7138705360889435, "loss_raw": 1.5592358112335205, "perplexity": 5.550402990864901, "muon_lr": 0.0017332651333069487, "adamw_lr": 5.199795399920846e-05, "tokens_per_sec": 694384.2821366856, "vram_gb": 2.785593032836914, "grad_norm": 0.10302734375, "tokens_processed": 16086466560, "tokens_processed_billions": 16.08646656}
+{"timestamp": 1787650509.7306516, "elapsed_hours": 6.459117708073722, "session_id": 2, "event": "train_step", "step": 122740, "epoch": 1, "loss": 1.6714695644378663, "loss_raw": 1.1768758296966553, "perplexity": 5.319980109005733, "muon_lr": 0.0017328470919538656, "adamw_lr": 5.198541275861596e-05, "tokens_per_sec": 694217.6950807583, "vram_gb": 2.785593032836914, "grad_norm": 0.0947265625, "tokens_processed": 16087777280, "tokens_processed_billions": 16.08777728}
+{"timestamp": 1787650517.187638, "elapsed_hours": 6.461189093258646, "session_id": 2, "event": "train_step", "step": 122750, "epoch": 1, "loss": 1.671658891439438, "loss_raw": 1.8189603090286255, "perplexity": 5.320987420240789, "muon_lr": 0.0017324291592357242, "adamw_lr": 5.197287477707171e-05, "tokens_per_sec": 694050.9218414053, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 16089088000, "tokens_processed_billions": 16.089088}
+{"timestamp": 1787650524.7340784, "elapsed_hours": 6.463285326560339, "session_id": 2, "event": "train_step", "step": 122760, "epoch": 1, "loss": 1.6471233594417571, "loss_raw": 1.523054599761963, "perplexity": 5.192022741217319, "muon_lr": 0.0017315936197524699, "adamw_lr": 5.194780859257409e-05, "tokens_per_sec": 693881.5764969361, "vram_gb": 2.363645076751709, "grad_norm": 0.0927734375, "tokens_processed": 16090398720, "tokens_processed_billions": 16.09039872}
+{"timestamp": 1787650532.1890671, "elapsed_hours": 6.465356156892247, "session_id": 2, "event": "train_step", "step": 122770, "epoch": 1, "loss": 1.6891503274440764, "loss_raw": 1.0090581178665161, "perplexity": 5.414877876834056, "muon_lr": 0.0017311760130114544, "adamw_lr": 5.193528039034362e-05, "tokens_per_sec": 693715.0803345336, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 16091709440, "tokens_processed_billions": 16.09170944}
+{"timestamp": 1787650539.653131, "elapsed_hours": 6.4674295078383555, "session_id": 2, "event": "train_step", "step": 122780, "epoch": 1, "loss": 1.6419310748577118, "loss_raw": 1.6037068367004395, "perplexity": 5.165134148599823, "muon_lr": 0.0017307585149535685, "adamw_lr": 5.192275544860705e-05, "tokens_per_sec": 693548.4177743425, "vram_gb": 2.785593032836914, "grad_norm": 0.099609375, "tokens_processed": 16093020160, "tokens_processed_billions": 16.09302016}
+{"timestamp": 1787650547.1083853, "elapsed_hours": 6.469500411815114, "session_id": 2, "event": "train_step", "step": 122790, "epoch": 1, "loss": 1.60834614276886, "loss_raw": 1.2172069549560547, "perplexity": 4.994544130492143, "muon_lr": 0.0017303411255908546, "adamw_lr": 5.191023376772563e-05, "tokens_per_sec": 693382.1258077825, "vram_gb": 2.785593032836914, "grad_norm": 0.091796875, "tokens_processed": 16094330880, "tokens_processed_billions": 16.09433088}
+{"timestamp": 1787650554.6419206, "elapsed_hours": 6.471593060493469, "session_id": 2, "event": "train_step", "step": 122800, "epoch": 1, "loss": 1.6453502070903778, "loss_raw": 1.7260029315948486, "perplexity": 5.182824651102824, "muon_lr": 0.0017295066729990823, "adamw_lr": 5.1885200189972464e-05, "tokens_per_sec": 693213.6021449642, "vram_gb": 2.363645076751709, "grad_norm": 0.0947265625, "tokens_processed": 16095641600, "tokens_processed_billions": 16.0956416}
+{"timestamp": 1787650554.6437254, "elapsed_hours": 6.471593561967214, "session_id": 2, "event": "lambda_stats", "step": 122800, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1892361111111112, "lambda_std": 0.20446532211957086, "lambda_all": [2.015625, 1.25, 1.1875, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787650562.1314962, "elapsed_hours": 6.473673498166932, "session_id": 2, "event": "train_step", "step": 122810, "epoch": 1, "loss": 1.6829491889476775, "loss_raw": 2.4927632808685303, "perplexity": 5.381403366485437, "muon_lr": 0.0017290896097940864, "adamw_lr": 5.187268829382258e-05, "tokens_per_sec": 693046.5018941225, "vram_gb": 2.785593032836914, "grad_norm": 0.0947265625, "tokens_processed": 16096952320, "tokens_processed_billions": 16.09695232}
+{"timestamp": 1787650569.5809076, "elapsed_hours": 6.475742779109213, "session_id": 2, "event": "train_step", "step": 122820, "epoch": 1, "loss": 1.6648723435401918, "loss_raw": 1.6899330615997314, "perplexity": 5.284998542492282, "muon_lr": 0.0017286726553323896, "adamw_lr": 5.186017965997168e-05, "tokens_per_sec": 692880.7060653489, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 16098263040, "tokens_processed_billions": 16.09826304}
+{"timestamp": 1787650577.0239847, "elapsed_hours": 6.47781030052238, "session_id": 2, "event": "train_step", "step": 122830, "epoch": 1, "loss": 1.6114140605926515, "loss_raw": 1.3219701051712036, "perplexity": 5.009890510130777, "muon_lr": 0.001728255809626014, "adamw_lr": 5.184767428878042e-05, "tokens_per_sec": 692715.2059845492, "vram_gb": 2.785593032836914, "grad_norm": 0.09765625, "tokens_processed": 16099573760, "tokens_processed_billions": 16.09957376}
+{"timestamp": 1787650584.5895123, "elapsed_hours": 6.4799118361208174, "session_id": 2, "event": "train_step", "step": 122840, "epoch": 1, "loss": 1.6372274899482726, "loss_raw": 1.2698837518692017, "perplexity": 5.140896548054778, "muon_lr": 0.0017274224445273075, "adamw_lr": 5.1822673335819216e-05, "tokens_per_sec": 692546.1637877432, "vram_gb": 2.363645076751709, "grad_norm": 0.09423828125, "tokens_processed": 16100884480, "tokens_processed_billions": 16.10088448}
+{"timestamp": 1787650592.0363898, "elapsed_hours": 6.481980413132244, "session_id": 2, "event": "train_step", "step": 122850, "epoch": 1, "loss": 1.5843069100379943, "loss_raw": 1.7023919820785522, "perplexity": 4.875910762307492, "muon_lr": 0.0017270059251590086, "adamw_lr": 5.181017775477025e-05, "tokens_per_sec": 692380.762698579, "vram_gb": 2.785593032836914, "grad_norm": 0.0908203125, "tokens_processed": 16102195200, "tokens_processed_billions": 16.1021952}
+{"timestamp": 1787650599.4832811, "elapsed_hours": 6.4840489939848585, "session_id": 2, "event": "train_step", "step": 122860, "epoch": 1, "loss": 1.5661937832832336, "loss_raw": 2.387303590774536, "perplexity": 4.78838782257873, "muon_lr": 0.0017265895145940964, "adamw_lr": 5.1797685437822887e-05, "tokens_per_sec": 692215.4715315171, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 16103505920, "tokens_processed_billions": 16.10350592}
+{"timestamp": 1787650606.9974933, "elapsed_hours": 6.486136275198724, "session_id": 2, "event": "train_step", "step": 122870, "epoch": 1, "loss": 1.5998534536361695, "loss_raw": 1.5606508255004883, "perplexity": 4.952306628686045, "muon_lr": 0.0017261732128445785, "adamw_lr": 5.178519638533735e-05, "tokens_per_sec": 692048.2774411244, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 16104816640, "tokens_processed_billions": 16.10481664}
+{"timestamp": 1787650614.5501995, "elapsed_hours": 6.488234249088499, "session_id": 2, "event": "train_step", "step": 122880, "epoch": 1, "loss": 1.6379739379882812, "loss_raw": 1.358487606048584, "perplexity": 5.144735392777775, "muon_lr": 0.0017253409358397409, "adamw_lr": 5.176022807519222e-05, "tokens_per_sec": 691880.0495745454, "vram_gb": 2.363645076751709, "grad_norm": 0.0947265625, "tokens_processed": 16106127360, "tokens_processed_billions": 16.10612736}
+{"timestamp": 1787650621.9968698, "elapsed_hours": 6.490302768614557, "session_id": 2, "event": "train_step", "step": 122890, "epoch": 1, "loss": 1.6581300210952759, "loss_raw": 1.8781219720840454, "perplexity": 5.249485233819896, "muon_lr": 0.0017249249606084219, "adamw_lr": 5.1747748818252655e-05, "tokens_per_sec": 691715.0828025286, "vram_gb": 2.785593032836914, "grad_norm": 0.0986328125, "tokens_processed": 16107438080, "tokens_processed_billions": 16.10743808}
+{"timestamp": 1787650629.4675343, "elapsed_hours": 6.492377953131993, "session_id": 2, "event": "train_step", "step": 122900, "epoch": 1, "loss": 1.7144749212265014, "loss_raw": 1.7393927574157715, "perplexity": 5.553758585873768, "muon_lr": 0.0017245090942404999, "adamw_lr": 5.1735272827214995e-05, "tokens_per_sec": 691549.5081813771, "vram_gb": 2.785593032836914, "grad_norm": 0.09912109375, "tokens_processed": 16108748800, "tokens_processed_billions": 16.1087488}
+{"timestamp": 1787650629.469169, "elapsed_hours": 6.492378407319387, "session_id": 2, "event": "lambda_stats", "step": 122900, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1892361111111112, "lambda_std": 0.20446532211957086, "lambda_all": [2.015625, 1.25, 1.1875, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}