Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -2879,3 +2879,47 @@
 {"timestamp": 1787647879.1484818, "elapsed_hours": 5.728400438626608, "session_id": 2, "event": "train_step", "step": 119500, "epoch": 1, "loss": 1.738738545179367, "loss_raw": 1.633450984954834, "perplexity": 5.6901610124420925, "muon_lr": 0.0019109043312418314, "adamw_lr": 5.732712993725494e-05, "tokens_per_sec": 762440.311671592, "vram_gb": 2.785593032836914, "grad_norm": 0.09912109375, "tokens_processed": 15663104000, "tokens_processed_billions": 15.663104}
 {"timestamp": 1787647879.1505268, "elapsed_hours": 5.7284010066588715, "session_id": 2, "event": "lambda_stats", "step": 119500, "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": 1787647904.8166227, "elapsed_hours": 5.735530477894677, "session_id": 2, "event": "validation", "step": 119500, "epoch": 1, "val_loss": 1.7519917166233063, "val_perplexity": 5.766075635465384, "is_best": false, "best_loss": 1.6233880519866943, "best_perplexity": 5.070239483980759, "train_loss": 1.633450984954834, "train_loss_avg": 1.738738545179367}
+{"timestamp": 1787647913.835603, "elapsed_hours": 5.738035750124189, "session_id": 2, "event": "train_step", "step": 119510, "epoch": 1, "loss": 1.7120494532585144, "loss_raw": 1.629027247428894, "perplexity": 5.54030444521118, "muon_lr": 0.0019104434216432107, "adamw_lr": 5.7313302649296315e-05, "tokens_per_sec": 761218.8090797847, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 15664414720, "tokens_processed_billions": 15.66441472}
+{"timestamp": 1787647921.36726, "elapsed_hours": 5.740127877063221, "session_id": 2, "event": "train_step", "step": 119520, "epoch": 1, "loss": 1.7149492359161378, "loss_raw": 1.6280888319015503, "perplexity": 5.556393439979348, "muon_lr": 0.0019095219129998244, "adamw_lr": 5.7285657389994726e-05, "tokens_per_sec": 761003.9729899382, "vram_gb": 2.363645076751709, "grad_norm": 0.09326171875, "tokens_processed": 15665725440, "tokens_processed_billions": 15.66572544}
+{"timestamp": 1787647928.8069367, "elapsed_hours": 5.742194453875224, "session_id": 2, "event": "train_step", "step": 119530, "epoch": 1, "loss": 1.7468876135349274, "loss_raw": 1.6477257013320923, "perplexity": 5.736719971903898, "muon_lr": 0.0019090613139816338, "adamw_lr": 5.7271839419449004e-05, "tokens_per_sec": 760792.69391687, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 15667036160, "tokens_processed_billions": 15.66703616}
+{"timestamp": 1787647936.2476542, "elapsed_hours": 5.744261319902208, "session_id": 2, "event": "train_step", "step": 119540, "epoch": 1, "loss": 1.7768056309223175, "loss_raw": 1.7698991298675537, "perplexity": 5.9109444905330255, "muon_lr": 0.001908600818516824, "adamw_lr": 5.725802455550472e-05, "tokens_per_sec": 760581.5283188357, "vram_gb": 2.785593032836914, "grad_norm": 0.0966796875, "tokens_processed": 15668346880, "tokens_processed_billions": 15.66834688}
+{"timestamp": 1787647943.679011, "elapsed_hours": 5.746325585643451, "session_id": 2, "event": "train_step", "step": 119550, "epoch": 1, "loss": 1.7105864572525025, "loss_raw": 1.5997711420059204, "perplexity": 5.532204928160948, "muon_lr": 0.0019081404266186736, "adamw_lr": 5.72442127985602e-05, "tokens_per_sec": 760370.8602323884, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15669657600, "tokens_processed_billions": 15.6696576}
+{"timestamp": 1787647951.2156715, "elapsed_hours": 5.7484191024303435, "session_id": 2, "event": "train_step", "step": 119560, "epoch": 1, "loss": 1.7846984374523163, "loss_raw": 1.7150720357894897, "perplexity": 5.957783032425244, "muon_lr": 0.0019072199535754516, "adamw_lr": 5.721659860726354e-05, "tokens_per_sec": 760156.4596328046, "vram_gb": 2.363645076751709, "grad_norm": 0.099609375, "tokens_processed": 15670968320, "tokens_processed_billions": 15.67096832}
+{"timestamp": 1787647958.6578877, "elapsed_hours": 5.750486384696431, "session_id": 2, "event": "train_step", "step": 119570, "epoch": 1, "loss": 1.7782040989398957, "loss_raw": 1.0985634326934814, "perplexity": 5.919216540106955, "muon_lr": 0.0019067598724569258, "adamw_lr": 5.720279617370776e-05, "tokens_per_sec": 759945.6994727267, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 15672279040, "tokens_processed_billions": 15.67227904}
+{"timestamp": 1787647966.1006765, "elapsed_hours": 5.752553826040692, "session_id": 2, "event": "train_step", "step": 119580, "epoch": 1, "loss": 1.7636000096797944, "loss_raw": 1.4193603992462158, "perplexity": 5.8333999353210855, "muon_lr": 0.0019062998949581464, "adamw_lr": 5.718899684874439e-05, "tokens_per_sec": 759735.0672086459, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 15673589760, "tokens_processed_billions": 15.67358976}
+{"timestamp": 1787647973.542842, "elapsed_hours": 5.754621094266573, "session_id": 2, "event": "train_step", "step": 119590, "epoch": 1, "loss": 1.7322165262699127, "loss_raw": 1.7242400646209717, "perplexity": 5.653170432473554, "muon_lr": 0.0019058400210923785, "adamw_lr": 5.717520063277135e-05, "tokens_per_sec": 759524.6107535154, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 15674900480, "tokens_processed_billions": 15.67490048}
+{"timestamp": 1787647981.0773215, "elapsed_hours": 5.756714005337821, "session_id": 2, "event": "train_step", "step": 119600, "epoch": 1, "loss": 1.7645412635803224, "loss_raw": 1.6856757402420044, "perplexity": 5.838893230650805, "muon_lr": 0.0019049205843129228, "adamw_lr": 5.714761752938768e-05, "tokens_per_sec": 759310.9095036827, "vram_gb": 2.363645076751709, "grad_norm": 0.09130859375, "tokens_processed": 15676211200, "tokens_processed_billions": 15.6762112}
+{"timestamp": 1787647981.079138, "elapsed_hours": 5.756714509858026, "session_id": 2, "event": "lambda_stats", "step": 119600, "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": 1787647988.5214667, "elapsed_hours": 5.75878182331721, "session_id": 2, "event": "train_step", "step": 119610, "epoch": 1, "loss": 1.7517254495620727, "loss_raw": 1.539360523223877, "perplexity": 5.764540523834979, "muon_lr": 0.0019044610214257475, "adamw_lr": 5.713383064277242e-05, "tokens_per_sec": 759100.6874917769, "vram_gb": 2.785593032836914, "grad_norm": 0.0986328125, "tokens_processed": 15677521920, "tokens_processed_billions": 15.67752192}
+{"timestamp": 1787647995.9659855, "elapsed_hours": 5.760849745207363, "session_id": 2, "event": "train_step", "step": 119620, "epoch": 1, "loss": 1.7904200148582459, "loss_raw": 2.4299895763397217, "perplexity": 5.99196865379281, "muon_lr": 0.0019040015622246134, "adamw_lr": 5.712004686673839e-05, "tokens_per_sec": 758890.601609648, "vram_gb": 2.785593032836914, "grad_norm": 0.0986328125, "tokens_processed": 15678832640, "tokens_processed_billions": 15.67883264}
+{"timestamp": 1787648003.4522278, "elapsed_hours": 5.762929256955783, "session_id": 2, "event": "train_step", "step": 119630, "epoch": 1, "loss": 1.8129222011566162, "loss_raw": 2.097790241241455, "perplexity": 6.128329503564315, "muon_lr": 0.0019035422067227691, "adamw_lr": 5.710626620168307e-05, "tokens_per_sec": 758679.1348616416, "vram_gb": 2.785593032836914, "grad_norm": 0.09912109375, "tokens_processed": 15680143360, "tokens_processed_billions": 15.68014336}
+{"timestamp": 1787648010.9963472, "elapsed_hours": 5.765024845732583, "session_id": 2, "event": "train_step", "step": 119640, "epoch": 1, "loss": 1.8033668935298919, "loss_raw": 1.4186655282974243, "perplexity": 6.070050311151505, "muon_lr": 0.001902623806869932, "adamw_lr": 5.707871420609795e-05, "tokens_per_sec": 758465.698391568, "vram_gb": 2.363645076751709, "grad_norm": 0.09619140625, "tokens_processed": 15681454080, "tokens_processed_billions": 15.68145408}
+{"timestamp": 1787648018.4478524, "elapsed_hours": 5.767094708217515, "session_id": 2, "event": "train_step", "step": 119650, "epoch": 1, "loss": 1.7843640172481536, "loss_raw": 1.6957662105560303, "perplexity": 5.955790962519961, "muon_lr": 0.0019021647625454235, "adamw_lr": 5.70649428763627e-05, "tokens_per_sec": 758255.8136380318, "vram_gb": 2.785593032836914, "grad_norm": 0.0908203125, "tokens_processed": 15682764800, "tokens_processed_billions": 15.6827648}
+{"timestamp": 1787648025.894555, "elapsed_hours": 5.769163236816724, "session_id": 2, "event": "train_step", "step": 119660, "epoch": 1, "loss": 1.756917062997818, "loss_raw": 2.023418426513672, "perplexity": 5.794545609879826, "muon_lr": 0.0019017058219731745, "adamw_lr": 5.7051174659195235e-05, "tokens_per_sec": 758046.2550728166, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15684075520, "tokens_processed_billions": 15.68407552}
+{"timestamp": 1787648033.344148, "elapsed_hours": 5.771232568158044, "session_id": 2, "event": "train_step", "step": 119670, "epoch": 1, "loss": 1.7374770057201385, "loss_raw": 2.1734697818756104, "perplexity": 5.68298717578596, "muon_lr": 0.0019012469851664188, "adamw_lr": 5.703740955499256e-05, "tokens_per_sec": 757836.7428328062, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15685386240, "tokens_processed_billions": 15.68538624}
+{"timestamp": 1787648040.8676195, "elapsed_hours": 5.773322421312332, "session_id": 2, "event": "train_step", "step": 119680, "epoch": 1, "loss": 1.7640478789806366, "loss_raw": 1.918121099472046, "perplexity": 5.836013121210839, "muon_lr": 0.001900329622902308, "adamw_lr": 5.700988868706923e-05, "tokens_per_sec": 757624.6750232136, "vram_gb": 2.363645076751709, "grad_norm": 0.0966796875, "tokens_processed": 15686696960, "tokens_processed_billions": 15.68669696}
+{"timestamp": 1787648048.3067465, "elapsed_hours": 5.775388845536444, "session_id": 2, "event": "train_step", "step": 119690, "epoch": 1, "loss": 1.7511043572425842, "loss_raw": 1.3378719091415405, "perplexity": 5.760961323612187, "muon_lr": 0.0018998710974714077, "adamw_lr": 5.699613292414223e-05, "tokens_per_sec": 757415.8488522839, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 15688007680, "tokens_processed_billions": 15.68800768}
+{"timestamp": 1787648055.7520535, "elapsed_hours": 5.777456986308098, "session_id": 2, "event": "train_step", "step": 119700, "epoch": 1, "loss": 1.7544680309295655, "loss_raw": 1.8493291139602661, "perplexity": 5.7803719448245845, "muon_lr": 0.0018994126758589106, "adamw_lr": 5.698238027576731e-05, "tokens_per_sec": 757206.9460507291, "vram_gb": 2.785593032836914, "grad_norm": 0.0927734375, "tokens_processed": 15689318400, "tokens_processed_billions": 15.6893184}
+{"timestamp": 1787648055.7538304, "elapsed_hours": 5.777457480033239, "session_id": 2, "event": "lambda_stats", "step": 119700, "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": 1787648063.1915743, "elapsed_hours": 5.779523519873619, "session_id": 2, "event": "train_step", "step": 119710, "epoch": 1, "loss": 1.7847069811820984, "loss_raw": 1.5429452657699585, "perplexity": 5.957833934331019, "muon_lr": 0.0018989543580780345, "adamw_lr": 5.696863074234103e-05, "tokens_per_sec": 756998.4035220876, "vram_gb": 2.785593032836914, "grad_norm": 0.1162109375, "tokens_processed": 15690629120, "tokens_processed_billions": 15.69062912}
+{"timestamp": 1787648070.7270632, "elapsed_hours": 5.781616711351607, "session_id": 2, "event": "train_step", "step": 119720, "epoch": 1, "loss": 1.8096466374397278, "loss_raw": 1.5008126497268677, "perplexity": 6.1082886103274285, "muon_lr": 0.0018980380340640095, "adamw_lr": 5.694114102192028e-05, "tokens_per_sec": 756786.5071491181, "vram_gb": 2.363645076751709, "grad_norm": 0.099609375, "tokens_processed": 15691939840, "tokens_processed_billions": 15.69193984}
+{"timestamp": 1787648078.1719518, "elapsed_hours": 5.7836847358279755, "session_id": 2, "event": "train_step", "step": 119730, "epoch": 1, "loss": 1.70642258644104, "loss_raw": 1.0998739004135132, "perplexity": 5.509217433229881, "muon_lr": 0.0018975800278572845, "adamw_lr": 5.692740083571853e-05, "tokens_per_sec": 756578.0731514749, "vram_gb": 2.785593032836914, "grad_norm": 0.099609375, "tokens_processed": 15693250560, "tokens_processed_billions": 15.69325056}
+{"timestamp": 1787648085.6160066, "elapsed_hours": 5.785752528839641, "session_id": 2, "event": "train_step", "step": 119740, "epoch": 1, "loss": 1.7321246266365051, "loss_raw": 1.4522500038146973, "perplexity": 5.6526509320545335, "muon_lr": 0.0018971221255350314, "adamw_lr": 5.691366376605094e-05, "tokens_per_sec": 756369.8161236655, "vram_gb": 2.785593032836914, "grad_norm": 0.09228515625, "tokens_processed": 15694561280, "tokens_processed_billions": 15.69456128}
+{"timestamp": 1787648093.0529077, "elapsed_hours": 5.787818334698677, "session_id": 2, "event": "train_step", "step": 119750, "epoch": 1, "loss": 1.7862697792053224, "loss_raw": 2.011336326599121, "perplexity": 5.967152104738925, "muon_lr": 0.001896664327110453, "adamw_lr": 5.689992981331358e-05, "tokens_per_sec": 756161.9699645785, "vram_gb": 2.785593032836914, "grad_norm": 0.09326171875, "tokens_processed": 15695872000, "tokens_processed_billions": 15.695872}
+{"timestamp": 1787648100.5862207, "elapsed_hours": 5.789910921653112, "session_id": 2, "event": "train_step", "step": 119760, "epoch": 1, "loss": 1.7327512907981872, "loss_raw": 1.8477728366851807, "perplexity": 5.656194355964577, "muon_lr": 0.0018957490420071234, "adamw_lr": 5.6872471260213694e-05, "tokens_per_sec": 755950.7609441937, "vram_gb": 2.363645076751709, "grad_norm": 0.09423828125, "tokens_processed": 15697182720, "tokens_processed_billions": 15.69718272}
+{"timestamp": 1787648108.0340254, "elapsed_hours": 5.7919797562890585, "session_id": 2, "event": "train_step", "step": 119770, "epoch": 1, "loss": 1.704569103717804, "loss_raw": 1.5355122089385986, "perplexity": 5.499015651232775, "muon_lr": 0.0018952915553547664, "adamw_lr": 5.6858746660642985e-05, "tokens_per_sec": 755742.8199040173, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 15698493440, "tokens_processed_billions": 15.69849344}
+{"timestamp": 1787648115.4721684, "elapsed_hours": 5.79404590719276, "session_id": 2, "event": "train_step", "step": 119780, "epoch": 1, "loss": 1.7707808327674865, "loss_raw": 1.3003418445587158, "perplexity": 5.875439306255286, "muon_lr": 0.0018948341726528753, "adamw_lr": 5.684502517958625e-05, "tokens_per_sec": 755535.3768033789, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15699804160, "tokens_processed_billions": 15.69980416}
+{"timestamp": 1787648122.9176204, "elapsed_hours": 5.796114088230663, "session_id": 2, "event": "train_step", "step": 119790, "epoch": 1, "loss": 1.7642555332183838, "loss_raw": 1.6451793909072876, "perplexity": 5.8372251199009835, "muon_lr": 0.0018943768939146373, "adamw_lr": 5.68313068174391e-05, "tokens_per_sec": 755327.8161714651, "vram_gb": 2.785593032836914, "grad_norm": 0.0947265625, "tokens_processed": 15701114880, "tokens_processed_billions": 15.70111488}
+{"timestamp": 1787648130.4508138, "elapsed_hours": 5.798206641938951, "session_id": 2, "event": "train_step", "step": 119800, "epoch": 1, "loss": 1.670184154510498, "loss_raw": 1.6170852184295654, "perplexity": 5.313146146922534, "muon_lr": 0.0018934626483818643, "adamw_lr": 5.680387945145592e-05, "tokens_per_sec": 755117.2169603297, "vram_gb": 2.363645076751709, "grad_norm": 0.08837890625, "tokens_processed": 15702425600, "tokens_processed_billions": 15.7024256}
+{"timestamp": 1787648130.452659, "elapsed_hours": 5.798207154538896, "session_id": 2, "event": "lambda_stats", "step": 119800, "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": 1787648137.8916194, "elapsed_hours": 5.80027353240384, "session_id": 2, "event": "train_step", "step": 119810, "epoch": 1, "loss": 1.6788443183898927, "loss_raw": 1.2391148805618286, "perplexity": 5.359358678496466, "muon_lr": 0.0018930056816136944, "adamw_lr": 5.679017044841083e-05, "tokens_per_sec": 754910.1271801579, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 15703736320, "tokens_processed_billions": 15.70373632}
+{"timestamp": 1787648145.3437796, "elapsed_hours": 5.80234357688162, "session_id": 2, "event": "train_step", "step": 119820, "epoch": 1, "loss": 1.6976992583274841, "loss_raw": 1.4032061100006104, "perplexity": 5.461367729675358, "muon_lr": 0.0018925488188619084, "adamw_lr": 5.677646456585724e-05, "tokens_per_sec": 754702.7700103046, "vram_gb": 2.785593032836914, "grad_norm": 0.09130859375, "tokens_processed": 15705047040, "tokens_processed_billions": 15.70504704}
+{"timestamp": 1787648152.7835932, "elapsed_hours": 5.804410191840596, "session_id": 2, "event": "train_step", "step": 119830, "epoch": 1, "loss": 1.648342580795288, "loss_raw": 0.7446180582046509, "perplexity": 5.198356826752964, "muon_lr": 0.001892092060139679, "adamw_lr": 5.6762761804190364e-05, "tokens_per_sec": 754496.0099834661, "vram_gb": 2.785593032836914, "grad_norm": 0.091796875, "tokens_processed": 15706357760, "tokens_processed_billions": 15.70635776}
+{"timestamp": 1787648160.3151965, "elapsed_hours": 5.8065023038122385, "session_id": 2, "event": "train_step", "step": 119840, "epoch": 1, "loss": 1.663401062488556, "loss_raw": 1.7917622327804565, "perplexity": 5.27722854160795, "muon_lr": 0.001891178854836575, "adamw_lr": 5.673536564509724e-05, "tokens_per_sec": 754286.0715282841, "vram_gb": 2.363645076751709, "grad_norm": 0.09912109375, "tokens_processed": 15707668480, "tokens_processed_billions": 15.70766848}
+{"timestamp": 1787648167.7613368, "elapsed_hours": 5.808570676181051, "session_id": 2, "event": "train_step", "step": 119850, "epoch": 1, "loss": 1.6728859806060792, "loss_raw": 1.2651097774505615, "perplexity": 5.327520753931703, "muon_lr": 0.0018907224082820363, "adamw_lr": 5.672167224846108e-05, "tokens_per_sec": 754079.3827289981, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 15708979200, "tokens_processed_billions": 15.7089792}
+{"timestamp": 1787648175.212437, "elapsed_hours": 5.810640426211887, "session_id": 2, "event": "train_step", "step": 119860, "epoch": 1, "loss": 1.6647558391094208, "loss_raw": 1.9157077074050903, "perplexity": 5.2843828526114605, "muon_lr": 0.0018902660658097236, "adamw_lr": 5.67079819742917e-05, "tokens_per_sec": 753872.6591128176, "vram_gb": 2.785593032836914, "grad_norm": 0.09130859375, "tokens_processed": 15710289920, "tokens_processed_billions": 15.71028992}
+{"timestamp": 1787648182.6550777, "elapsed_hours": 5.812707826428944, "session_id": 2, "event": "train_step", "step": 119870, "epoch": 1, "loss": 1.6416607701778412, "loss_raw": 0.7236124277114868, "perplexity": 5.1637381773445785, "muon_lr": 0.0018898098274327956, "adamw_lr": 5.669429482298386e-05, "tokens_per_sec": 753666.3882087668, "vram_gb": 2.785593032836914, "grad_norm": 0.1015625, "tokens_processed": 15711600640, "tokens_processed_billions": 15.71160064}
+{"timestamp": 1787648190.18446, "elapsed_hours": 5.814799321492513, "session_id": 2, "event": "train_step", "step": 119880, "epoch": 1, "loss": 1.664038335084915, "loss_raw": 1.0990113019943237, "perplexity": 5.280592646554336, "muon_lr": 0.0018888976630177235, "adamw_lr": 5.66669298905317e-05, "tokens_per_sec": 753457.1326676633, "vram_gb": 2.363645076751709, "grad_norm": 0.1171875, "tokens_processed": 15712911360, "tokens_processed_billions": 15.71291136}
+{"timestamp": 1787648197.651999, "elapsed_hours": 5.816873637835185, "session_id": 2, "event": "train_step", "step": 119890, "epoch": 1, "loss": 1.6226987183094024, "loss_raw": 1.7634453773498535, "perplexity": 5.066745601516292, "muon_lr": 0.001888441737005882, "adamw_lr": 5.6653252110176446e-05, "tokens_per_sec": 753250.2595323393, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 15714222080, "tokens_processed_billions": 15.71422208}
+{"timestamp": 1787648205.1339417, "elapsed_hours": 5.818951955371433, "session_id": 2, "event": "train_step", "step": 119900, "epoch": 1, "loss": 1.6772260510921477, "loss_raw": 2.2315988540649414, "perplexity": 5.350692817341199, "muon_lr": 0.001887985915142036, "adamw_lr": 5.663957745426107e-05, "tokens_per_sec": 753043.0152532146, "vram_gb": 2.785593032836914, "grad_norm": 0.09130859375, "tokens_processed": 15715532800, "tokens_processed_billions": 15.7155328}
+{"timestamp": 1787648205.1358461, "elapsed_hours": 5.818952484395769, "session_id": 2, "event": "lambda_stats", "step": 119900, "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}