Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -2498,3 +2498,46 @@
 {"timestamp": 1787645166.2439911, "elapsed_hours": 4.97481585794025, "session_id": 2, "event": "lambda_stats", "step": 116100, "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": 1787645173.6927178, "elapsed_hours": 4.976884948677487, "session_id": 2, "event": "train_step", "step": 116110, "epoch": 1, "loss": 1.8350332820415496, "loss_raw": 1.4731066226959229, "perplexity": 6.265342665360622, "muon_lr": 0.002115482522387196, "adamw_lr": 6.346447567161588e-05, "tokens_per_sec": 853169.1475171938, "vram_gb": 2.785593032836914, "grad_norm": 0.099609375, "tokens_processed": 15218769920, "tokens_processed_billions": 15.21876992}
 {"timestamp": 1787645181.2393918, "elapsed_hours": 4.978981246948242, "session_id": 2, "event": "train_step", "step": 116120, "epoch": 1, "loss": 1.7653577959537505, "loss_raw": 0.906911313533783, "perplexity": 5.8436628229969, "muon_lr": 0.0021144754912434376, "adamw_lr": 6.343426473730311e-05, "tokens_per_sec": 852881.7991289094, "vram_gb": 2.363645076751709, "grad_norm": 0.08837890625, "tokens_processed": 15220080640, "tokens_processed_billions": 15.22008064}
+{"timestamp": 1787645189.2787158, "elapsed_hours": 4.981214392582576, "session_id": 2, "event": "train_step", "step": 116130, "epoch": 1, "loss": 1.7652350962162018, "loss_raw": 2.4559998512268066, "perplexity": 5.842945851089227, "muon_lr": 0.002113972122108949, "adamw_lr": 6.341916366326845e-05, "tokens_per_sec": 852571.166732969, "vram_gb": 2.785593032836914, "grad_norm": 0.09228515625, "tokens_processed": 15221391360, "tokens_processed_billions": 15.22139136}
+{"timestamp": 1787645196.7260783, "elapsed_hours": 4.983283104366726, "session_id": 2, "event": "train_step", "step": 116140, "epoch": 1, "loss": 1.689493545293808, "loss_raw": 0.7785462141036987, "perplexity": 5.416736678544218, "muon_lr": 0.0021134688506187407, "adamw_lr": 6.340406551856222e-05, "tokens_per_sec": 852289.0708561733, "vram_gb": 2.785593032836914, "grad_norm": 0.09130859375, "tokens_processed": 15222702080, "tokens_processed_billions": 15.22270208}
+{"timestamp": 1787645204.206612, "elapsed_hours": 4.98536103049914, "session_id": 2, "event": "train_step", "step": 116150, "epoch": 1, "loss": 1.6216066634655, "loss_raw": 1.4340379238128662, "perplexity": 5.0612154575991575, "muon_lr": 0.00211296567678733, "adamw_lr": 6.338897030361989e-05, "tokens_per_sec": 852005.6090743063, "vram_gb": 2.785593032836914, "grad_norm": 0.08984375, "tokens_processed": 15224012800, "tokens_processed_billions": 15.2240128}
+{"timestamp": 1787645211.751574, "elapsed_hours": 4.98745685312483, "session_id": 2, "event": "train_step", "step": 116160, "epoch": 1, "loss": 1.5842113876342774, "loss_raw": 2.1752474308013916, "perplexity": 4.8754450258356545, "muon_lr": 0.0021119596221589295, "adamw_lr": 6.335878866476788e-05, "tokens_per_sec": 851719.3212051253, "vram_gb": 2.363645076751709, "grad_norm": 0.0908203125, "tokens_processed": 15225323520, "tokens_processed_billions": 15.22532352}
+{"timestamp": 1787645219.215136, "elapsed_hours": 4.989530064794752, "session_id": 2, "event": "train_step", "step": 116170, "epoch": 1, "loss": 1.63163818359375, "loss_raw": 1.6788967847824097, "perplexity": 5.1122426546015936, "muon_lr": 0.002111456741390954, "adamw_lr": 6.334370224172861e-05, "tokens_per_sec": 851437.1657451658, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15226634240, "tokens_processed_billions": 15.22663424}
+{"timestamp": 1787645226.6890523, "elapsed_hours": 4.991606152719921, "session_id": 2, "event": "train_step", "step": 116180, "epoch": 1, "loss": 1.618009202480316, "loss_raw": 1.3726776838302612, "perplexity": 5.043040643646007, "muon_lr": 0.0021109539583397965, "adamw_lr": 6.332861875019388e-05, "tokens_per_sec": 851154.7303468805, "vram_gb": 2.785593032836914, "grad_norm": 0.09423828125, "tokens_processed": 15227944960, "tokens_processed_billions": 15.22794496}
+{"timestamp": 1787645234.1374042, "elapsed_hours": 4.993675139281485, "session_id": 2, "event": "train_step", "step": 116190, "epoch": 1, "loss": 1.676438205242157, "loss_raw": 1.6209667921066284, "perplexity": 5.3464789563648845, "muon_lr": 0.0021104512730199584, "adamw_lr": 6.331353819059873e-05, "tokens_per_sec": 850873.7562398321, "vram_gb": 2.785593032836914, "grad_norm": 0.10498046875, "tokens_processed": 15229255680, "tokens_processed_billions": 15.22925568}
+{"timestamp": 1787645241.690888, "elapsed_hours": 4.995773329271211, "session_id": 2, "event": "train_step", "step": 116200, "epoch": 1, "loss": 1.7310111355781554, "loss_raw": 2.455929756164551, "perplexity": 5.646360258739926, "muon_lr": 0.002109446195632217, "adamw_lr": 6.32833858689665e-05, "tokens_per_sec": 850588.0209369442, "vram_gb": 2.363645076751709, "grad_norm": 0.0966796875, "tokens_processed": 15230566400, "tokens_processed_billions": 15.2305664}
+{"timestamp": 1787645241.6949873, "elapsed_hours": 4.995774467984836, "session_id": 2, "event": "lambda_stats", "step": 116200, "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": 1787645249.1873167, "elapsed_hours": 4.997855670518345, "session_id": 2, "event": "train_step", "step": 116210, "epoch": 1, "loss": 1.7785103976726533, "loss_raw": 1.046689510345459, "perplexity": 5.921029866327692, "muon_lr": 0.002108943803593299, "adamw_lr": 6.326831410779895e-05, "tokens_per_sec": 850305.2441552202, "vram_gb": 2.785593032836914, "grad_norm": 0.1044921875, "tokens_processed": 15231877120, "tokens_processed_billions": 15.23187712}
+{"timestamp": 1787645256.6516218, "elapsed_hours": 4.99992908861902, "session_id": 2, "event": "train_step", "step": 116220, "epoch": 1, "loss": 1.8244523072242738, "loss_raw": 2.0745036602020264, "perplexity": 6.1993987232318455, "muon_lr": 0.0021084415093436653, "adamw_lr": 6.325324528030994e-05, "tokens_per_sec": 850024.2189634135, "vram_gb": 2.785593032836914, "grad_norm": 0.10498046875, "tokens_processed": 15233187840, "tokens_processed_billions": 15.23318784}
+{"timestamp": 1787645264.1385078, "elapsed_hours": 5.0020087791813745, "session_id": 2, "event": "train_step", "step": 116230, "epoch": 1, "loss": 1.8544120585918427, "loss_raw": 1.5064787864685059, "perplexity": 6.387941411059904, "muon_lr": 0.0021079393128978037, "adamw_lr": 6.32381793869341e-05, "tokens_per_sec": 849742.3501869991, "vram_gb": 2.785593032836914, "grad_norm": 0.10498046875, "tokens_processed": 15234498560, "tokens_processed_billions": 15.23449856}
+{"timestamp": 1787645271.7293632, "elapsed_hours": 5.004117350180944, "session_id": 2, "event": "train_step", "step": 116240, "epoch": 1, "loss": 1.8802543008327484, "loss_raw": 2.1340816020965576, "perplexity": 6.555171635857008, "muon_lr": 0.002106935213475317, "adamw_lr": 6.32080564042595e-05, "tokens_per_sec": 849455.7986304476, "vram_gb": 2.363645076751709, "grad_norm": 0.10595703125, "tokens_processed": 15235809280, "tokens_processed_billions": 15.23580928}
+{"timestamp": 1787645279.2352884, "elapsed_hours": 5.006202329397201, "session_id": 2, "event": "train_step", "step": 116250, "epoch": 1, "loss": 1.8766420257091523, "loss_raw": 1.4355175495147705, "perplexity": 6.531535268658172, "muon_lr": 0.002106433310527649, "adamw_lr": 6.319299931582946e-05, "tokens_per_sec": 849173.5114909334, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 15237120000, "tokens_processed_billions": 15.23712}
+{"timestamp": 1787645286.6794305, "elapsed_hours": 5.008270146648089, "session_id": 2, "event": "train_step", "step": 116260, "epoch": 1, "loss": 1.7983631885051727, "loss_raw": 1.3636263608932495, "perplexity": 6.039753431445956, "muon_lr": 0.002105931505441661, "adamw_lr": 6.317794516324981e-05, "tokens_per_sec": 848894.3824985678, "vram_gb": 2.785593032836914, "grad_norm": 0.0908203125, "tokens_processed": 15238430720, "tokens_processed_billions": 15.23843072}
+{"timestamp": 1787645294.150083, "elapsed_hours": 5.010345328052838, "session_id": 2, "event": "train_step", "step": 116270, "epoch": 1, "loss": 1.709423998594284, "loss_raw": 2.0909371376037598, "perplexity": 5.525777705057505, "muon_lr": 0.0021054297982318272, "adamw_lr": 6.316289394695481e-05, "tokens_per_sec": 848614.2292456276, "vram_gb": 2.785593032836914, "grad_norm": 0.08984375, "tokens_processed": 15239741440, "tokens_processed_billions": 15.23974144}
+{"timestamp": 1787645301.6897283, "elapsed_hours": 5.0124396737416586, "session_id": 2, "event": "train_step", "step": 116280, "epoch": 1, "loss": 1.6709943556785583, "loss_raw": 1.4183390140533447, "perplexity": 5.317452608451354, "muon_lr": 0.0021044266774984857, "adamw_lr": 6.313280032495456e-05, "tokens_per_sec": 848331.0518350297, "vram_gb": 2.363645076751709, "grad_norm": 0.09228515625, "tokens_processed": 15241052160, "tokens_processed_billions": 15.24105216}
+{"timestamp": 1787645309.1352153, "elapsed_hours": 5.014507864647442, "session_id": 2, "event": "train_step", "step": 116290, "epoch": 1, "loss": 1.6959595417976379, "loss_raw": 2.660900831222534, "perplexity": 5.451874757894732, "muon_lr": 0.002103925264003905, "adamw_lr": 6.311775792011714e-05, "tokens_per_sec": 848052.5569628216, "vram_gb": 2.785593032836914, "grad_norm": 0.09912109375, "tokens_processed": 15242362880, "tokens_processed_billions": 15.24236288}
+{"timestamp": 1787645316.5741866, "elapsed_hours": 5.016574245558845, "session_id": 2, "event": "train_step", "step": 116300, "epoch": 1, "loss": 1.66844801902771, "loss_raw": 1.9295426607131958, "perplexity": 5.30392980809302, "muon_lr": 0.002103423948443334, "adamw_lr": 6.31027184533e-05, "tokens_per_sec": 847774.5988581671, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15243673600, "tokens_processed_billions": 15.2436736}
+{"timestamp": 1787645316.5761101, "elapsed_hours": 5.016574779881371, "session_id": 2, "event": "lambda_stats", "step": 116300, "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": 1787645324.0182776, "elapsed_hours": 5.018642048570845, "session_id": 2, "event": "train_step", "step": 116310, "epoch": 1, "loss": 1.7146147465705872, "loss_raw": 1.5359290838241577, "perplexity": 5.554535196372657, "muon_lr": 0.0021029227308312264, "adamw_lr": 6.308768192493678e-05, "tokens_per_sec": 847496.6285715568, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 15244984320, "tokens_processed_billions": 15.24498432}
+{"timestamp": 1787645331.5517986, "elapsed_hours": 5.020734693275558, "session_id": 2, "event": "train_step", "step": 116320, "epoch": 1, "loss": 1.7402876496315003, "loss_raw": 1.5386557579040527, "perplexity": 5.698982497136334, "muon_lr": 0.0021019205895102156, "adamw_lr": 6.305761768530646e-05, "tokens_per_sec": 847214.6759007457, "vram_gb": 2.363645076751709, "grad_norm": 0.09716796875, "tokens_processed": 15246295040, "tokens_processed_billions": 15.24629504}
+{"timestamp": 1787645338.9968581, "elapsed_hours": 5.022802765369415, "session_id": 2, "event": "train_step", "step": 116330, "epoch": 1, "loss": 1.7556979060173035, "loss_raw": 1.88978111743927, "perplexity": 5.787485453744434, "muon_lr": 0.0021014196658302123, "adamw_lr": 6.304258997490636e-05, "tokens_per_sec": 846937.1255901096, "vram_gb": 2.785593032836914, "grad_norm": 0.09716796875, "tokens_processed": 15247605760, "tokens_processed_billions": 15.24760576}
+{"timestamp": 1787645346.4353516, "elapsed_hours": 5.024869013561143, "session_id": 2, "event": "train_step", "step": 116340, "epoch": 1, "loss": 1.6630049228668213, "loss_raw": 1.5528607368469238, "perplexity": 5.275138436303765, "muon_lr": 0.00210091884015647, "adamw_lr": 6.302756520469409e-05, "tokens_per_sec": 846660.1117987237, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 15248916480, "tokens_processed_billions": 15.24891648}
+{"timestamp": 1787645353.8841507, "elapsed_hours": 5.026938124431505, "session_id": 2, "event": "train_step", "step": 116350, "epoch": 1, "loss": 1.6743361258506775, "loss_raw": 2.1248834133148193, "perplexity": 5.335252037205496, "muon_lr": 0.0021004181125034342, "adamw_lr": 6.301254337510302e-05, "tokens_per_sec": 846382.8410407804, "vram_gb": 2.785593032836914, "grad_norm": 0.091796875, "tokens_processed": 15250227200, "tokens_processed_billions": 15.2502272}
+{"timestamp": 1787645361.417103, "elapsed_hours": 5.0290306111839085, "session_id": 2, "event": "train_step", "step": 116360, "epoch": 1, "loss": 1.663865203857422, "loss_raw": 2.105548620223999, "perplexity": 5.279678490204335, "muon_lr": 0.0020994169513172322, "adamw_lr": 6.298250853951696e-05, "tokens_per_sec": 846101.8480209053, "vram_gb": 2.363645076751709, "grad_norm": 0.09765625, "tokens_processed": 15251537920, "tokens_processed_billions": 15.25153792}
+{"timestamp": 1787645368.8540702, "elapsed_hours": 5.031096435520384, "session_id": 2, "event": "train_step", "step": 116370, "epoch": 1, "loss": 1.6494638013839722, "loss_raw": 1.7780134677886963, "perplexity": 5.204188600195728, "muon_lr": 0.0020989165178129385, "adamw_lr": 6.296749553438815e-05, "tokens_per_sec": 845825.6001970865, "vram_gb": 2.785593032836914, "grad_norm": 0.0908203125, "tokens_processed": 15252848640, "tokens_processed_billions": 15.25284864}
+{"timestamp": 1787645376.301116, "elapsed_hours": 5.0331650592221155, "session_id": 2, "event": "train_step", "step": 116380, "epoch": 1, "loss": 1.6255150866508483, "loss_raw": 1.8510587215423584, "perplexity": 5.08103553683772, "muon_lr": 0.0020984161823870916, "adamw_lr": 6.295248547161274e-05, "tokens_per_sec": 845549.0982932064, "vram_gb": 2.785593032836914, "grad_norm": 0.09814453125, "tokens_processed": 15254159360, "tokens_processed_billions": 15.25415936}
+{"timestamp": 1787645383.7455611, "elapsed_hours": 5.035232960714234, "session_id": 2, "event": "train_step", "step": 116390, "epoch": 1, "loss": 1.6834299528598786, "loss_raw": 1.4464797973632812, "perplexity": 5.383991173033197, "muon_lr": 0.0020979159450541197, "adamw_lr": 6.29374783516236e-05, "tokens_per_sec": 845272.9513341876, "vram_gb": 2.785593032836914, "grad_norm": 0.09619140625, "tokens_processed": 15255470080, "tokens_processed_billions": 15.25547008}
+{"timestamp": 1787645391.2801785, "elapsed_hours": 5.037325909932455, "session_id": 2, "event": "train_step", "step": 116400, "epoch": 1, "loss": 1.7174978315830232, "loss_raw": 1.2725887298583984, "perplexity": 5.5705725008957785, "muon_lr": 0.0020969157647244994, "adamw_lr": 6.290747294173498e-05, "tokens_per_sec": 844992.8071124246, "vram_gb": 2.363645076751709, "grad_norm": 0.09423828125, "tokens_processed": 15256780800, "tokens_processed_billions": 15.2567808}
+{"timestamp": 1787645391.2818818, "elapsed_hours": 5.037326383127106, "session_id": 2, "event": "lambda_stats", "step": 116400, "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": 1787645398.7189257, "elapsed_hours": 5.039392228722573, "session_id": 2, "event": "train_step", "step": 116410, "epoch": 1, "loss": 1.7075550973415374, "loss_raw": 1.7867904901504517, "perplexity": 5.515460216368962, "muon_lr": 0.0020964158217566947, "adamw_lr": 6.289247465270082e-05, "tokens_per_sec": 844717.3844984043, "vram_gb": 2.785593032836914, "grad_norm": 0.09228515625, "tokens_processed": 15258091520, "tokens_processed_billions": 15.25809152}
+{"timestamp": 1787645406.160519, "elapsed_hours": 5.041459337804053, "session_id": 2, "event": "train_step", "step": 116420, "epoch": 1, "loss": 1.7356516897678376, "loss_raw": 2.20475697517395, "perplexity": 5.672623390106426, "muon_lr": 0.00209591597693945, "adamw_lr": 6.28774793081835e-05, "tokens_per_sec": 844442.054375357, "vram_gb": 2.785593032836914, "grad_norm": 0.10009765625, "tokens_processed": 15259402240, "tokens_processed_billions": 15.25940224}
+{"timestamp": 1787645413.607787, "elapsed_hours": 5.0435280234946145, "session_id": 2, "event": "train_step", "step": 116430, "epoch": 1, "loss": 1.7596453630924225, "loss_raw": 2.195080041885376, "perplexity": 5.8103764550437305, "muon_lr": 0.0020954162302871814, "adamw_lr": 6.286248690861543e-05, "tokens_per_sec": 844166.6864011781, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15260712960, "tokens_processed_billions": 15.26071296}
+{"timestamp": 1787645421.138361, "elapsed_hours": 5.0456198495626445, "session_id": 2, "event": "train_step", "step": 116440, "epoch": 1, "loss": 1.7196798372268676, "loss_raw": 1.0422755479812622, "perplexity": 5.582740792344566, "muon_lr": 0.002094417031535208, "adamw_lr": 6.283251094605624e-05, "tokens_per_sec": 843887.6563083823, "vram_gb": 2.363645076751709, "grad_norm": 0.0947265625, "tokens_processed": 15262023680, "tokens_processed_billions": 15.26202368}
+{"timestamp": 1787645428.5772092, "elapsed_hours": 5.0476861963007185, "session_id": 2, "event": "train_step", "step": 116450, "epoch": 1, "loss": 1.7050150156021118, "loss_raw": 1.1793406009674072, "perplexity": 5.501468274449936, "muon_lr": 0.0020939175794643192, "adamw_lr": 6.281752738392956e-05, "tokens_per_sec": 843613.1393799131, "vram_gb": 2.785593032836914, "grad_norm": 0.09375, "tokens_processed": 15263334400, "tokens_processed_billions": 15.2633344}
+{"timestamp": 1787645436.0236053, "elapsed_hours": 5.049754639665286, "session_id": 2, "event": "train_step", "step": 116460, "epoch": 1, "loss": 1.72848415851593, "loss_raw": 2.04038667678833, "perplexity": 5.6321100484413344, "muon_lr": 0.0020934182256160338, "adamw_lr": 6.280254676848101e-05, "tokens_per_sec": 843338.4936846207, "vram_gb": 2.785593032836914, "grad_norm": 0.09765625, "tokens_processed": 15264645120, "tokens_processed_billions": 15.26464512}
+{"timestamp": 1787645443.4614115, "elapsed_hours": 5.051820696923468, "session_id": 2, "event": "train_step", "step": 116470, "epoch": 1, "loss": 1.7619812750816346, "loss_raw": 1.4524809122085571, "perplexity": 5.823964847532629, "muon_lr": 0.0020929189700047528, "adamw_lr": 6.278756910014257e-05, "tokens_per_sec": 843064.4754514656, "vram_gb": 2.785593032836914, "grad_norm": 0.095703125, "tokens_processed": 15265955840, "tokens_processed_billions": 15.26595584}
+{"timestamp": 1787645450.9984965, "elapsed_hours": 5.05391433166133, "session_id": 2, "event": "train_step", "step": 116480, "epoch": 1, "loss": 1.7477490210533142, "loss_raw": 1.469327688217163, "perplexity": 5.741663754618675, "muon_lr": 0.002091920753550785, "adamw_lr": 6.275762260652353e-05, "tokens_per_sec": 842786.0563501037, "vram_gb": 2.363645076751709, "grad_norm": 0.095703125, "tokens_processed": 15267266560, "tokens_processed_billions": 15.26726656}
+{"timestamp": 1787645458.4446242, "elapsed_hours": 5.055982700387637, "session_id": 2, "event": "train_step", "step": 116490, "epoch": 1, "loss": 1.7387177801132203, "loss_raw": 0.9741166830062866, "perplexity": 5.69004285709904, "muon_lr": 0.0020914217927368835, "adamw_lr": 6.27426537821065e-05, "tokens_per_sec": 842512.1044724664, "vram_gb": 2.785593032836914, "grad_norm": 0.09228515625, "tokens_processed": 15268577280, "tokens_processed_billions": 15.26857728}
+{"timestamp": 1787645465.8852496, "elapsed_hours": 5.058049540850852, "session_id": 2, "event": "train_step", "step": 116500, "epoch": 1, "loss": 1.759345495700836, "loss_raw": 2.117434024810791, "perplexity": 5.808634373821751, "muon_lr": 0.0020909229302175595, "adamw_lr": 6.272768790652677e-05, "tokens_per_sec": 842238.632584897, "vram_gb": 2.785593032836914, "grad_norm": 0.09521484375, "tokens_processed": 15269888000, "tokens_processed_billions": 15.269888}
+{"timestamp": 1787645465.8868923, "elapsed_hours": 5.0580499972237485, "session_id": 2, "event": "lambda_stats", "step": 116500, "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": 1787645491.5499318, "elapsed_hours": 5.065178619159592, "session_id": 2, "event": "validation", "step": 116500, "epoch": 1, "val_loss": 1.754016865491867, "val_perplexity": 5.777764628995659, "is_best": false, "best_loss": 1.6233880519866943, "best_perplexity": 5.070239483980759, "train_loss": 2.117434024810791, "train_loss_avg": 1.759345495700836}