Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -5083,3 +5083,51 @@
 {"timestamp": 1787546353.7984495, "elapsed_hours": 5.480530539486144, "session_id": 2, "event": "train_step", "step": 45000, "epoch": 1, "loss": 1.8696460795402527, "loss_raw": 2.1941282749176025, "perplexity": 6.4860004647551674, "muon_lr": 0.008252115206017662, "adamw_lr": 0.0002475634561805298, "tokens_per_sec": 149559.76950524547, "vram_gb": 27.299458026885986, "grad_norm": 0.1484375, "tokens_processed": 2949120000, "tokens_processed_billions": 2.94912}
 {"timestamp": 1787546353.801489, "elapsed_hours": 5.480531384017732, "session_id": 2, "event": "lambda_stats", "step": 45000, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1896701388888888, "lambda_std": 0.2043699421791281, "lambda_all": [2.015625, 1.2421875, 1.203125, 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": 1787546367.2483637, "elapsed_hours": 5.484266627497143, "session_id": 2, "event": "validation", "step": 45000, "epoch": 1, "val_loss": 1.9518783861398696, "val_perplexity": 7.041902577470059, "is_best": true, "best_loss": 1.9518783861398696, "best_perplexity": 7.041902577470059, "train_loss": 2.1941282749176025, "train_loss_avg": 1.8696460795402527}
+{"timestamp": 1787546372.0178573, "elapsed_hours": 5.485591486162609, "session_id": 2, "event": "train_step", "step": 45010, "epoch": 1, "loss": 1.888098076581955, "loss_raw": 1.6340582370758057, "perplexity": 6.6067911133126485, "muon_lr": 0.008251510568077634, "adamw_lr": 0.00024754531704232897, "tokens_per_sec": 149454.9138569905, "vram_gb": 27.710740089416504, "grad_norm": 0.146484375, "tokens_processed": 2949775360, "tokens_processed_billions": 2.94977536}
+{"timestamp": 1787546375.9720793, "elapsed_hours": 5.486689881152577, "session_id": 2, "event": "train_step", "step": 45020, "epoch": 1, "loss": 1.8728492617607118, "loss_raw": 1.8644312620162964, "perplexity": 6.506809616085067, "muon_lr": 0.008250905850791163, "adamw_lr": 0.0002475271755237348, "tokens_per_sec": 149458.17543112504, "vram_gb": 27.710740089416504, "grad_norm": 0.1357421875, "tokens_processed": 2950430720, "tokens_processed_billions": 2.95043072}
+{"timestamp": 1787546379.9369547, "elapsed_hours": 5.487791235380702, "session_id": 2, "event": "train_step", "step": 45030, "epoch": 1, "loss": 1.8679721355438232, "loss_raw": 2.3041791915893555, "perplexity": 6.475152345321666, "muon_lr": 0.008250301054175685, "adamw_lr": 0.00024750903162527047, "tokens_per_sec": 149461.3545556796, "vram_gb": 27.710740089416504, "grad_norm": 0.1376953125, "tokens_processed": 2951086080, "tokens_processed_billions": 2.95108608}
+{"timestamp": 1787546384.0891922, "elapsed_hours": 5.488944634662734, "session_id": 2, "event": "train_step", "step": 45040, "epoch": 1, "loss": 1.8166419446468354, "loss_raw": 0.94803786277771, "perplexity": 6.151167767249003, "muon_lr": 0.008249091223027475, "adamw_lr": 0.00024747273669082417, "tokens_per_sec": 149463.11446276505, "vram_gb": 27.299458026885986, "grad_norm": 0.1376953125, "tokens_processed": 2951741440, "tokens_processed_billions": 2.95174144}
+{"timestamp": 1787546388.0377355, "elapsed_hours": 5.490041452248891, "session_id": 2, "event": "train_step", "step": 45050, "epoch": 1, "loss": 1.8976373481750488, "loss_raw": 1.7546155452728271, "perplexity": 6.6701166476309135, "muon_lr": 0.008248486188529632, "adamw_lr": 0.0002474545856558889, "tokens_per_sec": 149466.41546874004, "vram_gb": 27.710740089416504, "grad_norm": 0.1669921875, "tokens_processed": 2952396800, "tokens_processed_billions": 2.9523968}
+{"timestamp": 1787546391.9845061, "elapsed_hours": 5.491137777434455, "session_id": 2, "event": "train_step", "step": 45060, "epoch": 1, "loss": 1.8672123408317567, "loss_raw": 2.631917953491211, "perplexity": 6.470234427350488, "muon_lr": 0.008247881074772558, "adamw_lr": 0.0002474364322431767, "tokens_per_sec": 149469.72838033855, "vram_gb": 27.710740089416504, "grad_norm": 0.142578125, "tokens_processed": 2953052160, "tokens_processed_billions": 2.95305216}
+{"timestamp": 1787546395.9465315, "elapsed_hours": 5.49223834004667, "session_id": 2, "event": "train_step", "step": 45070, "epoch": 1, "loss": 1.9345861065387726, "loss_raw": 1.229803442955017, "perplexity": 6.921178831808058, "muon_lr": 0.008247275881773708, "adamw_lr": 0.0002474182764532112, "tokens_per_sec": 149472.92428790996, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 2953707520, "tokens_processed_billions": 2.95370752}
+{"timestamp": 1787546400.0986497, "elapsed_hours": 5.493391706347466, "session_id": 2, "event": "train_step", "step": 45080, "epoch": 1, "loss": 1.9136877977848052, "loss_raw": 1.6848158836364746, "perplexity": 6.778038801449493, "muon_lr": 0.008246065258120473, "adamw_lr": 0.00024738195774361415, "tokens_per_sec": 149474.68080407183, "vram_gb": 27.299458026885986, "grad_norm": 0.1474609375, "tokens_processed": 2954362880, "tokens_processed_billions": 2.95436288}
+{"timestamp": 1787546404.0831552, "elapsed_hours": 5.494498513208495, "session_id": 2, "event": "train_step", "step": 45090, "epoch": 1, "loss": 1.9381447458267211, "loss_raw": 1.356906771659851, "perplexity": 6.945852687354756, "muon_lr": 0.008245459827501005, "adamw_lr": 0.0002473637948250301, "tokens_per_sec": 149477.70504342965, "vram_gb": 27.710740089416504, "grad_norm": 0.14453125, "tokens_processed": 2955018240, "tokens_processed_billions": 2.95501824}
+{"timestamp": 1787546408.044013, "elapsed_hours": 5.4955987518363525, "session_id": 2, "event": "train_step", "step": 45100, "epoch": 1, "loss": 1.9218651580810546, "loss_raw": 3.146453857421875, "perplexity": 6.833692506905253, "muon_lr": 0.00824485431770958, "adamw_lr": 0.00024734562953128734, "tokens_per_sec": 149480.90704366812, "vram_gb": 27.710740089416504, "grad_norm": 0.1474609375, "tokens_processed": 2955673600, "tokens_processed_billions": 2.9556736}
+{"timestamp": 1787546408.0495443, "elapsed_hours": 5.495600288245413, "session_id": 2, "event": "lambda_stats", "step": 45100, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1896701388888888, "lambda_std": 0.2043699421791281, "lambda_all": [2.015625, 1.2421875, 1.203125, 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": 1787546412.0124123, "elapsed_hours": 5.496701084706518, "session_id": 2, "event": "train_step", "step": 45110, "epoch": 1, "loss": 1.8974647998809815, "loss_raw": 2.447131395339966, "perplexity": 6.668965829670827, "muon_lr": 0.008244248728763656, "adamw_lr": 0.00024732746186290966, "tokens_per_sec": 149484.05003491166, "vram_gb": 27.710740089416504, "grad_norm": 0.1513671875, "tokens_processed": 2956328960, "tokens_processed_billions": 2.95632896}
+{"timestamp": 1787546416.1645017, "elapsed_hours": 5.497854442927572, "session_id": 2, "event": "train_step", "step": 45120, "epoch": 1, "loss": 1.8714460277557372, "loss_raw": 2.275533676147461, "perplexity": 6.497685442740209, "muon_lr": 0.008243037313478183, "adamw_lr": 0.0002472911194043454, "tokens_per_sec": 149485.80315122035, "vram_gb": 27.299458026885986, "grad_norm": 0.15625, "tokens_processed": 2956984320, "tokens_processed_billions": 2.95698432}
+{"timestamp": 1787546420.1289155, "elapsed_hours": 5.498955669005712, "session_id": 2, "event": "train_step", "step": 45130, "epoch": 1, "loss": 1.8579390299320222, "loss_raw": 0.6888814568519592, "perplexity": 6.410511275575576, "muon_lr": 0.008242431487173563, "adamw_lr": 0.00024727294461520684, "tokens_per_sec": 149488.97449441347, "vram_gb": 27.710740089416504, "grad_norm": 0.142578125, "tokens_processed": 2957639680, "tokens_processed_billions": 2.95763968}
+{"timestamp": 1787546424.1056914, "elapsed_hours": 5.50006032884121, "session_id": 2, "event": "train_step", "step": 45140, "epoch": 1, "loss": 1.8357420253753662, "loss_raw": 1.516938328742981, "perplexity": 6.2697847591750175, "muon_lr": 0.008241825581784318, "adamw_lr": 0.00024725476745352947, "tokens_per_sec": 149492.05069242592, "vram_gb": 27.710740089416504, "grad_norm": 0.138671875, "tokens_processed": 2958295040, "tokens_processed_billions": 2.95829504}
+{"timestamp": 1787546428.0938094, "elapsed_hours": 5.501168139510685, "session_id": 2, "event": "train_step", "step": 45150, "epoch": 1, "loss": 1.8232556080818176, "loss_raw": 2.8418076038360596, "perplexity": 6.1919843453706624, "muon_lr": 0.008241219597327917, "adamw_lr": 0.00024723658791983745, "tokens_per_sec": 149495.040136135, "vram_gb": 27.710740089416504, "grad_norm": 0.1484375, "tokens_processed": 2958950400, "tokens_processed_billions": 2.9589504}
+{"timestamp": 1787546432.219181, "elapsed_hours": 5.5023140759600535, "session_id": 2, "event": "train_step", "step": 45160, "epoch": 1, "loss": 1.88487544298172, "loss_raw": 1.6032209396362305, "perplexity": 6.58553411643415, "muon_lr": 0.008240007391283551, "adamw_lr": 0.0002472002217385065, "tokens_per_sec": 149496.99124809031, "vram_gb": 27.299458026885986, "grad_norm": 0.15625, "tokens_processed": 2959605760, "tokens_processed_billions": 2.95960576}
+{"timestamp": 1787546436.1979458, "elapsed_hours": 5.50341928852929, "session_id": 2, "event": "train_step", "step": 45170, "epoch": 1, "loss": 1.948940815925598, "loss_raw": 1.8956232070922852, "perplexity": 7.0212468478884045, "muon_lr": 0.008239401169730546, "adamw_lr": 0.0002471820350919163, "tokens_per_sec": 149500.04978928654, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2960261120, "tokens_processed_billions": 2.96026112}
+{"timestamp": 1787546440.1697972, "elapsed_hours": 5.504522580504418, "session_id": 2, "event": "train_step", "step": 45180, "epoch": 1, "loss": 1.9768911039829253, "loss_raw": 0.9985365271568298, "perplexity": 7.220261015502503, "muon_lr": 0.008238794869180298, "adamw_lr": 0.0002471638460754089, "tokens_per_sec": 149503.15806361075, "vram_gb": 27.710740089416504, "grad_norm": 0.1669921875, "tokens_processed": 2960916480, "tokens_processed_billions": 2.96091648}
+{"timestamp": 1787546444.1328104, "elapsed_hours": 5.505623417496682, "session_id": 2, "event": "train_step", "step": 45190, "epoch": 1, "loss": 2.0331521141529083, "loss_raw": 1.7000091075897217, "perplexity": 7.6381246940379555, "muon_lr": 0.008238188489650293, "adamw_lr": 0.00024714565468950876, "tokens_per_sec": 149506.3324012187, "vram_gb": 27.710740089416504, "grad_norm": 0.146484375, "tokens_processed": 2961571840, "tokens_processed_billions": 2.96157184}
+{"timestamp": 1787546448.258733, "elapsed_hours": 5.506769507129987, "session_id": 2, "event": "train_step", "step": 45200, "epoch": 1, "loss": 1.9710051774978639, "loss_raw": 1.293487548828125, "perplexity": 7.1778879147717545, "muon_lr": 0.00823697549372096, "adamw_lr": 0.00024710926481162876, "tokens_per_sec": 149508.2756064893, "vram_gb": 27.299458026885986, "grad_norm": 0.1552734375, "tokens_processed": 2962227200, "tokens_processed_billions": 2.9622272}
+{"timestamp": 1787546448.2622263, "elapsed_hours": 5.506770477824741, "session_id": 2, "event": "lambda_stats", "step": 45200, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1896701388888888, "lambda_std": 0.2043699421791281, "lambda_all": [2.015625, 1.2421875, 1.203125, 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": 1787546452.3143675, "elapsed_hours": 5.507896072268486, "session_id": 2, "event": "train_step", "step": 45210, "epoch": 1, "loss": 1.951499238014221, "loss_raw": 1.5154045820236206, "perplexity": 7.03923315939122, "muon_lr": 0.008236368877356614, "adamw_lr": 0.00024709106632069835, "tokens_per_sec": 149510.74930431685, "vram_gb": 27.710740089416504, "grad_norm": 0.1474609375, "tokens_processed": 2962882560, "tokens_processed_billions": 2.96288256}
+{"timestamp": 1787546456.2887466, "elapsed_hours": 5.509000066518784, "session_id": 2, "event": "train_step", "step": 45220, "epoch": 1, "loss": 1.884490978717804, "loss_raw": 2.0543885231018066, "perplexity": 6.583002700558233, "muon_lr": 0.008235762182082468, "adamw_lr": 0.000247072865462474, "tokens_per_sec": 149513.83407417175, "vram_gb": 27.710740089416504, "grad_norm": 0.1337890625, "tokens_processed": 2963537920, "tokens_processed_billions": 2.96353792}
+{"timestamp": 1787546460.2666907, "elapsed_hours": 5.510105051000913, "session_id": 2, "event": "train_step", "step": 45230, "epoch": 1, "loss": 1.8100952565670014, "loss_raw": 1.2892287969589233, "perplexity": 6.111029520199257, "muon_lr": 0.00823515540791602, "adamw_lr": 0.00024705466223748053, "tokens_per_sec": 149516.89092517426, "vram_gb": 27.710740089416504, "grad_norm": 0.146484375, "tokens_processed": 2964193280, "tokens_processed_billions": 2.96419328}
+{"timestamp": 1787546464.3991592, "elapsed_hours": 5.511252959105704, "session_id": 2, "event": "train_step", "step": 45240, "epoch": 1, "loss": 1.75824143409729, "loss_raw": 1.0937647819519043, "perplexity": 5.802224822561376, "muon_lr": 0.00823394162297621, "adamw_lr": 0.0002470182486892863, "tokens_per_sec": 149518.7810994663, "vram_gb": 27.299458026885986, "grad_norm": 0.140625, "tokens_processed": 2964848640, "tokens_processed_billions": 2.96484864}
+{"timestamp": 1787546468.370714, "elapsed_hours": 5.512356168561512, "session_id": 2, "event": "train_step", "step": 45250, "epoch": 1, "loss": 1.7290899848937988, "loss_raw": 2.1563148498535156, "perplexity": 5.635523163044777, "muon_lr": 0.008233334612237853, "adamw_lr": 0.0002470000383671355, "tokens_per_sec": 149521.88351793095, "vram_gb": 27.710740089416504, "grad_norm": 0.9921875, "tokens_processed": 2965504000, "tokens_processed_billions": 2.965504}
+{"timestamp": 1787546472.3398724, "elapsed_hours": 5.513458712564574, "session_id": 2, "event": "train_step", "step": 45260, "epoch": 1, "loss": 1.7659644484519958, "loss_raw": 2.830909013748169, "perplexity": 5.8472089711784445, "muon_lr": 0.008232727522677197, "adamw_lr": 0.00024698182568031585, "tokens_per_sec": 149525.0039842417, "vram_gb": 27.710740089416504, "grad_norm": 0.146484375, "tokens_processed": 2966159360, "tokens_processed_billions": 2.96615936}
+{"timestamp": 1787546476.3179297, "elapsed_hours": 5.514563728504711, "session_id": 2, "event": "train_step", "step": 45270, "epoch": 1, "loss": 1.7709513759613038, "loss_raw": 2.073638677597046, "perplexity": 5.8764414078880325, "muon_lr": 0.00823212035431175, "adamw_lr": 0.0002469636106293525, "tokens_per_sec": 149528.05509511448, "vram_gb": 27.710740089416504, "grad_norm": 0.1494140625, "tokens_processed": 2966814720, "tokens_processed_billions": 2.96681472}
+{"timestamp": 1787546480.5124774, "elapsed_hours": 5.515728880763054, "session_id": 2, "event": "train_step", "step": 45280, "epoch": 1, "loss": 1.8452540254592895, "loss_raw": 2.138320207595825, "perplexity": 6.329707493047813, "muon_lr": 0.008230905781236527, "adamw_lr": 0.0002469271734370957, "tokens_per_sec": 149529.47318164707, "vram_gb": 27.299458026885986, "grad_norm": 0.138671875, "tokens_processed": 2967470080, "tokens_processed_billions": 2.96747008}
+{"timestamp": 1787546484.4873874, "elapsed_hours": 5.516833022236824, "session_id": 2, "event": "train_step", "step": 45290, "epoch": 1, "loss": 1.8963128995895386, "loss_raw": 1.553918719291687, "perplexity": 6.6612882687292805, "muon_lr": 0.008230298376561772, "adamw_lr": 0.00024690895129685314, "tokens_per_sec": 149532.54654950293, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2968125440, "tokens_processed_billions": 2.96812544}
+{"timestamp": 1787546488.4699054, "elapsed_hours": 5.517939277225071, "session_id": 2, "event": "train_step", "step": 45300, "epoch": 1, "loss": 1.9911220633983613, "loss_raw": 2.715916633605957, "perplexity": 7.323746861931258, "muon_lr": 0.00822969089315228, "adamw_lr": 0.00024689072679456837, "tokens_per_sec": 149535.56083414215, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2968780800, "tokens_processed_billions": 2.9687808}
+{"timestamp": 1787546488.4732232, "elapsed_hours": 5.51794019917647, "session_id": 2, "event": "lambda_stats", "step": 45300, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1896701388888888, "lambda_std": 0.2043699421791281, "lambda_all": [2.015625, 1.2421875, 1.203125, 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": 1787546492.424522, "elapsed_hours": 5.519037781821357, "session_id": 2, "event": "train_step", "step": 45310, "epoch": 1, "loss": 1.9117886984348298, "loss_raw": 1.021788239479065, "perplexity": 6.765178847397777, "muon_lr": 0.008229083331025566, "adamw_lr": 0.0002468724999307669, "tokens_per_sec": 149538.78339197938, "vram_gb": 27.710740089416504, "grad_norm": 0.13671875, "tokens_processed": 2969436160, "tokens_processed_billions": 2.96943616}
+{"timestamp": 1787546496.5353317, "elapsed_hours": 5.5201796735657584, "session_id": 2, "event": "train_step", "step": 45320, "epoch": 1, "loss": 1.9270205616950988, "loss_raw": 2.4647905826568604, "perplexity": 6.869013919790884, "muon_lr": 0.008227867970690552, "adamw_lr": 0.0002468360391207165, "tokens_per_sec": 149540.8302206376, "vram_gb": 27.299458026885986, "grad_norm": 0.1484375, "tokens_processed": 2970091520, "tokens_processed_billions": 2.97009152}
+{"timestamp": 1787546500.5587957, "elapsed_hours": 5.5212973025110035, "session_id": 2, "event": "train_step", "step": 45330, "epoch": 1, "loss": 1.9458444702625275, "loss_raw": 1.3883051872253418, "perplexity": 6.999540263548132, "muon_lr": 0.008227260172517304, "adamw_lr": 0.00024681780517551906, "tokens_per_sec": 149543.532190944, "vram_gb": 27.710740089416504, "grad_norm": 0.138671875, "tokens_processed": 2970746880, "tokens_processed_billions": 2.97074688}
+{"timestamp": 1787546504.5023098, "elapsed_hours": 5.522392722964287, "session_id": 2, "event": "train_step", "step": 45340, "epoch": 1, "loss": 1.8945965123176576, "loss_raw": 1.8138099908828735, "perplexity": 6.649864724747713, "muon_lr": 0.00822665229569693, "adamw_lr": 0.0002467995688709078, "tokens_per_sec": 149546.83533999513, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2971402240, "tokens_processed_billions": 2.97140224}
+{"timestamp": 1787546508.4517624, "elapsed_hours": 5.523489793075456, "session_id": 2, "event": "train_step", "step": 45350, "epoch": 1, "loss": 1.802307139635086, "loss_raw": 2.3302061557769775, "perplexity": 6.063620959059777, "muon_lr": 0.008226044340246957, "adamw_lr": 0.0002467813302074087, "tokens_per_sec": 149550.0932522037, "vram_gb": 27.710740089416504, "grad_norm": 0.126953125, "tokens_processed": 2972057600, "tokens_processed_billions": 2.9720576}
+{"timestamp": 1787546512.5677402, "elapsed_hours": 5.524633120232158, "session_id": 2, "event": "train_step", "step": 45360, "epoch": 1, "loss": 1.8162658596038819, "loss_raw": 2.3416967391967773, "perplexity": 6.148854840010975, "muon_lr": 0.008224828193528355, "adamw_lr": 0.00024674484580585056, "tokens_per_sec": 149552.0955680423, "vram_gb": 27.299458026885986, "grad_norm": 0.146484375, "tokens_processed": 2972712960, "tokens_processed_billions": 2.97271296}
+{"timestamp": 1787546516.5271127, "elapsed_hours": 5.525732945998509, "session_id": 2, "event": "train_step", "step": 45370, "epoch": 1, "loss": 1.7835809433460235, "loss_raw": 1.5407332181930542, "perplexity": 5.951128963633728, "muon_lr": 0.008224220002294794, "adamw_lr": 0.00024672660006884374, "tokens_per_sec": 149555.27617694155, "vram_gb": 27.710740089416504, "grad_norm": 0.1474609375, "tokens_processed": 2973368320, "tokens_processed_billions": 2.97336832}
+{"timestamp": 1787546520.4792995, "elapsed_hours": 5.5268307756053074, "session_id": 2, "event": "train_step", "step": 45380, "epoch": 1, "loss": 1.7774274289608, "loss_raw": 1.580030918121338, "perplexity": 5.914621047142206, "muon_lr": 0.008223611732501777, "adamw_lr": 0.0002467083519750532, "tokens_per_sec": 149558.50927446835, "vram_gb": 27.710740089416504, "grad_norm": 0.1416015625, "tokens_processed": 2974023680, "tokens_processed_billions": 2.97402368}
+{"timestamp": 1787546524.4740088, "elapsed_hours": 5.527940417461925, "session_id": 2, "event": "train_step", "step": 45390, "epoch": 1, "loss": 1.8215771174430848, "loss_raw": 1.0280494689941406, "perplexity": 6.18159987516783, "muon_lr": 0.008223003384166847, "adamw_lr": 0.00024669010152500535, "tokens_per_sec": 149561.42150189722, "vram_gb": 27.710740089416504, "grad_norm": 0.1728515625, "tokens_processed": 2974679040, "tokens_processed_billions": 2.97467904}
+{"timestamp": 1787546528.597397, "elapsed_hours": 5.529085802767012, "session_id": 2, "event": "train_step", "step": 45400, "epoch": 1, "loss": 1.8225671768188476, "loss_raw": 2.2872426509857178, "perplexity": 6.187723056737944, "muon_lr": 0.008221786451941415, "adamw_lr": 0.0002466535935582424, "tokens_per_sec": 149563.3649843329, "vram_gb": 27.299458026885986, "grad_norm": 0.2314453125, "tokens_processed": 2975334400, "tokens_processed_billions": 2.9753344}
+{"timestamp": 1787546528.6049693, "elapsed_hours": 5.529087906214926, "session_id": 2, "event": "lambda_stats", "step": 45400, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1896701388888888, "lambda_std": 0.2043699421791281, "lambda_all": [2.015625, 1.2421875, 1.203125, 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": 1787546532.635097, "elapsed_hours": 5.530207386149301, "session_id": 2, "event": "train_step", "step": 45410, "epoch": 1, "loss": 1.9361732876300812, "loss_raw": 2.134901285171509, "perplexity": 6.932172718315931, "muon_lr": 0.008221177868086009, "adamw_lr": 0.0002466353360425802, "tokens_per_sec": 149565.951369786, "vram_gb": 27.710740089416504, "grad_norm": 0.2041015625, "tokens_processed": 2975989760, "tokens_processed_billions": 2.97598976}
+{"timestamp": 1787546536.6046553, "elapsed_hours": 5.53131004108323, "session_id": 2, "event": "train_step", "step": 45420, "epoch": 1, "loss": 1.9713290274143218, "loss_raw": 1.7151119709014893, "perplexity": 7.1802128496179565, "muon_lr": 0.008220569205758872, "adamw_lr": 0.00024661707617276615, "tokens_per_sec": 149569.0487369863, "vram_gb": 27.710740089416504, "grad_norm": 0.1455078125, "tokens_processed": 2976645120, "tokens_processed_billions": 2.97664512}
+{"timestamp": 1787546540.6154113, "elapsed_hours": 5.532424139910274, "session_id": 2, "event": "train_step", "step": 45430, "epoch": 1, "loss": 1.9762127327919006, "loss_raw": 2.218130350112915, "perplexity": 7.215364659399137, "muon_lr": 0.008219960464977562, "adamw_lr": 0.0002465988139493268, "tokens_per_sec": 149571.83508718145, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2977300480, "tokens_processed_billions": 2.97730048}
+{"timestamp": 1787546544.7454917, "elapsed_hours": 5.533571384482913, "session_id": 2, "event": "train_step", "step": 45440, "epoch": 1, "loss": 2.0018708968162535, "loss_raw": 2.122863292694092, "perplexity": 7.40289320031961, "muon_lr": 0.008218742748122631, "adamw_lr": 0.0002465622824436789, "tokens_per_sec": 149573.72510410906, "vram_gb": 27.299458026885986, "grad_norm": 0.1552734375, "tokens_processed": 2977955840, "tokens_processed_billions": 2.97795584}