Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -4974,3 +4974,56 @@
 {"timestamp": 1787545936.8273687, "elapsed_hours": 5.364705239269469, "session_id": 2, "event": "train_step", "step": 44010, "epoch": 1, "loss": 1.9135370218753815, "loss_raw": 2.0727460384368896, "perplexity": 6.777016913525072, "muon_lr": 0.00832646972318226, "adamw_lr": 0.00024979409169546775, "tokens_per_sec": 149429.2872532583, "vram_gb": 27.710740089416504, "grad_norm": 0.138671875, "tokens_processed": 2884239360, "tokens_processed_billions": 2.88423936}
 {"timestamp": 1787545940.8919528, "elapsed_hours": 5.365834290385246, "session_id": 2, "event": "train_step", "step": 44020, "epoch": 1, "loss": 1.9076926362514497, "loss_raw": 1.0351283550262451, "perplexity": 6.73752492890617, "muon_lr": 0.00832587505859982, "adamw_lr": 0.00024977625175799455, "tokens_per_sec": 149431.7732074312, "vram_gb": 27.710740089416504, "grad_norm": 0.1474609375, "tokens_processed": 2884894720, "tokens_processed_billions": 2.88489472}
 {"timestamp": 1787545944.9694374, "elapsed_hours": 5.36696692513095, "session_id": 2, "event": "train_step", "step": 44030, "epoch": 1, "loss": 1.9391293299198151, "loss_raw": 2.5928168296813965, "perplexity": 6.952694831203964, "muon_lr": 0.008325280312526454, "adamw_lr": 0.00024975840937579356, "tokens_per_sec": 149434.1583967027, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2885550080, "tokens_processed_billions": 2.88555008}
+{"timestamp": 1787545950.3096294, "elapsed_hours": 5.3684503117534845, "session_id": 2, "event": "train_step", "step": 44040, "epoch": 1, "loss": 1.9973743426799775, "loss_raw": 2.0871591567993164, "perplexity": 7.3696804177774125, "muon_lr": 0.008324090575975558, "adamw_lr": 0.0002497227172792667, "tokens_per_sec": 149426.7723898771, "vram_gb": 27.299458026885986, "grad_norm": 0.15234375, "tokens_processed": 2886205440, "tokens_processed_billions": 2.88620544}
+{"timestamp": 1787545954.376395, "elapsed_hours": 5.369579968849818, "session_id": 2, "event": "train_step", "step": 44050, "epoch": 1, "loss": 2.0295571887493136, "loss_raw": 1.2518608570098877, "perplexity": 7.610715502058329, "muon_lr": 0.008323495585532335, "adamw_lr": 0.00024970486756596997, "tokens_per_sec": 149429.24152424716, "vram_gb": 27.710740089416504, "grad_norm": 0.1689453125, "tokens_processed": 2886860800, "tokens_processed_billions": 2.8868608}
+{"timestamp": 1787545958.500416, "elapsed_hours": 5.370725530187289, "session_id": 2, "event": "train_step", "step": 44060, "epoch": 1, "loss": 2.097292774915695, "loss_raw": 1.9952689409255981, "perplexity": 8.144092150757057, "muon_lr": 0.008322900513666805, "adamw_lr": 0.00024968701541000407, "tokens_per_sec": 149431.26520034237, "vram_gb": 27.710740089416504, "grad_norm": 0.1416015625, "tokens_processed": 2887516160, "tokens_processed_billions": 2.88751616}
+{"timestamp": 1787545962.5294762, "elapsed_hours": 5.37184471362167, "session_id": 2, "event": "train_step", "step": 44070, "epoch": 1, "loss": 2.0252710807323457, "loss_raw": 2.7960996627807617, "perplexity": 7.578164960729565, "muon_lr": 0.008322305360396128, "adamw_lr": 0.00024966916081188377, "tokens_per_sec": 149434.02163779634, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 2888171520, "tokens_processed_billions": 2.88817152}
+{"timestamp": 1787545966.6451786, "elapsed_hours": 5.372987964351972, "session_id": 2, "event": "train_step", "step": 44080, "epoch": 1, "loss": 2.067440085411072, "loss_raw": 1.5082578659057617, "perplexity": 7.90456219172005, "muon_lr": 0.008321114809707993, "adamw_lr": 0.00024963344429123973, "tokens_per_sec": 149436.1080439262, "vram_gb": 27.299458026885986, "grad_norm": 0.1455078125, "tokens_processed": 2888826880, "tokens_processed_billions": 2.88882688}
+{"timestamp": 1787545970.6918683, "elapsed_hours": 5.37411204483774, "session_id": 2, "event": "train_step", "step": 44090, "epoch": 1, "loss": 2.018800985813141, "loss_raw": 2.781024932861328, "perplexity": 7.529291791636226, "muon_lr": 0.008320519412324864, "adamw_lr": 0.0002496155823697458, "tokens_per_sec": 149438.7274868659, "vram_gb": 27.710740089416504, "grad_norm": 0.140625, "tokens_processed": 2889482240, "tokens_processed_billions": 2.88948224}
+{"timestamp": 1787545974.6638346, "elapsed_hours": 5.375215368602011, "session_id": 2, "event": "train_step", "step": 44100, "epoch": 1, "loss": 1.957155395746231, "loss_raw": 1.7922612428665161, "perplexity": 7.079160985045102, "muon_lr": 0.008319923933605252, "adamw_lr": 0.00024959771800815754, "tokens_per_sec": 149441.92185670833, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 2890137600, "tokens_processed_billions": 2.8901376}
+{"timestamp": 1787545974.6669827, "elapsed_hours": 5.375216243200832, "session_id": 2, "event": "lambda_stats", "step": 44100, "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": 1787545978.6925147, "elapsed_hours": 5.37633444653617, "session_id": 2, "event": "train_step", "step": 44110, "epoch": 1, "loss": 1.9346161222457885, "loss_raw": 2.152132034301758, "perplexity": 6.9213865790019025, "muon_lr": 0.008319328373566335, "adamw_lr": 0.00024957985120699, "tokens_per_sec": 149444.6787186826, "vram_gb": 27.710740089416504, "grad_norm": 0.1416015625, "tokens_processed": 2890792960, "tokens_processed_billions": 2.89079296}
+{"timestamp": 1787545982.8188443, "elapsed_hours": 5.377480649219619, "session_id": 2, "event": "train_step", "step": 44120, "epoch": 1, "loss": 2.0278435373306274, "loss_raw": 1.5529837608337402, "perplexity": 7.597684557078576, "muon_lr": 0.008318137009599266, "adamw_lr": 0.00024954411028797795, "tokens_per_sec": 149446.67854812945, "vram_gb": 27.299458026885986, "grad_norm": 0.154296875, "tokens_processed": 2891448320, "tokens_processed_billions": 2.89144832}
+{"timestamp": 1787545986.7780313, "elapsed_hours": 5.378580423394839, "session_id": 2, "event": "train_step", "step": 44130, "epoch": 1, "loss": 2.081037129163742, "loss_raw": 2.6342239379882812, "perplexity": 8.01277488888736, "muon_lr": 0.008317541205705475, "adamw_lr": 0.0002495262361711642, "tokens_per_sec": 149449.969123368, "vram_gb": 27.710740089416504, "grad_norm": 0.1630859375, "tokens_processed": 2892103680, "tokens_processed_billions": 2.89210368}
+{"timestamp": 1787545990.7467115, "elapsed_hours": 5.379682834744454, "session_id": 2, "event": "train_step", "step": 44140, "epoch": 1, "loss": 2.0443720245361328, "loss_raw": 1.2404149770736694, "perplexity": 7.724306339681906, "muon_lr": 0.008316945320561087, "adamw_lr": 0.0002495083596168326, "tokens_per_sec": 149453.184777647, "vram_gb": 27.710740089416504, "grad_norm": 0.1494140625, "tokens_processed": 2892759040, "tokens_processed_billions": 2.89275904}
+{"timestamp": 1787545994.7260518, "elapsed_hours": 5.380788206855456, "session_id": 2, "event": "train_step", "step": 44150, "epoch": 1, "loss": 1.9983444821834564, "loss_raw": 1.519646406173706, "perplexity": 7.376833505063964, "muon_lr": 0.008316349354183284, "adamw_lr": 0.00024949048062549844, "tokens_per_sec": 149456.31719415623, "vram_gb": 27.710740089416504, "grad_norm": 0.1494140625, "tokens_processed": 2893414400, "tokens_processed_billions": 2.8934144}
+{"timestamp": 1787545998.8622818, "elapsed_hours": 5.38193715956476, "session_id": 2, "event": "train_step", "step": 44160, "epoch": 1, "loss": 2.090783352851868, "loss_raw": 2.121457099914551, "perplexity": 8.091250986924493, "muon_lr": 0.008315157177796184, "adamw_lr": 0.00024945471533388547, "tokens_per_sec": 149458.2361943221, "vram_gb": 27.299458026885986, "grad_norm": 0.162109375, "tokens_processed": 2894069760, "tokens_processed_billions": 2.89406976}
+{"timestamp": 1787546002.9035096, "elapsed_hours": 5.383059722847409, "session_id": 2, "event": "train_step", "step": 44170, "epoch": 1, "loss": 2.0883253145217897, "loss_raw": 2.0721020698547363, "perplexity": 8.071386805321774, "muon_lr": 0.008314560967821268, "adamw_lr": 0.000249436829034638, "tokens_per_sec": 149460.888340775, "vram_gb": 27.710740089416504, "grad_norm": 0.1494140625, "tokens_processed": 2894725120, "tokens_processed_billions": 2.89472512}
+{"timestamp": 1787546006.9865108, "elapsed_hours": 5.384193889896075, "session_id": 2, "event": "train_step", "step": 44180, "epoch": 1, "loss": 2.005866072177887, "loss_raw": 1.1133052110671997, "perplexity": 7.432528215962417, "muon_lr": 0.008313964676681693, "adamw_lr": 0.00024941894030045073, "tokens_per_sec": 149463.2173583388, "vram_gb": 27.710740089416504, "grad_norm": 0.166015625, "tokens_processed": 2895380480, "tokens_processed_billions": 2.89538048}
+{"timestamp": 1787546011.010744, "elapsed_hours": 5.385311732358403, "session_id": 2, "event": "train_step", "step": 44190, "epoch": 1, "loss": 2.016747108697891, "loss_raw": 1.945765495300293, "perplexity": 7.513843421488934, "muon_lr": 0.00831336830439466, "adamw_lr": 0.00024940104913183977, "tokens_per_sec": 149465.99821224625, "vram_gb": 27.710740089416504, "grad_norm": 0.14453125, "tokens_processed": 2896035840, "tokens_processed_billions": 2.89603584}
+{"timestamp": 1787546015.1311443, "elapsed_hours": 5.386456287966834, "session_id": 2, "event": "train_step", "step": 44200, "epoch": 1, "loss": 2.0421189844608305, "loss_raw": 2.39070463180542, "perplexity": 7.706922758251029, "muon_lr": 0.008312175316447009, "adamw_lr": 0.00024936525949341025, "tokens_per_sec": 149468.03664490845, "vram_gb": 27.299458026885986, "grad_norm": 0.1533203125, "tokens_processed": 2896691200, "tokens_processed_billions": 2.8966912}
+{"timestamp": 1787546015.1343713, "elapsed_hours": 5.386457184619374, "session_id": 2, "event": "lambda_stats", "step": 44200, "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": 1787546019.2393441, "elapsed_hours": 5.387597454653846, "session_id": 2, "event": "train_step", "step": 44210, "epoch": 1, "loss": 2.0859067606925965, "loss_raw": 2.5904548168182373, "perplexity": 8.051889309232834, "muon_lr": 0.008311578700820794, "adamw_lr": 0.0002493473610246238, "tokens_per_sec": 149470.1681182661, "vram_gb": 27.710740089416504, "grad_norm": 0.166015625, "tokens_processed": 2897346560, "tokens_processed_billions": 2.89734656}
+{"timestamp": 1787546023.20307, "elapsed_hours": 5.388698489665985, "session_id": 2, "event": "train_step", "step": 44220, "epoch": 1, "loss": 2.0306456017494203, "loss_raw": 1.9606704711914062, "perplexity": 7.619003613376732, "muon_lr": 0.008310982004115924, "adamw_lr": 0.0002493294601234777, "tokens_per_sec": 149473.41267206654, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2898001920, "tokens_processed_billions": 2.89800192}
+{"timestamp": 1787546027.2452574, "elapsed_hours": 5.389821319580078, "session_id": 2, "event": "train_step", "step": 44230, "epoch": 1, "loss": 2.022205457687378, "loss_raw": 1.8801902532577515, "perplexity": 7.554968737192692, "muon_lr": 0.008310385226349606, "adamw_lr": 0.0002493115567904881, "tokens_per_sec": 149476.05072778507, "vram_gb": 27.710740089416504, "grad_norm": 0.142578125, "tokens_processed": 2898657280, "tokens_processed_billions": 2.89865728}
+{"timestamp": 1787546031.3657403, "elapsed_hours": 5.390965898036956, "session_id": 2, "event": "train_step", "step": 44240, "epoch": 1, "loss": 1.9536429297924043, "loss_raw": 1.2975953817367554, "perplexity": 7.054339291300197, "muon_lr": 0.008309191427701472, "adamw_lr": 0.0002492757428310441, "tokens_per_sec": 149478.08393711582, "vram_gb": 27.299458026885986, "grad_norm": 0.14453125, "tokens_processed": 2899312640, "tokens_processed_billions": 2.89931264}
+{"timestamp": 1787546035.4013522, "elapsed_hours": 5.392086901333597, "session_id": 2, "event": "train_step", "step": 44250, "epoch": 1, "loss": 1.9225523889064788, "loss_raw": 1.3936378955841064, "perplexity": 6.838390445146559, "muon_lr": 0.008308594406854079, "adamw_lr": 0.0002492578322056223, "tokens_per_sec": 149480.77134654063, "vram_gb": 27.710740089416504, "grad_norm": 0.1416015625, "tokens_processed": 2899968000, "tokens_processed_billions": 2.899968}
+{"timestamp": 1787546039.4621737, "elapsed_hours": 5.393214907315042, "session_id": 2, "event": "train_step", "step": 44260, "epoch": 1, "loss": 1.8867585241794587, "loss_raw": 2.566418170928955, "perplexity": 6.597946895373219, "muon_lr": 0.008307997305014091, "adamw_lr": 0.0002492399191504227, "tokens_per_sec": 149483.26281189715, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2900623360, "tokens_processed_billions": 2.90062336}
+{"timestamp": 1787546043.4324195, "elapsed_hours": 5.394317753646109, "session_id": 2, "event": "train_step", "step": 44270, "epoch": 1, "loss": 1.8362403452396392, "loss_raw": 0.6654891967773438, "perplexity": 6.2729098960594545, "muon_lr": 0.00830740012219873, "adamw_lr": 0.0002492220036659618, "tokens_per_sec": 149486.450647749, "vram_gb": 27.710740089416504, "grad_norm": 0.1357421875, "tokens_processed": 2901278720, "tokens_processed_billions": 2.90127872}
+{"timestamp": 1787546047.5631752, "elapsed_hours": 5.395465185642243, "session_id": 2, "event": "train_step", "step": 44280, "epoch": 1, "loss": 1.7846771788597107, "loss_raw": 0.6186764240264893, "perplexity": 5.957656379689158, "muon_lr": 0.00830620551371076, "adamw_lr": 0.00024918616541132275, "tokens_per_sec": 149488.40135587894, "vram_gb": 27.299458026885986, "grad_norm": 0.1376953125, "tokens_processed": 2901934080, "tokens_processed_billions": 2.90193408}
+{"timestamp": 1787546051.53638, "elapsed_hours": 5.396568853523996, "session_id": 2, "event": "train_step", "step": 44290, "epoch": 1, "loss": 1.775440945625305, "loss_raw": 1.0275630950927734, "perplexity": 5.902883413163277, "muon_lr": 0.008305608088072606, "adamw_lr": 0.0002491682426421781, "tokens_per_sec": 149491.56466760515, "vram_gb": 27.710740089416504, "grad_norm": 0.1435546875, "tokens_processed": 2902589440, "tokens_processed_billions": 2.90258944}
+{"timestamp": 1787546055.60565, "elapsed_hours": 5.397699206206534, "session_id": 2, "event": "train_step", "step": 44300, "epoch": 1, "loss": 1.7482738435268401, "loss_raw": 1.535160779953003, "perplexity": 5.744677899668875, "muon_lr": 0.008305010581527973, "adamw_lr": 0.00024915031744583916, "tokens_per_sec": 149493.98660568817, "vram_gb": 27.710740089416504, "grad_norm": 0.134765625, "tokens_processed": 2903244800, "tokens_processed_billions": 2.9032448}
+{"timestamp": 1787546055.6091843, "elapsed_hours": 5.397700188159942, "session_id": 2, "event": "lambda_stats", "step": 44300, "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": 1787546059.598621, "elapsed_hours": 5.398808364934392, "session_id": 2, "event": "train_step", "step": 44310, "epoch": 1, "loss": 1.7383746325969696, "loss_raw": 1.8346866369247437, "perplexity": 5.688090667988844, "muon_lr": 0.008304412994094092, "adamw_lr": 0.00024913238982282273, "tokens_per_sec": 149496.99446912177, "vram_gb": 27.710740089416504, "grad_norm": 0.1669921875, "tokens_processed": 2903900160, "tokens_processed_billions": 2.90390016}
+{"timestamp": 1787546063.769941, "elapsed_hours": 5.39996706492371, "session_id": 2, "event": "train_step", "step": 44320, "epoch": 1, "loss": 1.7554930877685546, "loss_raw": 2.1107594966888428, "perplexity": 5.786300192494849, "muon_lr": 0.008303217576627529, "adamw_lr": 0.0002490965272988258, "tokens_per_sec": 149498.62928397604, "vram_gb": 27.299458026885986, "grad_norm": 0.1396484375, "tokens_processed": 2904555520, "tokens_processed_billions": 2.90455552}
+{"timestamp": 1787546067.7875793, "elapsed_hours": 5.401083075602849, "session_id": 2, "event": "train_step", "step": 44330, "epoch": 1, "loss": 1.8364259254932405, "loss_raw": 1.9264098405838013, "perplexity": 6.27407413229507, "muon_lr": 0.008302619746629315, "adamw_lr": 0.00024907859239887937, "tokens_per_sec": 149501.44558400213, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2905210880, "tokens_processed_billions": 2.90521088}
+{"timestamp": 1787546071.7614858, "elapsed_hours": 5.402186938524246, "session_id": 2, "event": "train_step", "step": 44340, "epoch": 1, "loss": 1.9111655175685882, "loss_raw": 1.222915768623352, "perplexity": 6.760964230753999, "muon_lr": 0.0083020218358108, "adamw_lr": 0.00024906065507432395, "tokens_per_sec": 149504.59741958292, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2905866240, "tokens_processed_billions": 2.90586624}
+{"timestamp": 1787546075.7528944, "elapsed_hours": 5.403295663197835, "session_id": 2, "event": "train_step", "step": 44350, "epoch": 1, "loss": 1.9552981925010682, "loss_raw": 2.06394362449646, "perplexity": 7.066025745470937, "muon_lr": 0.008301423844189228, "adamw_lr": 0.0002490427153256768, "tokens_per_sec": 149507.61280257496, "vram_gb": 27.710740089416504, "grad_norm": 0.1513671875, "tokens_processed": 2906521600, "tokens_processed_billions": 2.9065216}
+{"timestamp": 1787546079.9378378, "elapsed_hours": 5.404458147419824, "session_id": 2, "event": "train_step", "step": 44360, "epoch": 1, "loss": 1.9199057078361512, "loss_raw": 1.9107145071029663, "perplexity": 6.820315336678817, "muon_lr": 0.008300227618605882, "adamw_lr": 0.0002490068285581764, "tokens_per_sec": 149509.1385884009, "vram_gb": 27.299458026885986, "grad_norm": 0.1572265625, "tokens_processed": 2907176960, "tokens_processed_billions": 2.90717696}
+{"timestamp": 1787546083.9039488, "elapsed_hours": 5.405559844772021, "session_id": 2, "event": "train_step", "step": 44370, "epoch": 1, "loss": 1.9828105175495148, "loss_raw": 1.3073817491531372, "perplexity": 7.263127473490424, "muon_lr": 0.008299629384678608, "adamw_lr": 0.0002489888815403582, "tokens_per_sec": 149512.34722016111, "vram_gb": 27.710740089416504, "grad_norm": 0.158203125, "tokens_processed": 2907832320, "tokens_processed_billions": 2.90783232}
+{"timestamp": 1787546087.96753, "elapsed_hours": 5.406688617335425, "session_id": 2, "event": "train_step", "step": 44380, "epoch": 1, "loss": 1.9543579959869384, "loss_raw": 0.8149181008338928, "perplexity": 7.059385414793326, "muon_lr": 0.008299031070017268, "adamw_lr": 0.00024897093210051797, "tokens_per_sec": 149514.80492189978, "vram_gb": 27.710740089416504, "grad_norm": 0.166015625, "tokens_processed": 2908487680, "tokens_processed_billions": 2.90848768}
+{"timestamp": 1787546092.018145, "elapsed_hours": 5.407813788255056, "session_id": 2, "event": "train_step", "step": 44390, "epoch": 1, "loss": 1.9393276512622832, "loss_raw": 2.559892177581787, "perplexity": 6.954073835715152, "muon_lr": 0.008298432674639113, "adamw_lr": 0.00024895298023917334, "tokens_per_sec": 149517.36072484238, "vram_gb": 27.710740089416504, "grad_norm": 0.1318359375, "tokens_processed": 2909143040, "tokens_processed_billions": 2.90914304}
+{"timestamp": 1787546096.2039769, "elapsed_hours": 5.408976519505183, "session_id": 2, "event": "train_step", "step": 44400, "epoch": 1, "loss": 1.9447069382667541, "loss_raw": 3.198793649673462, "perplexity": 6.991582589455177, "muon_lr": 0.00829723564180139, "adamw_lr": 0.00024891706925404166, "tokens_per_sec": 149518.8765503466, "vram_gb": 27.299458026885986, "grad_norm": 0.1474609375, "tokens_processed": 2909798400, "tokens_processed_billions": 2.9097984}
+{"timestamp": 1787546096.2073703, "elapsed_hours": 5.408977462251981, "session_id": 2, "event": "lambda_stats", "step": 44400, "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": 1787546100.179549, "elapsed_hours": 5.410080844958623, "session_id": 2, "event": "train_step", "step": 44410, "epoch": 1, "loss": 2.0266055607795717, "loss_raw": 3.0170247554779053, "perplexity": 7.5882846214050845, "muon_lr": 0.008296637004376344, "adamw_lr": 0.0002488991101312903, "tokens_per_sec": 149522.0072851967, "vram_gb": 27.710740089416504, "grad_norm": 0.15625, "tokens_processed": 2910453760, "tokens_processed_billions": 2.91045376}
+{"timestamp": 1787546104.1598547, "elapsed_hours": 5.411186485356755, "session_id": 2, "event": "train_step", "step": 44420, "epoch": 1, "loss": 2.01678827047348, "loss_raw": 0.8848177790641785, "perplexity": 7.514152710991068, "muon_lr": 0.008296038286303521, "adamw_lr": 0.0002488811485891056, "tokens_per_sec": 149525.10019288652, "vram_gb": 27.710740089416504, "grad_norm": 0.1591796875, "tokens_processed": 2911109120, "tokens_processed_billions": 2.91110912}
+{"timestamp": 1787546108.1293225, "elapsed_hours": 5.41228911532296, "session_id": 2, "event": "train_step", "step": 44430, "epoch": 1, "loss": 2.0931343495845796, "loss_raw": 1.758756399154663, "perplexity": 8.110295870015214, "muon_lr": 0.008295439487600191, "adamw_lr": 0.00024886318462800564, "tokens_per_sec": 149528.27503485334, "vram_gb": 27.710740089416504, "grad_norm": 0.1796875, "tokens_processed": 2911764480, "tokens_processed_billions": 2.91176448}
+{"timestamp": 1787546112.2455187, "elapsed_hours": 5.413432503210173, "session_id": 2, "event": "train_step", "step": 44440, "epoch": 1, "loss": 2.1593047893047332, "loss_raw": 2.5209646224975586, "perplexity": 8.66511148579572, "muon_lr": 0.008294241648371075, "adamw_lr": 0.0002488272494511322, "tokens_per_sec": 149530.32222464634, "vram_gb": 27.299458026885986, "grad_norm": 0.1533203125, "tokens_processed": 2912419840, "tokens_processed_billions": 2.91241984}
+{"timestamp": 1787546116.2088442, "elapsed_hours": 5.414533427158991, "session_id": 2, "event": "train_step", "step": 44450, "epoch": 1, "loss": 2.2493788063526154, "loss_raw": 2.2740752696990967, "perplexity": 9.481843945321023, "muon_lr": 0.008293642607879832, "adamw_lr": 0.0002488092782363949, "tokens_per_sec": 149533.54207609768, "vram_gb": 27.710740089416504, "grad_norm": 0.17578125, "tokens_processed": 2913075200, "tokens_processed_billions": 2.9130752}
+{"timestamp": 1787546120.1857927, "elapsed_hours": 5.415638134876887, "session_id": 2, "event": "train_step", "step": 44460, "epoch": 1, "loss": 2.1366991090774534, "loss_raw": 1.4801658391952515, "perplexity": 8.47142816589356, "muon_lr": 0.008293043486827165, "adamw_lr": 0.00024879130460481495, "tokens_per_sec": 149536.6547990879, "vram_gb": 27.710740089416504, "grad_norm": 0.1416015625, "tokens_processed": 2913730560, "tokens_processed_billions": 2.91373056}
+{"timestamp": 1787546124.228449, "elapsed_hours": 5.4167610948615605, "session_id": 2, "event": "train_step", "step": 44470, "epoch": 1, "loss": 2.1112526845932007, "loss_raw": 1.078071117401123, "perplexity": 8.258580203755264, "muon_lr": 0.008292444285230353, "adamw_lr": 0.00024877332855691056, "tokens_per_sec": 149539.26383867883, "vram_gb": 27.710740089416504, "grad_norm": 0.1376953125, "tokens_processed": 2914385920, "tokens_processed_billions": 2.91438592}
+{"timestamp": 1787546128.3676834, "elapsed_hours": 5.417910882234573, "session_id": 2, "event": "train_step", "step": 44480, "epoch": 1, "loss": 2.119839198589325, "loss_raw": 1.7638403177261353, "perplexity": 8.329797936730872, "muon_lr": 0.00829124564047341, "adamw_lr": 0.0002487373692142022, "tokens_per_sec": 149541.13052472135, "vram_gb": 27.299458026885986, "grad_norm": 0.1552734375, "tokens_processed": 2915041280, "tokens_processed_billions": 2.91504128}
+{"timestamp": 1787546132.3503957, "elapsed_hours": 5.419017191198137, "session_id": 2, "event": "train_step", "step": 44490, "epoch": 1, "loss": 2.041029460430145, "loss_raw": 1.5592721700668335, "perplexity": 7.698530453342396, "muon_lr": 0.008290646197347842, "adamw_lr": 0.0002487193859204352, "tokens_per_sec": 149544.19696394482, "vram_gb": 27.710740089416504, "grad_norm": 0.1474609375, "tokens_processed": 2915696640, "tokens_processed_billions": 2.91569664}
+{"timestamp": 1787546136.3839285, "elapsed_hours": 5.4201376171906785, "session_id": 2, "event": "train_step", "step": 44500, "epoch": 1, "loss": 1.952941210269928, "loss_raw": 2.321183681488037, "perplexity": 7.049390860109761, "muon_lr": 0.008290046673747258, "adamw_lr": 0.0002487014002124177, "tokens_per_sec": 149546.87208620502, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2916352000, "tokens_processed_billions": 2.916352}
+{"timestamp": 1787546136.387262, "elapsed_hours": 5.420138543181949, "session_id": 2, "event": "lambda_stats", "step": 44500, "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": 1787546149.915091, "elapsed_hours": 5.423896273440785, "session_id": 2, "event": "validation", "step": 44500, "epoch": 1, "val_loss": 1.9543376916646957, "val_perplexity": 7.05924208021219, "is_best": true, "best_loss": 1.9543376916646957, "best_perplexity": 7.05924208021219, "train_loss": 2.321183681488037, "train_loss_avg": 1.952941210269928}