Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -6896,3 +6896,54 @@
 {"timestamp": 1787553171.5439105, "elapsed_hours": 7.374348723159896, "session_id": 2, "event": "train_step", "step": 61030, "epoch": 1, "loss": 2.079086397886276, "loss_raw": 2.335613250732422, "perplexity": 7.997159354100259, "muon_lr": 0.006906646407866319, "adamw_lr": 0.00020719939223598952, "tokens_per_sec": 150723.5156188479, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 3999662080, "tokens_processed_billions": 3.99966208}
 {"timestamp": 1787553175.679274, "elapsed_hours": 7.37549743519889, "session_id": 2, "event": "train_step", "step": 61040, "epoch": 1, "loss": 2.055920784473419, "loss_raw": 1.264959692955017, "perplexity": 7.814029598242038, "muon_lr": 0.006905194571502029, "adamw_lr": 0.0002071558371450608, "tokens_per_sec": 150724.72325447793, "vram_gb": 27.299458026885986, "grad_norm": 0.140625, "tokens_processed": 4000317440, "tokens_processed_billions": 4.00031744}
 {"timestamp": 1787553179.7537098, "elapsed_hours": 7.376629222896364, "session_id": 2, "event": "train_step", "step": 61050, "epoch": 1, "loss": 1.9957792377471923, "loss_raw": 2.1078035831451416, "perplexity": 7.357934374766052, "muon_lr": 0.006904468592526154, "adamw_lr": 0.00020713405777578458, "tokens_per_sec": 150726.27720331462, "vram_gb": 27.710740089416504, "grad_norm": 0.1396484375, "tokens_processed": 4000972800, "tokens_processed_billions": 4.0009728}
+{"timestamp": 1787553184.905321, "elapsed_hours": 7.378060226109293, "session_id": 2, "event": "train_step", "step": 61060, "epoch": 1, "loss": 1.924622937440872, "loss_raw": 2.1530823707580566, "perplexity": 6.8525643332581705, "muon_lr": 0.006903742573049039, "adamw_lr": 0.0002071122771914711, "tokens_per_sec": 150721.71621373543, "vram_gb": 27.710740089416504, "grad_norm": 0.1357421875, "tokens_processed": 4001628160, "tokens_processed_billions": 4.00162816}
+{"timestamp": 1787553188.9194221, "elapsed_hours": 7.379175254305204, "session_id": 2, "event": "train_step", "step": 61070, "epoch": 1, "loss": 1.9142351400852204, "loss_raw": 1.8797603845596313, "perplexity": 6.781749724279257, "muon_lr": 0.006903016513091622, "adamw_lr": 0.00020709049539274858, "tokens_per_sec": 150723.61159184377, "vram_gb": 27.710740089416504, "grad_norm": 0.1298828125, "tokens_processed": 4002283520, "tokens_processed_billions": 4.00228352}
+{"timestamp": 1787553193.1546373, "elapsed_hours": 7.380351702769597, "session_id": 2, "event": "train_step", "step": 61080, "epoch": 1, "loss": 1.8298239982128144, "loss_raw": 2.991150140762329, "perplexity": 6.232789579878437, "muon_lr": 0.006901564271819626, "adamw_lr": 0.00020704692815458873, "tokens_per_sec": 150724.25144894063, "vram_gb": 27.299458026885986, "grad_norm": 0.14453125, "tokens_processed": 4002938880, "tokens_processed_billions": 4.00293888}
+{"timestamp": 1787553197.2076304, "elapsed_hours": 7.381477534174919, "session_id": 2, "event": "train_step", "step": 61090, "epoch": 1, "loss": 1.8092123985290527, "loss_raw": 2.8671717643737793, "perplexity": 6.1056367295517715, "muon_lr": 0.006900838090546929, "adamw_lr": 0.0002070251427164078, "tokens_per_sec": 150725.92654023238, "vram_gb": 27.710740089416504, "grad_norm": 0.1279296875, "tokens_processed": 4003594240, "tokens_processed_billions": 4.00359424}
+{"timestamp": 1787553201.190835, "elapsed_hours": 7.382583979898029, "session_id": 2, "event": "train_step", "step": 61100, "epoch": 1, "loss": 1.8127287197113038, "loss_raw": 1.591168999671936, "perplexity": 6.1271439002144374, "muon_lr": 0.006900111868877685, "adamw_lr": 0.00020700335606633048, "tokens_per_sec": 150727.99613291732, "vram_gb": 27.710740089416504, "grad_norm": 0.1494140625, "tokens_processed": 4004249600, "tokens_processed_billions": 4.0042496}
+{"timestamp": 1787553201.1963303, "elapsed_hours": 7.3825855065716635, "session_id": 2, "event": "lambda_stats", "step": 61100, "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": 1787553205.2898731, "elapsed_hours": 7.383722601599163, "session_id": 2, "event": "train_step", "step": 61110, "epoch": 1, "loss": 1.9617812931537628, "loss_raw": 3.0617194175720215, "perplexity": 7.111984317645928, "muon_lr": 0.0068993856068328375, "adamw_lr": 0.0002069815682049851, "tokens_per_sec": 150729.40841832574, "vram_gb": 27.710740089416504, "grad_norm": 0.1474609375, "tokens_processed": 4004904960, "tokens_processed_billions": 4.00490496}
+{"timestamp": 1787553209.4395058, "elapsed_hours": 7.384875277413262, "session_id": 2, "event": "train_step", "step": 61120, "epoch": 1, "loss": 1.9513473451137542, "loss_raw": 2.731715440750122, "perplexity": 7.038164031048135, "muon_lr": 0.006897932961700109, "adamw_lr": 0.00020693798885100322, "tokens_per_sec": 150730.53290368165, "vram_gb": 27.299458026885986, "grad_norm": 0.14453125, "tokens_processed": 4005560320, "tokens_processed_billions": 4.00556032}
+{"timestamp": 1787553213.4907756, "elapsed_hours": 7.386000630060832, "session_id": 2, "event": "train_step", "step": 61130, "epoch": 1, "loss": 1.967511214017868, "loss_raw": 2.635937452316284, "perplexity": 7.152852398592658, "muon_lr": 0.006897206578654119, "adamw_lr": 0.0002069161973596235, "tokens_per_sec": 150732.21557168514, "vram_gb": 27.710740089416504, "grad_norm": 0.140625, "tokens_processed": 4006215680, "tokens_processed_billions": 4.00621568}
+{"timestamp": 1787553217.458642, "elapsed_hours": 7.387102815243933, "session_id": 2, "event": "train_step", "step": 61140, "epoch": 1, "loss": 1.9786786925792694, "loss_raw": 1.9992033243179321, "perplexity": 7.233179414707961, "muon_lr": 0.006896480155316307, "adamw_lr": 0.00020689440465948913, "tokens_per_sec": 150734.37011415052, "vram_gb": 27.710740089416504, "grad_norm": 0.1396484375, "tokens_processed": 4006871040, "tokens_processed_billions": 4.00687104}
+{"timestamp": 1787553221.5445423, "elapsed_hours": 7.3882377876175775, "session_id": 2, "event": "train_step", "step": 61150, "epoch": 1, "loss": 2.018314949274063, "loss_raw": 1.5872551202774048, "perplexity": 7.525633169896059, "muon_lr": 0.006895753691707618, "adamw_lr": 0.0002068726107512285, "tokens_per_sec": 150735.85492736727, "vram_gb": 27.710740089416504, "grad_norm": 0.1494140625, "tokens_processed": 4007526400, "tokens_processed_billions": 4.0075264}
+{"timestamp": 1787553225.7359836, "elapsed_hours": 7.389402076734437, "session_id": 2, "event": "train_step", "step": 61160, "epoch": 1, "loss": 1.9670278072357177, "loss_raw": 2.3311383724212646, "perplexity": 7.149395496843651, "muon_lr": 0.00689430064376142, "adamw_lr": 0.00020682901931284254, "tokens_per_sec": 150736.7403252289, "vram_gb": 27.299458026885986, "grad_norm": 0.1357421875, "tokens_processed": 4008181760, "tokens_processed_billions": 4.00818176}
+{"timestamp": 1787553229.7182438, "elapsed_hours": 7.390508260130883, "session_id": 2, "event": "train_step", "step": 61170, "epoch": 1, "loss": 2.023648660182953, "loss_raw": 2.4440999031066895, "perplexity": 7.565879958585318, "muon_lr": 0.006893574059465811, "adamw_lr": 0.00020680722178397428, "tokens_per_sec": 150738.81209309067, "vram_gb": 27.710740089416504, "grad_norm": 0.1455078125, "tokens_processed": 4008837120, "tokens_processed_billions": 4.00883712}
+{"timestamp": 1787553233.8139708, "elapsed_hours": 7.391645962132348, "session_id": 2, "event": "train_step", "step": 61180, "epoch": 1, "loss": 1.9547193074226379, "loss_raw": 3.008833885192871, "perplexity": 7.061936512315302, "muon_lr": 0.0068928474349831325, "adamw_lr": 0.00020678542304949392, "tokens_per_sec": 150740.2398795942, "vram_gb": 27.710740089416504, "grad_norm": 0.13671875, "tokens_processed": 4009492480, "tokens_processed_billions": 4.00949248}
+{"timestamp": 1787553237.791407, "elapsed_hours": 7.3927508054176965, "session_id": 2, "event": "train_step", "step": 61190, "epoch": 1, "loss": 1.9846776854991912, "loss_raw": 1.8105149269104004, "perplexity": 7.276701620986266, "muon_lr": 0.006892120770334339, "adamw_lr": 0.0002067636231100301, "tokens_per_sec": 150742.33716968473, "vram_gb": 27.710740089416504, "grad_norm": 0.140625, "tokens_processed": 4010147840, "tokens_processed_billions": 4.01014784}
+{"timestamp": 1787553241.9000766, "elapsed_hours": 7.393892102705108, "session_id": 2, "event": "train_step", "step": 61200, "epoch": 1, "loss": 2.0042527151107787, "loss_raw": 2.0728700160980225, "perplexity": 7.420546561980719, "muon_lr": 0.006890667320622224, "adamw_lr": 0.00020672001961866666, "tokens_per_sec": 150743.69064063573, "vram_gb": 27.299458026885986, "grad_norm": 0.166015625, "tokens_processed": 4010803200, "tokens_processed_billions": 4.0108032}
+{"timestamp": 1787553241.9033382, "elapsed_hours": 7.39389300889439, "session_id": 2, "event": "lambda_stats", "step": 61200, "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": 1787553245.8737798, "elapsed_hours": 7.394995909002092, "session_id": 2, "event": "train_step", "step": 61210, "epoch": 1, "loss": 2.014929710626602, "loss_raw": 1.8629072904586792, "perplexity": 7.500200178303905, "muon_lr": 0.006889940535600818, "adamw_lr": 0.00020669821606802448, "tokens_per_sec": 150745.80857775494, "vram_gb": 27.710740089416504, "grad_norm": 0.1513671875, "tokens_processed": 4011458560, "tokens_processed_billions": 4.01145856}
+{"timestamp": 1787553249.8496802, "elapsed_hours": 7.3961003257830935, "session_id": 2, "event": "train_step", "step": 61220, "epoch": 1, "loss": 1.983899737596512, "loss_raw": 1.8967273235321045, "perplexity": 7.271042927591514, "muon_lr": 0.006889213710497123, "adamw_lr": 0.00020667641131491362, "tokens_per_sec": 150747.91262265606, "vram_gb": 27.710740089416504, "grad_norm": 0.1474609375, "tokens_processed": 4012113920, "tokens_processed_billions": 4.01211392}
+{"timestamp": 1787553253.8850205, "elapsed_hours": 7.3972212537791995, "session_id": 2, "event": "train_step", "step": 61230, "epoch": 1, "loss": 1.975616867542267, "loss_raw": 1.4135185480117798, "perplexity": 7.2110665550078625, "muon_lr": 0.0068884868453320995, "adamw_lr": 0.0002066546053599629, "tokens_per_sec": 150749.68000848065, "vram_gb": 27.710740089416504, "grad_norm": 0.1494140625, "tokens_processed": 4012769280, "tokens_processed_billions": 4.01276928}
+{"timestamp": 1787553258.0283651, "elapsed_hours": 7.398372182846069, "session_id": 2, "event": "train_step", "step": 61240, "epoch": 1, "loss": 1.8965709495544434, "loss_raw": 0.7707993984222412, "perplexity": 6.6630074357392, "muon_lr": 0.00688703299490191, "adamw_lr": 0.00020661098984705724, "tokens_per_sec": 150750.83473615986, "vram_gb": 27.299458026885986, "grad_norm": 0.1376953125, "tokens_processed": 4013424640, "tokens_processed_billions": 4.01342464}
+{"timestamp": 1787553262.0230815, "elapsed_hours": 7.399481826225917, "session_id": 2, "event": "train_step", "step": 61250, "epoch": 1, "loss": 1.8419535994529723, "loss_raw": 2.5572779178619385, "perplexity": 6.308851198086491, "muon_lr": 0.00688630600967867, "adamw_lr": 0.00020658918029036006, "tokens_per_sec": 150752.8316373964, "vram_gb": 27.710740089416504, "grad_norm": 0.1435546875, "tokens_processed": 4014080000, "tokens_processed_billions": 4.01408}
+{"timestamp": 1787553266.0301108, "elapsed_hours": 7.400594889852735, "session_id": 2, "event": "train_step", "step": 61260, "epoch": 1, "loss": 1.855030734539032, "loss_raw": 1.8823760747909546, "perplexity": 6.391894699539132, "muon_lr": 0.006885578984477955, "adamw_lr": 0.0002065673695343386, "tokens_per_sec": 150754.7571942688, "vram_gb": 27.710740089416504, "grad_norm": 0.1455078125, "tokens_processed": 4014735360, "tokens_processed_billions": 4.01473536}
+{"timestamp": 1787553270.0934227, "elapsed_hours": 7.401723587645424, "session_id": 2, "event": "train_step", "step": 61270, "epoch": 1, "loss": 1.792686243057251, "loss_raw": 1.962092399597168, "perplexity": 6.005563220500567, "muon_lr": 0.006884851919320725, "adamw_lr": 0.0002065455575796217, "tokens_per_sec": 150756.3659265808, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 4015390720, "tokens_processed_billions": 4.01539072}
+{"timestamp": 1787553274.309763, "elapsed_hours": 7.402894793417719, "session_id": 2, "event": "train_step", "step": 61280, "epoch": 1, "loss": 1.8203674685955047, "loss_raw": 1.1185202598571777, "perplexity": 6.174126830793633, "muon_lr": 0.006883397669220594, "adamw_lr": 0.00020650193007661777, "tokens_per_sec": 150757.10453541632, "vram_gb": 27.299458026885986, "grad_norm": 0.1533203125, "tokens_processed": 4016046080, "tokens_processed_billions": 4.01604608}
+{"timestamp": 1787553278.2884667, "elapsed_hours": 7.403999988701608, "session_id": 2, "event": "train_step", "step": 61290, "epoch": 1, "loss": 1.870261024236679, "loss_raw": 1.1566331386566162, "perplexity": 6.489990222956581, "muon_lr": 0.006882670484319632, "adamw_lr": 0.0002064801145295889, "tokens_per_sec": 150759.18892409527, "vram_gb": 27.710740089416504, "grad_norm": 0.1435546875, "tokens_processed": 4016701440, "tokens_processed_billions": 4.01670144}
+{"timestamp": 1787553282.3495378, "elapsed_hours": 7.405128064023124, "session_id": 2, "event": "train_step", "step": 61300, "epoch": 1, "loss": 1.8228340351581573, "loss_raw": 2.199108123779297, "perplexity": 6.18937452258089, "muon_lr": 0.00688194325954603, "adamw_lr": 0.00020645829778638086, "tokens_per_sec": 150760.80689012335, "vram_gb": 27.710740089416504, "grad_norm": 0.140625, "tokens_processed": 4017356800, "tokens_processed_billions": 4.0173568}
+{"timestamp": 1787553282.3529253, "elapsed_hours": 7.405129005379147, "session_id": 2, "event": "lambda_stats", "step": 61300, "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": 1787553286.3369384, "elapsed_hours": 7.406235675348176, "session_id": 2, "event": "train_step", "step": 61310, "epoch": 1, "loss": 1.8237952268123627, "loss_raw": 2.4412689208984375, "perplexity": 6.195326557781774, "muon_lr": 0.00688121599492076, "adamw_lr": 0.00020643647984762276, "tokens_per_sec": 150762.8412349769, "vram_gb": 27.710740089416504, "grad_norm": 0.140625, "tokens_processed": 4018012160, "tokens_processed_billions": 4.01801216}
+{"timestamp": 1787553290.4717784, "elapsed_hours": 7.407384242018064, "session_id": 2, "event": "train_step", "step": 61320, "epoch": 1, "loss": 1.7840387070178985, "loss_raw": 2.3361294269561768, "perplexity": 5.95385379789741, "muon_lr": 0.006879761346199111, "adamw_lr": 0.00020639284038597328, "tokens_per_sec": 150764.04058258646, "vram_gb": 27.299458026885986, "grad_norm": 0.1337890625, "tokens_processed": 4018667520, "tokens_processed_billions": 4.01866752}
+{"timestamp": 1787553294.4657643, "elapsed_hours": 7.408493682675892, "session_id": 2, "event": "train_step", "step": 61330, "epoch": 1, "loss": 1.8849386513233184, "loss_raw": 1.984636664390564, "perplexity": 6.585950390280041, "muon_lr": 0.006879033962144678, "adamw_lr": 0.00020637101886434027, "tokens_per_sec": 150766.037030906, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 4019322880, "tokens_processed_billions": 4.01932288}
+{"timestamp": 1787553298.4906847, "elapsed_hours": 7.40961171593931, "session_id": 2, "event": "train_step", "step": 61340, "epoch": 1, "loss": 1.8702511620521545, "loss_raw": 1.0586152076721191, "perplexity": 6.489926217791055, "muon_lr": 0.006878306538322473, "adamw_lr": 0.00020634919614967413, "tokens_per_sec": 150767.85712974187, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 4019978240, "tokens_processed_billions": 4.01997824}
+{"timestamp": 1787553302.4847424, "elapsed_hours": 7.410721176465352, "session_id": 2, "event": "train_step", "step": 61350, "epoch": 1, "loss": 1.9226373851299285, "loss_raw": 1.2408798933029175, "perplexity": 6.838971707211061, "muon_lr": 0.006877579074753476, "adamw_lr": 0.00020632737224260425, "tokens_per_sec": 150769.85202062115, "vram_gb": 27.710740089416504, "grad_norm": 0.1376953125, "tokens_processed": 4020633600, "tokens_processed_billions": 4.0206336}
+{"timestamp": 1787553306.6308734, "elapsed_hours": 7.411872879465421, "session_id": 2, "event": "train_step", "step": 61360, "epoch": 1, "loss": 1.9205713450908661, "loss_raw": 1.8556159734725952, "perplexity": 6.824856703939752, "muon_lr": 0.006876124028459011, "adamw_lr": 0.0002062837208537703, "tokens_per_sec": 150770.9856703434, "vram_gb": 27.299458026885986, "grad_norm": 0.1494140625, "tokens_processed": 4021288960, "tokens_processed_billions": 4.02128896}
+{"timestamp": 1787553310.627377, "elapsed_hours": 7.412983019418187, "session_id": 2, "event": "train_step", "step": 61370, "epoch": 1, "loss": 1.9491771805286406, "loss_raw": 2.4124410152435303, "perplexity": 7.022906618260224, "muon_lr": 0.006875396445775505, "adamw_lr": 0.00020626189337326508, "tokens_per_sec": 150772.96549224295, "vram_gb": 27.710740089416504, "grad_norm": 0.1337890625, "tokens_processed": 4021944320, "tokens_processed_billions": 4.02194432}
+{"timestamp": 1787553314.6126592, "elapsed_hours": 7.414090042180486, "session_id": 2, "event": "train_step", "step": 61380, "epoch": 1, "loss": 1.8424588072299957, "loss_raw": 2.0690643787384033, "perplexity": 6.312039284030945, "muon_lr": 0.006874668823429124, "adamw_lr": 0.00020624006470287365, "tokens_per_sec": 150775.0088525104, "vram_gb": 27.710740089416504, "grad_norm": 0.146484375, "tokens_processed": 4022599680, "tokens_processed_billions": 4.02259968}
+{"timestamp": 1787553318.6695552, "elapsed_hours": 7.415216957728068, "session_id": 2, "event": "train_step", "step": 61390, "epoch": 1, "loss": 1.8846793031692506, "loss_raw": 1.729149580001831, "perplexity": 6.584242557674729, "muon_lr": 0.006873941161440853, "adamw_lr": 0.00020621823484322552, "tokens_per_sec": 150776.64500960248, "vram_gb": 27.710740089416504, "grad_norm": 0.142578125, "tokens_processed": 4023255040, "tokens_processed_billions": 4.02325504}
+{"timestamp": 1787553322.952777, "elapsed_hours": 7.416406741539637, "session_id": 2, "event": "train_step", "step": 61400, "epoch": 1, "loss": 1.8225224995613098, "loss_raw": 1.259508490562439, "perplexity": 6.1874466124167995, "muon_lr": 0.006872485718622568, "adamw_lr": 0.00020617457155867698, "tokens_per_sec": 150777.0018465378, "vram_gb": 27.299458026885986, "grad_norm": 0.1318359375, "tokens_processed": 4023910400, "tokens_processed_billions": 4.0239104}
+{"timestamp": 1787553322.9608226, "elapsed_hours": 7.416408976978726, "session_id": 2, "event": "lambda_stats", "step": 61400, "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": 1787553327.0021954, "elapsed_hours": 7.417531580064032, "session_id": 2, "event": "train_step", "step": 61410, "epoch": 1, "loss": 1.7712776625156403, "loss_raw": 1.8645844459533691, "perplexity": 5.87835912455234, "muon_lr": 0.006871757937834528, "adamw_lr": 0.0002061527381350358, "tokens_per_sec": 150778.68089422796, "vram_gb": 27.710740089416504, "grad_norm": 0.1328125, "tokens_processed": 4024565760, "tokens_processed_billions": 4.02456576}
+{"timestamp": 1787553331.0618331, "elapsed_hours": 7.41865925722652, "session_id": 2, "event": "train_step", "step": 61420, "epoch": 1, "loss": 1.7970859253406524, "loss_raw": 2.0671331882476807, "perplexity": 6.032044001398791, "muon_lr": 0.006871030117488539, "adamw_lr": 0.00020613090352465611, "tokens_per_sec": 150780.3011367416, "vram_gb": 27.710740089416504, "grad_norm": 0.1357421875, "tokens_processed": 4025221120, "tokens_processed_billions": 4.02522112}
+{"timestamp": 1787553335.1019235, "elapsed_hours": 7.419781504538324, "session_id": 2, "event": "train_step", "step": 61430, "epoch": 1, "loss": 1.878912672996521, "loss_raw": 1.4006564617156982, "perplexity": 6.546382932022229, "muon_lr": 0.00687030225760559, "adamw_lr": 0.00020610906772816764, "tokens_per_sec": 150782.03114975232, "vram_gb": 27.710740089416504, "grad_norm": 0.142578125, "tokens_processed": 4025876480, "tokens_processed_billions": 4.02587648}
+{"timestamp": 1787553339.2355905, "elapsed_hours": 7.420929745435715, "session_id": 2, "event": "train_step", "step": 61440, "epoch": 1, "loss": 1.753529316186905, "loss_raw": 1.4776686429977417, "perplexity": 5.774948370454791, "muon_lr": 0.006868846419312766, "adamw_lr": 0.00020606539257938294, "tokens_per_sec": 150783.23192015194, "vram_gb": 27.299458026885986, "grad_norm": 0.1416015625, "tokens_processed": 4026531840, "tokens_processed_billions": 4.02653184}
+{"timestamp": 1787553343.2690883, "elapsed_hours": 7.422050161427921, "session_id": 2, "event": "train_step", "step": 61450, "epoch": 1, "loss": 1.7759775519371033, "loss_raw": 1.2793258428573608, "perplexity": 5.906051787669441, "muon_lr": 0.0068681184409448755, "adamw_lr": 0.0002060435532283462, "tokens_per_sec": 150784.99931008232, "vram_gb": 27.710740089416504, "grad_norm": 0.1357421875, "tokens_processed": 4027187200, "tokens_processed_billions": 4.0271872}
+{"timestamp": 1787553347.3429217, "elapsed_hours": 7.423181781768799, "session_id": 2, "event": "train_step", "step": 61460, "epoch": 1, "loss": 1.7762096774578096, "loss_raw": 2.0616297721862793, "perplexity": 5.907422892143987, "muon_lr": 0.006867390423123991, "adamw_lr": 0.00020602171269371966, "tokens_per_sec": 150786.53683786152, "vram_gb": 27.710740089416504, "grad_norm": 0.1337890625, "tokens_processed": 4027842560, "tokens_processed_billions": 4.02784256}
+{"timestamp": 1787553351.3306928, "elapsed_hours": 7.4242894960774315, "session_id": 2, "event": "train_step", "step": 61470, "epoch": 1, "loss": 1.8294231462478638, "loss_raw": 2.006009340286255, "perplexity": 6.230291654610793, "muon_lr": 0.006866662365871103, "adamw_lr": 0.00020599987097613304, "tokens_per_sec": 150788.559990165, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 4028497920, "tokens_processed_billions": 4.02849792}
+{"timestamp": 1787553355.5346842, "elapsed_hours": 7.4254572715361915, "session_id": 2, "event": "train_step", "step": 61480, "epoch": 1, "loss": 1.7473920667171479, "loss_raw": 1.4970406293869019, "perplexity": 5.739614608592193, "muon_lr": 0.0068652061331533065, "adamw_lr": 0.00020595618399459916, "tokens_per_sec": 150789.36295992965, "vram_gb": 27.299458026885986, "grad_norm": 0.1484375, "tokens_processed": 4029153280, "tokens_processed_billions": 4.02915328}
+{"timestamp": 1787553359.518965, "elapsed_hours": 7.426564016011026, "session_id": 2, "event": "train_step", "step": 61490, "epoch": 1, "loss": 1.9135690474510192, "loss_raw": 2.725071668624878, "perplexity": 6.777233954868252, "muon_lr": 0.0068644779577303906, "adamw_lr": 0.00020593433873191164, "tokens_per_sec": 150791.40485118813, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 4029808640, "tokens_processed_billions": 4.02980864}
+{"timestamp": 1787553363.683678, "elapsed_hours": 7.427720880773332, "session_id": 2, "event": "train_step", "step": 61500, "epoch": 1, "loss": 1.8962210810184479, "loss_raw": 2.2880043983459473, "perplexity": 6.660676666837458, "muon_lr": 0.006863749742959461, "adamw_lr": 0.00020591249228878378, "tokens_per_sec": 150792.42870690985, "vram_gb": 27.710740089416504, "grad_norm": 0.1484375, "tokens_processed": 4030464000, "tokens_processed_billions": 4.030464}
+{"timestamp": 1787553363.6871634, "elapsed_hours": 7.42772184934881, "session_id": 2, "event": "lambda_stats", "step": 61500, "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": 1787553377.3721118, "elapsed_hours": 7.431523223784235, "session_id": 2, "event": "validation", "step": 61500, "epoch": 1, "val_loss": 1.8882437086105346, "val_perplexity": 6.607753343768968, "is_best": true, "best_loss": 1.8882437086105346, "best_perplexity": 6.607753343768968, "train_loss": 2.2880043983459473, "train_loss_avg": 1.8962210810184479}