Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -627,3 +627,51 @@
 {"timestamp": 1787529040.500608, "elapsed_hours": 0.6712811391221153, "session_id": 2, "event": "train_step", "step": 5560, "epoch": 1, "loss": 2.5526931262016297, "loss_raw": 2.9633867740631104, "perplexity": 12.841641415654374, "muon_lr": 0.00998716463438461, "adamw_lr": 0.0002996149390315382, "tokens_per_sec": 151485.87175594547, "vram_gb": 27.299458026885986, "grad_norm": 0.232421875, "tokens_processed": 364380160, "tokens_processed_billions": 0.36438016}
 {"timestamp": 1787529044.4326587, "elapsed_hours": 0.6723733752965927, "session_id": 2, "event": "train_step", "step": 5570, "epoch": 1, "loss": 2.57450975894928, "loss_raw": 2.8271102905273438, "perplexity": 13.124881225477077, "muon_lr": 0.009987106893887383, "adamw_lr": 0.00029961320681662145, "tokens_per_sec": 151510.6516098949, "vram_gb": 27.710740089416504, "grad_norm": 0.1953125, "tokens_processed": 365035520, "tokens_processed_billions": 0.36503552}
 {"timestamp": 1787529048.3997927, "elapsed_hours": 0.6734753569629457, "session_id": 2, "event": "train_step", "step": 5580, "epoch": 1, "loss": 2.631426742076874, "loss_raw": 1.450372338294983, "perplexity": 13.893578320806187, "muon_lr": 0.009987049023993604, "adamw_lr": 0.00029961147071980805, "tokens_per_sec": 151533.15642863564, "vram_gb": 27.710740089416504, "grad_norm": 0.2412109375, "tokens_processed": 365690880, "tokens_processed_billions": 0.36569088}
+{"timestamp": 1787529053.0432007, "elapsed_hours": 0.6747651925351885, "session_id": 2, "event": "train_step", "step": 5590, "epoch": 1, "loss": 2.6282849717140198, "loss_raw": 1.6709707975387573, "perplexity": 13.849996386267781, "muon_lr": 0.009986991024704944, "adamw_lr": 0.0002996097307411483, "tokens_per_sec": 151513.19238068853, "vram_gb": 27.710740089416504, "grad_norm": 0.2255859375, "tokens_processed": 366346240, "tokens_processed_billions": 0.36634624}
+{"timestamp": 1787529057.1560814, "elapsed_hours": 0.6759076594644122, "session_id": 2, "event": "train_step", "step": 5600, "epoch": 1, "loss": 2.7610993242263793, "loss_raw": 2.8422958850860596, "perplexity": 15.817221649049609, "muon_lr": 0.009986874637949669, "adamw_lr": 0.00029960623913848994, "tokens_per_sec": 151526.48495033305, "vram_gb": 27.299458026885986, "grad_norm": 0.2275390625, "tokens_processed": 367001600, "tokens_processed_billions": 0.3670016}
+{"timestamp": 1787529057.1597254, "elapsed_hours": 0.6759086717499627, "session_id": 2, "event": "lambda_stats", "step": 5600, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.6953125, "lambda_mean": 1.2265625, "lambda_std": 0.13263366227340462, "lambda_all": [1.6953125, 1.390625, 1.296875, 1.296875, 1.171875, 1.171875, 1.140625, 1.25, 1.1875, 1.140625, 1.1484375, 1.140625, 1.140625, 1.1875, 1.140625, 1.21875, 1.140625, 1.21875], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787529061.1221213, "elapsed_hours": 0.6770093371470769, "session_id": 2, "event": "train_step", "step": 5610, "epoch": 1, "loss": 2.8103457927703857, "loss_raw": 1.9987895488739014, "perplexity": 16.615662801583813, "muon_lr": 0.009986816250486408, "adamw_lr": 0.0002996044875145922, "tokens_per_sec": 151548.91190262555, "vram_gb": 27.710740089416504, "grad_norm": 0.23828125, "tokens_processed": 367656960, "tokens_processed_billions": 0.36765696}
+{"timestamp": 1787529065.0781124, "elapsed_hours": 0.6781082236766816, "session_id": 2, "event": "train_step", "step": 5620, "epoch": 1, "loss": 2.91549809217453, "loss_raw": 4.002696514129639, "perplexity": 18.45800390075069, "muon_lr": 0.009986757733634978, "adamw_lr": 0.0002996027320090493, "tokens_per_sec": 151571.89012375724, "vram_gb": 27.710740089416504, "grad_norm": 0.2392578125, "tokens_processed": 368312320, "tokens_processed_billions": 0.36831232}
+{"timestamp": 1787529069.0271091, "elapsed_hours": 0.6792051671610938, "session_id": 2, "event": "train_step", "step": 5630, "epoch": 1, "loss": 2.9408719420433043, "loss_raw": 2.517866373062134, "perplexity": 18.93234702672729, "muon_lr": 0.009986699087397067, "adamw_lr": 0.00029960097262191196, "tokens_per_sec": 151595.22946201634, "vram_gb": 27.710740089416504, "grad_norm": 0.220703125, "tokens_processed": 368967680, "tokens_processed_billions": 0.36896768}
+{"timestamp": 1787529073.1216958, "elapsed_hours": 0.6803425523307588, "session_id": 2, "event": "train_step", "step": 5640, "epoch": 1, "loss": 3.013135371208191, "loss_raw": 2.600764751434326, "perplexity": 20.351108277331335, "muon_lr": 0.009986581406768561, "adamw_lr": 0.00029959744220305675, "tokens_per_sec": 151609.43285702952, "vram_gb": 27.299458026885986, "grad_norm": 0.2265625, "tokens_processed": 369623040, "tokens_processed_billions": 0.36962304}
+{"timestamp": 1787529077.0975776, "elapsed_hours": 0.6814469639460246, "session_id": 2, "event": "train_step", "step": 5650, "epoch": 1, "loss": 2.9013192796707155, "loss_raw": 1.6844837665557861, "perplexity": 18.1981379729587, "muon_lr": 0.009986522372381363, "adamw_lr": 0.0002995956711714408, "tokens_per_sec": 151630.96697196335, "vram_gb": 27.710740089416504, "grad_norm": 0.2451171875, "tokens_processed": 370278400, "tokens_processed_billions": 0.3702784}
+{"timestamp": 1787529081.0567267, "elapsed_hours": 0.6825467276573182, "session_id": 2, "event": "train_step", "step": 5660, "epoch": 1, "loss": 2.8240348434448244, "loss_raw": 2.3036370277404785, "perplexity": 16.84467938519757, "muon_lr": 0.009986463208614467, "adamw_lr": 0.0002995938962584339, "tokens_per_sec": 151653.46378000668, "vram_gb": 27.710740089416504, "grad_norm": 0.2109375, "tokens_processed": 370933760, "tokens_processed_billions": 0.37093376}
+{"timestamp": 1787529085.0285192, "elapsed_hours": 0.6836500032080545, "session_id": 2, "event": "train_step", "step": 5670, "epoch": 1, "loss": 2.6947672033309935, "loss_raw": 3.435635566711426, "perplexity": 14.8020724789952, "muon_lr": 0.009986403915469583, "adamw_lr": 0.0002995921174640874, "tokens_per_sec": 151675.10878911012, "vram_gb": 27.710740089416504, "grad_norm": 0.2421875, "tokens_processed": 371589120, "tokens_processed_billions": 0.37158912}
+{"timestamp": 1787529089.145238, "elapsed_hours": 0.6847935362656912, "session_id": 2, "event": "train_step", "step": 5680, "epoch": 1, "loss": 2.5937888860702514, "loss_raw": 2.398329973220825, "perplexity": 13.380372388914884, "muon_lr": 0.009986284941052689, "adamw_lr": 0.00029958854823158057, "tokens_per_sec": 151687.72304526647, "vram_gb": 27.299458026885986, "grad_norm": 0.2236328125, "tokens_processed": 372244480, "tokens_processed_billions": 0.37224448}
+{"timestamp": 1787529093.1175652, "elapsed_hours": 0.6858969604969025, "session_id": 2, "event": "train_step", "step": 5690, "epoch": 1, "loss": 2.629180850982666, "loss_raw": 3.981450080871582, "perplexity": 13.862409870562425, "muon_lr": 0.009986225259784109, "adamw_lr": 0.00029958675779352326, "tokens_per_sec": 151709.20699103252, "vram_gb": 27.710740089416504, "grad_norm": 0.228515625, "tokens_processed": 372899840, "tokens_processed_billions": 0.37289984}
+{"timestamp": 1787529097.1237628, "elapsed_hours": 0.6870097932550642, "session_id": 2, "event": "train_step", "step": 5700, "epoch": 1, "loss": 2.67482905626297, "loss_raw": 3.768643379211426, "perplexity": 14.50986925630238, "muon_lr": 0.009986165449144404, "adamw_lr": 0.00029958496347433204, "tokens_per_sec": 151728.534982461, "vram_gb": 27.710740089416504, "grad_norm": 0.224609375, "tokens_processed": 373555200, "tokens_processed_billions": 0.3735552}
+{"timestamp": 1787529097.1277142, "elapsed_hours": 0.6870108909739389, "session_id": 2, "event": "lambda_stats", "step": 5700, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.6953125, "lambda_mean": 1.2248263888888888, "lambda_std": 0.13310719846346855, "lambda_all": [1.6953125, 1.390625, 1.296875, 1.296875, 1.171875, 1.171875, 1.140625, 1.25, 1.1796875, 1.140625, 1.140625, 1.140625, 1.140625, 1.1875, 1.140625, 1.2109375, 1.140625, 1.2109375], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787529101.088792, "elapsed_hours": 0.688111190199852, "session_id": 2, "event": "train_step", "step": 5710, "epoch": 1, "loss": 2.6169377851486204, "loss_raw": 2.515712261199951, "perplexity": 13.69372618367243, "muon_lr": 0.009986105509135295, "adamw_lr": 0.0002995831652740588, "tokens_per_sec": 151750.33217684156, "vram_gb": 27.710740089416504, "grad_norm": 0.2333984375, "tokens_processed": 374210560, "tokens_processed_billions": 0.37421056}
+{"timestamp": 1787529105.199184, "elapsed_hours": 0.6892529657151965, "session_id": 2, "event": "train_step", "step": 5720, "epoch": 1, "loss": 2.774486801624298, "loss_raw": 2.867950201034546, "perplexity": 16.03039811003898, "muon_lr": 0.009985985241015783, "adamw_lr": 0.0002995795572304735, "tokens_per_sec": 151763.12594437102, "vram_gb": 27.299458026885986, "grad_norm": 0.248046875, "tokens_processed": 374865920, "tokens_processed_billions": 0.37486592}
+{"timestamp": 1787529109.154494, "elapsed_hours": 0.6903516629007127, "session_id": 2, "event": "train_step", "step": 5730, "epoch": 1, "loss": 2.7874553847312926, "loss_raw": 2.239328622817993, "perplexity": 16.239643535875842, "muon_lr": 0.00998592491290885, "adamw_lr": 0.00029957774738726545, "tokens_per_sec": 151785.3948534046, "vram_gb": 27.710740089416504, "grad_norm": 0.2255859375, "tokens_processed": 375521280, "tokens_processed_billions": 0.37552128}
+{"timestamp": 1787529113.0954454, "elapsed_hours": 0.6914463716745377, "session_id": 2, "event": "train_step", "step": 5740, "epoch": 1, "loss": 2.7630595421791075, "loss_raw": 2.575833559036255, "perplexity": 15.84825725923074, "muon_lr": 0.00998586445543945, "adamw_lr": 0.0002995759336631835, "tokens_per_sec": 151808.46424007852, "vram_gb": 27.710740089416504, "grad_norm": 0.2099609375, "tokens_processed": 376176640, "tokens_processed_billions": 0.37617664}
+{"timestamp": 1787529117.088472, "elapsed_hours": 0.6925555456346936, "session_id": 2, "event": "train_step", "step": 5750, "epoch": 1, "loss": 2.8010669708251954, "loss_raw": 3.2340190410614014, "perplexity": 16.462202093275295, "muon_lr": 0.009985803868609332, "adamw_lr": 0.0002995741160582799, "tokens_per_sec": 151828.2822447054, "vram_gb": 27.710740089416504, "grad_norm": 0.220703125, "tokens_processed": 376832000, "tokens_processed_billions": 0.376832}
+{"timestamp": 1787529121.1786022, "elapsed_hours": 0.6936916930145688, "session_id": 2, "event": "train_step", "step": 5760, "epoch": 1, "loss": 2.783833673000336, "loss_raw": 1.812042236328125, "perplexity": 16.180934605958242, "muon_lr": 0.009985682306873911, "adamw_lr": 0.00029957046920621726, "tokens_per_sec": 151842.10575999014, "vram_gb": 27.299458026885986, "grad_norm": 0.24609375, "tokens_processed": 377487360, "tokens_processed_billions": 0.37748736}
+{"timestamp": 1787529125.1633203, "elapsed_hours": 0.6947985592153337, "session_id": 2, "event": "train_step", "step": 5770, "epoch": 1, "loss": 2.675296809673309, "loss_raw": 1.1656614542007446, "perplexity": 14.516657884708858, "muon_lr": 0.009985621331972117, "adamw_lr": 0.00029956863995916346, "tokens_per_sec": 151862.31911765967, "vram_gb": 27.710740089416504, "grad_norm": 0.1884765625, "tokens_processed": 378142720, "tokens_processed_billions": 0.37814272}
+{"timestamp": 1787529129.153441, "elapsed_hours": 0.6959069261948267, "session_id": 2, "event": "train_step", "step": 5780, "epoch": 1, "loss": 2.696896109580994, "loss_raw": 2.694317579269409, "perplexity": 14.833618270710833, "muon_lr": 0.00998556022771661, "adamw_lr": 0.0002995668068314983, "tokens_per_sec": 151882.12433093338, "vram_gb": 27.710740089416504, "grad_norm": 0.2275390625, "tokens_processed": 378798080, "tokens_processed_billions": 0.37879808}
+{"timestamp": 1787529133.115336, "elapsed_hours": 0.6970074523157543, "session_id": 2, "event": "train_step", "step": 5790, "epoch": 1, "loss": 2.6420929503440855, "loss_raw": 3.5746042728424072, "perplexity": 14.042563260476463, "muon_lr": 0.009985498994109158, "adamw_lr": 0.0002995649698232747, "tokens_per_sec": 151903.59526423935, "vram_gb": 27.710740089416504, "grad_norm": 0.24609375, "tokens_processed": 379453440, "tokens_processed_billions": 0.37945344}
+{"timestamp": 1787529137.2107835, "elapsed_hours": 0.6981450766324997, "session_id": 2, "event": "train_step", "step": 5800, "epoch": 1, "loss": 2.5951610231399536, "loss_raw": 2.8673341274261475, "perplexity": 13.398744695655147, "muon_lr": 0.009985376138845464, "adamw_lr": 0.0002995612841653638, "tokens_per_sec": 151916.87614527476, "vram_gb": 27.299458026885986, "grad_norm": 0.2333984375, "tokens_processed": 380108800, "tokens_processed_billions": 0.3801088}
+{"timestamp": 1787529137.215473, "elapsed_hours": 0.6981463793913524, "session_id": 2, "event": "lambda_stats", "step": 5800, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.6953125, "lambda_mean": 1.2230902777777777, "lambda_std": 0.13360725804388837, "lambda_all": [1.6953125, 1.390625, 1.296875, 1.296875, 1.171875, 1.15625, 1.140625, 1.25, 1.1796875, 1.140625, 1.140625, 1.140625, 1.140625, 1.1875, 1.140625, 1.203125, 1.140625, 1.203125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787529141.1623282, "elapsed_hours": 0.6992427280876372, "session_id": 2, "event": "train_step", "step": 5810, "epoch": 1, "loss": 2.6573498320579527, "loss_raw": 2.1451241970062256, "perplexity": 14.258451691505364, "muon_lr": 0.00998531451719277, "adamw_lr": 0.000299559435515783, "tokens_per_sec": 151938.85042771004, "vram_gb": 27.710740089416504, "grad_norm": 0.20703125, "tokens_processed": 380764160, "tokens_processed_billions": 0.38076416}
+{"timestamp": 1787529145.1204293, "elapsed_hours": 0.7003422004646725, "session_id": 2, "event": "train_step", "step": 5820, "epoch": 1, "loss": 2.6559525537490845, "loss_raw": 2.8842952251434326, "perplexity": 14.23854257876366, "muon_lr": 0.009985252766195208, "adamw_lr": 0.0002995575829858562, "tokens_per_sec": 151960.35746471834, "vram_gb": 27.710740089416504, "grad_norm": 0.2353515625, "tokens_processed": 381419520, "tokens_processed_billions": 0.38141952}
+{"timestamp": 1787529149.0791123, "elapsed_hours": 0.7014418347676595, "session_id": 2, "event": "train_step", "step": 5830, "epoch": 1, "loss": 2.6637054800987245, "loss_raw": 2.370851516723633, "perplexity": 14.349361982949802, "muon_lr": 0.009985190885854568, "adamw_lr": 0.00029955572657563696, "tokens_per_sec": 151981.75683909547, "vram_gb": 27.710740089416504, "grad_norm": 0.2109375, "tokens_processed": 382074880, "tokens_processed_billions": 0.38207488}
+{"timestamp": 1787529153.1664178, "elapsed_hours": 0.7025771972868178, "session_id": 2, "event": "train_step", "step": 5840, "epoch": 1, "loss": 2.6388839745521544, "loss_raw": 2.5059814453125, "perplexity": 13.997573239468231, "muon_lr": 0.00998506673715117, "adamw_lr": 0.0002995520021145351, "tokens_per_sec": 151995.31891532414, "vram_gb": 27.299458026885986, "grad_norm": 0.212890625, "tokens_processed": 382730240, "tokens_processed_billions": 0.38273024}
+{"timestamp": 1787529157.1282012, "elapsed_hours": 0.703677692744467, "session_id": 2, "event": "train_step", "step": 5850, "epoch": 1, "loss": 2.63155770778656, "loss_raw": 1.904790997505188, "perplexity": 13.895398022307697, "muon_lr": 0.009985004468792, "adamw_lr": 0.00029955013406376, "tokens_per_sec": 152016.41103397854, "vram_gb": 27.710740089416504, "grad_norm": 0.216796875, "tokens_processed": 383385600, "tokens_processed_billions": 0.3833856}
+{"timestamp": 1787529161.1035151, "elapsed_hours": 0.7047819466061063, "session_id": 2, "event": "train_step", "step": 5860, "epoch": 1, "loss": 2.5916906094551084, "loss_raw": 2.898118734359741, "perplexity": 13.352326101154294, "muon_lr": 0.009984942071096906, "adamw_lr": 0.00029954826213290714, "tokens_per_sec": 152036.63246906523, "vram_gb": 27.710740089416504, "grad_norm": 0.296875, "tokens_processed": 384040960, "tokens_processed_billions": 0.38404096}
+{"timestamp": 1787529165.0700471, "elapsed_hours": 0.7058837611145443, "session_id": 2, "event": "train_step", "step": 5870, "epoch": 1, "loss": 2.534508562088013, "loss_raw": 2.6180601119995117, "perplexity": 12.610232178859482, "muon_lr": 0.009984879544067689, "adamw_lr": 0.0002995463863220306, "tokens_per_sec": 152057.29156725013, "vram_gb": 27.710740089416504, "grad_norm": 0.2080078125, "tokens_processed": 384696320, "tokens_processed_billions": 0.38469632}
+{"timestamp": 1787529169.1792417, "elapsed_hours": 0.7070252041021983, "session_id": 2, "event": "train_step", "step": 5880, "epoch": 1, "loss": 2.5503069519996644, "loss_raw": 2.436903953552246, "perplexity": 12.811035552180417, "muon_lr": 0.009984754102014094, "adamw_lr": 0.0002995426230604228, "tokens_per_sec": 152069.33091522122, "vram_gb": 27.299458026885986, "grad_norm": 0.2158203125, "tokens_processed": 385351680, "tokens_processed_billions": 0.38535168}
+{"timestamp": 1787529173.113945, "elapsed_hours": 0.7081181771887674, "session_id": 2, "event": "train_step", "step": 5890, "epoch": 1, "loss": 2.5490233492851258, "loss_raw": 2.7181479930877686, "perplexity": 12.794601821616569, "muon_lr": 0.009984691186993338, "adamw_lr": 0.00029954073560980006, "tokens_per_sec": 152091.8007300423, "vram_gb": 27.710740089416504, "grad_norm": 0.2021484375, "tokens_processed": 386007040, "tokens_processed_billions": 0.38600704}
+{"timestamp": 1787529177.070867, "elapsed_hours": 0.7092173221376207, "session_id": 2, "event": "train_step", "step": 5900, "epoch": 1, "loss": 2.505951752662659, "loss_raw": 3.3471362590789795, "perplexity": 12.255217352717498, "muon_lr": 0.009984628142645693, "adamw_lr": 0.0002995388442793707, "tokens_per_sec": 152112.86652807414, "vram_gb": 27.710740089416504, "grad_norm": 0.2265625, "tokens_processed": 386662400, "tokens_processed_billions": 0.3866624}
+{"timestamp": 1787529177.0740423, "elapsed_hours": 0.7092182041539086, "session_id": 2, "event": "lambda_stats", "step": 5900, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.703125, "lambda_mean": 1.2230902777777777, "lambda_std": 0.13492052094270396, "lambda_all": [1.703125, 1.390625, 1.296875, 1.2890625, 1.171875, 1.15625, 1.140625, 1.25, 1.1796875, 1.140625, 1.140625, 1.140625, 1.140625, 1.1875, 1.140625, 1.203125, 1.140625, 1.203125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787529181.0144374, "elapsed_hours": 0.7103127582867941, "session_id": 2, "event": "train_step", "step": 5910, "epoch": 1, "loss": 2.5830603551864626, "loss_raw": 2.9088261127471924, "perplexity": 13.237587953669882, "muon_lr": 0.009984564968972976, "adamw_lr": 0.00029953694906918916, "tokens_per_sec": 152134.66102080775, "vram_gb": 27.710740089416504, "grad_norm": 0.251953125, "tokens_processed": 387317760, "tokens_processed_billions": 0.38731776}
+{"timestamp": 1787529185.071851, "elapsed_hours": 0.7114398176802529, "session_id": 2, "event": "train_step", "step": 5920, "epoch": 1, "loss": 2.6321213960647585, "loss_raw": 3.076859474182129, "perplexity": 13.903232903301607, "muon_lr": 0.009984438233659618, "adamw_lr": 0.0002995331470097885, "tokens_per_sec": 152149.59015710786, "vram_gb": 27.299458026885986, "grad_norm": 0.2138671875, "tokens_processed": 387973120, "tokens_processed_billions": 0.38797312}
+{"timestamp": 1787529189.0293171, "elapsed_hours": 0.7125391138262219, "session_id": 2, "event": "train_step", "step": 5930, "epoch": 1, "loss": 2.573541269302368, "loss_raw": 4.208805084228516, "perplexity": 13.112176067293428, "muon_lr": 0.009984374672022636, "adamw_lr": 0.000299531240160679, "tokens_per_sec": 152170.4431645556, "vram_gb": 27.710740089416504, "grad_norm": 0.23046875, "tokens_processed": 388628480, "tokens_processed_billions": 0.38862848}
+{"timestamp": 1787529192.9881337, "elapsed_hours": 0.713638785150316, "session_id": 2, "event": "train_step", "step": 5940, "epoch": 1, "loss": 2.6247870087623597, "loss_raw": 1.908179759979248, "perplexity": 13.801634245827977, "muon_lr": 0.00998431098106789, "adamw_lr": 0.0002995293294320366, "tokens_per_sec": 152191.1378515868, "vram_gb": 27.710740089416504, "grad_norm": 0.2158203125, "tokens_processed": 389283840, "tokens_processed_billions": 0.38928384}
+{"timestamp": 1787529196.9429996, "elapsed_hours": 0.7147373589542177, "session_id": 2, "event": "train_step", "step": 5950, "epoch": 1, "loss": 2.663727719783783, "loss_raw": 2.5144379138946533, "perplexity": 14.34968111178974, "muon_lr": 0.009984247160797217, "adamw_lr": 0.0002995274148239165, "tokens_per_sec": 152212.00983519672, "vram_gb": 27.710740089416504, "grad_norm": 0.263671875, "tokens_processed": 389939200, "tokens_processed_billions": 0.3899392}
+{"timestamp": 1787529201.0440784, "elapsed_hours": 0.7158765474292967, "session_id": 2, "event": "train_step", "step": 5960, "epoch": 1, "loss": 2.7084370636940003, "loss_raw": 2.341425895690918, "perplexity": 15.0058040614916, "muon_lr": 0.009984119132315467, "adamw_lr": 0.000299523573969464, "tokens_per_sec": 152224.13879793684, "vram_gb": 27.299458026885986, "grad_norm": 0.22265625, "tokens_processed": 390594560, "tokens_processed_billions": 0.39059456}
+{"timestamp": 1787529204.9884388, "elapsed_hours": 0.7169722032546997, "session_id": 2, "event": "train_step", "step": 5970, "epoch": 1, "loss": 2.725248303413391, "loss_raw": 1.7501024007797241, "perplexity": 15.260202616866167, "muon_lr": 0.009984054924108081, "adamw_lr": 0.00029952164772324236, "tokens_per_sec": 152245.51651113233, "vram_gb": 27.710740089416504, "grad_norm": 0.232421875, "tokens_processed": 391249920, "tokens_processed_billions": 0.39124992}
+{"timestamp": 1787529208.9681997, "elapsed_hours": 0.7180776923232608, "session_id": 2, "event": "train_step", "step": 5980, "epoch": 1, "loss": 2.690529613494873, "loss_raw": 2.6132378578186035, "perplexity": 14.739480081224148, "muon_lr": 0.009983990586592151, "adamw_lr": 0.0002995197175977645, "tokens_per_sec": 152264.73256150336, "vram_gb": 27.710740089416504, "grad_norm": 0.2265625, "tokens_processed": 391905280, "tokens_processed_billions": 0.39190528}
+{"timestamp": 1787529212.9131165, "elapsed_hours": 0.7191735025246938, "session_id": 2, "event": "train_step", "step": 5990, "epoch": 1, "loss": 2.6689597153663636, "loss_raw": 2.6416139602661133, "perplexity": 14.424955325443168, "muon_lr": 0.009983926119769539, "adamw_lr": 0.0002995177835930861, "tokens_per_sec": 152285.9451918217, "vram_gb": 27.710740089416504, "grad_norm": 0.197265625, "tokens_processed": 392560640, "tokens_processed_billions": 0.39256064}
+{"timestamp": 1787529216.9518735, "elapsed_hours": 0.7202953794267443, "session_id": 2, "event": "train_step", "step": 6000, "epoch": 1, "loss": 2.829031686782837, "loss_raw": 3.256593704223633, "perplexity": 16.929060252580467, "muon_lr": 0.00998379679821169, "adamw_lr": 0.0002995139039463507, "tokens_per_sec": 152301.55571374664, "vram_gb": 27.299458026885986, "grad_norm": 0.27734375, "tokens_processed": 393216000, "tokens_processed_billions": 0.393216}
+{"timestamp": 1787529216.959604, "elapsed_hours": 0.7202975270483228, "session_id": 2, "event": "lambda_stats", "step": 6000, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.703125, "lambda_mean": 1.2230902777777777, "lambda_std": 0.13492052094270396, "lambda_all": [1.703125, 1.390625, 1.296875, 1.2890625, 1.171875, 1.15625, 1.140625, 1.25, 1.1796875, 1.140625, 1.140625, 1.140625, 1.140625, 1.1875, 1.140625, 1.203125, 1.140625, 1.203125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787529230.3547294, "elapsed_hours": 0.7240183952781889, "session_id": 2, "event": "validation", "step": 6000, "epoch": 1, "val_loss": 2.6934650886058806, "val_perplexity": 14.782811025490716, "is_best": true, "best_loss": 2.6934650886058806, "best_perplexity": 14.782811025490716, "train_loss": 3.256593704223633, "train_loss_avg": 2.829031686782837}