Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -507,3 +507,45 @@
 {"timestamp": 1787528575.5635412, "elapsed_hours": 0.5421319540341695, "session_id": 2, "event": "train_step", "step": 4500, "epoch": 1, "loss": 2.8711294078826906, "loss_raw": 2.2729012966156006, "perplexity": 17.656948840011868, "muon_lr": 0.009993690449545772, "adamw_lr": 0.00029981071348637314, "tokens_per_sec": 151982.28489176312, "vram_gb": 27.710740089416504, "grad_norm": 0.251953125, "tokens_processed": 294912000, "tokens_processed_billions": 0.294912}
 {"timestamp": 1787528575.5670547, "elapsed_hours": 0.5421329299608867, "session_id": 2, "event": "lambda_stats", "step": 4500, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.6484375, "lambda_mean": 1.2361111111111112, "lambda_std": 0.12279841691156958, "lambda_all": [1.6484375, 1.390625, 1.328125, 1.3203125, 1.1953125, 1.203125, 1.140625, 1.25, 1.203125, 1.140625, 1.171875, 1.140625, 1.140625, 1.1953125, 1.140625, 1.234375, 1.15625, 1.25], "local_layers": 9, "global_layers": 9}
 {"timestamp": 1787528588.9910204, "elapsed_hours": 0.5458618095848295, "session_id": 2, "event": "validation", "step": 4500, "epoch": 1, "val_loss": 2.9183415269851682, "val_perplexity": 18.510562719953008, "is_best": true, "best_loss": 2.9183415269851682, "best_perplexity": 18.510562719953008, "train_loss": 2.2729012966156006, "train_loss_avg": 2.8711294078826906}
+{"timestamp": 1787528593.4173126, "elapsed_hours": 0.5470913347270754, "session_id": 2, "event": "train_step", "step": 4510, "epoch": 1, "loss": 2.9056280136108397, "loss_raw": 2.797499656677246, "perplexity": 18.27671807649826, "muon_lr": 0.009993649932163874, "adamw_lr": 0.0002998094979649161, "tokens_per_sec": 150931.31346568096, "vram_gb": 27.710740089416504, "grad_norm": 0.2490234375, "tokens_processed": 295567360, "tokens_processed_billions": 0.29556736}
+{"timestamp": 1787528597.5669322, "elapsed_hours": 0.548244006898668, "session_id": 2, "event": "train_step", "step": 4520, "epoch": 1, "loss": 2.815598711967468, "loss_raw": 3.4845309257507324, "perplexity": 16.703173176927905, "muon_lr": 0.009993568508645537, "adamw_lr": 0.00029980705525936606, "tokens_per_sec": 150946.11354065503, "vram_gb": 27.299458026885986, "grad_norm": 0.3515625, "tokens_processed": 296222720, "tokens_processed_billions": 0.29622272}
+{"timestamp": 1787528601.4899118, "elapsed_hours": 0.5493337234523561, "session_id": 2, "event": "train_step", "step": 4530, "epoch": 1, "loss": 2.8174531984329225, "loss_raw": 3.288783311843872, "perplexity": 16.734177725386704, "muon_lr": 0.009993527602511449, "adamw_lr": 0.0002998058280753434, "tokens_per_sec": 150978.25149489712, "vram_gb": 27.710740089416504, "grad_norm": 0.25390625, "tokens_processed": 296878080, "tokens_processed_billions": 0.29687808}
+{"timestamp": 1787528605.4552264, "elapsed_hours": 0.5504351997375488, "session_id": 2, "event": "train_step", "step": 4540, "epoch": 1, "loss": 2.8480233407020568, "loss_raw": 2.771050453186035, "perplexity": 17.25364353686476, "muon_lr": 0.009993486566795655, "adamw_lr": 0.00029980459700386955, "tokens_per_sec": 151007.0273276135, "vram_gb": 27.710740089416504, "grad_norm": 0.2392578125, "tokens_processed": 297533440, "tokens_processed_billions": 0.29753344}
+{"timestamp": 1787528609.400929, "elapsed_hours": 0.5515312282244365, "session_id": 2, "event": "train_step", "step": 4550, "epoch": 1, "loss": 2.872327251434326, "loss_raw": 1.6409143209457397, "perplexity": 17.67811177473345, "muon_lr": 0.00999344540149933, "adamw_lr": 0.0002998033620449799, "tokens_per_sec": 151037.18008970167, "vram_gb": 27.710740089416504, "grad_norm": 0.255859375, "tokens_processed": 298188800, "tokens_processed_billions": 0.2981888}
+{"timestamp": 1787528613.5139997, "elapsed_hours": 0.5526737479368845, "session_id": 2, "event": "train_step", "step": 4560, "epoch": 1, "loss": 2.7958014154434205, "loss_raw": 2.207730770111084, "perplexity": 16.375747272524826, "muon_lr": 0.009993362682169868, "adamw_lr": 0.000299800880465096, "tokens_per_sec": 151054.42953651273, "vram_gb": 27.299458026885986, "grad_norm": 0.240234375, "tokens_processed": 298844160, "tokens_processed_billions": 0.29884416}
+{"timestamp": 1787528617.457788, "elapsed_hours": 0.553769244617886, "session_id": 2, "event": "train_step", "step": 4570, "epoch": 1, "loss": 2.8607885694503783, "loss_raw": 1.9484530687332153, "perplexity": 17.475301993859983, "muon_lr": 0.00999332112813911, "adamw_lr": 0.0002997996338441732, "tokens_per_sec": 151084.51322560047, "vram_gb": 27.710740089416504, "grad_norm": 0.22265625, "tokens_processed": 299499520, "tokens_processed_billions": 0.29949952}
+{"timestamp": 1787528621.4677055, "elapsed_hours": 0.554883110721906, "session_id": 2, "event": "train_step", "step": 4580, "epoch": 1, "loss": 2.9338729310035707, "loss_raw": 2.6871378421783447, "perplexity": 18.80030195237699, "muon_lr": 0.009993279444532596, "adamw_lr": 0.0002997983833359778, "tokens_per_sec": 151109.44443056962, "vram_gb": 27.710740089416504, "grad_norm": 0.25, "tokens_processed": 300154880, "tokens_processed_billions": 0.30015488}
+{"timestamp": 1787528625.42885, "elapsed_hours": 0.555983428425259, "session_id": 2, "event": "train_step", "step": 4590, "epoch": 1, "loss": 2.938056321144104, "loss_raw": 1.7719242572784424, "perplexity": 18.879115689567556, "muon_lr": 0.009993237631351532, "adamw_lr": 0.0002997971289405459, "tokens_per_sec": 151137.98256076008, "vram_gb": 27.710740089416504, "grad_norm": 0.298828125, "tokens_processed": 300810240, "tokens_processed_billions": 0.30081024}
+{"timestamp": 1787528629.5342727, "elapsed_hours": 0.5571238238281674, "session_id": 2, "event": "train_step", "step": 4600, "epoch": 1, "loss": 2.854786367416382, "loss_raw": 2.5697879791259766, "perplexity": 17.37072585819338, "muon_lr": 0.00999315361627057, "adamw_lr": 0.0002997946084881171, "tokens_per_sec": 151155.46222446204, "vram_gb": 27.299458026885986, "grad_norm": 0.2314453125, "tokens_processed": 301465600, "tokens_processed_billions": 0.3014656}
+{"timestamp": 1787528629.5390866, "elapsed_hours": 0.5571251610252592, "session_id": 2, "event": "lambda_stats", "step": 4600, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.6484375, "lambda_mean": 1.2348090277777777, "lambda_std": 0.12279611580034608, "lambda_all": [1.6484375, 1.390625, 1.328125, 1.3125, 1.1875, 1.203125, 1.140625, 1.25, 1.1953125, 1.140625, 1.171875, 1.140625, 1.140625, 1.1953125, 1.140625, 1.234375, 1.15625, 1.25], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787528633.4706895, "elapsed_hours": 0.5582172728247112, "session_id": 2, "event": "train_step", "step": 4610, "epoch": 1, "loss": 2.909910328388214, "loss_raw": 3.103571891784668, "perplexity": 18.35515255710447, "muon_lr": 0.0099931114143731, "adamw_lr": 0.0002997933424311929, "tokens_per_sec": 151185.67429252013, "vram_gb": 27.710740089416504, "grad_norm": 0.271484375, "tokens_processed": 302120960, "tokens_processed_billions": 0.30212096}
+{"timestamp": 1787528637.4318976, "elapsed_hours": 0.5593176084094578, "session_id": 2, "event": "train_step", "step": 4620, "epoch": 1, "loss": 2.890389199256897, "loss_raw": 2.464655876159668, "perplexity": 18.000313947231007, "muon_lr": 0.009993069082905922, "adamw_lr": 0.00029979207248717756, "tokens_per_sec": 151213.8780689946, "vram_gb": 27.710740089416504, "grad_norm": 0.314453125, "tokens_processed": 302776320, "tokens_processed_billions": 0.30277632}
+{"timestamp": 1787528641.438729, "elapsed_hours": 0.5604306170675489, "session_id": 2, "event": "train_step", "step": 4630, "epoch": 1, "loss": 2.8072239685058595, "loss_raw": 2.7242162227630615, "perplexity": 16.56387250444321, "muon_lr": 0.009993026621870255, "adamw_lr": 0.00029979079865610764, "tokens_per_sec": 151238.53151631678, "vram_gb": 27.710740089416504, "grad_norm": 0.2294921875, "tokens_processed": 303431680, "tokens_processed_billions": 0.30343168}
+{"timestamp": 1787528645.542951, "elapsed_hours": 0.5615706789493561, "session_id": 2, "event": "train_step", "step": 4640, "epoch": 1, "loss": 2.8746698188781736, "loss_raw": 2.7686376571655273, "perplexity": 17.71957248714312, "muon_lr": 0.009992941311098371, "adamw_lr": 0.00029978823933295113, "tokens_per_sec": 151255.7691798841, "vram_gb": 27.299458026885986, "grad_norm": 0.251953125, "tokens_processed": 304087040, "tokens_processed_billions": 0.30408704}
+{"timestamp": 1787528649.5164037, "elapsed_hours": 0.5626744157075882, "session_id": 2, "event": "train_step", "step": 4650, "epoch": 1, "loss": 2.963240101337433, "loss_raw": 2.650409698486328, "perplexity": 19.360600546621356, "muon_lr": 0.00999289846136461, "adamw_lr": 0.0002997869538409383, "tokens_per_sec": 151282.74871253164, "vram_gb": 27.710740089416504, "grad_norm": 0.240234375, "tokens_processed": 304742400, "tokens_processed_billions": 0.3047424}
+{"timestamp": 1787528653.4932141, "elapsed_hours": 0.5637790851460562, "session_id": 2, "event": "train_step", "step": 4660, "epoch": 1, "loss": 2.980713746547699, "loss_raw": 2.5502960681915283, "perplexity": 19.701873771940562, "muon_lr": 0.009992855482067287, "adamw_lr": 0.0002997856644620185, "tokens_per_sec": 151309.36675014682, "vram_gb": 27.710740089416504, "grad_norm": 0.26171875, "tokens_processed": 305397760, "tokens_processed_billions": 0.30539776}
+{"timestamp": 1787528657.4360783, "elapsed_hours": 0.5648743252621756, "session_id": 2, "event": "train_step", "step": 4670, "epoch": 1, "loss": 2.929356894493103, "loss_raw": 2.0267581939697266, "perplexity": 18.71559052626997, "muon_lr": 0.009992812373207635, "adamw_lr": 0.00029978437119622896, "tokens_per_sec": 151338.4322199087, "vram_gb": 27.710740089416504, "grad_norm": 0.2421875, "tokens_processed": 306053120, "tokens_processed_billions": 0.30605312}
+{"timestamp": 1787528661.5464637, "elapsed_hours": 0.5660160989893808, "session_id": 2, "event": "train_step", "step": 4680, "epoch": 1, "loss": 2.9782615375518797, "loss_raw": 3.0640459060668945, "perplexity": 19.653619848375794, "muon_lr": 0.009992725766806328, "adamw_lr": 0.0002997817730041898, "tokens_per_sec": 151354.85903218426, "vram_gb": 27.299458026885986, "grad_norm": 0.283203125, "tokens_processed": 306708480, "tokens_processed_billions": 0.30670848}
+{"timestamp": 1787528665.5122898, "elapsed_hours": 0.5671177173985376, "session_id": 2, "event": "train_step", "step": 4690, "epoch": 1, "loss": 2.893753161430359, "loss_raw": 2.2388269901275635, "perplexity": 18.060968284710153, "muon_lr": 0.00999268226926717, "adamw_lr": 0.0002997804680780151, "tokens_per_sec": 151382.01449394433, "vram_gb": 27.710740089416504, "grad_norm": 0.232421875, "tokens_processed": 307363840, "tokens_processed_billions": 0.30736384}
+{"timestamp": 1787528669.45672, "elapsed_hours": 0.5682133924298817, "session_id": 2, "event": "train_step", "step": 4700, "epoch": 1, "loss": 2.845223002433777, "loss_raw": 1.3901619911193848, "perplexity": 17.205395086122653, "muon_lr": 0.009992638642170683, "adamw_lr": 0.0002997791592651205, "tokens_per_sec": 151410.6392528794, "vram_gb": 27.710740089416504, "grad_norm": 0.2578125, "tokens_processed": 308019200, "tokens_processed_billions": 0.3080192}
+{"timestamp": 1787528669.463165, "elapsed_hours": 0.5682151829534107, "session_id": 2, "event": "lambda_stats", "step": 4700, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.6484375, "lambda_mean": 1.2335069444444444, "lambda_std": 0.12289042602828579, "lambda_all": [1.6484375, 1.390625, 1.3203125, 1.3125, 1.1875, 1.1953125, 1.140625, 1.25, 1.1953125, 1.140625, 1.171875, 1.140625, 1.140625, 1.1953125, 1.140625, 1.234375, 1.1484375, 1.25], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787528673.4132118, "elapsed_hours": 0.5693124178383085, "session_id": 2, "event": "train_step", "step": 4710, "epoch": 1, "loss": 2.825435039997101, "loss_raw": 2.1993660926818848, "perplexity": 16.86828176732783, "muon_lr": 0.009992594885518125, "adamw_lr": 0.00029977784656554367, "tokens_per_sec": 151438.26542076262, "vram_gb": 27.710740089416504, "grad_norm": 0.2734375, "tokens_processed": 308674560, "tokens_processed_billions": 0.30867456}
+{"timestamp": 1787528677.4956582, "elapsed_hours": 0.5704464308420817, "session_id": 2, "event": "train_step", "step": 4720, "epoch": 1, "loss": 2.785420305728912, "loss_raw": 2.645745038986206, "perplexity": 16.206628184128938, "muon_lr": 0.009992506983549834, "adamw_lr": 0.00029977520950649495, "tokens_per_sec": 151456.4335341258, "vram_gb": 27.299458026885986, "grad_norm": 0.3203125, "tokens_processed": 309329920, "tokens_processed_billions": 0.30932992}
+{"timestamp": 1787528681.4510944, "elapsed_hours": 0.5715451631281111, "session_id": 2, "event": "train_step", "step": 4730, "epoch": 1, "loss": 2.761707892417908, "loss_raw": 2.55880069732666, "perplexity": 15.826850436613322, "muon_lr": 0.00999246283823664, "adamw_lr": 0.00029977388514709915, "tokens_per_sec": 151483.94344860676, "vram_gb": 27.710740089416504, "grad_norm": 0.54296875, "tokens_processed": 309985280, "tokens_processed_billions": 0.30998528}
+{"timestamp": 1787528685.4214032, "elapsed_hours": 0.5726480266120699, "session_id": 2, "event": "train_step", "step": 4740, "epoch": 1, "loss": 2.745821011066437, "loss_raw": 2.13590669631958, "perplexity": 15.577397899638996, "muon_lr": 0.009992418563372442, "adamw_lr": 0.0002997725569011732, "tokens_per_sec": 151510.24290962602, "vram_gb": 27.710740089416504, "grad_norm": 0.2314453125, "tokens_processed": 310640640, "tokens_processed_billions": 0.31064064}
+{"timestamp": 1787528689.3791332, "elapsed_hours": 0.5737473960055245, "session_id": 2, "event": "train_step", "step": 4750, "epoch": 1, "loss": 2.7943074107170105, "loss_raw": 1.5333151817321777, "perplexity": 16.35130009534758, "muon_lr": 0.009992374158958518, "adamw_lr": 0.0002997712247687555, "tokens_per_sec": 151537.36808158515, "vram_gb": 27.710740089416504, "grad_norm": 0.2421875, "tokens_processed": 311296000, "tokens_processed_billions": 0.311296}
+{"timestamp": 1787528693.4676614, "elapsed_hours": 0.5748830984036127, "session_id": 2, "event": "train_step", "step": 4760, "epoch": 1, "loss": 2.7658472418785096, "loss_raw": 1.8024557828903198, "perplexity": 15.892499079030967, "muon_lr": 0.009992284961486617, "adamw_lr": 0.0002997685488445984, "tokens_per_sec": 151554.7524840614, "vram_gb": 27.299458026885986, "grad_norm": 0.2353515625, "tokens_processed": 311951360, "tokens_processed_billions": 0.31195136}
+{"timestamp": 1787528697.4176683, "elapsed_hours": 0.5759803224934472, "session_id": 2, "event": "train_step", "step": 4770, "epoch": 1, "loss": 2.8285454869270326, "loss_raw": 2.45346999168396, "perplexity": 16.920831346535326, "muon_lr": 0.009992240168431213, "adamw_lr": 0.00029976720505293626, "tokens_per_sec": 151582.25923475187, "vram_gb": 27.710740089416504, "grad_norm": 0.2255859375, "tokens_processed": 312606720, "tokens_processed_billions": 0.31260672}
+{"timestamp": 1787528701.366259, "elapsed_hours": 0.5770771534575356, "session_id": 2, "event": "train_step", "step": 4780, "epoch": 1, "loss": 2.763144836425781, "loss_raw": 1.9055057764053345, "perplexity": 15.84960908204529, "muon_lr": 0.009992195245831222, "adamw_lr": 0.00029976585737493663, "tokens_per_sec": 151609.76389242403, "vram_gb": 27.710740089416504, "grad_norm": 0.30078125, "tokens_processed": 313262080, "tokens_processed_billions": 0.31326208}
+{"timestamp": 1787528705.3113065, "elapsed_hours": 0.578172999686665, "session_id": 2, "event": "train_step", "step": 4790, "epoch": 1, "loss": 2.7378903293609618, "loss_raw": 3.2647194862365723, "perplexity": 15.454347098413123, "muon_lr": 0.00999215019368795, "adamw_lr": 0.0002997645058106384, "tokens_per_sec": 151637.42072977393, "vram_gb": 27.710740089416504, "grad_norm": 0.2353515625, "tokens_processed": 313917440, "tokens_processed_billions": 0.31391744}
+{"timestamp": 1787528709.4126768, "elapsed_hours": 0.5793122692902883, "session_id": 2, "event": "train_step", "step": 4800, "epoch": 1, "loss": 2.7025021171569823, "loss_raw": 2.140148878097534, "perplexity": 14.917009173736885, "muon_lr": 0.00999205970077674, "adamw_lr": 0.0002997617910233021, "tokens_per_sec": 151653.5294997819, "vram_gb": 27.299458026885986, "grad_norm": 0.2314453125, "tokens_processed": 314572800, "tokens_processed_billions": 0.3145728}
+{"timestamp": 1787528709.4158788, "elapsed_hours": 0.5793131589227253, "session_id": 2, "event": "lambda_stats", "step": 4800, "epoch": 1, "lambda_min": 1.140625, "lambda_max": 1.6484375, "lambda_mean": 1.2317708333333333, "lambda_std": 0.12371847416895523, "lambda_all": [1.6484375, 1.390625, 1.3203125, 1.3125, 1.1875, 1.1875, 1.140625, 1.25, 1.1875, 1.140625, 1.1640625, 1.140625, 1.140625, 1.1953125, 1.140625, 1.234375, 1.140625, 1.25], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787528713.3566437, "elapsed_hours": 0.5804078157080544, "session_id": 2, "event": "train_step", "step": 4810, "epoch": 1, "loss": 2.6937646317481994, "loss_raw": 3.24676251411438, "perplexity": 14.787239778425766, "muon_lr": 0.009992014260011416, "adamw_lr": 0.0002997604278003424, "tokens_per_sec": 151681.08138917922, "vram_gb": 27.710740089416504, "grad_norm": 0.248046875, "tokens_processed": 315228160, "tokens_processed_billions": 0.31522816}
+{"timestamp": 1787528717.3222337, "elapsed_hours": 0.5815093684196472, "session_id": 2, "event": "train_step", "step": 4820, "epoch": 1, "loss": 2.69484126329422, "loss_raw": 2.1614396572113037, "perplexity": 14.803168760533454, "muon_lr": 0.009991968689708024, "adamw_lr": 0.00029975906069124064, "tokens_per_sec": 151706.94708902607, "vram_gb": 27.710740089416504, "grad_norm": 0.2451171875, "tokens_processed": 315883520, "tokens_processed_billions": 0.31588352}
+{"timestamp": 1787528721.2742865, "elapsed_hours": 0.5826071608066559, "session_id": 2, "event": "train_step", "step": 4830, "epoch": 1, "loss": 2.8388656163215638, "loss_raw": 3.7284178733825684, "perplexity": 17.096360700422515, "muon_lr": 0.009991922989867879, "adamw_lr": 0.0002997576896960363, "tokens_per_sec": 151733.69562437103, "vram_gb": 27.710740089416504, "grad_norm": 0.251953125, "tokens_processed": 316538880, "tokens_processed_billions": 0.31653888}
+{"timestamp": 1787528725.3650162, "elapsed_hours": 0.5837434748146269, "session_id": 2, "event": "train_step", "step": 4840, "epoch": 1, "loss": 2.8474251294136046, "loss_raw": 3.124166250228882, "perplexity": 17.243325299084944, "muon_lr": 0.009991831201582604, "adamw_lr": 0.00029975493604747803, "tokens_per_sec": 151750.2749821604, "vram_gb": 27.299458026885986, "grad_norm": 0.2294921875, "tokens_processed": 317194240, "tokens_processed_billions": 0.31719424}
+{"timestamp": 1787528729.2753098, "elapsed_hours": 0.5848296673430337, "session_id": 2, "event": "train_step", "step": 4850, "epoch": 1, "loss": 2.868427126407623, "loss_raw": 3.1600844860076904, "perplexity": 17.609299204614896, "muon_lr": 0.009991785113140118, "adamw_lr": 0.0002997535533942035, "tokens_per_sec": 151779.86787249497, "vram_gb": 27.710740089416504, "grad_norm": 0.2412109375, "tokens_processed": 317849600, "tokens_processed_billions": 0.3178496}
+{"timestamp": 1787528733.2442048, "elapsed_hours": 0.5859321381648381, "session_id": 2, "event": "train_step", "step": 4860, "epoch": 1, "loss": 2.792389817237854, "loss_raw": 2.290922164916992, "perplexity": 16.31997499291405, "muon_lr": 0.009991738895166177, "adamw_lr": 0.00029975216685498526, "tokens_per_sec": 151805.10788309248, "vram_gb": 27.710740089416504, "grad_norm": 0.2197265625, "tokens_processed": 318504960, "tokens_processed_billions": 0.31850496}
+{"timestamp": 1787528737.1936774, "elapsed_hours": 0.5870292139715618, "session_id": 2, "event": "train_step", "step": 4870, "epoch": 1, "loss": 2.839409728050232, "loss_raw": 2.9670908451080322, "perplexity": 17.105665562009754, "muon_lr": 0.009991692547662107, "adamw_lr": 0.0002997507764298631, "tokens_per_sec": 151831.66428765864, "vram_gb": 27.710740089416504, "grad_norm": 0.2109375, "tokens_processed": 319160320, "tokens_processed_billions": 0.31916032}
+{"timestamp": 1787528741.2884698, "elapsed_hours": 0.5881666562292311, "session_id": 2, "event": "train_step", "step": 4880, "epoch": 1, "loss": 2.829335081577301, "loss_raw": 2.06535005569458, "perplexity": 16.934197220561533, "muon_lr": 0.009991599464068936, "adamw_lr": 0.00029974798392206806, "tokens_per_sec": 151847.6314814435, "vram_gb": 27.299458026885986, "grad_norm": 0.2490234375, "tokens_processed": 319815680, "tokens_processed_billions": 0.31981568}
+{"timestamp": 1787528745.2312198, "elapsed_hours": 0.5892618645562065, "session_id": 2, "event": "train_step", "step": 4890, "epoch": 1, "loss": 2.860183184146881, "loss_raw": 3.443005323410034, "perplexity": 17.464725904486265, "muon_lr": 0.009991552727982522, "adamw_lr": 0.00029974658183947557, "tokens_per_sec": 151874.49196729137, "vram_gb": 27.710740089416504, "grad_norm": 0.25, "tokens_processed": 320471040, "tokens_processed_billions": 0.32047104}