Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -3671,3 +3671,59 @@
 {"timestamp": 1787540809.7900612, "elapsed_hours": 3.940528209540579, "session_id": 2, "event": "train_step", "step": 32500, "epoch": 1, "loss": 1.9808004879951477, "loss_raw": 1.344995379447937, "perplexity": 7.2485430350980335, "muon_lr": 0.00908967178958298, "adamw_lr": 0.0002726901536874893, "tokens_per_sec": 150262.4767691014, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 2129920000, "tokens_processed_billions": 2.12992}
 {"timestamp": 1787540809.7933142, "elapsed_hours": 3.9405291131469937, "session_id": 2, "event": "lambda_stats", "step": 32500, "epoch": 1, "lambda_min": 1.109375, "lambda_max": 2.015625, "lambda_mean": 1.1970486111111112, "lambda_std": 0.20254918979414713, "lambda_all": [2.015625, 1.265625, 1.2109375, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.125, 1.140625, 1.109375, 1.140625, 1.109375, 1.125, 1.1328125], "local_layers": 9, "global_layers": 9}
 {"timestamp": 1787540823.3487804, "elapsed_hours": 3.944294520417849, "session_id": 2, "event": "validation", "step": 32500, "epoch": 1, "val_loss": 2.0221471470594405, "val_perplexity": 7.554528215065266, "is_best": true, "best_loss": 2.0221471470594405, "best_perplexity": 7.554528215065266, "train_loss": 1.344995379447937, "train_loss_avg": 1.9808004879951477}
+{"timestamp": 1787540828.2460687, "elapsed_hours": 3.945654878218969, "session_id": 2, "event": "train_step", "step": 32510, "epoch": 1, "loss": 1.9364805948734283, "loss_raw": 2.631622791290283, "perplexity": 6.934303352567303, "muon_lr": 0.00908921090586149, "adamw_lr": 0.00027267632717584465, "tokens_per_sec": 150113.2578488247, "vram_gb": 27.710740089416504, "grad_norm": 0.1357421875, "tokens_processed": 2130575360, "tokens_processed_billions": 2.13057536}
+{"timestamp": 1787540832.410012, "elapsed_hours": 3.9468115291992825, "session_id": 2, "event": "train_step", "step": 32520, "epoch": 1, "loss": 1.8257967293262483, "loss_raw": 2.393092632293701, "perplexity": 6.207738937021964, "muon_lr": 0.009088288827921145, "adamw_lr": 0.0002726486648376343, "tokens_per_sec": 150115.39150936657, "vram_gb": 27.299458026885986, "grad_norm": 0.1455078125, "tokens_processed": 2131230720, "tokens_processed_billions": 2.13123072}
+{"timestamp": 1787540836.3984635, "elapsed_hours": 3.947919432255957, "session_id": 2, "event": "train_step", "step": 32530, "epoch": 1, "loss": 1.8407606303691864, "loss_raw": 2.050765037536621, "perplexity": 6.3013294211686945, "muon_lr": 0.009087827633728878, "adamw_lr": 0.00027263482901186627, "tokens_per_sec": 150119.37968732143, "vram_gb": 27.710740089416504, "grad_norm": 0.1416015625, "tokens_processed": 2131886080, "tokens_processed_billions": 2.13188608}
+{"timestamp": 1787540840.4789157, "elapsed_hours": 3.9490528912014433, "session_id": 2, "event": "train_step", "step": 32540, "epoch": 1, "loss": 1.8423860383033752, "loss_raw": 1.5989477634429932, "perplexity": 6.311579980419178, "muon_lr": 0.009087366336072947, "adamw_lr": 0.00027262099008218833, "tokens_per_sec": 150122.39276471385, "vram_gb": 27.710740089416504, "grad_norm": 0.158203125, "tokens_processed": 2132541440, "tokens_processed_billions": 2.13254144}
+{"timestamp": 1787540844.4598892, "elapsed_hours": 3.950158717367384, "session_id": 2, "event": "train_step", "step": 32550, "epoch": 1, "loss": 1.9256216013431549, "loss_raw": 1.475899577140808, "perplexity": 6.8594111601662595, "muon_lr": 0.009086904934966649, "adamw_lr": 0.00027260714804899945, "tokens_per_sec": 150126.4548361771, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 2133196800, "tokens_processed_billions": 2.1331968}
+{"timestamp": 1787540848.6568694, "elapsed_hours": 3.9513245450788075, "session_id": 2, "event": "train_step", "step": 32560, "epoch": 1, "loss": 1.9581900227069855, "loss_raw": 1.0466222763061523, "perplexity": 7.086489066121179, "muon_lr": 0.0090859818224562, "adamw_lr": 0.0002725794546736859, "tokens_per_sec": 150128.2325580841, "vram_gb": 27.299458026885986, "grad_norm": 0.1669921875, "tokens_processed": 2133852160, "tokens_processed_billions": 2.13385216}
+{"timestamp": 1787540852.7336278, "elapsed_hours": 3.9524569780959022, "session_id": 2, "event": "train_step", "step": 32570, "epoch": 1, "loss": 1.9808998024463653, "loss_raw": 1.7260730266571045, "perplexity": 7.24926295592037, "muon_lr": 0.009085520111078663, "adamw_lr": 0.0002725656033323598, "tokens_per_sec": 150131.28165179558, "vram_gb": 27.710740089416504, "grad_norm": 0.1474609375, "tokens_processed": 2134507520, "tokens_processed_billions": 2.13450752}
+{"timestamp": 1787540856.8581483, "elapsed_hours": 3.9536026783784233, "session_id": 2, "event": "train_step", "step": 32580, "epoch": 1, "loss": 1.9674711167812347, "loss_raw": 0.9238407611846924, "perplexity": 7.152565594727487, "muon_lr": 0.009085058296304006, "adamw_lr": 0.0002725517488891201, "tokens_per_sec": 150134.71752005623, "vram_gb": 27.710740089416504, "grad_norm": 0.142578125, "tokens_processed": 2135162880, "tokens_processed_billions": 2.13516288}
+{"timestamp": 1787540860.873229, "elapsed_hours": 3.954717978305287, "session_id": 2, "event": "train_step", "step": 32590, "epoch": 1, "loss": 2.048658355474472, "loss_raw": 2.8795289993286133, "perplexity": 7.757486332337119, "muon_lr": 0.009084596378145543, "adamw_lr": 0.0002725378913443662, "tokens_per_sec": 150137.51622130894, "vram_gb": 27.710740089416504, "grad_norm": 0.158203125, "tokens_processed": 2135818240, "tokens_processed_billions": 2.13581824}
+{"timestamp": 1787540865.0078042, "elapsed_hours": 3.955866471330325, "session_id": 2, "event": "train_step", "step": 32600, "epoch": 1, "loss": 2.104238065481186, "loss_raw": 1.9839328527450562, "perplexity": 8.20085211620147, "muon_lr": 0.00908367223173049, "adamw_lr": 0.00027251016695191465, "tokens_per_sec": 150139.9475838938, "vram_gb": 27.299458026885986, "grad_norm": 0.16015625, "tokens_processed": 2136473600, "tokens_processed_billions": 2.1364736}
+{"timestamp": 1787540865.0113995, "elapsed_hours": 3.955867470304171, "session_id": 2, "event": "lambda_stats", "step": 32600, "epoch": 1, "lambda_min": 1.109375, "lambda_max": 2.015625, "lambda_mean": 1.1970486111111112, "lambda_std": 0.20254918979414713, "lambda_all": [2.015625, 1.265625, 1.2109375, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.125, 1.140625, 1.109375, 1.140625, 1.109375, 1.125, 1.1328125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787540869.130389, "elapsed_hours": 3.957011633846495, "session_id": 2, "event": "train_step", "step": 32610, "epoch": 1, "loss": 2.1298364174365996, "loss_raw": 1.092157244682312, "perplexity": 8.413490398538086, "muon_lr": 0.009083210003500548, "adamw_lr": 0.0002724963001050164, "tokens_per_sec": 150142.50503560036, "vram_gb": 27.710740089416504, "grad_norm": 0.1640625, "tokens_processed": 2137128960, "tokens_processed_billions": 2.13712896}
+{"timestamp": 1787540873.179849, "elapsed_hours": 3.958136483695772, "session_id": 2, "event": "train_step", "step": 32620, "epoch": 1, "loss": 2.091362017393112, "loss_raw": 1.5851211547851562, "perplexity": 8.095934461914696, "muon_lr": 0.009082747671940103, "adamw_lr": 0.00027248243015820305, "tokens_per_sec": 150145.83183869265, "vram_gb": 27.710740089416504, "grad_norm": 0.14453125, "tokens_processed": 2137784320, "tokens_processed_billions": 2.13778432}
+{"timestamp": 1787540877.2354004, "elapsed_hours": 3.9592630259195962, "session_id": 2, "event": "train_step", "step": 32630, "epoch": 1, "loss": 2.1383860409259796, "loss_raw": 1.5522990226745605, "perplexity": 8.485730948386932, "muon_lr": 0.009082285237062487, "adamw_lr": 0.0002724685571118746, "tokens_per_sec": 150149.09314441928, "vram_gb": 27.710740089416504, "grad_norm": 0.240234375, "tokens_processed": 2138439680, "tokens_processed_billions": 2.13843968}
+{"timestamp": 1787540881.3928087, "elapsed_hours": 3.9604178615411123, "session_id": 2, "event": "train_step", "step": 32640, "epoch": 1, "loss": 2.0366184389591218, "loss_raw": 1.9054018259048462, "perplexity": 7.664646855795247, "muon_lr": 0.009081360057409081, "adamw_lr": 0.00027244080172227243, "tokens_per_sec": 150151.27665994235, "vram_gb": 27.299458026885986, "grad_norm": 0.138671875, "tokens_processed": 2139095040, "tokens_processed_billions": 2.13909504}
+{"timestamp": 1787540885.3804429, "elapsed_hours": 3.961525537636545, "session_id": 2, "event": "train_step", "step": 32650, "epoch": 1, "loss": 1.9288603568077087, "loss_raw": 1.9993997812271118, "perplexity": 6.881663130438961, "muon_lr": 0.009080897312659972, "adamw_lr": 0.00027242691937979916, "tokens_per_sec": 150155.25042343885, "vram_gb": 27.710740089416504, "grad_norm": 0.1416015625, "tokens_processed": 2139750400, "tokens_processed_billions": 2.1397504}
+{"timestamp": 1787540889.4819179, "elapsed_hours": 3.962664836380217, "session_id": 2, "event": "train_step", "step": 32660, "epoch": 1, "loss": 1.8600894558429717, "loss_raw": 1.768833041191101, "perplexity": 6.424311437920305, "muon_lr": 0.009080434464647052, "adamw_lr": 0.0002724130339394115, "tokens_per_sec": 150158.0214227387, "vram_gb": 27.710740089416504, "grad_norm": 0.1357421875, "tokens_processed": 2140405760, "tokens_processed_billions": 2.14040576}
+{"timestamp": 1787540893.5219438, "elapsed_hours": 3.9637870656781726, "session_id": 2, "event": "train_step", "step": 32670, "epoch": 1, "loss": 1.9108333921432494, "loss_raw": 1.4855002164840698, "perplexity": 6.75871911548373, "muon_lr": 0.009079971513383668, "adamw_lr": 0.00027239914540150994, "tokens_per_sec": 150161.43813529535, "vram_gb": 27.710740089416504, "grad_norm": 0.1533203125, "tokens_processed": 2141061120, "tokens_processed_billions": 2.14106112}
+{"timestamp": 1787540897.6559508, "elapsed_hours": 3.9649354012144937, "session_id": 2, "event": "train_step", "step": 32680, "epoch": 1, "loss": 1.7716320741176605, "loss_raw": 1.7185417413711548, "perplexity": 5.8804428524537835, "muon_lr": 0.009079045301158904, "adamw_lr": 0.00027237135903476707, "tokens_per_sec": 150163.86327539125, "vram_gb": 27.299458026885986, "grad_norm": 0.13671875, "tokens_processed": 2141716480, "tokens_processed_billions": 2.14171648}
+{"timestamp": 1787540901.6386242, "elapsed_hours": 3.9660416991180845, "session_id": 2, "event": "train_step", "step": 32690, "epoch": 1, "loss": 1.8275599026679992, "loss_raw": 1.7066264152526855, "perplexity": 6.21869391174992, "muon_lr": 0.009078582040224239, "adamw_lr": 0.00027235746120672703, "tokens_per_sec": 150167.87975457872, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2142371840, "tokens_processed_billions": 2.14237184}
+{"timestamp": 1787540905.6499522, "elapsed_hours": 3.9671559568246204, "session_id": 2, "event": "train_step", "step": 32700, "epoch": 1, "loss": 1.8832092201709747, "loss_raw": 2.609663724899292, "perplexity": 6.574570285897245, "muon_lr": 0.009078118676092525, "adamw_lr": 0.00027234356028277566, "tokens_per_sec": 150171.59329639448, "vram_gb": 27.710740089416504, "grad_norm": 0.1533203125, "tokens_processed": 2143027200, "tokens_processed_billions": 2.1430272}
+{"timestamp": 1787540905.653362, "elapsed_hours": 3.967156904273563, "session_id": 2, "event": "lambda_stats", "step": 32700, "epoch": 1, "lambda_min": 1.109375, "lambda_max": 2.015625, "lambda_mean": 1.1970486111111112, "lambda_std": 0.20254918979414713, "lambda_all": [2.015625, 1.265625, 1.2109375, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.125, 1.140625, 1.109375, 1.140625, 1.109375, 1.125, 1.1328125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787540909.6286654, "elapsed_hours": 3.96826115515497, "session_id": 2, "event": "train_step", "step": 32710, "epoch": 1, "loss": 1.8826595330238343, "loss_raw": 3.5210120677948, "perplexity": 6.570957322203942, "muon_lr": 0.009077655208777128, "adamw_lr": 0.00027232965626331374, "tokens_per_sec": 150175.647556542, "vram_gb": 27.710740089416504, "grad_norm": 0.134765625, "tokens_processed": 2143682560, "tokens_processed_billions": 2.14368256}
+{"timestamp": 1787540913.8365612, "elapsed_hours": 3.9694300150209, "session_id": 2, "event": "train_step", "step": 32720, "epoch": 1, "loss": 1.8769878017902375, "loss_raw": 1.8819468021392822, "perplexity": 6.533794107830611, "muon_lr": 0.009076727964648746, "adamw_lr": 0.0002723018389394623, "tokens_per_sec": 150177.28870779497, "vram_gb": 27.299458026885986, "grad_norm": 0.1455078125, "tokens_processed": 2144337920, "tokens_processed_billions": 2.14433792}
+{"timestamp": 1787540917.8333123, "elapsed_hours": 3.970540223585235, "session_id": 2, "event": "train_step", "step": 32730, "epoch": 1, "loss": 2.0461481416225435, "loss_raw": 2.454550266265869, "perplexity": 7.738037802885932, "muon_lr": 0.0090762641878625, "adamw_lr": 0.000272287925635875, "tokens_per_sec": 150181.15001350958, "vram_gb": 27.710740089416504, "grad_norm": 0.154296875, "tokens_processed": 2144993280, "tokens_processed_billions": 2.14499328}
+{"timestamp": 1787540921.8427205, "elapsed_hours": 3.9716539482275643, "session_id": 2, "event": "train_step", "step": 32740, "epoch": 1, "loss": 1.987230430841446, "loss_raw": 2.0259768962860107, "perplexity": 7.295300916688747, "muon_lr": 0.009075800307946053, "adamw_lr": 0.0002722740092383815, "tokens_per_sec": 150184.87503883074, "vram_gb": 27.710740089416504, "grad_norm": 0.1484375, "tokens_processed": 2145648640, "tokens_processed_billions": 2.14564864}
+{"timestamp": 1787540925.8382738, "elapsed_hours": 3.972763823999299, "session_id": 2, "event": "train_step", "step": 32750, "epoch": 1, "loss": 1.9887670576572418, "loss_raw": 0.9301151037216187, "perplexity": 7.306519689032343, "muon_lr": 0.009075336324912774, "adamw_lr": 0.00027226008974738316, "tokens_per_sec": 150188.74454790461, "vram_gb": 27.710740089416504, "grad_norm": 0.1494140625, "tokens_processed": 2146304000, "tokens_processed_billions": 2.146304}
+{"timestamp": 1787540930.0265234, "elapsed_hours": 3.973927226993773, "session_id": 2, "event": "train_step", "step": 32760, "epoch": 1, "loss": 2.021172081232071, "loss_raw": 1.5918755531311035, "perplexity": 7.5471656428404765, "muon_lr": 0.009074408049549256, "adamw_lr": 0.00027223224148647764, "tokens_per_sec": 150190.58608615285, "vram_gb": 27.299458026885986, "grad_norm": 0.138671875, "tokens_processed": 2146959360, "tokens_processed_billions": 2.14695936}
+{"timestamp": 1787540934.0184734, "elapsed_hours": 3.9750361018710665, "session_id": 2, "event": "train_step", "step": 32770, "epoch": 1, "loss": 2.0477692925930024, "loss_raw": 3.3133902549743652, "perplexity": 7.750592504162992, "muon_lr": 0.009073943757245788, "adamw_lr": 0.00027221831271737355, "tokens_per_sec": 150194.4886585054, "vram_gb": 27.710740089416504, "grad_norm": 0.158203125, "tokens_processed": 2147614720, "tokens_processed_billions": 2.14761472}
+{"timestamp": 1787540938.002885, "elapsed_hours": 3.976142882704735, "session_id": 2, "event": "train_step", "step": 32780, "epoch": 1, "loss": 1.9724226093292236, "loss_raw": 2.624375581741333, "perplexity": 7.188069295586083, "muon_lr": 0.009073479361879029, "adamw_lr": 0.00027220438085637076, "tokens_per_sec": 150198.46892790555, "vram_gb": 27.710740089416504, "grad_norm": 0.16015625, "tokens_processed": 2148270080, "tokens_processed_billions": 2.14827008}
+{"timestamp": 1787540942.0390937, "elapsed_hours": 3.977264051834742, "session_id": 2, "event": "train_step", "step": 32790, "epoch": 1, "loss": 2.0811575627326966, "loss_raw": 3.4540693759918213, "perplexity": 8.013739954076414, "muon_lr": 0.009073014863462372, "adamw_lr": 0.00027219044590387107, "tokens_per_sec": 150201.90234461462, "vram_gb": 27.710740089416504, "grad_norm": 0.1689453125, "tokens_processed": 2148925440, "tokens_processed_billions": 2.14892544}
+{"timestamp": 1787540946.2387369, "elapsed_hours": 3.9784306193060344, "session_id": 2, "event": "train_step", "step": 32800, "epoch": 1, "loss": 1.9880993044376374, "loss_raw": 1.1467196941375732, "perplexity": 7.301642365591334, "muon_lr": 0.009072085557532947, "adamw_lr": 0.0002721625667259883, "tokens_per_sec": 150203.61832269785, "vram_gb": 27.299458026885986, "grad_norm": 0.1337890625, "tokens_processed": 2149580800, "tokens_processed_billions": 2.1495808}
+{"timestamp": 1787540946.2441165, "elapsed_hours": 3.9784321138593888, "session_id": 2, "event": "lambda_stats", "step": 32800, "epoch": 1, "lambda_min": 1.109375, "lambda_max": 2.015625, "lambda_mean": 1.1970486111111112, "lambda_std": 0.20254918979414713, "lambda_all": [2.015625, 1.265625, 1.2109375, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.125, 1.140625, 1.109375, 1.140625, 1.109375, 1.125, 1.1328125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787540950.2064266, "elapsed_hours": 3.9795327553484174, "session_id": 2, "event": "train_step", "step": 32810, "epoch": 1, "loss": 2.0103102123737333, "loss_raw": 2.570335865020752, "perplexity": 7.465632919850488, "muon_lr": 0.009071620750046975, "adamw_lr": 0.00027214862250140925, "tokens_per_sec": 150207.76799603968, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2150236160, "tokens_processed_billions": 2.15023616}
+{"timestamp": 1787540954.187534, "elapsed_hours": 3.980638618535466, "session_id": 2, "event": "train_step", "step": 32820, "epoch": 1, "loss": 1.9658287453651428, "loss_raw": 2.4543697834014893, "perplexity": 7.140828066771575, "muon_lr": 0.009071155839564707, "adamw_lr": 0.0002721346751869411, "tokens_per_sec": 150211.7747032099, "vram_gb": 27.710740089416504, "grad_norm": 0.146484375, "tokens_processed": 2150891520, "tokens_processed_billions": 2.15089152}
+{"timestamp": 1787540958.2285254, "elapsed_hours": 3.9817611161867776, "session_id": 2, "event": "train_step", "step": 32830, "epoch": 1, "loss": 2.017038209438324, "loss_raw": 1.610858678817749, "perplexity": 7.516031025263516, "muon_lr": 0.009070690826099542, "adamw_lr": 0.0002721207247829862, "tokens_per_sec": 150215.1504193272, "vram_gb": 27.710740089416504, "grad_norm": 0.16015625, "tokens_processed": 2151546880, "tokens_processed_billions": 2.15154688}
+{"timestamp": 1787540962.3683596, "elapsed_hours": 3.9829110701216592, "session_id": 2, "event": "train_step", "step": 32840, "epoch": 1, "loss": 1.9814494109153749, "loss_raw": 2.6092445850372314, "perplexity": 7.2532483073261425, "muon_lr": 0.009069760490274179, "adamw_lr": 0.0002720928147082253, "tokens_per_sec": 150217.48773177707, "vram_gb": 27.299458026885986, "grad_norm": 0.154296875, "tokens_processed": 2152202240, "tokens_processed_billions": 2.15220224}
+{"timestamp": 1787540966.3574584, "elapsed_hours": 3.9840191531181337, "session_id": 2, "event": "train_step", "step": 32850, "epoch": 1, "loss": 2.100079361200333, "loss_raw": 1.682412028312683, "perplexity": 8.166818015330799, "muon_lr": 0.009069295167940805, "adamw_lr": 0.00027207885503822407, "tokens_per_sec": 150221.4043079582, "vram_gb": 27.710740089416504, "grad_norm": 0.1513671875, "tokens_processed": 2152857600, "tokens_processed_billions": 2.1528576}
+{"timestamp": 1787540970.4008818, "elapsed_hours": 3.9851423262225256, "session_id": 2, "event": "train_step", "step": 32860, "epoch": 1, "loss": 2.134668494462967, "loss_raw": 2.6068456172943115, "perplexity": 8.454243413763697, "muon_lr": 0.009068829742678194, "adamw_lr": 0.0002720648922803458, "tokens_per_sec": 150224.74981121972, "vram_gb": 27.710740089416504, "grad_norm": 0.171875, "tokens_processed": 2153512960, "tokens_processed_billions": 2.15351296}
+{"timestamp": 1787540974.369765, "elapsed_hours": 3.9862447938654157, "session_id": 2, "event": "train_step", "step": 32870, "epoch": 1, "loss": 2.119124884605408, "loss_raw": 2.705247640609741, "perplexity": 8.323849970192208, "muon_lr": 0.009068364214499771, "adamw_lr": 0.00027205092643499304, "tokens_per_sec": 150228.8735064317, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2154168320, "tokens_processed_billions": 2.15416832}
+{"timestamp": 1787540978.4856007, "elapsed_hours": 3.9873880816168255, "session_id": 2, "event": "train_step", "step": 32880, "epoch": 1, "loss": 2.0609302163124084, "loss_raw": 2.070396900177002, "perplexity": 7.853271655043682, "muon_lr": 0.009067432849449178, "adamw_lr": 0.0002720229854834753, "tokens_per_sec": 150231.45623539496, "vram_gb": 27.299458026885986, "grad_norm": 0.1533203125, "tokens_processed": 2154823680, "tokens_processed_billions": 2.15482368}
+{"timestamp": 1787540982.463088, "elapsed_hours": 3.9884929392072888, "session_id": 2, "event": "train_step", "step": 32890, "epoch": 1, "loss": 1.9845286989212036, "loss_raw": 3.026585340499878, "perplexity": 7.275617570869183, "muon_lr": 0.009066967012603868, "adamw_lr": 0.000272009010378116, "tokens_per_sec": 150235.4864523236, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 2155479040, "tokens_processed_billions": 2.15547904}
+{"timestamp": 1787540986.496808, "elapsed_hours": 3.9896134168571895, "session_id": 2, "event": "train_step", "step": 32900, "epoch": 1, "loss": 1.9154006338119507, "loss_raw": 1.8522733449935913, "perplexity": 6.789658418910901, "muon_lr": 0.00906650107289646, "adamw_lr": 0.00027199503218689375, "tokens_per_sec": 150238.92478137265, "vram_gb": 27.710740089416504, "grad_norm": 0.1337890625, "tokens_processed": 2156134400, "tokens_processed_billions": 2.1561344}
+{"timestamp": 1787540986.5053935, "elapsed_hours": 3.989615801771482, "session_id": 2, "event": "lambda_stats", "step": 32900, "epoch": 1, "lambda_min": 1.109375, "lambda_max": 2.015625, "lambda_mean": 1.1970486111111112, "lambda_std": 0.20254918979414713, "lambda_all": [2.015625, 1.265625, 1.2109375, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.125, 1.140625, 1.109375, 1.140625, 1.109375, 1.125, 1.1328125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787540990.4873154, "elapsed_hours": 3.9907218911250433, "session_id": 2, "event": "train_step", "step": 32910, "epoch": 1, "loss": 1.8251568675041199, "loss_raw": 1.066137433052063, "perplexity": 6.203768112399442, "muon_lr": 0.00906603503034039, "adamw_lr": 0.0002719810509102116, "tokens_per_sec": 150242.8140904788, "vram_gb": 27.710740089416504, "grad_norm": 0.1416015625, "tokens_processed": 2156789760, "tokens_processed_billions": 2.15678976}
+{"timestamp": 1787540994.6124544, "elapsed_hours": 3.991867763068941, "session_id": 2, "event": "train_step", "step": 32920, "epoch": 1, "loss": 1.8930151391029357, "loss_raw": 2.0068211555480957, "perplexity": 6.639357117204767, "muon_lr": 0.009065102636736024, "adamw_lr": 0.00027195307910208064, "tokens_per_sec": 150245.29154737893, "vram_gb": 27.299458026885986, "grad_norm": 0.1494140625, "tokens_processed": 2157445120, "tokens_processed_billions": 2.15744512}
+{"timestamp": 1787540998.590451, "elapsed_hours": 3.992972762187322, "session_id": 2, "event": "train_step", "step": 32930, "epoch": 1, "loss": 1.9849536538124084, "loss_raw": 3.0317819118499756, "perplexity": 7.278710037175262, "muon_lr": 0.009064636285714616, "adamw_lr": 0.0002719390885714384, "tokens_per_sec": 150249.30891892343, "vram_gb": 27.710740089416504, "grad_norm": 0.162109375, "tokens_processed": 2158100480, "tokens_processed_billions": 2.15810048}
+{"timestamp": 1787541002.5809147, "elapsed_hours": 3.9940812242693373, "session_id": 2, "event": "train_step", "step": 32940, "epoch": 1, "loss": 1.9546616172790527, "loss_raw": 1.9867630004882812, "perplexity": 7.061529119935293, "muon_lr": 0.009064169831898321, "adamw_lr": 0.00027192509495694954, "tokens_per_sec": 150253.1916761656, "vram_gb": 27.710740089416504, "grad_norm": 0.1640625, "tokens_processed": 2158755840, "tokens_processed_billions": 2.15875584}
+{"timestamp": 1787541006.604837, "elapsed_hours": 3.995198980437385, "session_id": 2, "event": "train_step", "step": 32950, "epoch": 1, "loss": 2.0160335540771483, "loss_raw": 1.755122423171997, "perplexity": 7.50848379621609, "muon_lr": 0.00906370327530059, "adamw_lr": 0.0002719110982590177, "tokens_per_sec": 150256.72338103526, "vram_gb": 27.710740089416504, "grad_norm": 0.15625, "tokens_processed": 2159411200, "tokens_processed_billions": 2.1594112}
+{"timestamp": 1787541010.7216978, "elapsed_hours": 3.9963425529003143, "session_id": 2, "event": "train_step", "step": 32960, "epoch": 1, "loss": 2.063615200519562, "loss_raw": 2.5205368995666504, "perplexity": 7.874385898432356, "muon_lr": 0.009062769853814646, "adamw_lr": 0.0002718830956144393, "tokens_per_sec": 150259.28077946446, "vram_gb": 27.299458026885986, "grad_norm": 0.1484375, "tokens_processed": 2160066560, "tokens_processed_billions": 2.16006656}
+{"timestamp": 1787541014.6790926, "elapsed_hours": 3.997441829310523, "session_id": 2, "event": "train_step", "step": 32970, "epoch": 1, "loss": 2.0427563285827635, "loss_raw": 2.0648319721221924, "perplexity": 7.711836285806818, "muon_lr": 0.009062302988953348, "adamw_lr": 0.0002718690896686004, "tokens_per_sec": 150263.50469073752, "vram_gb": 27.710740089416504, "grad_norm": 0.15625, "tokens_processed": 2160721920, "tokens_processed_billions": 2.16072192}
+{"timestamp": 1787541018.7023902, "elapsed_hours": 3.9985594118303722, "session_id": 2, "event": "train_step", "step": 32980, "epoch": 1, "loss": 1.9287954270839691, "loss_raw": 1.670169711112976, "perplexity": 6.8812163204588135, "muon_lr": 0.009061836021364452, "adamw_lr": 0.0002718550806409335, "tokens_per_sec": 150267.0366672689, "vram_gb": 27.710740089416504, "grad_norm": 0.134765625, "tokens_processed": 2161377280, "tokens_processed_billions": 2.16137728}
+{"timestamp": 1787541022.7173629, "elapsed_hours": 3.999674682219823, "session_id": 2, "event": "train_step", "step": 32990, "epoch": 1, "loss": 1.9952406919002532, "loss_raw": 2.086113929748535, "perplexity": 7.353972856591799, "muon_lr": 0.009061368951061417, "adamw_lr": 0.0002718410685318425, "tokens_per_sec": 150270.65403047352, "vram_gb": 27.710740089416504, "grad_norm": 0.17578125, "tokens_processed": 2162032640, "tokens_processed_billions": 2.16203264}
+{"timestamp": 1787541026.833551, "elapsed_hours": 4.000818067656623, "session_id": 2, "event": "train_step", "step": 33000, "epoch": 1, "loss": 2.0204446506500244, "loss_raw": 2.1008999347686768, "perplexity": 7.54167760007118, "muon_lr": 0.009060434502366831, "adamw_lr": 0.0002718130350710049, "tokens_per_sec": 150273.21141929785, "vram_gb": 27.299458026885986, "grad_norm": 0.1591796875, "tokens_processed": 2162688000, "tokens_processed_billions": 2.162688}
+{"timestamp": 1787541026.8368824, "elapsed_hours": 4.000818993316757, "session_id": 2, "event": "lambda_stats", "step": 33000, "epoch": 1, "lambda_min": 1.109375, "lambda_max": 2.015625, "lambda_mean": 1.1966145833333333, "lambda_std": 0.20269468977080954, "lambda_all": [2.015625, 1.265625, 1.2109375, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.125, 1.140625, 1.109375, 1.140625, 1.109375, 1.125, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787541040.3042943, "elapsed_hours": 4.0045599412255815, "session_id": 2, "event": "validation", "step": 33000, "epoch": 1, "val_loss": 2.0201363903284073, "val_perplexity": 7.5393531583936255, "is_best": true, "best_loss": 2.0201363903284073, "best_perplexity": 7.5393531583936255, "train_loss": 2.1008999347686768, "train_loss_avg": 2.0204446506500244}