Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -5027,3 +5027,28 @@
 {"timestamp": 1787546136.3839285, "elapsed_hours": 5.4201376171906785, "session_id": 2, "event": "train_step", "step": 44500, "epoch": 1, "loss": 1.952941210269928, "loss_raw": 2.321183681488037, "perplexity": 7.049390860109761, "muon_lr": 0.008290046673747258, "adamw_lr": 0.0002487014002124177, "tokens_per_sec": 149546.87208620502, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2916352000, "tokens_processed_billions": 2.916352}
 {"timestamp": 1787546136.387262, "elapsed_hours": 5.420138543181949, "session_id": 2, "event": "lambda_stats", "step": 44500, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1896701388888888, "lambda_std": 0.2043699421791281, "lambda_all": [2.015625, 1.2421875, 1.203125, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
 {"timestamp": 1787546149.915091, "elapsed_hours": 5.423896273440785, "session_id": 2, "event": "validation", "step": 44500, "epoch": 1, "val_loss": 1.9543376916646957, "val_perplexity": 7.05924208021219, "is_best": true, "best_loss": 1.9543376916646957, "best_perplexity": 7.05924208021219, "train_loss": 2.321183681488037, "train_loss_avg": 1.952941210269928}
+{"timestamp": 1787546154.8475273, "elapsed_hours": 5.425266394350264, "session_id": 2, "event": "train_step", "step": 44510, "epoch": 1, "loss": 1.9641323459148408, "loss_raw": 3.3540098667144775, "perplexity": 7.128724638991093, "muon_lr": 0.008289447069688951, "adamw_lr": 0.0002486834120906685, "tokens_per_sec": 149438.9904345551, "vram_gb": 27.710740089416504, "grad_norm": 0.13671875, "tokens_processed": 2917007360, "tokens_processed_billions": 2.91700736}
+{"timestamp": 1787546158.9884818, "elapsed_hours": 5.426416659620073, "session_id": 2, "event": "train_step", "step": 44520, "epoch": 1, "loss": 1.9231116831302644, "loss_raw": 1.3216519355773926, "perplexity": 6.842216187180503, "muon_lr": 0.008288247620268321, "adamw_lr": 0.00024864742860804956, "tokens_per_sec": 149440.86223751766, "vram_gb": 27.299458026885986, "grad_norm": 0.1416015625, "tokens_processed": 2917662720, "tokens_processed_billions": 2.91766272}
+{"timestamp": 1787546163.0455613, "elapsed_hours": 5.427543626162741, "session_id": 2, "event": "train_step", "step": 44530, "epoch": 1, "loss": 1.794276189804077, "loss_raw": 0.787689745426178, "perplexity": 6.015119341054153, "muon_lr": 0.00828764777494059, "adamw_lr": 0.0002486294332482176, "tokens_per_sec": 149443.37467917724, "vram_gb": 27.710740089416504, "grad_norm": 0.13671875, "tokens_processed": 2918318080, "tokens_processed_billions": 2.91831808}
+{"timestamp": 1787546167.0360377, "elapsed_hours": 5.428652091821035, "session_id": 2, "event": "train_step", "step": 44540, "epoch": 1, "loss": 1.8095910382270812, "loss_raw": 1.6907140016555786, "perplexity": 6.107949003731094, "muon_lr": 0.008287047849224308, "adamw_lr": 0.00024861143547672915, "tokens_per_sec": 149446.39590349578, "vram_gb": 27.710740089416504, "grad_norm": 0.1416015625, "tokens_processed": 2918973440, "tokens_processed_billions": 2.91897344}
+{"timestamp": 1787546171.023447, "elapsed_hours": 5.429759705464045, "session_id": 2, "event": "train_step", "step": 44550, "epoch": 1, "loss": 1.8288348007202149, "loss_raw": 1.379653811454773, "perplexity": 6.226627168477574, "muon_lr": 0.00828644784313678, "adamw_lr": 0.00024859343529410336, "tokens_per_sec": 149449.43960243525, "vram_gb": 27.710740089416504, "grad_norm": 0.1484375, "tokens_processed": 2919628800, "tokens_processed_billions": 2.9196288}
+{"timestamp": 1787546175.1770492, "elapsed_hours": 5.43091348376539, "session_id": 2, "event": "train_step", "step": 44560, "epoch": 1, "loss": 1.803877420425415, "loss_raw": 1.7649575471878052, "perplexity": 6.073150026269169, "muon_lr": 0.00828524758991719, "adamw_lr": 0.00024855742769751564, "tokens_per_sec": 149451.20844116213, "vram_gb": 27.299458026885986, "grad_norm": 0.146484375, "tokens_processed": 2920284160, "tokens_processed_billions": 2.92028416}
+{"timestamp": 1787546179.211944, "elapsed_hours": 5.432034288114972, "session_id": 2, "event": "train_step", "step": 44570, "epoch": 1, "loss": 1.9196841406822205, "loss_raw": 3.253981590270996, "perplexity": 6.818804346219868, "muon_lr": 0.008284647342819742, "adamw_lr": 0.0002485394202845922, "tokens_per_sec": 149453.88840557725, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2920939520, "tokens_processed_billions": 2.92093952}
+{"timestamp": 1787546183.2426436, "elapsed_hours": 5.4331539267963835, "session_id": 2, "event": "train_step", "step": 44580, "epoch": 1, "loss": 2.029372240304947, "loss_raw": 3.078707695007324, "perplexity": 7.609308042223472, "muon_lr": 0.00828404701542027, "adamw_lr": 0.00024852141046260805, "tokens_per_sec": 149456.59771438446, "vram_gb": 27.710740089416504, "grad_norm": 0.1591796875, "tokens_processed": 2921594880, "tokens_processed_billions": 2.92159488}
+{"timestamp": 1787546187.216381, "elapsed_hours": 5.434257742696338, "session_id": 2, "event": "train_step", "step": 44590, "epoch": 1, "loss": 2.0108351147174837, "loss_raw": 2.1608266830444336, "perplexity": 7.469552676722483, "muon_lr": 0.008283446607736087, "adamw_lr": 0.0002485033982320825, "tokens_per_sec": 149459.74113971955, "vram_gb": 27.710740089416504, "grad_norm": 0.1640625, "tokens_processed": 2922250240, "tokens_processed_billions": 2.92225024}
+{"timestamp": 1787546191.3690677, "elapsed_hours": 5.4354112668169865, "session_id": 2, "event": "train_step", "step": 44600, "epoch": 1, "loss": 1.9331106793880464, "loss_raw": 2.7726049423217773, "perplexity": 6.910974666247145, "muon_lr": 0.008282245551582844, "adamw_lr": 0.0002484673665474853, "tokens_per_sec": 149461.51478957772, "vram_gb": 27.299458026885986, "grad_norm": 0.1396484375, "tokens_processed": 2922905600, "tokens_processed_billions": 2.9229056}
+{"timestamp": 1787546191.3724556, "elapsed_hours": 5.435412208305465, "session_id": 2, "event": "lambda_stats", "step": 44600, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1896701388888888, "lambda_std": 0.2043699421791281, "lambda_all": [2.015625, 1.2421875, 1.203125, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787546195.4365966, "elapsed_hours": 5.436541135973401, "session_id": 2, "event": "train_step", "step": 44610, "epoch": 1, "loss": 1.9831676828861236, "loss_raw": 1.9152473211288452, "perplexity": 7.265722074182458, "muon_lr": 0.00828164490314842, "adamw_lr": 0.00024844934709445255, "tokens_per_sec": 149463.93955827615, "vram_gb": 27.710740089416504, "grad_norm": 0.1494140625, "tokens_processed": 2923560960, "tokens_processed_billions": 2.92356096}
+{"timestamp": 1787546199.4132383, "elapsed_hours": 5.437645758655336, "session_id": 2, "event": "train_step", "step": 44620, "epoch": 1, "loss": 1.9462054526805879, "loss_raw": 1.4831629991531372, "perplexity": 7.002067430621777, "muon_lr": 0.008281044174498552, "adamw_lr": 0.00024843132523495655, "tokens_per_sec": 149467.0574551552, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2924216320, "tokens_processed_billions": 2.92421632}
+{"timestamp": 1787546203.3876033, "elapsed_hours": 5.438749748931991, "session_id": 2, "event": "train_step", "step": 44630, "epoch": 1, "loss": 1.8515805780887604, "loss_raw": 1.810439944267273, "perplexity": 6.369879662311549, "muon_lr": 0.00828044336565057, "adamw_lr": 0.00024841330096951706, "tokens_per_sec": 149470.19103776745, "vram_gb": 27.710740089416504, "grad_norm": 0.1435546875, "tokens_processed": 2924871680, "tokens_processed_billions": 2.92487168}
+{"timestamp": 1787546207.6523132, "elapsed_hours": 5.439934390584628, "session_id": 2, "event": "train_step", "step": 44640, "epoch": 1, "loss": 1.909278566837311, "loss_raw": 1.3385010957717896, "perplexity": 6.748218653274818, "muon_lr": 0.008279241507429556, "adamw_lr": 0.00024837724522288663, "tokens_per_sec": 149471.10656185914, "vram_gb": 27.299458026885986, "grad_norm": 0.138671875, "tokens_processed": 2925527040, "tokens_processed_billions": 2.92552704}
+{"timestamp": 1787546211.7076569, "elapsed_hours": 5.441060874991947, "session_id": 2, "event": "train_step", "step": 44650, "epoch": 1, "loss": 1.9838343608379363, "loss_raw": 1.3850027322769165, "perplexity": 7.270567585911762, "muon_lr": 0.008278640458091185, "adamw_lr": 0.0002483592137427355, "tokens_per_sec": 149473.61964618936, "vram_gb": 27.710740089416504, "grad_norm": 0.146484375, "tokens_processed": 2926182400, "tokens_processed_billions": 2.9261824}
+{"timestamp": 1787546215.6784048, "elapsed_hours": 5.442163860532973, "session_id": 2, "event": "train_step", "step": 44660, "epoch": 1, "loss": 1.9711168754100798, "loss_raw": 1.4248474836349487, "perplexity": 7.178689714644808, "muon_lr": 0.008278039328624011, "adamw_lr": 0.0002483411798587203, "tokens_per_sec": 149476.77802211052, "vram_gb": 27.710740089416504, "grad_norm": 0.142578125, "tokens_processed": 2926837760, "tokens_processed_billions": 2.92683776}
+{"timestamp": 1787546219.6915205, "elapsed_hours": 5.4432786147462, "session_id": 2, "event": "train_step", "step": 44670, "epoch": 1, "loss": 1.896399155855179, "loss_raw": 2.0202133655548096, "perplexity": 6.661862871360877, "muon_lr": 0.008277438119045374, "adamw_lr": 0.0002483231435713612, "tokens_per_sec": 149479.61172788937, "vram_gb": 27.710740089416504, "grad_norm": 0.1435546875, "tokens_processed": 2927493120, "tokens_processed_billions": 2.92749312}
+{"timestamp": 1787546223.9878783, "elapsed_hours": 5.444472047554122, "session_id": 2, "event": "train_step", "step": 44680, "epoch": 1, "loss": 1.9034269440174103, "loss_raw": 1.1163804531097412, "perplexity": 6.708845932513601, "muon_lr": 0.008276235459623053, "adamw_lr": 0.00024828706378869154, "tokens_per_sec": 149480.28187276766, "vram_gb": 27.299458026885986, "grad_norm": 0.142578125, "tokens_processed": 2928148480, "tokens_processed_billions": 2.92814848}
+{"timestamp": 1787546228.0274684, "elapsed_hours": 5.445594156053331, "session_id": 2, "event": "train_step", "step": 44690, "epoch": 1, "loss": 1.8443445587158203, "loss_raw": 1.0722544193267822, "perplexity": 6.323953451538306, "muon_lr": 0.00827563400981405, "adamw_lr": 0.00024826902029442143, "tokens_per_sec": 149482.9121316911, "vram_gb": 27.710740089416504, "grad_norm": 0.134765625, "tokens_processed": 2928803840, "tokens_processed_billions": 2.92880384}
+{"timestamp": 1787546232.1172726, "elapsed_hours": 5.446730212635464, "session_id": 2, "event": "train_step", "step": 44700, "epoch": 1, "loss": 1.7989446806907654, "loss_raw": 2.8729169368743896, "perplexity": 6.043266522187664, "muon_lr": 0.008275032479962944, "adamw_lr": 0.0002482509743988883, "tokens_per_sec": 149485.15726853668, "vram_gb": 27.710740089416504, "grad_norm": 0.138671875, "tokens_processed": 2929459200, "tokens_processed_billions": 2.9294592}
+{"timestamp": 1787546232.1206386, "elapsed_hours": 5.446731147766113, "session_id": 2, "event": "lambda_stats", "step": 44700, "epoch": 1, "lambda_min": 1.078125, "lambda_max": 2.015625, "lambda_mean": 1.1896701388888888, "lambda_std": 0.2043699421791281, "lambda_all": [2.015625, 1.2421875, 1.203125, 1.140625, 1.140625, 1.140625, 1.140625, 1.21875, 1.140625, 1.109375, 1.140625, 1.1171875, 1.140625, 1.109375, 1.109375, 1.078125, 1.1015625, 1.125], "local_layers": 9, "global_layers": 9}
+{"timestamp": 1787546236.1768198, "elapsed_hours": 5.44785786456532, "session_id": 2, "event": "train_step", "step": 44710, "epoch": 1, "loss": 1.7858576107025146, "loss_raw": 2.2908401489257812, "perplexity": 5.964693139378736, "muon_lr": 0.008274430870087083, "adamw_lr": 0.00024823292610261244, "tokens_per_sec": 149487.632359739, "vram_gb": 27.710740089416504, "grad_norm": 0.150390625, "tokens_processed": 2930114560, "tokens_processed_billions": 2.93011456}
+{"timestamp": 1787546240.2647903, "elapsed_hours": 5.448993411925104, "session_id": 2, "event": "train_step", "step": 44720, "epoch": 1, "loss": 1.856523985862732, "loss_raw": 1.7565408945083618, "perplexity": 6.401446534630478, "muon_lr": 0.008273227410330493, "adamw_lr": 0.0002481968223099148, "tokens_per_sec": 149489.88891330487, "vram_gb": 27.299458026885986, "grad_norm": 0.1494140625, "tokens_processed": 2930769920, "tokens_processed_billions": 2.93076992}
+{"timestamp": 1787546244.1915607, "elapsed_hours": 5.450084181427956, "session_id": 2, "event": "train_step", "step": 44730, "epoch": 1, "loss": 1.845556037425995, "loss_raw": 0.8913135528564453, "perplexity": 6.331619429155718, "muon_lr": 0.008272625560484467, "adamw_lr": 0.00024817876681453396, "tokens_per_sec": 149493.37509771384, "vram_gb": 27.710740089416504, "grad_norm": 0.1435546875, "tokens_processed": 2931425280, "tokens_processed_billions": 2.93142528}