Upload metrics_session_002.jsonl with huggingface_hub

metrics_session_002.jsonl

@@ -3823,3 +3823,19 @@
 {"timestamp": 1787541400.6087234, "elapsed_hours": 4.104644504586855, "session_id": 2, "event": "train_step", "step": 33850, "epoch": 1, "loss": 2.051640108823776, "loss_raw": 2.5566468238830566, "perplexity": 7.780651762827354, "muon_lr": 0.00901032370689199, "adamw_lr": 0.0002703097112067597, "tokens_per_sec": 150241.87017006474, "vram_gb": 27.710740089416504, "grad_norm": 0.1552734375, "tokens_processed": 2218393600, "tokens_processed_billions": 2.2183936}
 {"timestamp": 1787541404.6381867, "elapsed_hours": 4.105763800011741, "session_id": 2, "event": "train_step", "step": 33860, "epoch": 1, "loss": 2.045316025018692, "loss_raw": 2.2412502765655518, "perplexity": 7.731601531384495, "muon_lr": 0.009009845522132517, "adamw_lr": 0.00027029536566397543, "tokens_per_sec": 150245.25240108997, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2219048960, "tokens_processed_billions": 2.21904896}
 {"timestamp": 1787541408.6118424, "elapsed_hours": 4.106867593063249, "session_id": 2, "event": "train_step", "step": 33870, "epoch": 1, "loss": 2.118738993406296, "loss_raw": 3.0878119468688965, "perplexity": 8.320638489426917, "muon_lr": 0.00900936723615818, "adamw_lr": 0.00027028101708474533, "tokens_per_sec": 150249.20182105686, "vram_gb": 27.710740089416504, "grad_norm": 0.15625, "tokens_processed": 2219704320, "tokens_processed_billions": 2.21970432}
+{"timestamp": 1787541413.3771486, "elapsed_hours": 4.108191289438142, "session_id": 2, "event": "train_step", "step": 33880, "epoch": 1, "loss": 2.166109277009964, "loss_raw": 1.5550730228424072, "perplexity": 8.724274188030812, "muon_lr": 0.00900841036062009, "adamw_lr": 0.0002702523108186027, "tokens_per_sec": 150245.09807375114, "vram_gb": 27.299458026885986, "grad_norm": 0.15625, "tokens_processed": 2220359680, "tokens_processed_billions": 2.22035968}
+{"timestamp": 1787541417.3421197, "elapsed_hours": 4.109292670024766, "session_id": 2, "event": "train_step", "step": 33890, "epoch": 1, "loss": 2.1800753617286683, "loss_raw": 1.746841311454773, "perplexity": 8.846972956774312, "muon_lr": 0.009007931771083933, "adamw_lr": 0.00027023795313251795, "tokens_per_sec": 150249.13345798381, "vram_gb": 27.710740089416504, "grad_norm": 0.138671875, "tokens_processed": 2221015040, "tokens_processed_billions": 2.22101504}
+{"timestamp": 1787541421.3116095, "elapsed_hours": 4.110395306282573, "session_id": 2, "event": "train_step", "step": 33900, "epoch": 1, "loss": 2.134861433506012, "loss_raw": 2.7205810546875, "perplexity": 8.455874724764348, "muon_lr": 0.009007453080388101, "adamw_lr": 0.00027022359241164295, "tokens_per_sec": 150253.1193909179, "vram_gb": 27.710740089416504, "grad_norm": 0.15234375, "tokens_processed": 2221670400, "tokens_processed_billions": 2.2216704}
+{"timestamp": 1787541421.3150542, "elapsed_hours": 4.110396263135804, "session_id": 2, "event": "lambda_stats", "step": 33900, "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": 1787541425.2954657, "elapsed_hours": 4.111501932872666, "session_id": 2, "event": "train_step", "step": 33910, "epoch": 1, "loss": 2.0678900861740113, "loss_raw": 1.5691440105438232, "perplexity": 7.908120051196725, "muon_lr": 0.009006974288546396, "adamw_lr": 0.0002702092286563919, "tokens_per_sec": 150256.95870754746, "vram_gb": 27.710740089416504, "grad_norm": 0.138671875, "tokens_processed": 2222325760, "tokens_processed_billions": 2.22232576}
+{"timestamp": 1787541429.4320168, "elapsed_hours": 4.112650975121392, "session_id": 2, "event": "train_step", "step": 33920, "epoch": 1, "loss": 1.9725482630729676, "loss_raw": 2.4810750484466553, "perplexity": 7.188972560151463, "muon_lr": 0.009006016401480608, "adamw_lr": 0.0002701804920444182, "tokens_per_sec": 150259.24381386774, "vram_gb": 27.299458026885986, "grad_norm": 0.1416015625, "tokens_processed": 2222981120, "tokens_processed_billions": 2.22298112}
+{"timestamp": 1787541433.4364672, "elapsed_hours": 4.113763322234154, "session_id": 2, "event": "train_step", "step": 33930, "epoch": 1, "loss": 1.9697084403038025, "loss_raw": 2.49200177192688, "perplexity": 7.1685861128375725, "muon_lr": 0.009005537306284145, "adamw_lr": 0.00027016611918852435, "tokens_per_sec": 150262.8701034819, "vram_gb": 27.710740089416504, "grad_norm": 0.1533203125, "tokens_processed": 2223636480, "tokens_processed_billions": 2.22363648}
+{"timestamp": 1787541437.420485, "elapsed_hours": 4.114869993858867, "session_id": 2, "event": "train_step", "step": 33940, "epoch": 1, "loss": 1.9999594867229462, "loss_raw": 1.5016261339187622, "perplexity": 7.38875675011759, "muon_lr": 0.00900505810999706, "adamw_lr": 0.00027015174329991174, "tokens_per_sec": 150266.70145337706, "vram_gb": 27.710740089416504, "grad_norm": 0.1455078125, "tokens_processed": 2224291840, "tokens_processed_billions": 2.22429184}
+{"timestamp": 1787541441.4920971, "elapsed_hours": 4.116000997092989, "session_id": 2, "event": "train_step", "step": 33950, "epoch": 1, "loss": 2.0067415058612825, "loss_raw": 1.885798692703247, "perplexity": 7.439037750432595, "muon_lr": 0.009004578812633168, "adamw_lr": 0.00027013736437899496, "tokens_per_sec": 150269.64120810395, "vram_gb": 27.710740089416504, "grad_norm": 0.154296875, "tokens_processed": 2224947200, "tokens_processed_billions": 2.2249472}
+{"timestamp": 1787541445.6890683, "elapsed_hours": 4.117166822552681, "session_id": 2, "event": "train_step", "step": 33960, "epoch": 1, "loss": 2.0638575422763825, "loss_raw": 1.8843598365783691, "perplexity": 7.876294422193026, "muon_lr": 0.009003619914730254, "adamw_lr": 0.0002701085974419076, "tokens_per_sec": 150271.30804092274, "vram_gb": 27.299458026885986, "grad_norm": 0.1533203125, "tokens_processed": 2225602560, "tokens_processed_billions": 2.22560256}
+{"timestamp": 1787541449.7097557, "elapsed_hours": 4.118283680147595, "session_id": 2, "event": "train_step", "step": 33970, "epoch": 1, "loss": 2.146084917783737, "loss_raw": 1.3111958503723145, "perplexity": 8.55131367875066, "muon_lr": 0.009003140314218887, "adamw_lr": 0.0002700942094265665, "tokens_per_sec": 150274.7615714534, "vram_gb": 27.710740089416504, "grad_norm": 0.1748046875, "tokens_processed": 2226257920, "tokens_processed_billions": 2.22625792}
+{"timestamp": 1787541453.6929023, "elapsed_hours": 4.11939010977745, "session_id": 2, "event": "train_step", "step": 33980, "epoch": 1, "loss": 2.080144414901733, "loss_raw": 2.148475170135498, "perplexity": 8.005624962361312, "muon_lr": 0.009002660612686016, "adamw_lr": 0.00027007981838058037, "tokens_per_sec": 150278.59443147972, "vram_gb": 27.710740089416504, "grad_norm": 0.1337890625, "tokens_processed": 2226913280, "tokens_processed_billions": 2.22691328}
+{"timestamp": 1787541457.6652288, "elapsed_hours": 4.12049353380998, "session_id": 2, "event": "train_step", "step": 33990, "epoch": 1, "loss": 2.0307699286937715, "loss_raw": 1.846954107284546, "perplexity": 7.619950919701614, "muon_lr": 0.009002180810145477, "adamw_lr": 0.0002700654243043642, "tokens_per_sec": 150282.53476788112, "vram_gb": 27.710740089416504, "grad_norm": 0.14453125, "tokens_processed": 2227568640, "tokens_processed_billions": 2.22756864}
+{"timestamp": 1787541461.7821665, "elapsed_hours": 4.1216371275318995, "session_id": 2, "event": "train_step", "step": 34000, "epoch": 1, "loss": 1.9854538583755492, "loss_raw": 2.439256191253662, "perplexity": 7.2823517918848655, "muon_lr": 0.009001220902096738, "adamw_lr": 0.0002700366270629021, "tokens_per_sec": 150285.00676724713, "vram_gb": 27.299458026885986, "grad_norm": 0.1455078125, "tokens_processed": 2228224000, "tokens_processed_billions": 2.228224}
+{"timestamp": 1787541461.7856736, "elapsed_hours": 4.121638101869159, "session_id": 2, "event": "lambda_stats", "step": 34000, "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": 1787541475.2430763, "elapsed_hours": 4.12537626961867, "session_id": 2, "event": "validation", "step": 34000, "epoch": 1, "val_loss": 2.0120910525321962, "val_perplexity": 7.4789398640159845, "is_best": true, "best_loss": 2.0120910525321962, "best_perplexity": 7.4789398640159845, "train_loss": 2.439256191253662, "train_loss_avg": 1.9854538583755492}