codedp-ase26's picture
Initial commit
903307f
{"timestamp": 1773845500.3890991, "event": "train_step", "step": 10, "epoch": 1, "metrics": {"train/step_loss": 1.5491975618131233, "train/step_real_loss": 1.1542506143450737, "train/lr": 5.2631578947368424e-05, "train/step_canary_loss": 14.1875, "perf/step_duration_sec": 7.20499980985187, "perf/samples_per_sec": 9.160305585262314, "perf/tokens_per_sec": 6590.423491069632, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 47484.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773845576.5618713, "event": "train_step", "step": 20, "epoch": 1, "metrics": {"train/step_loss": 1.749317905796108, "train/step_real_loss": 1.195575013756752, "train/lr": 9.999795083071328e-05, "train/step_canary_loss": 13.5625, "perf/step_duration_sec": 7.156173000810668, "perf/samples_per_sec": 9.362546153147791, "perf/tokens_per_sec": 6812.859330605484, "perf/logical_batch_size": 67.0, "perf/logical_token_count": 48754.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 17.044722080230713, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773845647.392718, "event": "train_step", "step": 30, "epoch": 1, "metrics": {"train/step_loss": 1.0699390769004822, "train/step_real_loss": 1.0699390769004822, "train/lr": 9.975225368753412e-05, "perf/step_duration_sec": 6.728532065171748, "perf/samples_per_sec": 9.511732927792234, "perf/tokens_per_sec": 7888.793496987684, "perf/logical_batch_size": 64.0, "perf/logical_token_count": 53080.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773845720.600243, "event": "train_step", "step": 40, "epoch": 1, "metrics": {"train/step_loss": 1.095807023346424, "train/step_real_loss": 1.095807023346424, "train/lr": 9.909902907826884e-05, "perf/step_duration_sec": 6.983004813082516, "perf/samples_per_sec": 9.165108962848961, "perf/tokens_per_sec": 7604.176342613747, "perf/logical_batch_size": 64.0, "perf/logical_token_count": 53100.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773845792.8609428, "event": "train_step", "step": 50, "epoch": 1, "metrics": {"train/step_loss": 1.0978227560336773, "train/step_real_loss": 1.0427106022834778, "train/lr": 9.804362765439688e-05, "train/step_canary_loss": 4.625, "perf/step_duration_sec": 7.137449341127649, "perf/samples_per_sec": 9.106894759372206, "perf/tokens_per_sec": 6859.453238832368, "perf/logical_batch_size": 65.0, "perf/logical_token_count": 48959.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773845810.9490905, "event": "eval_step", "step": 50, "epoch": 1, "metrics": {"eval/loss": 0.9920933229195606, "eval/duration_sec": 18.071402312023565}}
{"timestamp": 1773845883.288467, "event": "train_step", "step": 60, "epoch": 1, "metrics": {"train/step_loss": 1.3474635359118967, "train/step_real_loss": 1.0449612587690353, "train/lr": 9.659469435229992e-05, "train/step_canary_loss": 6.1875, "perf/step_duration_sec": 7.404544448945671, "perf/samples_per_sec": 9.183549436276593, "perf/tokens_per_sec": 7116.845656521583, "perf/logical_batch_size": 68.0, "perf/logical_token_count": 52697.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773845955.9986777, "event": "train_step", "step": 70, "epoch": 1, "metrics": {"train/step_loss": 1.2259749336676165, "train/step_real_loss": 0.9742475748062134, "train/lr": 9.476409758141405e-05, "train/step_canary_loss": 9.28125, "perf/step_duration_sec": 7.936319092987105, "perf/samples_per_sec": 8.316197877970987, "perf/tokens_per_sec": 6486.50834181922, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 51479.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 17.044722080230713, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846033.6489208, "event": "train_step", "step": 80, "epoch": 1, "metrics": {"train/step_loss": 1.075090453028679, "train/step_real_loss": 1.075090453028679, "train/lr": 9.256683200847638e-05, "perf/step_duration_sec": 8.235574607970193, "perf/samples_per_sec": 7.771163889167165, "perf/tokens_per_sec": 6101.2136240466025, "perf/logical_batch_size": 64.0, "perf/logical_token_count": 50247.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846114.691344, "event": "train_step", "step": 90, "epoch": 1, "metrics": {"train/step_loss": 1.0333750322461128, "train/step_real_loss": 1.0333750322461128, "train/lr": 9.002089573417356e-05, "perf/step_duration_sec": 8.228110728086904, "perf/samples_per_sec": 7.778213263651652, "perf/tokens_per_sec": 6766.3163318834695, "perf/logical_batch_size": 64.0, "perf/logical_token_count": 55674.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846196.510411, "event": "train_step", "step": 100, "epoch": 1, "metrics": {"train/step_loss": 1.3118896195382783, "train/step_real_loss": 1.0975150763988495, "train/lr": 8.714714286825512e-05, "train/step_canary_loss": 8.171875, "perf/step_duration_sec": 8.322382139973342, "perf/samples_per_sec": 7.930421709788419, "perf/tokens_per_sec": 5789.32800605024, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 48181.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846218.1695988, "event": "eval_step", "step": 100, "epoch": 1, "metrics": {"eval/loss": 0.9644145566921729, "eval/duration_sec": 21.646080798935145}}
{"timestamp": 1773846300.9108367, "event": "train_step", "step": 110, "epoch": 1, "metrics": {"train/step_loss": 1.1254502712790646, "train/step_real_loss": 0.9916823133826256, "train/lr": 8.39691127106884e-05, "train/step_canary_loss": 3.9791666666666665, "perf/step_duration_sec": 8.406522131990641, "perf/samples_per_sec": 7.970002213523535, "perf/tokens_per_sec": 6148.7972301049485, "perf/logical_batch_size": 67.0, "perf/logical_token_count": 51690.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846382.1696744, "event": "train_step", "step": 120, "epoch": 1, "metrics": {"train/step_loss": 0.9898425564169884, "train/step_real_loss": 0.9898425564169884, "train/lr": 8.051283693805624e-05, "perf/step_duration_sec": 7.923252637963742, "perf/samples_per_sec": 8.077490763496323, "perf/tokens_per_sec": 6447.73079116776, "perf/logical_batch_size": 64.0, "perf/logical_token_count": 51087.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846465.942783, "event": "train_step", "step": 130, "epoch": 1, "metrics": {"train/step_loss": 1.0856322862885215, "train/step_real_loss": 1.0018824934959412, "train/lr": 7.680662637455889e-05, "train/step_canary_loss": 3.765625, "perf/step_duration_sec": 8.345965349115431, "perf/samples_per_sec": 7.908012703047609, "perf/tokens_per_sec": 6140.811500665045, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 51251.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846547.3686187, "event": "train_step", "step": 140, "epoch": 1, "metrics": {"train/step_loss": 1.11777756430886, "train/step_real_loss": 0.9652081280946732, "train/lr": 7.288083909420865e-05, "train/step_canary_loss": 6.0, "perf/step_duration_sec": 7.923310180194676, "perf/samples_per_sec": 8.329851854717921, "perf/tokens_per_sec": 6493.371940505792, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 51449.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846630.8803382, "event": "train_step", "step": 150, "epoch": 1, "metrics": {"train/step_loss": 1.0294065218705397, "train/step_real_loss": 1.005940206348896, "train/lr": 6.876763175372305e-05, "train/step_canary_loss": 2.53125, "perf/step_duration_sec": 8.249663470080122, "perf/samples_per_sec": 7.879109279516914, "perf/tokens_per_sec": 6361.229181084441, "perf/logical_batch_size": 65.0, "perf/logical_token_count": 52478.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846652.1329367, "event": "eval_step", "step": 150, "epoch": 1, "metrics": {"eval/loss": 0.9538889107379046, "eval/duration_sec": 21.243608912918717}}
{"timestamp": 1773846734.8427546, "event": "train_step", "step": 160, "epoch": 1, "metrics": {"train/step_loss": 1.02191624045372, "train/step_real_loss": 1.02191624045372, "train/lr": 6.450069619298299e-05, "perf/step_duration_sec": 8.291792104020715, "perf/samples_per_sec": 7.718476198766031, "perf/tokens_per_sec": 6133.052947063246, "perf/logical_batch_size": 64.0, "perf/logical_token_count": 50854.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846816.398206, "event": "train_step", "step": 170, "epoch": 1, "metrics": {"train/step_loss": 0.9282315596938133, "train/step_real_loss": 0.9282315596938133, "train/lr": 6.011498346059712e-05, "perf/step_duration_sec": 7.786464124917984, "perf/samples_per_sec": 8.219391879709473, "perf/tokens_per_sec": 6340.875551201497, "perf/logical_batch_size": 64.0, "perf/logical_token_count": 49373.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846898.7244656, "event": "train_step", "step": 180, "epoch": 1, "metrics": {"train/step_loss": 1.1359728827620998, "train/step_real_loss": 1.0057005435228348, "train/lr": 5.564641752511637e-05, "train/step_canary_loss": 5.3046875, "perf/step_duration_sec": 8.632995563792065, "perf/samples_per_sec": 7.645086750283158, "perf/tokens_per_sec": 5906.756191775592, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 50993.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 64.87937498092651}}
{"timestamp": 1773846945.409436, "event": "train_epoch", "step": 183, "epoch": 1, "metrics": {"train/epoch_loss": 1.1745113145391142, "train/epoch_real_loss": 1.045788711171007, "train/epoch_canary_loss": 7.361844605070935, "perf/epoch_duration_sec": 1529.87840606994, "perf/epoch_samples_per_sec": 31.26849807814928, "perf/epoch_tokens_per_sec": 24337.648568848304, "perf/epoch_samples": 47837.0, "perf/epoch_tokens": 37233643.0, "system/cuda_epoch_peak_memory_gb": 64.87937498092651, "eval/loss": 0.9498279011481768, "eval/duration_sec": 21.582291743019596}}
{"timestamp": 1773846954.5938265, "event": "audit_epoch", "step": 183, "epoch": 1, "metrics": {"audit/delta": 1e-05, "audit/num_canaries": 500.0, "audit/num_members": 250.0, "audit/paper_guess_fraction": 0.2, "audit/paper_guess_steps": 20.0, "audit/loss/auc": 0.996728, "audit/loss/empirical_epsilon/0.05": 3.4791953936219215, "audit/loss/empirical_epsilon/0.01": 3.023197554051876, "audit/loss/empirical_epsilon_details/0.05/epsilon": 3.4791953936219215, "audit/loss/empirical_epsilon_details/0.05/num_guesses": 100.0, "audit/loss/empirical_epsilon_details/0.05/correct_guesses": 100.0, "audit/loss/empirical_epsilon_details/0.01/epsilon": 3.023197554051876, "audit/loss/empirical_epsilon_details/0.01/num_guesses": 100.0, "audit/loss/empirical_epsilon_details/0.01/correct_guesses": 100.0, "audit/embedding/auc": 0.996, "audit/embedding/empirical_epsilon/0.05": 3.4791953936219215, "audit/embedding/empirical_epsilon/0.01": 3.023197554051876, "audit/embedding/empirical_epsilon_details/0.05/epsilon": 3.4791953936219215, "audit/embedding/empirical_epsilon_details/0.05/num_guesses": 100.0, "audit/embedding/empirical_epsilon_details/0.05/correct_guesses": 100.0, "audit/embedding/empirical_epsilon_details/0.01/epsilon": 3.023197554051876, "audit/embedding/empirical_epsilon_details/0.01/num_guesses": 100.0, "audit/embedding/empirical_epsilon_details/0.01/correct_guesses": 100.0, "perf/audit_duration_sec": 7.048272565938532}}
{"timestamp": 1773847012.740762, "event": "train_step", "step": 190, "epoch": 2, "metrics": {"train/step_loss": 1.0674721995989482, "train/step_real_loss": 1.0078131332993507, "train/lr": 5.113160101692939e-05, "train/step_canary_loss": 2.9765625, "perf/step_duration_sec": 8.316238466184586, "perf/samples_per_sec": 7.9362803590070925, "perf/tokens_per_sec": 5525.09408993421, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 45948.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 56.812254428863525}}
{"timestamp": 1773847096.9887106, "event": "train_step", "step": 200, "epoch": 2, "metrics": {"train/step_loss": 1.1259089543269232, "train/step_real_loss": 1.0414504930377007, "train/lr": 4.660751541114641e-05, "train/step_canary_loss": 6.53125, "perf/step_duration_sec": 8.248281789012253, "perf/samples_per_sec": 7.880429119988136, "perf/tokens_per_sec": 5337.475261653503, "perf/logical_batch_size": 65.0, "perf/logical_token_count": 44025.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 56.81271266937256}}
{"timestamp": 1773847117.9581919, "event": "eval_step", "step": 200, "epoch": 2, "metrics": {"eval/loss": 0.9487124373773477, "eval/duration_sec": 20.96032243501395}}
{"timestamp": 1773847200.0371358, "event": "train_step", "step": 210, "epoch": 2, "metrics": {"train/step_loss": 1.1445510784784954, "train/step_real_loss": 1.0787557885050774, "train/lr": 4.2111218107314846e-05, "train/step_canary_loss": 3.25, "perf/step_duration_sec": 8.08854101691395, "perf/samples_per_sec": 8.159691576266644, "perf/tokens_per_sec": 6864.773249451238, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 55526.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 17.044722080230713, "system/cuda_max_memory_allocated_gb": 56.81271266937256}}
{"timestamp": 1773847279.7295897, "event": "train_step", "step": 220, "epoch": 2, "metrics": {"train/step_loss": 1.0096144080162048, "train/step_real_loss": 0.9559597969055176, "train/lr": 3.7679538887227244e-05, "train/step_canary_loss": 2.7265625, "perf/step_duration_sec": 8.052449005888775, "perf/samples_per_sec": 8.196264260939007, "perf/tokens_per_sec": 6205.4413462857765, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 49969.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84816789627075}}
{"timestamp": 1773847361.656875, "event": "train_step", "step": 230, "epoch": 2, "metrics": {"train/step_loss": 0.9799742924444603, "train/step_real_loss": 0.9675686806440353, "train/lr": 3.334877823717737e-05, "train/step_canary_loss": 1.376953125, "perf/step_duration_sec": 8.545294485986233, "perf/samples_per_sec": 7.7235489201964915, "perf/tokens_per_sec": 6140.923532367137, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 52476.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84816789627075}}
{"timestamp": 1773847448.8259733, "event": "train_step", "step": 240, "epoch": 2, "metrics": {"train/step_loss": 1.0699740214781328, "train/step_real_loss": 0.9959888234734535, "train/lr": 2.9154410005747583e-05, "train/step_canary_loss": 3.4375, "perf/step_duration_sec": 8.351536879083142, "perf/samples_per_sec": 7.902737059726149, "perf/tokens_per_sec": 6316.322464206273, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 52751.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84816789627075}}
{"timestamp": 1773847527.7394445, "event": "train_step", "step": 250, "epoch": 2, "metrics": {"train/step_loss": 1.0023332100648146, "train/step_real_loss": 0.9728286564350128, "train/lr": 2.5130790832697737e-05, "train/step_canary_loss": 2.890625, "perf/step_duration_sec": 8.24364663194865, "perf/samples_per_sec": 7.884860050658815, "perf/tokens_per_sec": 5500.114454721866, "perf/logical_batch_size": 65.0, "perf/logical_token_count": 45341.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84816789627075}}
{"timestamp": 1773847548.8306634, "event": "eval_step", "step": 250, "epoch": 2, "metrics": {"eval/loss": 0.9464517607317342, "eval/duration_sec": 21.085849778959528}}
{"timestamp": 1773847631.4214828, "event": "train_step", "step": 260, "epoch": 2, "metrics": {"train/step_loss": 1.0758904056115584, "train/step_real_loss": 0.9906154796481133, "train/lr": 2.1310878729063642e-05, "train/step_canary_loss": 3.8046875, "perf/step_duration_sec": 8.149904164019972, "perf/samples_per_sec": 8.098254736709105, "perf/tokens_per_sec": 6327.313666786037, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 51567.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 17.044722080230713, "system/cuda_max_memory_allocated_gb": 60.84816789627075}}
{"timestamp": 1773847712.8631833, "event": "train_step", "step": 270, "epoch": 2, "metrics": {"train/step_loss": 1.1142344745722683, "train/step_real_loss": 0.9749820455908775, "train/lr": 1.7725963113612998e-05, "train/step_canary_loss": 5.5703125, "perf/step_duration_sec": 7.97826447407715, "perf/samples_per_sec": 8.272475826596892, "perf/tokens_per_sec": 6396.37858155898, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 51032.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84816789627075}}
{"timestamp": 1773847793.1259449, "event": "train_step", "step": 280, "epoch": 2, "metrics": {"train/step_loss": 1.0694914211088151, "train/step_real_loss": 1.0146433636546135, "train/lr": 1.4405408516967328e-05, "train/step_canary_loss": 2.2395833333333335, "perf/step_duration_sec": 8.121065079933032, "perf/samples_per_sec": 8.25014937579499, "perf/tokens_per_sec": 6357.909891349593, "perf/logical_batch_size": 67.0, "perf/logical_token_count": 51633.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773847869.4765995, "event": "train_step", "step": 290, "epoch": 2, "metrics": {"train/step_loss": 0.9731774203705065, "train/step_real_loss": 0.9380374550819397, "train/lr": 1.1376414052744056e-05, "train/step_canary_loss": 2.09765625, "perf/step_duration_sec": 7.720702049089596, "perf/samples_per_sec": 8.54844541083962, "perf/tokens_per_sec": 6774.642988090398, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 52305.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773847946.6124372, "event": "train_step", "step": 300, "epoch": 2, "metrics": {"train/step_loss": 1.0307691061135493, "train/step_real_loss": 1.0040206909179688, "train/lr": 8.66379062592345e-06, "train/step_canary_loss": 1.88671875, "perf/step_duration_sec": 7.759747894015163, "perf/samples_per_sec": 8.505430962635218, "perf/tokens_per_sec": 6224.171282323573, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 48298.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 17.044722080230713, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773847966.612822, "event": "eval_step", "step": 300, "epoch": 2, "metrics": {"eval/loss": 0.9457971912312817, "eval/duration_sec": 19.98325987579301}}
{"timestamp": 1773848043.95005, "event": "train_step", "step": 310, "epoch": 2, "metrics": {"train/step_loss": 1.0016437012757828, "train/step_real_loss": 0.9807856976985931, "train/lr": 6.2897577033565965e-06, "train/step_canary_loss": 1.4466145833333333, "perf/step_duration_sec": 7.719462159788236, "perf/samples_per_sec": 8.679361154072678, "perf/tokens_per_sec": 6271.6804613921595, "perf/logical_batch_size": 67.0, "perf/logical_token_count": 48414.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773848122.1587205, "event": "train_step", "step": 320, "epoch": 2, "metrics": {"train/step_loss": 0.9575878876906175, "train/step_real_loss": 0.9505775347352028, "train/lr": 4.2737613110951925e-06, "train/step_canary_loss": 1.40625, "perf/step_duration_sec": 7.7270688829012215, "perf/samples_per_sec": 8.411986612910194, "perf/tokens_per_sec": 6291.130665027544, "perf/logical_batch_size": 65.0, "perf/logical_token_count": 48612.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773848200.6989753, "event": "train_step", "step": 330, "epoch": 2, "metrics": {"train/step_loss": 1.0191495253489569, "train/step_real_loss": 0.9818510636687279, "train/lr": 2.6323147493514833e-06, "train/step_canary_loss": 3.40625, "perf/step_duration_sec": 7.586550263920799, "perf/samples_per_sec": 8.567794022155125, "perf/tokens_per_sec": 6317.495875290012, "perf/logical_batch_size": 65.0, "perf/logical_token_count": 47928.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773848281.642083, "event": "train_step", "step": 340, "epoch": 2, "metrics": {"train/step_loss": 1.0784487796552253, "train/step_real_loss": 1.0184003114700317, "train/lr": 1.3788633298137288e-06, "train/step_canary_loss": 3.0, "perf/step_duration_sec": 8.599153365008533, "perf/samples_per_sec": 7.675174194306803, "perf/tokens_per_sec": 5509.263294777043, "perf/logical_batch_size": 66.0, "perf/logical_token_count": 47375.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773848363.9098017, "event": "train_step", "step": 350, "epoch": 2, "metrics": {"train/step_loss": 1.1414880389752595, "train/step_real_loss": 0.950454406440258, "train/lr": 5.236742432724262e-07, "train/step_canary_loss": 3.58671875, "perf/step_duration_sec": 8.310993053019047, "perf/samples_per_sec": 8.302256969753463, "perf/tokens_per_sec": 6413.433347852161, "perf/logical_batch_size": 69.0, "perf/logical_token_count": 53302.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773848384.8834276, "event": "eval_step", "step": 350, "epoch": 2, "metrics": {"eval/loss": 0.9456373546030614, "eval/duration_sec": 20.951530331978574}}
{"timestamp": 1773848463.9385724, "event": "train_step", "step": 360, "epoch": 2, "metrics": {"train/step_loss": 1.0201530621601984, "train/step_real_loss": 0.9811613112688065, "train/lr": 7.375245966623756e-08, "train/step_canary_loss": 3.515625, "perf/step_duration_sec": 7.831112121930346, "perf/samples_per_sec": 8.300225943384616, "perf/tokens_per_sec": 6854.454279830759, "perf/logical_batch_size": 65.0, "perf/logical_token_count": 53678.0, "perf/gradient_accumulation_steps": 8.0, "system/cuda_memory_allocated_gb": 16.85233783721924, "system/cuda_max_memory_allocated_gb": 60.84869050979614}}
{"timestamp": 1773848532.3133028, "event": "train_epoch", "step": 366, "epoch": 2, "metrics": {"train/epoch_loss": 1.0387111434896155, "train/epoch_real_loss": 0.9917015731131118, "train/epoch_canary_loss": 3.26868664640822, "perf/epoch_duration_sec": 1556.666436379077, "perf/epoch_samples_per_sec": 30.74068977250494, "perf/epoch_tokens_per_sec": 23919.13266056493, "perf/epoch_samples": 47853.0, "perf/epoch_tokens": 37234111.0, "system/cuda_epoch_peak_memory_gb": 60.84869050979614, "eval/loss": 0.9456853411801449, "eval/duration_sec": 21.018074851948768}}
{"timestamp": 1773848539.6229737, "event": "audit_epoch", "step": 366, "epoch": 2, "metrics": {"audit/delta": 1e-05, "audit/num_canaries": 500.0, "audit/num_members": 250.0, "audit/paper_guess_fraction": 0.2, "audit/paper_guess_steps": 20.0, "audit/loss/auc": 1.0, "audit/loss/empirical_epsilon/0.05": 3.4791953936219215, "audit/loss/empirical_epsilon/0.01": 3.023197554051876, "audit/loss/empirical_epsilon_details/0.05/epsilon": 3.4791953936219215, "audit/loss/empirical_epsilon_details/0.05/num_guesses": 100.0, "audit/loss/empirical_epsilon_details/0.05/correct_guesses": 100.0, "audit/loss/empirical_epsilon_details/0.01/epsilon": 3.023197554051876, "audit/loss/empirical_epsilon_details/0.01/num_guesses": 100.0, "audit/loss/empirical_epsilon_details/0.01/correct_guesses": 100.0, "audit/embedding/auc": 1.0, "audit/embedding/empirical_epsilon/0.05": 3.4791953936219215, "audit/embedding/empirical_epsilon/0.01": 3.023197554051876, "audit/embedding/empirical_epsilon_details/0.05/epsilon": 3.4791953936219215, "audit/embedding/empirical_epsilon_details/0.05/num_guesses": 100.0, "audit/embedding/empirical_epsilon_details/0.05/correct_guesses": 100.0, "audit/embedding/empirical_epsilon_details/0.01/epsilon": 3.023197554051876, "audit/embedding/empirical_epsilon_details/0.01/num_guesses": 100.0, "audit/embedding/empirical_epsilon_details/0.01/correct_guesses": 100.0, "perf/audit_duration_sec": 5.547323588980362}}
{"timestamp": 1773848547.60095, "event": "audit_final", "step": 366, "epoch": 2, "metrics": {"audit/delta": 1e-05, "audit/num_canaries": 500.0, "audit/num_members": 250.0, "audit/paper_guess_fraction": 0.2, "audit/paper_guess_steps": 20.0, "audit/loss/auc": 0.999984, "audit/loss/empirical_epsilon/0.05": 3.4791953936219215, "audit/loss/empirical_epsilon/0.01": 3.023197554051876, "audit/loss/empirical_epsilon_details/0.05/epsilon": 3.4791953936219215, "audit/loss/empirical_epsilon_details/0.05/num_guesses": 100.0, "audit/loss/empirical_epsilon_details/0.05/correct_guesses": 100.0, "audit/loss/empirical_epsilon_details/0.01/epsilon": 3.023197554051876, "audit/loss/empirical_epsilon_details/0.01/num_guesses": 100.0, "audit/loss/empirical_epsilon_details/0.01/correct_guesses": 100.0, "audit/embedding/auc": 1.0, "audit/embedding/empirical_epsilon/0.05": 3.4791953936219215, "audit/embedding/empirical_epsilon/0.01": 3.023197554051876, "audit/embedding/empirical_epsilon_details/0.05/epsilon": 3.4791953936219215, "audit/embedding/empirical_epsilon_details/0.05/num_guesses": 100.0, "audit/embedding/empirical_epsilon_details/0.05/correct_guesses": 100.0, "audit/embedding/empirical_epsilon_details/0.01/epsilon": 3.023197554051876, "audit/embedding/empirical_epsilon_details/0.01/num_guesses": 100.0, "audit/embedding/empirical_epsilon_details/0.01/correct_guesses": 100.0}}
{"timestamp": 1773848548.142728, "event": "energy_final", "step": 366, "epoch": null, "metrics": {"energy/codecarbon/duration": 3257.8297634990886, "energy/codecarbon/emissions": 0.046521798865538776, "energy/codecarbon/emissions_rate": 1.427999688221026e-05, "energy/codecarbon/cpu_power": 84.78240267963766, "energy/codecarbon/gpu_power": 1356.0935826800528, "energy/codecarbon/ram_power": 38.0, "energy/codecarbon/cpu_energy": 0.0740143041223195, "energy/codecarbon/gpu_energy": 1.2279951440620138, "energy/codecarbon/ram_energy": 0.033174056839308075, "energy/codecarbon/energy_consumed": 1.3351835050236422, "energy/codecarbon/water_consumed": 0.0, "energy/codecarbon/cpu_count": 8.0, "energy/codecarbon/gpu_count": 5.0, "energy/codecarbon/longitude": 8.212, "energy/codecarbon/latitude": 47.4843, "energy/codecarbon/ram_total_size": 128.0, "energy/codecarbon/cpu_utilization_percent": 12.756927339901509, "energy/codecarbon/gpu_utilization_percent": 63.19950738916256, "energy/codecarbon/ram_utilization_percent": 10.355603448275856, "energy/codecarbon/ram_used_gb": 208.74299130651164, "energy/codecarbon/pue": 1.0, "energy/codecarbon/wue": 0.0}}