diff --git a/dk1-teleop/outputs/vla_pickupcube_200_lora/molmoact2_lora_pickupcube_200_20260603_205822/telemetry.jsonl b/dk1-teleop/outputs/vla_pickupcube_200_lora/molmoact2_lora_pickupcube_200_20260603_205822/telemetry.jsonl
new file mode 100644
index 0000000000000000000000000000000000000000..04a775e1e879196f4fa61486f43df4b87327aee6
--- /dev/null
+++ b/dk1-teleop/outputs/vla_pickupcube_200_lora/molmoact2_lora_pickupcube_200_20260603_205822/telemetry.jsonl
@@ -0,0 +1,187 @@
+{"batch_size": 16, "cuda_allocated_gib": 0.0, "cuda_device": 0, "cuda_max_allocated_gib": 0.0, "cuda_max_reserved_gib": 0.0, "cuda_reserved_gib": 0.0, "event": "start", "grad_accum_steps": 2, "job_name": "molmoact2_lora_dk1_pickupcube_200_pickupcube_200_20260603_205822", "num_workers": 8, "output_dir": "/home/andon/clawmachine/dk1-teleop/outputs/vla_pickupcube_200_lora/molmoact2_lora_pickupcube_200_20260603_205822", "steps": 100000, "time": 1780557643.4669285, "val_batches": 32, "val_freq": 0}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 75.87912461199448, "epoch": 0.02242152466367713, "event": "heartbeat", "loss": 0.16273221373558044, "lr": 1.8456375838926177e-06, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08923116577347606, "mfu_sample_tflops_s": 18.693929229543233, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34019381951137756, "mfu_text_token_tflops_s": 71.27060518763359, "mfu_tokens_per_sample_assumed": 128.0, "step": 10, "time": 1780557824.0293305}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 148.19974685297348, "epoch": 0.04484304932735426, "event": "heartbeat", "loss": 0.20182295143604279, "lr": 3.523489932885904e-06, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08912293166069805, "mfu_sample_tflops_s": 18.671254182916243, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3397811769564113, "mfu_text_token_tflops_s": 71.18415657236817, "mfu_tokens_per_sample_assumed": 128.0, "step": 20, "time": 1780557896.3499527}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 220.54937872098526, "epoch": 0.06726457399103139, "event": "heartbeat", "loss": 0.17528212815523148, "lr": 5.201342281879196e-06, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08905096562766289, "mfu_sample_tflops_s": 18.656177298995374, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33950680645546477, "mfu_text_token_tflops_s": 71.12667595241987, "mfu_tokens_per_sample_assumed": 128.0, "step": 30, "time": 1780557968.6995845}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 292.90159650999703, "epoch": 0.08968609865470852, "event": "heartbeat", "loss": 0.13821755349636078, "lr": 6.879194630872485e-06, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08911775278499157, "mfu_sample_tflops_s": 18.670169208455736, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3397614324927804, "mfu_text_token_tflops_s": 71.1800201072375, "mfu_tokens_per_sample_assumed": 128.0, "step": 40, "time": 1780558041.0518024}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 365.3448000259814, "epoch": 0.11210762331838565, "event": "heartbeat", "loss": 0.12421151623129845, "lr": 8.55704697986577e-06, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08896838554001552, "mfu_sample_tflops_s": 18.63887677063325, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3391919698713091, "mfu_text_token_tflops_s": 71.06071768803926, "mfu_tokens_per_sample_assumed": 128.0, "step": 50, "time": 1780558113.4950054}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 437.79761897897697, "epoch": 0.13452914798206278, "event": "heartbeat", "loss": 0.10932466015219688, "lr": 1.0234899328859063e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08818484810433322, "mfu_sample_tflops_s": 18.47472567785781, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33620473339777035, "mfu_text_token_tflops_s": 70.4348916468329, "mfu_tokens_per_sample_assumed": 128.0, "step": 60, "time": 1780558185.947825}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 510.219397184992, "epoch": 0.15695067264573992, "event": "heartbeat", "loss": 0.11781858280301094, "lr": 1.1912751677852351e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08826956227017284, "mfu_sample_tflops_s": 18.49247329560121, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3365277061550339, "mfu_text_token_tflops_s": 70.50255443947961, "mfu_tokens_per_sample_assumed": 128.0, "step": 70, "time": 1780558258.3696032}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 582.6088502539787, "epoch": 0.17937219730941703, "event": "heartbeat", "loss": 0.10611231252551079, "lr": 1.3590604026845638e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08892433944323089, "mfu_sample_tflops_s": 18.62964911335687, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33902404412731774, "mfu_text_token_tflops_s": 71.02553724467306, "mfu_tokens_per_sample_assumed": 128.0, "step": 80, "time": 1780558330.759056}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 655.0527288599988, "epoch": 0.20179372197309417, "event": "heartbeat", "loss": 0.09352846071124077, "lr": 1.5268456375838928e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08893779058802737, "mfu_sample_tflops_s": 18.632467128191735, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33907532661685436, "mfu_text_token_tflops_s": 71.03628092623099, "mfu_tokens_per_sample_assumed": 128.0, "step": 90, "time": 1780558403.2029347}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 727.5073104319745, "epoch": 0.2242152466367713, "event": "heartbeat", "loss": 0.07884152978658676, "lr": 1.6946308724832215e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08897479607573647, "mfu_sample_tflops_s": 18.64021977786679, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3392164100387453, "mfu_text_token_tflops_s": 71.06583790311714, "mfu_tokens_per_sample_assumed": 128.0, "step": 100, "time": 1780558475.6575162}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 799.9862646809779, "epoch": 0.24663677130044842, "event": "heartbeat", "loss": 0.08228205889463425, "lr": 1.86241610738255e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08879234553825241, "mfu_sample_tflops_s": 18.601996390263878, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33852081736458733, "mfu_text_token_tflops_s": 70.92011123788104, "mfu_tokens_per_sample_assumed": 128.0, "step": 110, "time": 1780558548.1364706}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 872.4708758249762, "epoch": 0.26905829596412556, "event": "heartbeat", "loss": 0.07274128496646881, "lr": 2.0302013422818795e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08890926667289549, "mfu_sample_tflops_s": 18.626491367971607, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33896657919041406, "mfu_text_token_tflops_s": 71.01349834039175, "mfu_tokens_per_sample_assumed": 128.0, "step": 120, "time": 1780558620.6210818}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 944.9448035409732, "epoch": 0.2914798206278027, "event": "heartbeat", "loss": 0.07454699650406837, "lr": 2.197986577181208e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08818118658829245, "mfu_sample_tflops_s": 18.47395859024727, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.336190773867865, "mfu_text_token_tflops_s": 70.43196712531771, "mfu_tokens_per_sample_assumed": 128.0, "step": 130, "time": 1780558693.0950096}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 1017.3884529059869, "epoch": 0.31390134529147984, "event": "heartbeat", "loss": 0.052559759467840195, "lr": 2.365771812080537e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0881983723518858, "mfu_sample_tflops_s": 18.477559007720075, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3362562945915646, "mfu_text_token_tflops_s": 70.44569371693278, "mfu_tokens_per_sample_assumed": 128.0, "step": 140, "time": 1780558765.5386589}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 1089.869263872999, "epoch": 0.336322869955157, "event": "heartbeat", "loss": 0.06125841103494167, "lr": 2.4937261951818945e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08822172925022578, "mfu_sample_tflops_s": 18.482452277922302, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33634534276648576, "mfu_text_token_tflops_s": 70.46434930957876, "mfu_tokens_per_sample_assumed": 128.0, "step": 150, "time": 1780558838.0194697}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 1162.4605138009938, "epoch": 0.35874439461883406, "event": "heartbeat", "loss": 0.07511032745242119, "lr": 2.492862719372501e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08826709478957075, "mfu_sample_tflops_s": 18.49195635841507, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33651829888523843, "mfu_text_token_tflops_s": 70.50058361645745, "mfu_tokens_per_sample_assumed": 128.0, "step": 160, "time": 1780558910.6107194}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 1234.955599326, "epoch": 0.3811659192825112, "event": "heartbeat", "loss": 0.04980505257844925, "lr": 2.491943778858424e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0888366148474545, "mfu_sample_tflops_s": 18.61127081054172, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33868959410592037, "mfu_text_token_tflops_s": 70.95546996519032, "mfu_tokens_per_sample_assumed": 128.0, "step": 170, "time": 1780558983.1058054}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 1307.448330130981, "epoch": 0.40358744394618834, "event": "heartbeat", "loss": 0.060537759214639664, "lr": 2.4909694192344957e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08884659338898775, "mfu_sample_tflops_s": 18.613361314992932, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3387276372955158, "mfu_text_token_tflops_s": 70.96344001341056, "mfu_tokens_per_sample_assumed": 128.0, "step": 180, "time": 1780559055.598536}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 1379.9469464809808, "epoch": 0.4260089686098655, "event": "heartbeat", "loss": 0.06373971700668335, "lr": 2.4899396888452655e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08887968760283094, "mfu_sample_tflops_s": 18.620294552793084, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.338853808985793, "mfu_text_token_tflops_s": 70.98987298252362, "mfu_tokens_per_sample_assumed": 128.0, "step": 190, "time": 1780559128.0971522}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 1452.440961521992, "epoch": 0.4484304932735426, "event": "heartbeat", "loss": 0.06044390797615051, "lr": 2.4888546387825988e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08881695411702112, "mfu_sample_tflops_s": 18.607151887515926, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3386146375711431, "mfu_text_token_tflops_s": 70.93976657115446, "mfu_tokens_per_sample_assumed": 128.0, "step": 200, "time": 1780559200.5911674}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 1524.9405840419931, "epoch": 0.47085201793721976, "event": "heartbeat", "loss": 0.05409246310591698, "lr": 2.487714322883144e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08880346419342183, "mfu_sample_tflops_s": 18.604325748521877, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3385632072374208, "mfu_text_token_tflops_s": 70.92899191623965, "mfu_tokens_per_sample_assumed": 128.0, "step": 210, "time": 1780559273.0907898}
+{"cuda_allocated_gib": 12.896815299987793, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "elapsed_s": 1597.454962799995, "epoch": 0.49327354260089684, "event": "heartbeat", "loss": 0.07631662487983704, "lr": 2.4865187977256584e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08892656902334534, "mfu_sample_tflops_s": 18.63011621039085, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3390325444015041, "mfu_text_token_tflops_s": 71.0273180521151, "mfu_tokens_per_sample_assumed": 128.0, "step": 220, "time": 1780559345.6051686}
+{"best_val_loss": 0.06796903302893043, "checkpoint_dir": "/home/andon/clawmachine/dk1-teleop/outputs/vla_pickupcube_200_lora/molmoact2_lora_pickupcube_200_20260603_205822/checkpoints/best", "cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.299659252166748, "cuda_max_reserved_gib": 26.556640625, "cuda_reserved_gib": 26.556640625, "event": "best_checkpoint", "step": 223}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 1702.490436921973, "epoch": 0.515695067264574, "event": "heartbeat", "loss": 0.06531723216176033, "lr": 2.485268122628204e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08825647950736967, "mfu_sample_tflops_s": 18.489732456793945, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3364778281218469, "mfu_text_token_tflops_s": 70.49210499152692, "mfu_tokens_per_sample_assumed": 128.0, "step": 230, "time": 1780559450.6406422}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 1774.983699637989, "epoch": 0.5381165919282511, "event": "heartbeat", "loss": 0.060765430331230164, "lr": 2.4839623596452034e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08824619628587652, "mfu_sample_tflops_s": 18.48757812189113, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33643862333990415, "mfu_text_token_tflops_s": 70.48389158970993, "mfu_tokens_per_sample_assumed": 128.0, "step": 240, "time": 1780559523.1339056}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 1847.5253467749862, "epoch": 0.5605381165919282, "event": "heartbeat", "loss": 0.048268526792526245, "lr": 2.482601573564359e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08809277289651656, "mfu_sample_tflops_s": 18.455435921820218, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3358536966679693, "mfu_text_token_tflops_s": 70.36134945193957, "mfu_tokens_per_sample_assumed": 128.0, "step": 250, "time": 1780559595.6755528}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 1920.0289345049823, "epoch": 0.5829596412556054, "event": "heartbeat", "loss": 0.06180444918572903, "lr": 2.4811858319034415e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0888652840169499, "mfu_sample_tflops_s": 18.617277001551003, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3387988953146215, "mfu_text_token_tflops_s": 70.9783685684132, "mfu_tokens_per_sample_assumed": 128.0, "step": 260, "time": 1780559668.1791403}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 1992.5212754949753, "epoch": 0.6053811659192825, "event": "heartbeat", "loss": 0.06007571332156658, "lr": 2.4797152049069377e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08886748355843517, "mfu_sample_tflops_s": 18.61773780549217, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33880728106653407, "mfu_text_token_tflops_s": 70.98012538343889, "mfu_tokens_per_sample_assumed": 128.0, "step": 270, "time": 1780559740.6714814}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2065.0067132619733, "epoch": 0.6278026905829597, "event": "heartbeat", "loss": 0.05703458935022354, "lr": 2.478189765542566e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0888619326510303, "mfu_sample_tflops_s": 18.616574890390847, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.338786118232053, "mfu_text_token_tflops_s": 70.97569176961511, "mfu_tokens_per_sample_assumed": 128.0, "step": 280, "time": 1780559813.1569192}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2137.4861969019985, "epoch": 0.6502242152466368, "event": "heartbeat", "loss": 0.04821855016052723, "lr": 2.476609589497656e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.088867967356171, "mfu_sample_tflops_s": 18.617839161117825, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33880912554540193, "mfu_text_token_tflops_s": 70.98051180176171, "mfu_tokens_per_sample_assumed": 128.0, "step": 290, "time": 1780559885.6364026}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2210.039365022996, "epoch": 0.672645739910314, "event": "heartbeat", "loss": 0.06311702355742455, "lr": 2.474974755175392e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08888947023933164, "mfu_sample_tflops_s": 18.622344015139976, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3388911052874518, "mfu_text_token_tflops_s": 70.99768655772115, "mfu_tokens_per_sample_assumed": 128.0, "step": 300, "time": 1780559958.1895711}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2282.5918388449936, "epoch": 0.695067264573991, "event": "heartbeat", "loss": 0.05991560034453869, "lr": 2.4732853436909257e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08878605585303857, "mfu_sample_tflops_s": 18.60067870121158, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3384968379397096, "mfu_text_token_tflops_s": 70.91508754836916, "mfu_tokens_per_sample_assumed": 128.0, "step": 310, "time": 1780560030.7420447}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2355.175637900975, "epoch": 0.7174887892376681, "event": "heartbeat", "loss": 0.05349511653184891, "lr": 2.4715414388673477e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08794084547052912, "mfu_sample_tflops_s": 18.42360712607585, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33527447335639227, "mfu_text_token_tflops_s": 70.24000216816418, "mfu_tokens_per_sample_assumed": 128.0, "step": 320, "time": 1780560103.3258438}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2427.744242113986, "epoch": 0.7399103139013453, "event": "heartbeat", "loss": 0.062456075102090836, "lr": 2.4697431272315322e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08799880904529885, "mfu_sample_tflops_s": 18.43575049499011, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3354954594852018, "mfu_text_token_tflops_s": 70.28629876214978, "mfu_tokens_per_sample_assumed": 128.0, "step": 330, "time": 1780560175.894448}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2500.25496400299, "epoch": 0.7623318385650224, "event": "heartbeat", "loss": 0.06566721946001053, "lr": 2.467890498009841e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08886142382263253, "mfu_sample_tflops_s": 18.616468290841517, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33878417832378654, "mfu_text_token_tflops_s": 70.97528535883328, "mfu_tokens_per_sample_assumed": 128.0, "step": 340, "time": 1780560248.40517}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2572.8255053279863, "epoch": 0.7847533632286996, "event": "heartbeat", "loss": 0.06766223534941673, "lr": 2.4659836431236975e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08889098588012062, "mfu_sample_tflops_s": 18.622661541885268, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3388968836679599, "mfu_text_token_tflops_s": 70.99889712843759, "mfu_tokens_per_sample_assumed": 128.0, "step": 350, "time": 1780560320.9757113}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2645.4017023519846, "epoch": 0.8071748878923767, "event": "heartbeat", "loss": 0.05323897860944271, "lr": 2.4640226571850265e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08874774844053296, "mfu_sample_tflops_s": 18.592653298291655, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3383507909295319, "mfu_text_token_tflops_s": 70.88449069973694, "mfu_tokens_per_sample_assumed": 128.0, "step": 360, "time": 1780560393.5519083}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2718.0005548089975, "epoch": 0.8295964125560538, "event": "heartbeat", "loss": 0.05836499109864235, "lr": 2.462007637491559e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08881784201324795, "mfu_sample_tflops_s": 18.607337901775445, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33861802267550783, "mfu_text_token_tflops_s": 70.94047575051889, "mfu_tokens_per_sample_assumed": 128.0, "step": 370, "time": 1780560466.1507607}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2790.672373543988, "epoch": 0.852017937219731, "event": "heartbeat", "loss": 0.053474023938179016, "lr": 2.459938684022004e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08879315424787579, "mfu_sample_tflops_s": 18.602165814929975, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3385239005700264, "mfu_text_token_tflops_s": 70.92075716942053, "mfu_tokens_per_sample_assumed": 128.0, "step": 380, "time": 1780560538.8225791}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2863.290342419001, "epoch": 0.874439461883408, "event": "heartbeat", "loss": 0.05181376449763775, "lr": 2.4578158994310902e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08797145913856252, "mfu_sample_tflops_s": 18.43002068952885, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3353911879657696, "mfu_text_token_tflops_s": 70.26445387882873, "mfu_tokens_per_sample_assumed": 128.0, "step": 390, "time": 1780560611.4405484}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 2935.888936866977, "epoch": 0.8968609865470852, "event": "heartbeat", "loss": 0.054231710731983185, "lr": 2.4556393890444692e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08808134414963266, "mfu_sample_tflops_s": 18.453041599348044, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33581012457047454, "mfu_text_token_tflops_s": 70.35222109751442, "mfu_tokens_per_sample_assumed": 128.0, "step": 400, "time": 1780560684.0391428}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3008.424638465978, "epoch": 0.9192825112107623, "event": "heartbeat", "loss": 0.04886617884039879, "lr": 2.4534092608534935e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08880670373412806, "mfu_sample_tflops_s": 18.60500443229983, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33857555798636324, "mfu_text_token_tflops_s": 70.9315793981431, "mfu_tokens_per_sample_assumed": 128.0, "step": 410, "time": 1780560756.5748444}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3080.993321696995, "epoch": 0.9417040358744395, "event": "heartbeat", "loss": 0.047981562092900276, "lr": 2.4511256255098548e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08881093078002744, "mfu_sample_tflops_s": 18.60588999841575, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3385916735988546, "mfu_text_token_tflops_s": 70.93495561896005, "mfu_tokens_per_sample_assumed": 128.0, "step": 420, "time": 1780560829.1435277}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3153.6336244459962, "epoch": 0.9641255605381166, "event": "heartbeat", "loss": 0.045002661645412445, "lr": 2.4487885963200953e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08907811546379633, "mfu_sample_tflops_s": 18.66186518966533, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33961031520572355, "mfu_text_token_tflops_s": 71.14836103559908, "mfu_tokens_per_sample_assumed": 128.0, "step": 430, "time": 1780560901.7838302}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3225.6921656039776, "epoch": 0.9865470852017937, "event": "heartbeat", "loss": 0.03951455466449261, "lr": 2.4463982892399868e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08948744393187216, "mfu_sample_tflops_s": 18.74761950372722, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34117087999026263, "mfu_text_token_tflops_s": 71.47529935796001, "mfu_tokens_per_sample_assumed": 128.0, "step": 440, "time": 1780560973.8423717}
+{"best_val_loss": 0.06323498976416886, "checkpoint_dir": "/home/andon/clawmachine/dk1-teleop/outputs/vla_pickupcube_200_lora/molmoact2_lora_pickupcube_200_20260603_205822/checkpoints/best", "cuda_allocated_gib": 12.885714530944824, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "event": "best_checkpoint", "step": 446}
+{"best_step": 446, "best_val_loss": 0.06323498976416886, "checkpoint_dir": "/home/andon/clawmachine/dk1-teleop/outputs/vla_pickupcube_200_lora/molmoact2_lora_pickupcube_200_20260603_205822/checkpoints/000446", "cuda_allocated_gib": 12.885714530944824, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "epoch": 1.0, "event": "checkpoint", "step": 446}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3329.987460059987, "epoch": 1.0089686098654709, "event": "heartbeat", "loss": 0.041508451104164124, "lr": 2.4439548228687766e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08962248164796587, "mfu_sample_tflops_s": 18.77590990524885, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34168571128286984, "mfu_text_token_tflops_s": 71.58315651376124, "mfu_tokens_per_sample_assumed": 128.0, "step": 450, "time": 1780561078.1376657}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3401.958114508976, "epoch": 1.031390134529148, "event": "heartbeat", "loss": 0.05392448976635933, "lr": 2.4414583184433003e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0895646263592734, "mfu_sample_tflops_s": 18.76378922226778, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3414651379947298, "mfu_text_token_tflops_s": 71.53694640989589, "mfu_tokens_per_sample_assumed": 128.0, "step": 460, "time": 1780561150.1083205}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3474.0104559939937, "epoch": 1.053811659192825, "event": "heartbeat", "loss": 0.04331132769584656, "lr": 2.4389088998319724e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08966415860463806, "mfu_sample_tflops_s": 18.78464122767167, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3418446046801826, "mfu_text_token_tflops_s": 71.61644468049825, "mfu_tokens_per_sample_assumed": 128.0, "step": 470, "time": 1780561222.1606615}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3546.0195067789755, "epoch": 1.0762331838565022, "event": "heartbeat", "loss": 0.06076354905962944, "lr": 2.4363066935286344e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08869634635850983, "mfu_sample_tflops_s": 18.58188456210781, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33815482049181866, "mfu_text_token_tflops_s": 70.84343489303602, "mfu_tokens_per_sample_assumed": 128.0, "step": 480, "time": 1780561294.1697125}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3618.4305298119725, "epoch": 1.0986547085201794, "event": "heartbeat", "loss": 0.043325064703822136, "lr": 2.4336518286462808e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08789862122297405, "mfu_sample_tflops_s": 18.414761146213063, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3351134934125885, "mfu_text_token_tflops_s": 70.2062768699373, "mfu_tokens_per_sample_assumed": 128.0, "step": 490, "time": 1780561366.5807354}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3690.9269709409855, "epoch": 1.1210762331838564, "event": "heartbeat", "loss": 0.04628417082130909, "lr": 2.430944436910656e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08901504249804355, "mfu_sample_tflops_s": 18.648651403340125, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33936984952379107, "mfu_text_token_tflops_s": 71.09798347523423, "mfu_tokens_per_sample_assumed": 128.0, "step": 500, "time": 1780561439.0771768}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3763.504651282972, "epoch": 1.1434977578475336, "event": "heartbeat", "loss": 0.04163338243961334, "lr": 2.4281846526537118e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0887836008721345, "mfu_sample_tflops_s": 18.600164382712176, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33848747832501275, "mfu_text_token_tflops_s": 70.91312670909018, "mfu_tokens_per_sample_assumed": 128.0, "step": 510, "time": 1780561511.654857}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3836.0943090459914, "epoch": 1.1659192825112108, "event": "heartbeat", "loss": 0.04538264125585556, "lr": 2.425372612806949e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0888284882486913, "mfu_sample_tflops_s": 18.609568288100828, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3386586114481356, "mfu_text_token_tflops_s": 70.94897909838441, "mfu_tokens_per_sample_assumed": 128.0, "step": 520, "time": 1780561584.2445152}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3908.685649209976, "epoch": 1.188340807174888, "event": "heartbeat", "loss": 0.05257779732346535, "lr": 2.4225084568946195e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08879534141327292, "mfu_sample_tflops_s": 18.602624026080676, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.338532239138103, "mfu_text_token_tflops_s": 70.92250409943257, "mfu_tokens_per_sample_assumed": 128.0, "step": 530, "time": 1780561656.835855}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 3981.257533081982, "epoch": 1.210762331838565, "event": "heartbeat", "loss": 0.059687547385692596, "lr": 2.4195923270268043e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0888188124964959, "mfu_sample_tflops_s": 18.60754121801589, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3386217226428906, "mfu_text_token_tflops_s": 70.94125089368558, "mfu_tokens_per_sample_assumed": 128.0, "step": 540, "time": 1780561729.407739}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4053.8953889569966, "epoch": 1.2331838565022422, "event": "heartbeat", "loss": 0.052585216239094734, "lr": 2.4166243678923633e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08791428356039482, "mfu_sample_tflops_s": 18.418042405902714, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33517320607400525, "mfu_text_token_tflops_s": 70.2187866725041, "mfu_tokens_per_sample_assumed": 128.0, "step": 550, "time": 1780561802.045595}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4126.469486963993, "epoch": 1.2556053811659194, "event": "heartbeat", "loss": 0.052396468818187714, "lr": 2.4136047267517548e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08801093196702715, "mfu_sample_tflops_s": 18.438290247092187, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.335541678124291, "mfu_text_token_tflops_s": 70.29598156703896, "mfu_tokens_per_sample_assumed": 128.0, "step": 560, "time": 1780561874.6196928}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4199.06877292, "epoch": 1.2780269058295963, "event": "heartbeat", "loss": 0.050728848204016685, "lr": 2.41053355342973e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08801119079447839, "mfu_sample_tflops_s": 18.438344471443223, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33554266490394885, "mfu_text_token_tflops_s": 70.29618829737728, "mfu_tokens_per_sample_assumed": 128.0, "step": 570, "time": 1780561947.218979}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4271.735841503978, "epoch": 1.3004484304932735, "event": "heartbeat", "loss": 0.052281010895967484, "lr": 2.4074110003079e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08793672325414376, "mfu_sample_tflops_s": 18.42274352174312, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3352587574064231, "mfu_text_token_tflops_s": 70.23670967664563, "mfu_tokens_per_sample_assumed": 128.0, "step": 580, "time": 1780562019.8860476}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4344.301724625984, "epoch": 1.3228699551569507, "event": "heartbeat", "loss": 0.0495888814330101, "lr": 2.4042372223171734e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08882101194732268, "mfu_sample_tflops_s": 18.608002002964103, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3386301080491677, "mfu_text_token_tflops_s": 70.94300763630063, "mfu_tokens_per_sample_assumed": 128.0, "step": 590, "time": 1780562092.4519305}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4416.921260749979, "epoch": 1.345291479820628, "event": "heartbeat", "loss": 0.039218440651893616, "lr": 2.401012376930069e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08881593720168504, "mfu_sample_tflops_s": 18.606938843753017, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33861076058142414, "mfu_text_token_tflops_s": 70.93895434180835, "mfu_tokens_per_sample_assumed": 128.0, "step": 600, "time": 1780562165.0714667}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4489.599471269001, "epoch": 1.3677130044843049, "event": "heartbeat", "loss": 0.03964243270456791, "lr": 2.3977366241529052e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08880844278032123, "mfu_sample_tflops_s": 18.605368762477298, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33858218809997465, "mfu_text_token_tflops_s": 70.93296840694468, "mfu_tokens_per_sample_assumed": 128.0, "step": 610, "time": 1780562237.749677}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4562.279338597, "epoch": 1.390134529147982, "event": "heartbeat", "loss": 0.04266204684972763, "lr": 2.394410126517857e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08878419407340363, "mfu_sample_tflops_s": 18.60028865837806, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33848973990485137, "mfu_text_token_tflops_s": 70.91360051006636, "mfu_tokens_per_sample_assumed": 128.0, "step": 620, "time": 1780562310.429544}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4634.911294585996, "epoch": 1.4125560538116593, "event": "heartbeat", "loss": 0.04109710454940796, "lr": 2.3910330490748957e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08873720051165666, "mfu_sample_tflops_s": 18.59044350719207, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.338310576950691, "mfu_text_token_tflops_s": 70.87606587116977, "mfu_tokens_per_sample_assumed": 128.0, "step": 630, "time": 1780562383.0615003}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4707.538536574983, "epoch": 1.4349775784753362, "event": "heartbeat", "loss": 0.05941282957792282, "lr": 2.387605559383597e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0887790804592268, "mfu_sample_tflops_s": 18.599217356208015, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3384702442508022, "mfu_text_token_tflops_s": 70.90951617054306, "mfu_tokens_per_sample_assumed": 128.0, "step": 640, "time": 1780562455.6887424}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4780.195060043974, "epoch": 1.4573991031390134, "event": "heartbeat", "loss": 0.05350662022829056, "lr": 2.3841278275048286e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08874092453163875, "mfu_sample_tflops_s": 18.591223689378317, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33832477477687267, "mfu_text_token_tflops_s": 70.87904031575482, "mfu_tokens_per_sample_assumed": 128.0, "step": 650, "time": 1780562528.3452659}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4852.802307373, "epoch": 1.4798206278026906, "event": "heartbeat", "loss": 0.04696689546108246, "lr": 2.3806000259923104e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08787517089541336, "mfu_sample_tflops_s": 18.409848302589097, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33502408903876346, "mfu_text_token_tflops_s": 70.18754665362094, "mfu_tokens_per_sample_assumed": 128.0, "step": 660, "time": 1780562600.9525132}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 4956.075024190999, "epoch": 1.5022421524663676, "event": "heartbeat", "loss": 0.04908699542284012, "lr": 2.377022329884056e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08785714752713968, "mfu_sample_tflops_s": 18.406072406935763, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33495537494722, "mfu_text_token_tflops_s": 70.17315105144259, "mfu_tokens_per_sample_assumed": 128.0, "step": 670, "time": 1780562704.22523}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5028.702355780988, "epoch": 1.5246636771300448, "event": "heartbeat", "loss": 0.04156528599560261, "lr": 2.373394916693685e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08789957446251505, "mfu_sample_tflops_s": 18.414960849896904, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3351171276383387, "mfu_text_token_tflops_s": 70.20703824023195, "mfu_tokens_per_sample_assumed": 128.0, "step": 680, "time": 1780562776.8525617}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5101.2794548409875, "epoch": 1.547085201793722, "event": "heartbeat", "loss": 0.05380230024456978, "lr": 2.3697179664016175e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08866561792879628, "mfu_sample_tflops_s": 18.57544695608282, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3380376683535358, "mfu_text_token_tflops_s": 70.81889152006575, "mfu_tokens_per_sample_assumed": 128.0, "step": 690, "time": 1780562849.4296606}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5173.923063347, "epoch": 1.5695067264573992, "event": "heartbeat", "loss": 0.05241483449935913, "lr": 2.3659916614461428e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08888082479391485, "mfu_sample_tflops_s": 18.62053279432516, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33885814452680035, "mfu_text_token_tflops_s": 70.99078127836466, "mfu_tokens_per_sample_assumed": 128.0, "step": 700, "time": 1780562922.0732691}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5246.588991175988, "epoch": 1.5919282511210762, "event": "heartbeat", "loss": 0.043882379308342934, "lr": 2.362216186714367e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08877864194201945, "mfu_sample_tflops_s": 18.599125486853076, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3384685724039491, "mfu_text_token_tflops_s": 70.90916591862734, "mfu_tokens_per_sample_assumed": 128.0, "step": 710, "time": 1780562994.739197}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5319.265775041975, "epoch": 1.6143497757847534, "event": "heartbeat", "loss": 0.03695201873779297, "lr": 2.3583917295330417e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08864781665262589, "mfu_sample_tflops_s": 18.571717588725125, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3379698009881363, "mfu_text_token_tflops_s": 70.80467330701455, "mfu_tokens_per_sample_assumed": 128.0, "step": 720, "time": 1780563067.415981}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5391.906740072998, "epoch": 1.6367713004484306, "event": "heartbeat", "loss": 0.04480621591210365, "lr": 2.354518479659266e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0887656863217331, "mfu_sample_tflops_s": 18.596411284403082, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3384191791016074, "mfu_text_token_tflops_s": 70.89881802178675, "mfu_tokens_per_sample_assumed": 128.0, "step": 730, "time": 1780563140.056946}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5464.562145423988, "epoch": 1.6591928251121075, "event": "heartbeat", "loss": 0.041824402287602425, "lr": 2.3505966292710756e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08781451071717294, "mfu_sample_tflops_s": 18.39713999524773, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33479282210922184, "mfu_text_token_tflops_s": 70.13909623188198, "mfu_tokens_per_sample_assumed": 128.0, "step": 740, "time": 1780563212.7123513}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5537.188496143994, "epoch": 1.6816143497757847, "event": "heartbeat", "loss": 0.02920028381049633, "lr": 2.3466263729579044e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08783542618096407, "mfu_sample_tflops_s": 18.401521784911974, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3348725623149255, "mfu_text_token_tflops_s": 70.15580180497689, "mfu_tokens_per_sample_assumed": 128.0, "step": 750, "time": 1780563285.338702}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5609.904558133974, "epoch": 1.704035874439462, "event": "heartbeat", "loss": 0.03902561031281948, "lr": 2.3426079077109297e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08765889233192983, "mfu_sample_tflops_s": 18.3645379435393, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3341995270154825, "mfu_text_token_tflops_s": 70.01480090974358, "mfu_tokens_per_sample_assumed": 128.0, "step": 760, "time": 1780563358.054764}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5682.515464719996, "epoch": 1.7264573991031391, "event": "heartbeat", "loss": 0.03841349482536316, "lr": 2.3385414329133e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08880809200614874, "mfu_sample_tflops_s": 18.60529527528816, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3385808507734421, "mfu_text_token_tflops_s": 70.9326882370361, "mfu_tokens_per_sample_assumed": 128.0, "step": 770, "time": 1780563430.6656706}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5755.161268854979, "epoch": 1.748878923766816, "event": "heartbeat", "loss": 0.0506245419383049, "lr": 2.334427150330241e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08875472312685943, "mfu_sample_tflops_s": 18.59411449507705, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3383773819211516, "mfu_text_token_tflops_s": 70.89006151248125, "mfu_tokens_per_sample_assumed": 128.0, "step": 780, "time": 1780563503.3114748}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5827.791562967992, "epoch": 1.7713004484304933, "event": "heartbeat", "loss": 0.03960032761096954, "lr": 2.3302652640990443e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08872383584545036, "mfu_sample_tflops_s": 18.58764360962185, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33825962416077954, "mfu_text_token_tflops_s": 70.86539126168331, "mfu_tokens_per_sample_assumed": 128.0, "step": 790, "time": 1780563575.941769}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5900.432488333987, "epoch": 1.7937219730941705, "event": "heartbeat", "loss": 0.03217124566435814, "lr": 2.3260559807189402e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08882234217097272, "mfu_sample_tflops_s": 18.608280684818784, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3386351795268335, "mfu_text_token_tflops_s": 70.9440701108716, "mfu_tokens_per_sample_assumed": 128.0, "step": 800, "time": 1780563648.5826936}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 5973.098674898996, "epoch": 1.8161434977578474, "event": "heartbeat", "loss": 0.04570855759084225, "lr": 2.321799509040851e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0886956288861066, "mfu_sample_tflops_s": 18.581734251639332, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3381520851282814, "mfu_text_token_tflops_s": 70.84286183437496, "mfu_tokens_per_sample_assumed": 128.0, "step": 810, "time": 1780563721.2488809}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6045.764787437976, "epoch": 1.8385650224215246, "event": "heartbeat", "loss": 0.044831911101937294, "lr": 2.3174960602570275e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08878725204167079, "mfu_sample_tflops_s": 18.60092930273003, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33850139840886995, "mfu_text_token_tflops_s": 70.91604296665825, "mfu_tokens_per_sample_assumed": 128.0, "step": 820, "time": 1780563793.9149935}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6118.412295644986, "epoch": 1.8609865470852018, "event": "heartbeat", "loss": 0.04300197958946228, "lr": 2.3131458478905725e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08783953902575023, "mfu_sample_tflops_s": 18.402383425894673, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3348882425356727, "mfu_text_token_tflops_s": 70.15908681122343, "mfu_tokens_per_sample_assumed": 128.0, "step": 830, "time": 1780563866.5625017}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6191.047590923001, "epoch": 1.883408071748879, "event": "heartbeat", "loss": 0.04124167189002037, "lr": 2.308749087784845e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08796439282499635, "mfu_sample_tflops_s": 18.428540296836733, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33536424764529854, "mfu_text_token_tflops_s": 70.25880988169004, "mfu_tokens_per_sample_assumed": 128.0, "step": 840, "time": 1780563939.1977968}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6263.611315944989, "epoch": 1.905829596412556, "event": "heartbeat", "loss": 0.044270774349570274, "lr": 2.304305998092752e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08880360835737902, "mfu_sample_tflops_s": 18.604355950870907, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33856375686250756, "mfu_text_token_tflops_s": 70.92910706269532, "mfu_tokens_per_sample_assumed": 128.0, "step": 850, "time": 1780564011.7615216}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6336.260283166979, "epoch": 1.9282511210762332, "event": "heartbeat", "loss": 0.04110747668892145, "lr": 2.2998167992659208e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08874543283894444, "mfu_sample_tflops_s": 18.592168179758858, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33834196269847566, "mfu_text_token_tflops_s": 70.88264118533066, "mfu_tokens_per_sample_assumed": 128.0, "step": 860, "time": 1780564084.4104888}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6408.875551857986, "epoch": 1.9506726457399104, "event": "heartbeat", "loss": 0.0518640149384737, "lr": 2.2952817140437664e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08888143958588088, "mfu_sample_tflops_s": 18.620661593242044, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3388604884211709, "mfu_text_token_tflops_s": 70.9912723242353, "mfu_tokens_per_sample_assumed": 128.0, "step": 870, "time": 1780564157.0257578}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6481.298214328999, "epoch": 1.9730941704035874, "event": "heartbeat", "loss": 0.05363352410495281, "lr": 2.2907009674424367e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08934875964751313, "mfu_sample_tflops_s": 18.718565146154, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3406421461561438, "mfu_text_token_tflops_s": 71.36452961971213, "mfu_tokens_per_sample_assumed": 128.0, "step": 880, "time": 1780564229.4484203}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6553.554822618986, "epoch": 1.9955156950672646, "event": "heartbeat", "loss": 0.037669871002435684, "lr": 2.286074786743648e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08930287222202878, "mfu_sample_tflops_s": 18.70895173051503, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3404672003464847, "mfu_text_token_tflops_s": 71.32787847258855, "mfu_tokens_per_sample_assumed": 128.0, "step": 890, "time": 1780564301.7050285}
+{"best_step": 446, "best_val_loss": 0.06323498976416886, "checkpoint_dir": "/home/andon/clawmachine/dk1-teleop/outputs/vla_pickupcube_200_lora/molmoact2_lora_pickupcube_200_20260603_205822/checkpoints/000892", "cuda_allocated_gib": 12.885714530944824, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "epoch": 2.0, "event": "checkpoint", "step": 892}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6657.638691133994, "epoch": 2.0179372197309418, "event": "heartbeat", "loss": 0.047734448686242104, "lr": 2.281403401483408e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0887197988399962, "mfu_sample_tflops_s": 18.586797856979203, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3382442330774855, "mfu_text_token_tflops_s": 70.86216682973321, "mfu_tokens_per_sample_assumed": 128.0, "step": 900, "time": 1780564405.788897}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6730.359133513994, "epoch": 2.0403587443946187, "event": "heartbeat", "loss": 0.03568059392273426, "lr": 2.2766870434406277e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0887343236018926, "mfu_sample_tflops_s": 18.5898407945965, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33829960873221554, "mfu_text_token_tflops_s": 70.87376802939916, "mfu_tokens_per_sample_assumed": 128.0, "step": 910, "time": 1780564478.5093393}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6803.039009430999, "epoch": 2.062780269058296, "event": "heartbeat", "loss": 0.04543314874172211, "lr": 2.2719259466256217e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08786544482049229, "mfu_sample_tflops_s": 18.407810689893132, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33498700837812684, "mfu_text_token_tflops_s": 70.17977825521757, "mfu_tokens_per_sample_assumed": 128.0, "step": 920, "time": 1780564551.1892152}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6875.729024367989, "epoch": 2.085201793721973, "event": "heartbeat", "loss": 0.042640384286642075, "lr": 2.2671203472684956e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08784226752734327, "mfu_sample_tflops_s": 18.402955046978413, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3348986449479962, "mfu_text_token_tflops_s": 70.1612661166052, "mfu_tokens_per_sample_assumed": 128.0, "step": 930, "time": 1780564623.8792303}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 6948.301611348987, "epoch": 2.10762331838565, "event": "heartbeat", "loss": 0.03102225624024868, "lr": 2.2622704838074273e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08874426196163512, "mfu_sample_tflops_s": 18.59192288096256, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.338337498728734, "mfu_text_token_tflops_s": 70.88170598366976, "mfu_tokens_per_sample_assumed": 128.0, "step": 940, "time": 1780564696.4518173}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7020.967812315997, "epoch": 2.1300448430493275, "event": "heartbeat", "loss": 0.03879321366548538, "lr": 2.257376596876834e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08872737169229154, "mfu_sample_tflops_s": 18.588384369535078, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3382731045768615, "mfu_text_token_tflops_s": 70.86821540885248, "mfu_tokens_per_sample_assumed": 128.0, "step": 950, "time": 1780564769.1180184}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7093.627317255974, "epoch": 2.1524663677130045, "event": "heartbeat", "loss": 0.043583646416664124, "lr": 2.252438929295435e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08865116768352008, "mfu_sample_tflops_s": 18.572419629697457, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33798257679342036, "mfu_text_token_tflops_s": 70.80734983822157, "mfu_tokens_per_sample_assumed": 128.0, "step": 960, "time": 1780564841.7775233}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7166.290617310995, "epoch": 2.1748878923766815, "event": "heartbeat", "loss": 0.04426608979701996, "lr": 2.247457726054202e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08881411182918877, "mfu_sample_tflops_s": 18.606556428215047, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3386038013487822, "mfu_text_token_tflops_s": 70.93749638256988, "mfu_tokens_per_sample_assumed": 128.0, "step": 970, "time": 1780564914.4408233}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7238.927970295976, "epoch": 2.197309417040359, "event": "heartbeat", "loss": 0.038305964320898056, "lr": 2.242433234304204e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08873383646941024, "mfu_sample_tflops_s": 18.589738740341446, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33829775153962655, "mfu_text_token_tflops_s": 70.87337894755177, "mfu_tokens_per_sample_assumed": 128.0, "step": 980, "time": 1780564987.0781763}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7311.615302476974, "epoch": 2.219730941704036, "event": "heartbeat", "loss": 0.02718463446944952, "lr": 2.2373657033443464e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08774253628812358, "mfu_sample_tflops_s": 18.38206135236189, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3345184195984711, "mfu_text_token_tflops_s": 70.0816089058797, "mfu_tokens_per_sample_assumed": 128.0, "step": 990, "time": 1780565059.7655084}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7384.224069425982, "epoch": 2.242152466367713, "event": "heartbeat", "loss": 0.02911452203989029, "lr": 2.2322553846089986e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08785994790347726, "mfu_sample_tflops_s": 18.406659085778486, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.334966051382007, "mfu_text_token_tflops_s": 70.17538776453047, "mfu_tokens_per_sample_assumed": 128.0, "step": 1000, "time": 1780565132.3742754}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7456.921434794, "epoch": 2.2645739910313902, "event": "heartbeat", "loss": 0.038771284744143486, "lr": 2.2271025316555195e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08788941227125088, "mfu_sample_tflops_s": 18.41283187082706, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33507838428414394, "mfu_text_token_tflops_s": 70.19892150752815, "mfu_tokens_per_sample_assumed": 128.0, "step": 1010, "time": 1780565205.0716405}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7529.5144732609915, "epoch": 2.286995515695067, "event": "heartbeat", "loss": 0.049255795776844025, "lr": 2.2219074001516786e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08876015746258445, "mfu_sample_tflops_s": 18.595252988411442, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33839810032610323, "mfu_text_token_tflops_s": 70.89440201831863, "mfu_tokens_per_sample_assumed": 128.0, "step": 1020, "time": 1780565277.664679}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7602.118413338991, "epoch": 2.309417040358744, "event": "heartbeat", "loss": 0.04904281534254551, "lr": 2.2166702478629687e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08886340291885343, "mfu_sample_tflops_s": 18.616882911499793, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33879172362812865, "mfu_text_token_tflops_s": 70.97686610009295, "mfu_tokens_per_sample_assumed": 128.0, "step": 1030, "time": 1780565350.2686193}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7674.726922429982, "epoch": 2.3318385650224216, "event": "heartbeat", "loss": 0.0366715993732214, "lr": 2.2113913346398167e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0887948235348145, "mfu_sample_tflops_s": 18.602515530543638, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33853026472648023, "mfu_text_token_tflops_s": 70.92209046019761, "mfu_tokens_per_sample_assumed": 128.0, "step": 1040, "time": 1780565422.8771284}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7747.338390672987, "epoch": 2.3542600896860986, "event": "heartbeat", "loss": 0.036855774000287056, "lr": 2.2060709224046904e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0888161984416004, "mfu_sample_tflops_s": 18.606993573515282, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3386117565586015, "mfu_text_token_tflops_s": 70.93916299902702, "mfu_tokens_per_sample_assumed": 128.0, "step": 1050, "time": 1780565495.4885964}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7820.007652130997, "epoch": 2.376681614349776, "event": "heartbeat", "loss": 0.040342748165130615, "lr": 2.2007092751391055e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08876807122319487, "mfu_sample_tflops_s": 18.596910921259326, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3384282715384304, "mfu_text_token_tflops_s": 70.90072288730117, "mfu_tokens_per_sample_assumed": 128.0, "step": 1060, "time": 1780565568.1578574}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7892.674134659988, "epoch": 2.399103139013453, "event": "heartbeat", "loss": 0.04980650916695595, "lr": 2.195306658870523e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08875486156904545, "mfu_sample_tflops_s": 18.594143498715024, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33837790973198584, "mfu_text_token_tflops_s": 70.89017208885103, "mfu_tokens_per_sample_assumed": 128.0, "step": 1070, "time": 1780565640.8243406}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 7965.330672381999, "epoch": 2.42152466367713, "event": "heartbeat", "loss": 0.034005146473646164, "lr": 2.1898633416591536e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08769289438812389, "mfu_sample_tflops_s": 18.37166137431195, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3343291598547223, "mfu_text_token_tflops_s": 70.04195898956432, "mfu_tokens_per_sample_assumed": 128.0, "step": 1080, "time": 1780565713.4808784}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8037.958380721975, "epoch": 2.4439461883408073, "event": "heartbeat", "loss": 0.035620834678411484, "lr": 2.1843795935846566e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08784810249603506, "mfu_sample_tflops_s": 18.404177472919343, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33492089076613357, "mfu_text_token_tflops_s": 70.16592661550499, "mfu_tokens_per_sample_assumed": 128.0, "step": 1090, "time": 1780565786.1085868}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8110.604984466976, "epoch": 2.4663677130044843, "event": "heartbeat", "loss": 0.03294756915420294, "lr": 2.178855686732738e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08789898329945016, "mfu_sample_tflops_s": 18.414837001234808, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3351148738291537, "mfu_text_token_tflops_s": 70.2065660672077, "mfu_tokens_per_sample_assumed": 128.0, "step": 1100, "time": 1780565858.7551906}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8183.309729704983, "epoch": 2.4887892376681613, "event": "heartbeat", "loss": 0.029998847283422947, "lr": 2.1732918951816506e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08784980446458258, "mfu_sample_tflops_s": 18.40453403533005, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33492737952122104, "mfu_text_token_tflops_s": 70.16728600969581, "mfu_tokens_per_sample_assumed": 128.0, "step": 1110, "time": 1780565931.4599354}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8287.056680787995, "epoch": 2.5112107623318387, "event": "heartbeat", "loss": 0.039045022800564766, "lr": 2.1676884949885985e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08784024576327232, "mfu_sample_tflops_s": 18.40253148740555, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33489093697247574, "mfu_text_token_tflops_s": 70.15965129573367, "mfu_tokens_per_sample_assumed": 128.0, "step": 1120, "time": 1780566035.2068868}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8359.653670236992, "epoch": 2.5336322869955157, "event": "heartbeat", "loss": 0.04086139425635338, "lr": 2.1620457641760353e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08879241999966762, "mfu_sample_tflops_s": 18.602011989930368, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33852110124873286, "mfu_text_token_tflops_s": 70.92017071160953, "mfu_tokens_per_sample_assumed": 128.0, "step": 1130, "time": 1780566107.8038762}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8432.29744245598, "epoch": 2.5560538116591927, "event": "heartbeat", "loss": 0.03651626780629158, "lr": 2.156363982717873e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08871059456907719, "mfu_sample_tflops_s": 18.584869562221673, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3382091417946068, "mfu_text_token_tflops_s": 70.85481520597013, "mfu_tokens_per_sample_assumed": 128.0, "step": 1140, "time": 1780566180.4476483}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8504.901748380973, "epoch": 2.57847533632287, "event": "heartbeat", "loss": 0.04855440929532051, "lr": 2.150643432525589e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08877337780195649, "mfu_sample_tflops_s": 18.598022649509883, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33844850286995914, "mfu_text_token_tflops_s": 70.90496135125643, "mfu_tokens_per_sample_assumed": 128.0, "step": 1150, "time": 1780566253.0519545}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8577.523456067982, "epoch": 2.600896860986547, "event": "heartbeat", "loss": 0.03528209775686264, "lr": 2.144884397434238e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08874792999097228, "mfu_sample_tflops_s": 18.592691333108693, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33835148309058183, "mfu_text_token_tflops_s": 70.88463570747689, "mfu_tokens_per_sample_assumed": 128.0, "step": 1160, "time": 1780566325.673662}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8650.122398050997, "epoch": 2.623318385650224, "event": "heartbeat", "loss": 0.03971759416162968, "lr": 2.1390871631883715e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08875917446048456, "mfu_sample_tflops_s": 18.595047049471514, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3383943526305974, "mfu_text_token_tflops_s": 70.89361687611016, "mfu_tokens_per_sample_assumed": 128.0, "step": 1170, "time": 1780566398.272604}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8722.722414930991, "epoch": 2.6457399103139014, "event": "heartbeat", "loss": 0.03338553290814161, "lr": 2.1332520174278585e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08868532285707639, "mfu_sample_tflops_s": 18.579575138557505, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33811279339260375, "mfu_text_token_tflops_s": 70.83463021575048, "mfu_tokens_per_sample_assumed": 128.0, "step": 1180, "time": 1780566470.8726203}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8795.345394236996, "epoch": 2.6681614349775784, "event": "heartbeat", "loss": 0.03176060225814581, "lr": 2.1273792496736117e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08787221862652614, "mfu_sample_tflops_s": 18.409229802257226, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33501283351363087, "mfu_text_token_tflops_s": 70.18518862110567, "mfu_tokens_per_sample_assumed": 128.0, "step": 1190, "time": 1780566543.4956002}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8867.954057361989, "epoch": 2.690582959641256, "event": "heartbeat", "loss": 0.03587989695370197, "lr": 2.121469151313227e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08786289725804983, "mfu_sample_tflops_s": 18.407276975561437, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3349772957963149, "mfu_text_token_tflops_s": 70.17774346932798, "mfu_tokens_per_sample_assumed": 128.0, "step": 1200, "time": 1780566616.104263}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 8940.584938193992, "epoch": 2.713004484304933, "event": "heartbeat", "loss": 0.032913253642618656, "lr": 2.1155220155865214e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0878809043271873, "mfu_sample_tflops_s": 18.411049456545737, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33504594774740154, "mfu_text_token_tflops_s": 70.19212605308063, "mfu_tokens_per_sample_assumed": 128.0, "step": 1210, "time": 1780566688.7351441}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9013.299184658972, "epoch": 2.7354260089686098, "event": "heartbeat", "loss": 0.03460179083049297, "lr": 2.1095381375709872e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08784892947860175, "mfu_sample_tflops_s": 18.404350725767067, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3349240436371692, "mfu_text_token_tflops_s": 70.16658714198695, "mfu_tokens_per_sample_assumed": 128.0, "step": 1220, "time": 1780566761.4493906}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9085.948513082985, "epoch": 2.757847533632287, "event": "heartbeat", "loss": 0.03745702933520079, "lr": 2.1035178141671476e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08882709291963846, "mfu_sample_tflops_s": 18.609275966664256, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33865329175612163, "mfu_text_token_tflops_s": 70.94786462290749, "mfu_tokens_per_sample_assumed": 128.0, "step": 1230, "time": 1780566834.0987184}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9158.648767770996, "epoch": 2.780269058295964, "event": "heartbeat", "loss": 0.043826354667544365, "lr": 2.0974613440838277e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08871432227067137, "mfu_sample_tflops_s": 18.58565051570565, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3382233536569347, "mfu_text_token_tflops_s": 70.85779259112782, "mfu_tokens_per_sample_assumed": 128.0, "step": 1240, "time": 1780566906.798973}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9231.322812122991, "epoch": 2.802690582959641, "event": "heartbeat", "loss": 0.03910966031253338, "lr": 2.091369027823334e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08861064036250921, "mfu_sample_tflops_s": 18.56392915594568, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33782806638206636, "mfu_text_token_tflops_s": 70.7749799070429, "mfu_tokens_per_sample_assumed": 128.0, "step": 1250, "time": 1780566979.4730182}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9303.96697489798, "epoch": 2.8251121076233185, "event": "heartbeat", "loss": 0.04258435778319836, "lr": 2.0852411676665413e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08838847537320027, "mfu_sample_tflops_s": 18.517385590685457, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.336981062360326, "mfu_text_token_tflops_s": 70.59753256448829, "mfu_tokens_per_sample_assumed": 128.0, "step": 1260, "time": 1780567052.1171808}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9376.602887975983, "epoch": 2.8475336322869955, "event": "heartbeat", "loss": 0.03953726775944233, "lr": 2.079078067657899e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08878783196088387, "mfu_sample_tflops_s": 18.601050795805172, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33850360935086976, "mfu_text_token_tflops_s": 70.91650615900721, "mfu_tokens_per_sample_assumed": 128.0, "step": 1270, "time": 1780567124.753094}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9449.228645880998, "epoch": 2.8699551569506725, "event": "heartbeat", "loss": 0.042893427424132824, "lr": 2.0728800335903425e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08876088976759493, "mfu_sample_tflops_s": 18.595406406311138, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3384008922389557, "mfu_text_token_tflops_s": 70.89498692406121, "mfu_tokens_per_sample_assumed": 128.0, "step": 1280, "time": 1780567197.378852}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9521.822257074993, "epoch": 2.89237668161435, "event": "heartbeat", "loss": 0.03858952224254608, "lr": 2.0666473729901205e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08798238809165235, "mfu_sample_tflops_s": 18.432310305201167, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33543285459942457, "mfu_text_token_tflops_s": 70.27318303857945, "mfu_tokens_per_sample_assumed": 128.0, "step": 1290, "time": 1780567269.9724631}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9594.487056802987, "epoch": 2.914798206278027, "event": "heartbeat", "loss": 0.02582953032106161, "lr": 2.060380395101537e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0877928521975303, "mfu_sample_tflops_s": 18.392602535382597, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3347102490030842, "mfu_text_token_tflops_s": 70.12179716614615, "mfu_tokens_per_sample_assumed": 128.0, "step": 1300, "time": 1780567342.6372628}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9667.020059588976, "epoch": 2.937219730941704, "event": "heartbeat", "loss": 0.03209305368363857, "lr": 2.05407941087161e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08882981021448359, "mfu_sample_tflops_s": 18.609845239934312, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33866365144271865, "mfu_text_token_tflops_s": 70.95003497724956, "mfu_tokens_per_sample_assumed": 128.0, "step": 1310, "time": 1780567415.1702657}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9739.610927852977, "epoch": 2.9596412556053813, "event": "heartbeat", "loss": 0.039677247405052185, "lr": 2.04774473293464e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0888112027639168, "mfu_sample_tflops_s": 18.60594697904057, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3385927105374328, "mfu_text_token_tflops_s": 70.93517285759218, "mfu_tokens_per_sample_assumed": 128.0, "step": 1320, "time": 1780567487.761134}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9811.883092359989, "epoch": 2.9820627802690582, "event": "heartbeat", "loss": 0.03372473735362291, "lr": 2.0413766755966996e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08918904640194891, "mfu_sample_tflops_s": 18.685105221208296, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3400332394074302, "mfu_text_token_tflops_s": 71.23696365585663, "mfu_tokens_per_sample_assumed": 128.0, "step": 1330, "time": 1780567560.0332983}
+{"best_step": 446, "best_val_loss": 0.06323498976416886, "checkpoint_dir": "/home/andon/clawmachine/dk1-teleop/outputs/vla_pickupcube_200_lora/molmoact2_lora_pickupcube_200_20260603_205822/checkpoints/001338", "cuda_allocated_gib": 12.885714530944824, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "epoch": 3.0, "event": "checkpoint", "step": 1338}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9915.592196982994, "epoch": 3.004484304932735, "event": "heartbeat", "loss": 0.03405451215803623, "lr": 2.034975554820038e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08880856061731984, "mfu_sample_tflops_s": 18.605393449328506, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33858263735353195, "mfu_text_token_tflops_s": 70.93306252556494, "mfu_tokens_per_sample_assumed": 128.0, "step": 1340, "time": 1780567663.7424028}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 9988.083727709978, "epoch": 3.0269058295964126, "event": "heartbeat", "loss": 0.035285165533423424, "lr": 2.0285416882074054e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08887918219633618, "mfu_sample_tflops_s": 18.62018867013243, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3388518821235317, "mfu_text_token_tflops_s": 70.98946930487989, "mfu_tokens_per_sample_assumed": 128.0, "step": 1350, "time": 1780567736.2339334}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10060.490030037996, "epoch": 3.0493273542600896, "event": "heartbeat", "loss": 0.029025699011981487, "lr": 2.0220753949862933e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08900015128428342, "mfu_sample_tflops_s": 18.645531694057375, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33931307677133055, "mfu_text_token_tflops_s": 71.08608958359375, "mfu_tokens_per_sample_assumed": 128.0, "step": 1360, "time": 1780567808.6402361}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10132.989581018977, "epoch": 3.071748878923767, "event": "heartbeat", "loss": 0.02668639086186886, "lr": 2.0155769959930956e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08894464534934828, "mfu_sample_tflops_s": 18.633903200688465, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3391014603943904, "mfu_text_token_tflops_s": 71.04175595262478, "mfu_tokens_per_sample_assumed": 128.0, "step": 1370, "time": 1780567881.139787}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10205.42592899999, "epoch": 3.094170403587444, "event": "heartbeat", "loss": 0.03325581829994917, "lr": 2.0090468136571895e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08883404958763541, "mfu_sample_tflops_s": 18.610733388609617, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33867981405286, "mfu_text_token_tflops_s": 70.95342104407418, "mfu_tokens_per_sample_assumed": 128.0, "step": 1380, "time": 1780567953.576135}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10277.909340675978, "epoch": 3.116591928251121, "event": "heartbeat", "loss": 0.03197987377643585, "lr": 2.00248517198494e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08892219587265268, "mfu_sample_tflops_s": 18.62920003532074, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33901587176448844, "mfu_text_token_tflops_s": 71.02382513466033, "mfu_tokens_per_sample_assumed": 128.0, "step": 1390, "time": 1780568026.0595467}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10350.323916433, "epoch": 3.1390134529147984, "event": "heartbeat", "loss": 0.03327016718685627, "lr": 1.9958923965436192e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08922759442867599, "mfu_sample_tflops_s": 18.693181032807622, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3401802037593273, "mfu_text_token_tflops_s": 71.26775268757906, "mfu_tokens_per_sample_assumed": 128.0, "step": 1400, "time": 1780568098.4741223}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10422.669521772972, "epoch": 3.1614349775784754, "event": "heartbeat", "loss": 0.029416805133223534, "lr": 1.9892688144452573e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08826371425438984, "mfu_sample_tflops_s": 18.491248136294672, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3365054105948613, "mfu_text_token_tflops_s": 70.49788351962344, "mfu_tokens_per_sample_assumed": 128.0, "step": 1410, "time": 1780568170.8197277}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10494.912384342984, "epoch": 3.1838565022421523, "event": "heartbeat", "loss": 0.029380789026618004, "lr": 1.9826147543304095e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08848766823399042, "mfu_sample_tflops_s": 18.53816649502099, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33735923514208843, "mfu_text_token_tflops_s": 70.67675976226754, "mfu_tokens_per_sample_assumed": 128.0, "step": 1420, "time": 1780568243.0625901}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10567.255050443986, "epoch": 3.2062780269058297, "event": "heartbeat", "loss": 0.03492097184062004, "lr": 1.975930546351852e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08760380416442635, "mfu_sample_tflops_s": 18.352996972447322, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33398950337687544, "mfu_text_token_tflops_s": 69.97080095745541, "mfu_tokens_per_sample_assumed": 128.0, "step": 1430, "time": 1780568315.4052565}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10639.758255371999, "epoch": 3.2286995515695067, "event": "heartbeat", "loss": 0.03714005649089813, "lr": 1.9692165221581978e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08819924211429479, "mfu_sample_tflops_s": 18.477741222944758, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3362596105607489, "mfu_text_token_tflops_s": 70.44638841247689, "mfu_tokens_per_sample_assumed": 128.0, "step": 1440, "time": 1780568387.9084613}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10711.992963707977, "epoch": 3.2511210762331837, "event": "heartbeat", "loss": 0.027534188702702522, "lr": 1.9624730148774443e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08927558290936066, "mfu_sample_tflops_s": 18.703234619511058, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3403631598419376, "mfu_text_token_tflops_s": 71.30608198688593, "mfu_tokens_per_sample_assumed": 128.0, "step": 1450, "time": 1780568460.1431696}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10784.229503091978, "epoch": 3.273542600896861, "event": "heartbeat", "loss": 0.03555257432162762, "lr": 1.9557003591004443e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08929402482839989, "mfu_sample_tflops_s": 18.707098201549776, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34043346965827453, "mfu_text_token_tflops_s": 71.32081189340852, "mfu_tokens_per_sample_assumed": 128.0, "step": 1460, "time": 1780568532.3797088}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10856.670933067973, "epoch": 3.295964125560538, "event": "heartbeat", "loss": 0.030280286446213722, "lr": 1.948898890864303e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08927603639306567, "mfu_sample_tflops_s": 18.70332962434726, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34036488874856285, "mfu_text_token_tflops_s": 71.30644419282392, "mfu_tokens_per_sample_assumed": 128.0, "step": 1470, "time": 1780568604.8211389}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 10928.888995274989, "epoch": 3.318385650224215, "event": "heartbeat", "loss": 0.02917201817035675, "lr": 1.9420689476357064e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08925910214935939, "mfu_sample_tflops_s": 18.699781900290795, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34030032694443274, "mfu_text_token_tflops_s": 71.29291849485865, "mfu_tokens_per_sample_assumed": 128.0, "step": 1480, "time": 1780568677.0392013}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11001.13966499499, "epoch": 3.3408071748878925, "event": "heartbeat", "loss": 0.03320280462503433, "lr": 1.9352108682941767e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08924479316107765, "mfu_sample_tflops_s": 18.69678416724577, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3402457739266086, "mfu_text_token_tflops_s": 71.2814896376245, "mfu_tokens_per_sample_assumed": 128.0, "step": 1490, "time": 1780568749.2898707}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11073.279157006997, "epoch": 3.3632286995515694, "event": "heartbeat", "loss": 0.030707716941833496, "lr": 1.9283249931152578e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08939623431059249, "mfu_sample_tflops_s": 18.728511088069126, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3408231433091338, "mfu_text_token_tflops_s": 71.40244852326353, "mfu_tokens_per_sample_assumed": 128.0, "step": 1500, "time": 1780568821.429363}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11145.455858249974, "epoch": 3.3856502242152464, "event": "heartbeat", "loss": 0.033593567088246346, "lr": 1.9214116637536336e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08900993612797406, "mfu_sample_tflops_s": 18.647581618810566, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3393503814879011, "mfu_text_token_tflops_s": 71.09390492171528, "mfu_tokens_per_sample_assumed": 128.0, "step": 1510, "time": 1780568893.606064}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11217.555890071992, "epoch": 3.408071748878924, "event": "heartbeat", "loss": 0.03946463577449322, "lr": 1.914471223226174e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08857907782974349, "mfu_sample_tflops_s": 18.55731680533126, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33770773422589706, "mfu_text_token_tflops_s": 70.74977032032544, "mfu_tokens_per_sample_assumed": 128.0, "step": 1520, "time": 1780568965.7060962}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11289.671758527984, "epoch": 3.430493273542601, "event": "heartbeat", "loss": 0.04596183821558952, "lr": 1.9075040158949177e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08853622112187426, "mfu_sample_tflops_s": 18.548338325032656, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3375443430271456, "mfu_text_token_tflops_s": 70.715539864187, "mfu_tokens_per_sample_assumed": 128.0, "step": 1530, "time": 1780569037.8219645}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11361.76603542699, "epoch": 3.4529147982062782, "event": "heartbeat", "loss": 0.02291794028133154, "lr": 1.9005103874499827e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08937396632951547, "mfu_sample_tflops_s": 18.723845946033492, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3407382466312777, "mfu_text_token_tflops_s": 71.38466266925269, "mfu_tokens_per_sample_assumed": 128.0, "step": 1540, "time": 1780569109.9162414}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11433.967211451993, "epoch": 3.475336322869955, "event": "heartbeat", "loss": 0.031040897592902184, "lr": 1.8934906848924196e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08933809726020842, "mfu_sample_tflops_s": 18.716331376013663, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3406014958045446, "mfu_text_token_tflops_s": 71.3560133710521, "mfu_tokens_per_sample_assumed": 128.0, "step": 1550, "time": 1780569182.117417}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11506.068423738994, "epoch": 3.497757847533632, "event": "heartbeat", "loss": 0.036005277186632156, "lr": 1.8864452565169894e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08932219927434923, "mfu_sample_tflops_s": 18.713000747976164, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34054088473345645, "mfu_text_token_tflops_s": 71.34331535165913, "mfu_tokens_per_sample_assumed": 128.0, "step": 1560, "time": 1780569254.2186296}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11608.919457485987, "epoch": 3.5201793721973096, "event": "heartbeat", "loss": 0.03518286719918251, "lr": 1.8793744518948843e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08941540610153893, "mfu_sample_tflops_s": 18.732527578272407, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3408962357621172, "mfu_text_token_tflops_s": 71.41776139216354, "mfu_tokens_per_sample_assumed": 128.0, "step": 1570, "time": 1780569357.0696635}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11680.943027599977, "epoch": 3.5426008968609866, "event": "heartbeat", "loss": 0.03665084205567837, "lr": 1.8722786218563853e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08961221323913371, "mfu_sample_tflops_s": 18.77375867359851, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34164656297419727, "mfu_text_token_tflops_s": 71.57495494309433, "mfu_tokens_per_sample_assumed": 128.0, "step": 1580, "time": 1780569429.0932333}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11752.91172482699, "epoch": 3.5650224215246635, "event": "heartbeat", "loss": 0.03781296871602535, "lr": 1.86515811847345e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08955885806530349, "mfu_sample_tflops_s": 18.762580764681083, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34144314637396955, "mfu_text_token_tflops_s": 71.53233916534663, "mfu_tokens_per_sample_assumed": 128.0, "step": 1590, "time": 1780569501.0619307}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11825.043571498973, "epoch": 3.587443946188341, "event": "heartbeat", "loss": 0.022224420681595802, "lr": 1.858013295042249e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08842317594356293, "mfu_sample_tflops_s": 18.524655360176432, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3371133582848337, "mfu_text_token_tflops_s": 70.62524856067266, "mfu_tokens_per_sample_assumed": 128.0, "step": 1600, "time": 1780569573.1937773}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11897.010995682998, "epoch": 3.609865470852018, "event": "heartbeat", "loss": 0.02829230483621359, "lr": 1.850844506065634e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0886519576234199, "mfu_sample_tflops_s": 18.572585122106467, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33798558843928833, "mfu_text_token_tflops_s": 70.80798077803091, "mfu_tokens_per_sample_assumed": 128.0, "step": 1610, "time": 1780569645.1612017}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 11968.900263018993, "epoch": 3.632286995515695, "event": "heartbeat", "loss": 0.038494981825351715, "lr": 1.8436521072355485e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08972919695706133, "mfu_sample_tflops_s": 18.798266762504348, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3420925633987963, "mfu_text_token_tflops_s": 71.66839203204783, "mfu_tokens_per_sample_assumed": 128.0, "step": 1620, "time": 1780569717.050469}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12040.855082263995, "epoch": 3.6547085201793723, "event": "heartbeat", "loss": 0.02980615198612213, "lr": 1.83643645541538e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08954550803838036, "mfu_sample_tflops_s": 18.759783934040687, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3413922493963252, "mfu_text_token_tflops_s": 71.52167624853013, "mfu_tokens_per_sample_assumed": 128.0, "step": 1630, "time": 1780569789.0052881}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12112.939885070984, "epoch": 3.6771300448430493, "event": "heartbeat", "loss": 0.0340274553745985, "lr": 1.8291979086222546e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0895163137301833, "mfu_sample_tflops_s": 18.753667726473402, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3412809460963238, "mfu_text_token_tflops_s": 71.49835820717985, "mfu_tokens_per_sample_assumed": 128.0, "step": 1640, "time": 1780569861.090091}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12184.981799994974, "epoch": 3.6995515695067263, "event": "heartbeat", "loss": 0.024558844044804573, "lr": 1.8219368260092708e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08960917075955396, "mfu_sample_tflops_s": 18.773121274126556, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34163496352079953, "mfu_text_token_tflops_s": 71.5725248576075, "mfu_tokens_per_sample_assumed": 128.0, "step": 1650, "time": 1780569933.1320057}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12256.937546722998, "epoch": 3.7219730941704037, "event": "heartbeat", "loss": 0.029732894152402878, "lr": 1.8146535678476825e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08962079600432703, "mfu_sample_tflops_s": 18.77555676290651, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3416792847664968, "mfu_text_token_tflops_s": 71.58181015858108, "mfu_tokens_per_sample_assumed": 128.0, "step": 1660, "time": 1780570005.0877528}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12328.859561715974, "epoch": 3.7443946188340806, "event": "heartbeat", "loss": 0.03343381639569998, "lr": 1.8073484955090206e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08954683263282423, "mfu_sample_tflops_s": 18.760061436576677, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3413972994126423, "mfu_text_token_tflops_s": 71.52273422694856, "mfu_tokens_per_sample_assumed": 128.0, "step": 1670, "time": 1780570077.0097678}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12400.861705958989, "epoch": 3.766816143497758, "event": "heartbeat", "loss": 0.03821006044745445, "lr": 1.8000219714471657e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08957681199500139, "mfu_sample_tflops_s": 18.766342112952792, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3415115957309428, "mfu_text_token_tflops_s": 71.54667930563251, "mfu_tokens_per_sample_assumed": 128.0, "step": 1680, "time": 1780570149.0119119}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12472.87911422798, "epoch": 3.789237668161435, "event": "heartbeat", "loss": 0.03357852064073086, "lr": 1.792674359180362e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08853960569748034, "mfu_sample_tflops_s": 18.549047393622132, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3375572467216439, "mfu_text_token_tflops_s": 70.71824318818439, "mfu_tokens_per_sample_assumed": 128.0, "step": 1690, "time": 1780570221.0293202}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12544.784301683976, "epoch": 3.811659192825112, "event": "heartbeat", "loss": 0.03548250161111355, "lr": 1.7853060232731827e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08883356345905277, "mfu_sample_tflops_s": 18.610631544671556, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33867796068763867, "mfu_text_token_tflops_s": 70.9530327640603, "mfu_tokens_per_sample_assumed": 128.0, "step": 1700, "time": 1780570292.9345074}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12616.603595896973, "epoch": 3.8340807174887894, "event": "heartbeat", "loss": 0.029920971021056175, "lr": 1.777917329318439e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08967240624754887, "mfu_sample_tflops_s": 18.78636910886149, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34187604881878003, "mfu_text_token_tflops_s": 71.62303222753442, "mfu_tokens_per_sample_assumed": 128.0, "step": 1710, "time": 1780570364.7538018}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12688.542980866972, "epoch": 3.8565022421524664, "event": "heartbeat", "loss": 0.02956908382475376, "lr": 1.7705086439190438e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0897219871167982, "mfu_sample_tflops_s": 18.79675630096922, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3420650758827931, "mfu_text_token_tflops_s": 71.66263339744516, "mfu_tokens_per_sample_assumed": 128.0, "step": 1720, "time": 1780570436.6931863}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12760.532329479, "epoch": 3.8789237668161434, "event": "heartbeat", "loss": 0.03187971282750368, "lr": 1.7630803346698184e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.0895495261514713, "mfu_sample_tflops_s": 18.76062572873324, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3414075684524843, "mfu_text_token_tflops_s": 71.52488559079546, "mfu_tokens_per_sample_assumed": 128.0, "step": 1730, "time": 1780570508.6825354}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12832.435421387985, "epoch": 3.901345291479821, "event": "heartbeat", "loss": 0.03740593418478966, "lr": 1.755632770139258e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08969189293513004, "mfu_sample_tflops_s": 18.79045156990974, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34195034181518325, "mfu_text_token_tflops_s": 71.63859661028089, "mfu_tokens_per_sample_assumed": 128.0, "step": 1740, "time": 1780570580.585627}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12904.314320035977, "epoch": 3.9237668161434978, "event": "heartbeat", "loss": 0.032284852117300034, "lr": 1.74816631985124e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08968413728453511, "mfu_sample_tflops_s": 18.788826761110105, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3419207733972901, "mfu_text_token_tflops_s": 71.63240202673228, "mfu_tokens_per_sample_assumed": 128.0, "step": 1750, "time": 1780570652.4645262}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 12976.171731597977, "epoch": 3.9461883408071747, "event": "heartbeat", "loss": 0.03337559290230274, "lr": 1.7406813542666924e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08882607116880392, "mfu_sample_tflops_s": 18.609061909864423, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.33864939633106494, "mfu_text_token_tflops_s": 70.9470485313581, "mfu_tokens_per_sample_assumed": 128.0, "step": 1760, "time": 1780570724.3219373}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 13047.952010321984, "epoch": 3.968609865470852, "event": "heartbeat", "loss": 0.03852874878793955, "lr": 1.733178244765212e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08932105520363333, "mfu_sample_tflops_s": 18.712761065161185, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3405365229638521, "mfu_text_token_tflops_s": 71.34240156092702, "mfu_tokens_per_sample_assumed": 128.0, "step": 1770, "time": 1780570796.1022162}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 13119.419539709983, "epoch": 3.991031390134529, "event": "heartbeat", "loss": 0.028546319343149662, "lr": 1.725657363626637e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.09009925794147017, "mfu_sample_tflops_s": 18.875794538738, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34350342090185504, "mfu_text_token_tflops_s": 71.96396667893862, "mfu_tokens_per_sample_assumed": 128.0, "step": 1780, "time": 1780570867.5697458}
+{"best_step": 446, "best_val_loss": 0.06323498976416886, "checkpoint_dir": "/home/andon/clawmachine/dk1-teleop/outputs/vla_pickupcube_200_lora/molmoact2_lora_pickupcube_200_20260603_205822/checkpoints/001784", "cuda_allocated_gib": 12.885714530944824, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "epoch": 4.0, "event": "checkpoint", "step": 1784}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 13222.749508445995, "epoch": 4.013452914798206, "event": "heartbeat", "loss": 0.039656249806284904, "lr": 1.7181190840125763e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08973635051845658, "mfu_sample_tflops_s": 18.799765433616653, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.3421198363516157, "mfu_text_token_tflops_s": 71.67410571566349, "mfu_tokens_per_sample_assumed": 128.0, "step": 1790, "time": 1780570970.899714}
+{"cuda_allocated_gib": 13.006690979003906, "cuda_device": 0, "cuda_max_allocated_gib": 25.40953493118286, "cuda_max_reserved_gib": 26.732421875, "cuda_reserved_gib": 26.732421875, "elapsed_s": 13294.854200691974, "epoch": 4.0358744394618835, "event": "heartbeat", "loss": 0.025000717490911484, "lr": 1.7105637799478958e-05, "mfu_batch_items": 32.0, "mfu_estimate_method": "6*total_params*effective_token_units/update_time/gpu_peak_tflops", "mfu_observed_text_tokens": 15616.0, "mfu_peak_tflops": 209.5, "mfu_sample_pct": 0.08967143603516667, "mfu_sample_tflops_s": 18.786165849367418, "mfu_sample_token_units": 4096.0, "mfu_text_token_pct": 0.34187234988407295, "mfu_text_token_tflops_s": 71.62225730071329, "mfu_tokens_per_sample_assumed": 128.0, "step": 1800, "time": 1780571043.0044065}