Upload exp_phase8_bridge_recipeA_300step_035043/log.jsonl with huggingface_hub
Browse files
exp_phase8_bridge_recipeA_300step_035043/log.jsonl
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 0, "loss": 1.2816545963287354, "loss_mdlm": 1.2816545963287354, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 0.0, "gnorm": 0.00057798414491117, "bridge_gate_avg": 1.0, "bridge_out_proj_avg": 0.0, "elapsed_s": 3.4989242553710938}
|
| 2 |
+
{"step": 25, "loss": 0.7683903574943542, "loss_mdlm": 0.7683903574943542, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 4.166666666666667e-05, "gnorm": 0.011605962179601192, "bridge_gate_avg": 1.000287264585495, "bridge_out_proj_avg": 0.0002059817488770932, "elapsed_s": 13.977524280548096}
|
| 3 |
+
{"step": 50, "loss": 1.3804423809051514, "loss_mdlm": 1.3804423809051514, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 4.9326121764495596e-05, "gnorm": 0.002503998577594757, "bridge_gate_avg": 1.0005621910095215, "bridge_out_proj_avg": 0.00039976220432436094, "elapsed_s": 24.16152572631836}
|
| 4 |
+
{"step": 75, "loss": 1.3444643020629883, "loss_mdlm": 1.3444643020629883, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 4.665063509461097e-05, "gnorm": 0.002093251096084714, "bridge_gate_avg": 1.0015327334403992, "bridge_out_proj_avg": 0.0008522827993147075, "elapsed_s": 34.82144379615784}
|
| 5 |
+
{"step": 100, "loss": 1.3629400730133057, "loss_mdlm": 1.3629400730133057, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 4.215604094671835e-05, "gnorm": 0.004405131563544273, "bridge_gate_avg": 1.002137154340744, "bridge_out_proj_avg": 0.001173498952994123, "elapsed_s": 45.45208787918091}
|
| 6 |
+
{"step": 125, "loss": 1.3012900352478027, "loss_mdlm": 1.3012900352478027, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 3.621997950501156e-05, "gnorm": 0.0025559659115970135, "bridge_gate_avg": 1.0025257170200348, "bridge_out_proj_avg": 0.0013335345720406622, "elapsed_s": 56.23868727684021}
|
| 7 |
+
{"step": 150, "loss": 1.3193247318267822, "loss_mdlm": 1.3193247318267822, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 2.9341204441673266e-05, "gnorm": 0.0031902703922241926, "bridge_gate_avg": 1.0030110776424408, "bridge_out_proj_avg": 0.0015056452248245478, "elapsed_s": 67.16778349876404}
|
| 8 |
+
{"step": 175, "loss": 1.3244844675064087, "loss_mdlm": 1.3244844675064087, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 2.2097677146869242e-05, "gnorm": 0.003336072899401188, "bridge_gate_avg": 1.00327467918396, "bridge_out_proj_avg": 0.001578078867169097, "elapsed_s": 77.65710639953613}
|
| 9 |
+
{"step": 200, "loss": 1.322136640548706, "loss_mdlm": 1.322136640548706, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 1.509800584902108e-05, "gnorm": 0.0045806486159563065, "bridge_gate_avg": 1.0033027827739716, "bridge_out_proj_avg": 0.0015888980706222355, "elapsed_s": 88.65838170051575}
|
| 10 |
+
{"step": 225, "loss": 1.3281638622283936, "loss_mdlm": 1.3281638622283936, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 8.930309757836517e-06, "gnorm": 0.004369310103356838, "bridge_gate_avg": 1.0032944977283478, "bridge_out_proj_avg": 0.0015881792642176151, "elapsed_s": 99.25430107116699}
|
| 11 |
+
{"step": 250, "loss": 1.2803382873535156, "loss_mdlm": 1.2803382873535156, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 4.112804714676594e-06, "gnorm": 0.002672883216291666, "bridge_gate_avg": 1.003298133611679, "bridge_out_proj_avg": 0.0015902798040769994, "elapsed_s": 110.16259908676147}
|
| 12 |
+
{"step": 275, "loss": 1.3376786708831787, "loss_mdlm": 1.3376786708831787, "loss_lm": 0.0, "loss_ct": 0.0, "lr": 1.0502621921127776e-06, "gnorm": 0.006582462694495916, "bridge_gate_avg": 1.0033127665519714, "bridge_out_proj_avg": 0.0015922577586025, "elapsed_s": 120.98979616165161}
|