Upload exp_phase8_bridge_recipeAB_2000step_035847/log.jsonl with huggingface_hub
Browse files
exp_phase8_bridge_recipeAB_2000step_035847/log.jsonl
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 0, "loss": 2.214611291885376, "loss_mdlm": 1.2915422916412354, "loss_lm": 1.8461381196975708, "loss_ct": 0.0, "lr": 0.0, "gnorm": 19.865812301635742, "bridge_gate_avg": 1.0, "bridge_out_proj_avg": 0.0, "elapsed_s": 3.5323541164398193}
|
| 2 |
+
{"step": 100, "loss": 1.9302589893341064, "loss_mdlm": 1.3533483743667603, "loss_lm": 1.153821349143982, "loss_ct": 0.0, "lr": 2.5e-05, "gnorm": 3.5818185806274414, "bridge_gate_avg": 1.0008090436458588, "bridge_out_proj_avg": 0.00031053891143528745, "elapsed_s": 53.055164098739624}
|
| 3 |
+
{"step": 200, "loss": 2.0620169639587402, "loss_mdlm": 1.3085517883300781, "loss_lm": 1.5069301128387451, "loss_ct": 0.0, "lr": 5e-05, "gnorm": 4.170899868011475, "bridge_gate_avg": 1.0035446286201477, "bridge_out_proj_avg": 0.001585339952725917, "elapsed_s": 100.86480164527893}
|
| 4 |
+
{"step": 300, "loss": 1.6456503868103027, "loss_mdlm": 1.3180292844772339, "loss_lm": 0.6552423238754272, "loss_ct": 0.0, "lr": 4.962019382530521e-05, "gnorm": 2.7423763275146484, "bridge_gate_avg": 1.0047843754291534, "bridge_out_proj_avg": 0.001951522019226104, "elapsed_s": 148.4580671787262}
|
| 5 |
+
{"step": 400, "loss": 1.7750089168548584, "loss_mdlm": 1.3292547464370728, "loss_lm": 0.8915082216262817, "loss_ct": 0.0, "lr": 4.849231551964771e-05, "gnorm": 3.321453094482422, "bridge_gate_avg": 1.005309671163559, "bridge_out_proj_avg": 0.0020799324265681207, "elapsed_s": 196.04585528373718}
|
| 6 |
+
{"step": 500, "loss": 1.7376114130020142, "loss_mdlm": 1.3520991802215576, "loss_lm": 0.7710245251655579, "loss_ct": 0.0, "lr": 4.665063509461097e-05, "gnorm": 3.1260738372802734, "bridge_gate_avg": 1.0064088106155396, "bridge_out_proj_avg": 0.002219506073743105, "elapsed_s": 243.22749829292297}
|
| 7 |
+
{"step": 600, "loss": 1.7461450099945068, "loss_mdlm": 1.3908817768096924, "loss_lm": 0.7105265259742737, "loss_ct": 0.0, "lr": 4.415111107797445e-05, "gnorm": 2.411102294921875, "bridge_gate_avg": 1.008141428232193, "bridge_out_proj_avg": 0.00230307161109522, "elapsed_s": 290.7805826663971}
|
| 8 |
+
{"step": 700, "loss": 1.670566439628601, "loss_mdlm": 1.3158471584320068, "loss_lm": 0.7094385027885437, "loss_ct": 0.0, "lr": 4.1069690242163484e-05, "gnorm": 2.4969120025634766, "bridge_gate_avg": 1.0087928473949432, "bridge_out_proj_avg": 0.00234654254745692, "elapsed_s": 338.1300573348999}
|
| 9 |
+
{"step": 800, "loss": 1.81608247756958, "loss_mdlm": 1.297305703163147, "loss_lm": 1.0375534296035767, "loss_ct": 0.0, "lr": 3.7500000000000003e-05, "gnorm": 3.391765594482422, "bridge_gate_avg": 1.0095359683036804, "bridge_out_proj_avg": 0.0024170801043510437, "elapsed_s": 385.7148995399475}
|
| 10 |
+
{"step": 900, "loss": 1.6422744989395142, "loss_mdlm": 1.2794607877731323, "loss_lm": 0.7256274819374084, "loss_ct": 0.0, "lr": 3.355050358314172e-05, "gnorm": 2.8085825443267822, "bridge_gate_avg": 1.0102248787879944, "bridge_out_proj_avg": 0.002477060304954648, "elapsed_s": 433.3227746486664}
|
| 11 |
+
{"step": 1000, "loss": 1.632951259613037, "loss_mdlm": 1.3078991174697876, "loss_lm": 0.6501043438911438, "loss_ct": 0.0, "lr": 2.9341204441673266e-05, "gnorm": 2.1434926986694336, "bridge_gate_avg": 1.0108878314495087, "bridge_out_proj_avg": 0.0025306580937467515, "elapsed_s": 480.8728461265564}
|
| 12 |
+
{"step": 1100, "loss": 1.4884339570999146, "loss_mdlm": 1.3199772834777832, "loss_lm": 0.3369132876396179, "loss_ct": 0.0, "lr": 2.5e-05, "gnorm": 1.690338134765625, "bridge_gate_avg": 1.0113024711608887, "bridge_out_proj_avg": 0.002564622147474438, "elapsed_s": 527.4669098854065}
|
| 13 |
+
{"step": 1200, "loss": 1.6875197887420654, "loss_mdlm": 1.3392366170883179, "loss_lm": 0.6965664029121399, "loss_ct": 0.0, "lr": 2.0658795558326743e-05, "gnorm": 2.3736460208892822, "bridge_gate_avg": 1.0116801857948303, "bridge_out_proj_avg": 0.002592337492387742, "elapsed_s": 574.6518707275391}
|
| 14 |
+
{"step": 1300, "loss": 1.6600041389465332, "loss_mdlm": 1.3473459482192993, "loss_lm": 0.6253165006637573, "loss_ct": 0.0, "lr": 1.6449496416858284e-05, "gnorm": 2.866643190383911, "bridge_gate_avg": 1.0121097266674042, "bridge_out_proj_avg": 0.0026152997743338346, "elapsed_s": 621.6165583133698}
|
| 15 |
+
{"step": 1400, "loss": 1.5996967554092407, "loss_mdlm": 1.250421404838562, "loss_lm": 0.6985507607460022, "loss_ct": 0.0, "lr": 1.2500000000000006e-05, "gnorm": 3.1556034088134766, "bridge_gate_avg": 1.0123736560344696, "bridge_out_proj_avg": 0.0026302755577489734, "elapsed_s": 668.891455411911}
|
| 16 |
+
{"step": 1500, "loss": 1.7236825227737427, "loss_mdlm": 1.3213104009628296, "loss_lm": 0.8047441840171814, "loss_ct": 0.0, "lr": 8.930309757836517e-06, "gnorm": 2.5364813804626465, "bridge_gate_avg": 1.0125497579574585, "bridge_out_proj_avg": 0.002643879852257669, "elapsed_s": 716.3971345424652}
|
| 17 |
+
{"step": 1600, "loss": 1.7214505672454834, "loss_mdlm": 1.3184363842010498, "loss_lm": 0.8060283064842224, "loss_ct": 0.0, "lr": 5.848888922025553e-06, "gnorm": 2.6006975173950195, "bridge_gate_avg": 1.012660175561905, "bridge_out_proj_avg": 0.0026557748788036406, "elapsed_s": 763.3245782852173}
|
| 18 |
+
{"step": 1700, "loss": 1.5210776329040527, "loss_mdlm": 1.3097103834152222, "loss_lm": 0.42273443937301636, "loss_ct": 0.0, "lr": 3.3493649053890326e-06, "gnorm": 2.291513681411743, "bridge_gate_avg": 1.0126990377902985, "bridge_out_proj_avg": 0.0026603754959069192, "elapsed_s": 810.6708629131317}
|
| 19 |
+
{"step": 1800, "loss": 1.7654281854629517, "loss_mdlm": 1.3385063409805298, "loss_lm": 0.8538437485694885, "loss_ct": 0.0, "lr": 1.5076844803522922e-06, "gnorm": 3.2395219802856445, "bridge_gate_avg": 1.0127519071102142, "bridge_out_proj_avg": 0.0026628166087903082, "elapsed_s": 857.8076863288879}
|
| 20 |
+
{"step": 1900, "loss": 1.7146199941635132, "loss_mdlm": 1.3556466102600098, "loss_lm": 0.7179467082023621, "loss_ct": 0.0, "lr": 3.7980617469479953e-07, "gnorm": 3.086524724960327, "bridge_gate_avg": 1.0127664506435394, "bridge_out_proj_avg": 0.002664009458385408, "elapsed_s": 896.1228673458099}
|