Upload exp_phase8_bridge_recipeABC_2000step_035847/log.jsonl with huggingface_hub
Browse files
exp_phase8_bridge_recipeABC_2000step_035847/log.jsonl
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 0, "loss": 2.2065634727478027, "loss_mdlm": 1.282113790512085, "loss_lm": 1.8488996028900146, "loss_ct": 0.0, "lr": 0.0, "gnorm": 2.6921541690826416, "bridge_gate_avg": 1.0, "bridge_out_proj_avg": 0.0, "elapsed_s": 3.3606250286102295}
|
| 2 |
+
{"step": 100, "loss": 1.9266400337219238, "loss_mdlm": 1.338171124458313, "loss_lm": 1.1769376993179321, "loss_ct": 0.0, "lr": 2.5e-05, "gnorm": 3.6858625411987305, "bridge_gate_avg": 1.0007699131965637, "bridge_out_proj_avg": 0.0003300538592156954, "elapsed_s": 52.99418520927429}
|
| 3 |
+
{"step": 200, "loss": 2.0699241161346436, "loss_mdlm": 1.317544937133789, "loss_lm": 1.5047584772109985, "loss_ct": 0.0, "lr": 5e-05, "gnorm": 4.146609783172607, "bridge_gate_avg": 1.0030298829078674, "bridge_out_proj_avg": 0.001296733389608562, "elapsed_s": 101.2505898475647}
|
| 4 |
+
{"step": 300, "loss": 1.6431007385253906, "loss_mdlm": 1.3042664527893066, "loss_lm": 0.677668571472168, "loss_ct": 0.0, "lr": 4.962019382530521e-05, "gnorm": 2.769054651260376, "bridge_gate_avg": 1.004978746175766, "bridge_out_proj_avg": 0.0017723908240441233, "elapsed_s": 148.8913860321045}
|
| 5 |
+
{"step": 400, "loss": 1.7841801643371582, "loss_mdlm": 1.3372082710266113, "loss_lm": 0.8939437866210938, "loss_ct": 0.0, "lr": 4.849231551964771e-05, "gnorm": 3.2198431491851807, "bridge_gate_avg": 1.0055984854698181, "bridge_out_proj_avg": 0.0019099045312032104, "elapsed_s": 196.44344210624695}
|
| 6 |
+
{"step": 500, "loss": 1.7111378908157349, "loss_mdlm": 1.337371826171875, "loss_lm": 0.7475321292877197, "loss_ct": 0.0, "lr": 4.665063509461097e-05, "gnorm": 3.152147054672241, "bridge_gate_avg": 1.0074205994606018, "bridge_out_proj_avg": 0.002055648306850344, "elapsed_s": 243.69884204864502}
|
| 7 |
+
{"step": 600, "loss": 1.7072166204452515, "loss_mdlm": 1.3591734170913696, "loss_lm": 0.6960864067077637, "loss_ct": 0.0, "lr": 4.415111107797445e-05, "gnorm": 2.370748996734619, "bridge_gate_avg": 1.0090404748916626, "bridge_out_proj_avg": 0.0021234675659798086, "elapsed_s": 291.176798582077}
|
| 8 |
+
{"step": 700, "loss": 1.6800892353057861, "loss_mdlm": 1.317908763885498, "loss_lm": 0.7243608236312866, "loss_ct": 0.0, "lr": 4.1069690242163484e-05, "gnorm": 2.6385953426361084, "bridge_gate_avg": 1.0102164149284363, "bridge_out_proj_avg": 0.002204728196375072, "elapsed_s": 338.58538818359375}
|
| 9 |
+
{"step": 800, "loss": 1.8069911003112793, "loss_mdlm": 1.2863882780075073, "loss_lm": 1.041205644607544, "loss_ct": 0.0, "lr": 3.7500000000000003e-05, "gnorm": 3.3424949645996094, "bridge_gate_avg": 1.010936200618744, "bridge_out_proj_avg": 0.0022694036597386003, "elapsed_s": 386.14202642440796}
|
| 10 |
+
{"step": 900, "loss": 1.6764284372329712, "loss_mdlm": 1.318030595779419, "loss_lm": 0.7167956829071045, "loss_ct": 0.0, "lr": 3.355050358314172e-05, "gnorm": 2.7551939487457275, "bridge_gate_avg": 1.011419266462326, "bridge_out_proj_avg": 0.002309815026819706, "elapsed_s": 433.74856758117676}
|
| 11 |
+
{"step": 1000, "loss": 1.6800734996795654, "loss_mdlm": 1.3500856161117554, "loss_lm": 0.6599758267402649, "loss_ct": 0.0, "lr": 2.9341204441673266e-05, "gnorm": 2.0982542037963867, "bridge_gate_avg": 1.0121433734893799, "bridge_out_proj_avg": 0.002354038064368069, "elapsed_s": 481.3239505290985}
|
| 12 |
+
{"step": 1100, "loss": 1.450477957725525, "loss_mdlm": 1.2842185497283936, "loss_lm": 0.3325188457965851, "loss_ct": 0.0, "lr": 2.5e-05, "gnorm": 1.6457056999206543, "bridge_gate_avg": 1.0126015543937683, "bridge_out_proj_avg": 0.002403747756034136, "elapsed_s": 528.8631024360657}
|
| 13 |
+
{"step": 1200, "loss": 1.637521505355835, "loss_mdlm": 1.2975401878356934, "loss_lm": 0.6799626350402832, "loss_ct": 0.0, "lr": 2.0658795558326743e-05, "gnorm": 2.2360928058624268, "bridge_gate_avg": 1.0129742920398712, "bridge_out_proj_avg": 0.0024374136119149625, "elapsed_s": 576.4955825805664}
|
| 14 |
+
{"step": 1300, "loss": 1.6057547330856323, "loss_mdlm": 1.2936149835586548, "loss_lm": 0.6242794990539551, "loss_ct": 0.0, "lr": 1.6449496416858284e-05, "gnorm": 2.8162965774536133, "bridge_gate_avg": 1.013480544090271, "bridge_out_proj_avg": 0.0024709103163331747, "elapsed_s": 623.9490346908569}
|
| 15 |
+
{"step": 1400, "loss": 1.5927040576934814, "loss_mdlm": 1.2519505023956299, "loss_lm": 0.6815072298049927, "loss_ct": 0.0, "lr": 1.2500000000000006e-05, "gnorm": 3.0524463653564453, "bridge_gate_avg": 1.0137510895729065, "bridge_out_proj_avg": 0.0024953692336566746, "elapsed_s": 671.2380285263062}
|
| 16 |
+
{"step": 1500, "loss": 1.7208943367004395, "loss_mdlm": 1.316318392753601, "loss_lm": 0.8091519474983215, "loss_ct": 0.0, "lr": 8.930309757836517e-06, "gnorm": 2.526066780090332, "bridge_gate_avg": 1.0138984322547913, "bridge_out_proj_avg": 0.0025060652405954897, "elapsed_s": 718.6398675441742}
|
| 17 |
+
{"step": 1600, "loss": 1.7422128915786743, "loss_mdlm": 1.3382477760314941, "loss_lm": 0.8079302310943604, "loss_ct": 0.0, "lr": 5.848888922025553e-06, "gnorm": 2.6460342407226562, "bridge_gate_avg": 1.0140718817710876, "bridge_out_proj_avg": 0.0025181745877489448, "elapsed_s": 766.0576133728027}
|
| 18 |
+
{"step": 1700, "loss": 1.5278103351593018, "loss_mdlm": 1.3192803859710693, "loss_lm": 0.41705983877182007, "loss_ct": 0.0, "lr": 3.3493649053890326e-06, "gnorm": 2.2176411151885986, "bridge_gate_avg": 1.0141394436359406, "bridge_out_proj_avg": 0.0025267652235925198, "elapsed_s": 813.5030243396759}
|
| 19 |
+
{"step": 1800, "loss": 1.7481588125228882, "loss_mdlm": 1.3209232091903687, "loss_lm": 0.8544712662696838, "loss_ct": 0.0, "lr": 1.5076844803522922e-06, "gnorm": 3.188605546951294, "bridge_gate_avg": 1.0141636431217194, "bridge_out_proj_avg": 0.00252881390042603, "elapsed_s": 861.0775835514069}
|
| 20 |
+
{"step": 1900, "loss": 1.683616042137146, "loss_mdlm": 1.3264950513839722, "loss_lm": 0.7142419219017029, "loss_ct": 0.0, "lr": 3.7980617469479953e-07, "gnorm": 3.0813472270965576, "bridge_gate_avg": 1.0141851305961609, "bridge_out_proj_avg": 0.002530483645386994, "elapsed_s": 898.9086828231812}
|