Upload exp_phase8_bridge_T3_recipeABC_111944/log.jsonl with huggingface_hub
Browse files
exp_phase8_bridge_T3_recipeABC_111944/log.jsonl
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"step": 0, "loss": 1.542597770690918, "loss_mdlm": 1.281509518623352, "loss_lm": 2.61088228225708, "loss_ct": 0.0, "lr": 0.0, "gnorm": 1.6943250894546509, "bridge_gate_avg": 1.0, "bridge_out_proj_avg": 0.0, "elapsed_s": 3.3380026817321777}
|
| 2 |
+
{"step": 100, "loss": 1.5235168933868408, "loss_mdlm": 1.3427046537399292, "loss_lm": 1.8081226348876953, "loss_ct": 0.0, "lr": 2.5e-05, "gnorm": 0.9331293106079102, "bridge_gate_avg": 1.0005252063274384, "bridge_out_proj_avg": 0.0002232895203633234, "elapsed_s": 39.34641909599304}
|
| 3 |
+
{"step": 200, "loss": 1.5069659948349, "loss_mdlm": 1.3362873792648315, "loss_lm": 1.7067866325378418, "loss_ct": 0.0, "lr": 5e-05, "gnorm": 0.9870486259460449, "bridge_gate_avg": 1.0028347373008728, "bridge_out_proj_avg": 0.0007693533261772245, "elapsed_s": 73.05906653404236}
|
| 4 |
+
{"step": 300, "loss": 1.073629379272461, "loss_mdlm": 0.961466908454895, "loss_lm": 1.1216249465942383, "loss_ct": 0.0, "lr": 4.962019382530521e-05, "gnorm": 0.9526773691177368, "bridge_gate_avg": 1.004991352558136, "bridge_out_proj_avg": 0.0012081714230589569, "elapsed_s": 106.41354727745056}
|
| 5 |
+
{"step": 400, "loss": 1.4774343967437744, "loss_mdlm": 1.353020191192627, "loss_lm": 1.2441420555114746, "loss_ct": 0.0, "lr": 4.849231551964771e-05, "gnorm": 0.9193490743637085, "bridge_gate_avg": 1.006550371646881, "bridge_out_proj_avg": 0.0014543899742420763, "elapsed_s": 139.05843424797058}
|
| 6 |
+
{"step": 500, "loss": 1.2641733884811401, "loss_mdlm": 1.1470997333526611, "loss_lm": 1.1707361936569214, "loss_ct": 0.0, "lr": 4.665063509461097e-05, "gnorm": 0.675282895565033, "bridge_gate_avg": 1.0073183178901672, "bridge_out_proj_avg": 0.0015557292208541185, "elapsed_s": 171.0628147125244}
|
| 7 |
+
{"step": 600, "loss": 1.0715910196304321, "loss_mdlm": 0.9747456908226013, "loss_lm": 0.968453049659729, "loss_ct": 0.0, "lr": 4.415111107797445e-05, "gnorm": 0.7520832419395447, "bridge_gate_avg": 1.0087614357471466, "bridge_out_proj_avg": 0.0016444586508441716, "elapsed_s": 203.00878524780273}
|
| 8 |
+
{"step": 700, "loss": 1.478933334350586, "loss_mdlm": 1.3748466968536377, "loss_lm": 1.0408666133880615, "loss_ct": 0.0, "lr": 4.1069690242163484e-05, "gnorm": 0.7312368154525757, "bridge_gate_avg": 1.009645015001297, "bridge_out_proj_avg": 0.0016992974269669503, "elapsed_s": 234.72299194335938}
|
| 9 |
+
{"step": 800, "loss": 1.4035696983337402, "loss_mdlm": 1.331003189086914, "loss_lm": 0.7256649136543274, "loss_ct": 0.0, "lr": 3.7500000000000003e-05, "gnorm": 0.7493772506713867, "bridge_gate_avg": 1.0100778937339783, "bridge_out_proj_avg": 0.0017440000956412405, "elapsed_s": 266.52451181411743}
|
| 10 |
+
{"step": 900, "loss": 1.4200632572174072, "loss_mdlm": 1.3291752338409424, "loss_lm": 0.9088806509971619, "loss_ct": 0.0, "lr": 3.355050358314172e-05, "gnorm": 0.6524292826652527, "bridge_gate_avg": 1.0108168721199036, "bridge_out_proj_avg": 0.001800286874640733, "elapsed_s": 297.9191884994507}
|
| 11 |
+
{"step": 1000, "loss": 1.286969542503357, "loss_mdlm": 1.2108865976333618, "loss_lm": 0.7608296871185303, "loss_ct": 0.0, "lr": 2.9341204441673266e-05, "gnorm": 0.6426944136619568, "bridge_gate_avg": 1.0113093256950378, "bridge_out_proj_avg": 0.0018184005457442254, "elapsed_s": 329.5657699108124}
|
| 12 |
+
{"step": 1100, "loss": 1.3956308364868164, "loss_mdlm": 1.2909878492355347, "loss_lm": 1.0464292764663696, "loss_ct": 0.0, "lr": 2.5e-05, "gnorm": 0.637485682964325, "bridge_gate_avg": 1.0118263363838196, "bridge_out_proj_avg": 0.0018582225020509213, "elapsed_s": 361.34737396240234}
|
| 13 |
+
{"step": 1200, "loss": 1.3978227376937866, "loss_mdlm": 1.2971391677856445, "loss_lm": 1.0068355798721313, "loss_ct": 0.0, "lr": 2.0658795558326743e-05, "gnorm": 0.6686980724334717, "bridge_gate_avg": 1.012482911348343, "bridge_out_proj_avg": 0.0018914339016191661, "elapsed_s": 392.3453314304352}
|
| 14 |
+
{"step": 1300, "loss": 1.4036799669265747, "loss_mdlm": 1.3334825038909912, "loss_lm": 0.7019744515419006, "loss_ct": 0.0, "lr": 1.6449496416858284e-05, "gnorm": 0.5687031745910645, "bridge_gate_avg": 1.012610912322998, "bridge_out_proj_avg": 0.0019027607049793005, "elapsed_s": 423.5393934249878}
|
| 15 |
+
{"step": 1400, "loss": 1.4413402080535889, "loss_mdlm": 1.3778777122497559, "loss_lm": 0.6346246004104614, "loss_ct": 0.0, "lr": 1.2500000000000006e-05, "gnorm": 0.6606745719909668, "bridge_gate_avg": 1.0128365755081177, "bridge_out_proj_avg": 0.0019035480509046465, "elapsed_s": 454.91157507896423}
|
| 16 |
+
{"step": 1500, "loss": 1.232775330543518, "loss_mdlm": 1.1631433963775635, "loss_lm": 0.6963193416595459, "loss_ct": 0.0, "lr": 8.930309757836517e-06, "gnorm": 0.5367175340652466, "bridge_gate_avg": 1.0129780769348145, "bridge_out_proj_avg": 0.001907850819407031, "elapsed_s": 485.892502784729}
|
| 17 |
+
{"step": 1600, "loss": 1.4369968175888062, "loss_mdlm": 1.3665934801101685, "loss_lm": 0.704033374786377, "loss_ct": 0.0, "lr": 5.848888922025553e-06, "gnorm": 0.6619663834571838, "bridge_gate_avg": 1.013072669506073, "bridge_out_proj_avg": 0.0019127265259157866, "elapsed_s": 516.7589004039764}
|
| 18 |
+
{"step": 1700, "loss": 1.3729010820388794, "loss_mdlm": 1.3075687885284424, "loss_lm": 0.6533230543136597, "loss_ct": 0.0, "lr": 3.3493649053890326e-06, "gnorm": 0.5693467259407043, "bridge_gate_avg": 1.0131173133850098, "bridge_out_proj_avg": 0.001913795422296971, "elapsed_s": 547.7233114242554}
|
| 19 |
+
{"step": 1800, "loss": 1.3958165645599365, "loss_mdlm": 1.2831954956054688, "loss_lm": 1.1262105703353882, "loss_ct": 0.0, "lr": 1.5076844803522922e-06, "gnorm": 0.7861090302467346, "bridge_gate_avg": 1.0131647884845734, "bridge_out_proj_avg": 0.001915005937917158, "elapsed_s": 578.6705870628357}
|
| 20 |
+
{"step": 1900, "loss": 1.3560594320297241, "loss_mdlm": 1.3020555973052979, "loss_lm": 0.5400379300117493, "loss_ct": 0.0, "lr": 3.7980617469479953e-07, "gnorm": 0.6299136877059937, "bridge_gate_avg": 1.0131788551807404, "bridge_out_proj_avg": 0.0019152238091919571, "elapsed_s": 609.5615367889404}
|