explcre commited on
Commit
2ab5bea
·
verified ·
1 Parent(s): 3357730

Upload exp_phase8_bridge_HYBRID_v7delta_095628/log.jsonl with huggingface_hub

Browse files
exp_phase8_bridge_HYBRID_v7delta_095628/log.jsonl ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"step": 0, "loss": 1.459015130996704, "loss_mdlm": 1.2730238437652588, "loss_lm": 1.8599127531051636, "loss_ct": 0.0, "lr": 0.0, "gnorm": 1.6453849077224731, "bridge_gate_avg": 1.0, "bridge_out_proj_avg": 0.0, "elapsed_s": 3.5633385181427}
2
+ {"step": 100, "loss": 1.466135025024414, "loss_mdlm": 1.34710693359375, "loss_lm": 1.1902804374694824, "loss_ct": 0.0, "lr": 2.5e-05, "gnorm": 1.6559585332870483, "bridge_gate_avg": 1.000984936952591, "bridge_out_proj_avg": 0.00043443984031910077, "elapsed_s": 52.029250621795654}
3
+ {"step": 200, "loss": 1.4735294580459595, "loss_mdlm": 1.3186297416687012, "loss_lm": 1.5489975214004517, "loss_ct": 0.0, "lr": 5e-05, "gnorm": 1.4552229642868042, "bridge_gate_avg": 1.003785490989685, "bridge_out_proj_avg": 0.001641204085899517, "elapsed_s": 98.84114074707031}
4
+ {"step": 300, "loss": 1.3819844722747803, "loss_mdlm": 1.3156907558441162, "loss_lm": 0.6629365682601929, "loss_ct": 0.0, "lr": 4.962019382530521e-05, "gnorm": 3.867669105529785, "bridge_gate_avg": 1.005186915397644, "bridge_out_proj_avg": 0.001975860068341717, "elapsed_s": 144.70700311660767}
5
+ {"step": 400, "loss": 1.438403606414795, "loss_mdlm": 1.3489402532577515, "loss_lm": 0.8946332335472107, "loss_ct": 0.0, "lr": 4.849231551964771e-05, "gnorm": 2.2528419494628906, "bridge_gate_avg": 1.0056349337100983, "bridge_out_proj_avg": 0.002028127491939813, "elapsed_s": 190.3928611278534}
6
+ {"step": 500, "loss": 1.3966467380523682, "loss_mdlm": 1.319581151008606, "loss_lm": 0.7706555128097534, "loss_ct": 0.0, "lr": 4.665063509461097e-05, "gnorm": 2.22143816947937, "bridge_gate_avg": 1.0067057609558105, "bridge_out_proj_avg": 0.0021065184846520424, "elapsed_s": 235.10117030143738}
7
+ {"step": 600, "loss": 1.392098069190979, "loss_mdlm": 1.3202756643295288, "loss_lm": 0.7182244062423706, "loss_ct": 0.0, "lr": 4.415111107797445e-05, "gnorm": 6.555563926696777, "bridge_gate_avg": 1.0079008042812347, "bridge_out_proj_avg": 0.0021309235016815364, "elapsed_s": 279.6745367050171}
8
+ {"step": 700, "loss": 1.3760024309158325, "loss_mdlm": 1.3010886907577515, "loss_lm": 0.7491374015808105, "loss_ct": 0.0, "lr": 4.1069690242163484e-05, "gnorm": 3.100015163421631, "bridge_gate_avg": 1.0088341534137726, "bridge_out_proj_avg": 0.0021577103761956096, "elapsed_s": 324.5792045593262}
9
+ {"step": 800, "loss": 1.3876187801361084, "loss_mdlm": 1.2835594415664673, "loss_lm": 1.040592908859253, "loss_ct": 0.0, "lr": 3.7500000000000003e-05, "gnorm": 1.832913875579834, "bridge_gate_avg": 1.0091339945793152, "bridge_out_proj_avg": 0.0021757923532277346, "elapsed_s": 369.2768669128418}
10
+ {"step": 900, "loss": 1.3426506519317627, "loss_mdlm": 1.2705268859863281, "loss_lm": 0.7212380170822144, "loss_ct": 0.0, "lr": 3.355050358314172e-05, "gnorm": 2.9468846321105957, "bridge_gate_avg": 1.0096003413200378, "bridge_out_proj_avg": 0.0022130668512545526, "elapsed_s": 414.591406583786}
11
+ {"step": 1000, "loss": 1.383650541305542, "loss_mdlm": 1.3159191608428955, "loss_lm": 0.6773139834403992, "loss_ct": 0.0, "lr": 2.9341204441673266e-05, "gnorm": 3.5543723106384277, "bridge_gate_avg": 1.0098629593849182, "bridge_out_proj_avg": 0.0022431802935898304, "elapsed_s": 459.28358936309814}
12
+ {"step": 1100, "loss": 1.3432815074920654, "loss_mdlm": 1.3093326091766357, "loss_lm": 0.33948925137519836, "loss_ct": 0.0, "lr": 2.5e-05, "gnorm": 3.0365848541259766, "bridge_gate_avg": 1.010056048631668, "bridge_out_proj_avg": 0.002257610089145601, "elapsed_s": 503.7448818683624}
13
+ {"step": 1200, "loss": 1.370711326599121, "loss_mdlm": 1.298753261566162, "loss_lm": 0.7195807695388794, "loss_ct": 0.0, "lr": 2.0658795558326743e-05, "gnorm": 3.5778613090515137, "bridge_gate_avg": 1.0101981461048126, "bridge_out_proj_avg": 0.0022730896016582847, "elapsed_s": 548.4395108222961}
14
+ {"step": 1300, "loss": 1.3849544525146484, "loss_mdlm": 1.3224542140960693, "loss_lm": 0.6250021457672119, "loss_ct": 0.0, "lr": 1.6449496416858284e-05, "gnorm": 1.7962719202041626, "bridge_gate_avg": 1.0103366672992706, "bridge_out_proj_avg": 0.0022854958078823984, "elapsed_s": 593.1945672035217}
15
+ {"step": 1400, "loss": 1.3018178939819336, "loss_mdlm": 1.2309952974319458, "loss_lm": 0.708226203918457, "loss_ct": 0.0, "lr": 1.2500000000000006e-05, "gnorm": 4.803203582763672, "bridge_gate_avg": 1.0104908347129822, "bridge_out_proj_avg": 0.0022991644218564034, "elapsed_s": 637.2096564769745}
16
+ {"step": 1500, "loss": 1.4332282543182373, "loss_mdlm": 1.352264404296875, "loss_lm": 0.8096380233764648, "loss_ct": 0.0, "lr": 8.930309757836517e-06, "gnorm": 2.584556818008423, "bridge_gate_avg": 1.0105810463428497, "bridge_out_proj_avg": 0.0023000610526651144, "elapsed_s": 681.4569041728973}
17
+ {"step": 1600, "loss": 1.3832874298095703, "loss_mdlm": 1.3020309209823608, "loss_lm": 0.8125651478767395, "loss_ct": 0.0, "lr": 5.848888922025553e-06, "gnorm": 2.1836297512054443, "bridge_gate_avg": 1.010673701763153, "bridge_out_proj_avg": 0.0023096834775060415, "elapsed_s": 725.4119546413422}
18
+ {"step": 1700, "loss": 1.3376214504241943, "loss_mdlm": 1.2946021556854248, "loss_lm": 0.43019336462020874, "loss_ct": 0.0, "lr": 3.3493649053890326e-06, "gnorm": 2.0750133991241455, "bridge_gate_avg": 1.0106693804264069, "bridge_out_proj_avg": 0.0023101582773961127, "elapsed_s": 769.5402901172638}
19
+ {"step": 1800, "loss": 1.4180700778961182, "loss_mdlm": 1.331863284111023, "loss_lm": 0.8620678782463074, "loss_ct": 0.0, "lr": 1.5076844803522922e-06, "gnorm": 2.2978219985961914, "bridge_gate_avg": 1.0106909573078156, "bridge_out_proj_avg": 0.002312287804670632, "elapsed_s": 813.5067446231842}
20
+ {"step": 1900, "loss": 1.4467514753341675, "loss_mdlm": 1.3725568056106567, "loss_lm": 0.7419468760490417, "loss_ct": 0.0, "lr": 3.7980617469479953e-07, "gnorm": 1.8832848072052002, "bridge_gate_avg": 1.0107081234455109, "bridge_out_proj_avg": 0.0023133897921070457, "elapsed_s": 857.7561085224152}