|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 4.0, |
|
"eval_steps": 100, |
|
"global_step": 1540, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 1.282051282051282e-07, |
|
"logits/chosen": -1.7278180122375488, |
|
"logits/rejected": -1.7377450466156006, |
|
"logps/chosen": -29.553977966308594, |
|
"logps/rejected": -42.813133239746094, |
|
"loss": 1.0, |
|
"rewards/accuracies": 0.0, |
|
"rewards/chosen": 0.0, |
|
"rewards/margins": 0.0, |
|
"rewards/rejected": 0.0, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 1.282051282051282e-06, |
|
"logits/chosen": -1.8664803504943848, |
|
"logits/rejected": -1.8707994222640991, |
|
"logps/chosen": -36.978511810302734, |
|
"logps/rejected": -33.66939163208008, |
|
"loss": 0.9993, |
|
"rewards/accuracies": 0.5694444179534912, |
|
"rewards/chosen": 0.00028087408281862736, |
|
"rewards/margins": 0.0006740752141922712, |
|
"rewards/rejected": -0.00039320107316598296, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 2.564102564102564e-06, |
|
"logits/chosen": -1.9984451532363892, |
|
"logits/rejected": -2.0010995864868164, |
|
"logps/chosen": -29.63176918029785, |
|
"logps/rejected": -29.05954933166504, |
|
"loss": 0.9999, |
|
"rewards/accuracies": 0.48750001192092896, |
|
"rewards/chosen": 0.00010425634536659345, |
|
"rewards/margins": 6.528960511786863e-05, |
|
"rewards/rejected": 3.8966707506915554e-05, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 3.846153846153847e-06, |
|
"logits/chosen": -1.9210799932479858, |
|
"logits/rejected": -1.9183847904205322, |
|
"logps/chosen": -31.414783477783203, |
|
"logps/rejected": -33.19659423828125, |
|
"loss": 1.0002, |
|
"rewards/accuracies": 0.4749999940395355, |
|
"rewards/chosen": 1.3138540452928282e-05, |
|
"rewards/margins": -0.00020548875909298658, |
|
"rewards/rejected": 0.00021862727589905262, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 4.999896948438434e-06, |
|
"logits/chosen": -2.0177221298217773, |
|
"logits/rejected": -2.008965492248535, |
|
"logps/chosen": -32.57322311401367, |
|
"logps/rejected": -32.500308990478516, |
|
"loss": 1.0001, |
|
"rewards/accuracies": 0.4749999940395355, |
|
"rewards/chosen": 3.584356090868823e-05, |
|
"rewards/margins": -8.777440234553069e-05, |
|
"rewards/rejected": 0.00012361796689219773, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 4.987541037542187e-06, |
|
"logits/chosen": -1.8622690439224243, |
|
"logits/rejected": -1.851509690284729, |
|
"logps/chosen": -33.547603607177734, |
|
"logps/rejected": -35.463592529296875, |
|
"loss": 0.9998, |
|
"rewards/accuracies": 0.5, |
|
"rewards/chosen": 9.876764670480043e-05, |
|
"rewards/margins": 0.0002112251240760088, |
|
"rewards/rejected": -0.00011245747737120837, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 4.954691471941119e-06, |
|
"logits/chosen": -1.9400131702423096, |
|
"logits/rejected": -1.9419806003570557, |
|
"logps/chosen": -32.52842330932617, |
|
"logps/rejected": -33.22877883911133, |
|
"loss": 0.9987, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.0007175664068199694, |
|
"rewards/margins": 0.0012915965635329485, |
|
"rewards/rejected": -0.00057403021492064, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 4.901618883413549e-06, |
|
"logits/chosen": -2.070552349090576, |
|
"logits/rejected": -2.0755274295806885, |
|
"logps/chosen": -34.00461959838867, |
|
"logps/rejected": -36.64922332763672, |
|
"loss": 0.9994, |
|
"rewards/accuracies": 0.612500011920929, |
|
"rewards/chosen": -0.00016131921438500285, |
|
"rewards/margins": 0.0006422021542675793, |
|
"rewards/rejected": -0.0008035213686525822, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.828760511501322e-06, |
|
"logits/chosen": -1.9306777715682983, |
|
"logits/rejected": -1.9338254928588867, |
|
"logps/chosen": -34.32624816894531, |
|
"logps/rejected": -34.661468505859375, |
|
"loss": 0.9983, |
|
"rewards/accuracies": 0.612500011920929, |
|
"rewards/chosen": 0.0010156143689528108, |
|
"rewards/margins": 0.0017062196275219321, |
|
"rewards/rejected": -0.0006906053749844432, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 4.7367166013034295e-06, |
|
"logits/chosen": -1.9389193058013916, |
|
"logits/rejected": -1.9434226751327515, |
|
"logps/chosen": -32.38957214355469, |
|
"logps/rejected": -32.348140716552734, |
|
"loss": 0.9993, |
|
"rewards/accuracies": 0.5874999761581421, |
|
"rewards/chosen": 0.0007785108755342662, |
|
"rewards/margins": 0.0006708315922878683, |
|
"rewards/rejected": 0.00010767912317533046, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 4.626245458345211e-06, |
|
"logits/chosen": -2.0358777046203613, |
|
"logits/rejected": -2.0339014530181885, |
|
"logps/chosen": -32.13254165649414, |
|
"logps/rejected": -31.29019546508789, |
|
"loss": 0.9987, |
|
"rewards/accuracies": 0.612500011920929, |
|
"rewards/chosen": 0.0010829826351255178, |
|
"rewards/margins": 0.001322855008766055, |
|
"rewards/rejected": -0.0002398724900558591, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"eval_logits/chosen": -2.2312774658203125, |
|
"eval_logits/rejected": -2.226422071456909, |
|
"eval_logps/chosen": -34.04991149902344, |
|
"eval_logps/rejected": -37.55283737182617, |
|
"eval_loss": 0.9997907280921936, |
|
"eval_rewards/accuracies": 0.5336378812789917, |
|
"eval_rewards/chosen": -0.0001535558985779062, |
|
"eval_rewards/margins": 0.00020861340453848243, |
|
"eval_rewards/rejected": -0.00036216925946064293, |
|
"eval_runtime": 146.0254, |
|
"eval_samples_per_second": 2.349, |
|
"eval_steps_per_second": 0.294, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 4.498257201263691e-06, |
|
"logits/chosen": -1.9907060861587524, |
|
"logits/rejected": -1.9883339405059814, |
|
"logps/chosen": -33.13169860839844, |
|
"logps/rejected": -34.033958435058594, |
|
"loss": 0.9988, |
|
"rewards/accuracies": 0.6875, |
|
"rewards/chosen": 0.0011211589444428682, |
|
"rewards/margins": 0.0011670273961499333, |
|
"rewards/rejected": -4.586850991472602e-05, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 4.353806263777678e-06, |
|
"logits/chosen": -2.002023458480835, |
|
"logits/rejected": -1.993699312210083, |
|
"logps/chosen": -32.341697692871094, |
|
"logps/rejected": -32.16511917114258, |
|
"loss": 0.9989, |
|
"rewards/accuracies": 0.612500011920929, |
|
"rewards/chosen": 0.0010320657165721059, |
|
"rewards/margins": 0.001068194629624486, |
|
"rewards/rejected": -3.612901855376549e-05, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 4.1940827077152755e-06, |
|
"logits/chosen": -2.028505802154541, |
|
"logits/rejected": -2.020526885986328, |
|
"logps/chosen": -30.3519287109375, |
|
"logps/rejected": -32.101314544677734, |
|
"loss": 0.9983, |
|
"rewards/accuracies": 0.612500011920929, |
|
"rewards/chosen": 0.0012378758983686566, |
|
"rewards/margins": 0.001720982021652162, |
|
"rewards/rejected": -0.0004831062688026577, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 4.0204024186666215e-06, |
|
"logits/chosen": -1.9588673114776611, |
|
"logits/rejected": -1.9690834283828735, |
|
"logps/chosen": -31.205490112304688, |
|
"logps/rejected": -32.55961608886719, |
|
"loss": 0.9976, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.0021417266689240932, |
|
"rewards/margins": 0.0023890691809356213, |
|
"rewards/rejected": -0.00024734257021918893, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 3.834196265035119e-06, |
|
"logits/chosen": -1.8695415258407593, |
|
"logits/rejected": -1.8707062005996704, |
|
"logps/chosen": -33.88127899169922, |
|
"logps/rejected": -34.7686653137207, |
|
"loss": 0.9968, |
|
"rewards/accuracies": 0.574999988079071, |
|
"rewards/chosen": 0.0031141345389187336, |
|
"rewards/margins": 0.003228238318115473, |
|
"rewards/rejected": -0.00011410393926780671, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 3.636998309800573e-06, |
|
"logits/chosen": -1.9212032556533813, |
|
"logits/rejected": -1.9178003072738647, |
|
"logps/chosen": -35.99773406982422, |
|
"logps/rejected": -32.705848693847656, |
|
"loss": 0.9984, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.0017394202295690775, |
|
"rewards/margins": 0.0015704210381954908, |
|
"rewards/rejected": 0.00016899927868507802, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 3.4304331721118078e-06, |
|
"logits/chosen": -2.0206995010375977, |
|
"logits/rejected": -2.0133931636810303, |
|
"logps/chosen": -33.504085540771484, |
|
"logps/rejected": -31.432220458984375, |
|
"loss": 0.9963, |
|
"rewards/accuracies": 0.637499988079071, |
|
"rewards/chosen": 0.0029377774335443974, |
|
"rewards/margins": 0.003741443855687976, |
|
"rewards/rejected": -0.0008036663057282567, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 3.2162026428305436e-06, |
|
"logits/chosen": -2.0269291400909424, |
|
"logits/rejected": -2.032160997390747, |
|
"logps/chosen": -32.24355697631836, |
|
"logps/rejected": -32.431182861328125, |
|
"loss": 0.9975, |
|
"rewards/accuracies": 0.625, |
|
"rewards/chosen": 0.003012270200997591, |
|
"rewards/margins": 0.0024727012496441603, |
|
"rewards/rejected": 0.0005395688931457698, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 2.996071664294641e-06, |
|
"logits/chosen": -2.027367115020752, |
|
"logits/rejected": -2.0246078968048096, |
|
"logps/chosen": -31.290613174438477, |
|
"logps/rejected": -31.361133575439453, |
|
"loss": 0.9976, |
|
"rewards/accuracies": 0.625, |
|
"rewards/chosen": 0.0019833946134895086, |
|
"rewards/margins": 0.002412599278613925, |
|
"rewards/rejected": -0.00042920451960526407, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 2.7718537898066833e-06, |
|
"logits/chosen": -1.8985168933868408, |
|
"logits/rejected": -1.903148889541626, |
|
"logps/chosen": -31.30405616760254, |
|
"logps/rejected": -32.838443756103516, |
|
"loss": 0.9965, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": 0.0030222723726183176, |
|
"rewards/margins": 0.0034973658621311188, |
|
"rewards/rejected": -0.0004750936641357839, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"eval_logits/chosen": -2.225177764892578, |
|
"eval_logits/rejected": -2.2203547954559326, |
|
"eval_logps/chosen": -34.06184387207031, |
|
"eval_logps/rejected": -37.579010009765625, |
|
"eval_loss": 0.9996482133865356, |
|
"eval_rewards/accuracies": 0.5070598125457764, |
|
"eval_rewards/chosen": -0.0002729461120907217, |
|
"eval_rewards/margins": 0.00035095339990220964, |
|
"eval_rewards/rejected": -0.0006238995119929314, |
|
"eval_runtime": 145.704, |
|
"eval_samples_per_second": 2.354, |
|
"eval_steps_per_second": 0.295, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 2.5453962426402006e-06, |
|
"logits/chosen": -2.011120557785034, |
|
"logits/rejected": -2.021751880645752, |
|
"logps/chosen": -31.745685577392578, |
|
"logps/rejected": -33.96772003173828, |
|
"loss": 0.9967, |
|
"rewards/accuracies": 0.637499988079071, |
|
"rewards/chosen": 0.0022561827208846807, |
|
"rewards/margins": 0.003306365106254816, |
|
"rewards/rejected": -0.0010501822689548135, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 2.3185646976551794e-06, |
|
"logits/chosen": -1.903857946395874, |
|
"logits/rejected": -1.918621301651001, |
|
"logps/chosen": -29.797290802001953, |
|
"logps/rejected": -31.628814697265625, |
|
"loss": 0.9962, |
|
"rewards/accuracies": 0.7124999761581421, |
|
"rewards/chosen": 0.0031425058841705322, |
|
"rewards/margins": 0.0038488968275487423, |
|
"rewards/rejected": -0.0007063907687552273, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 2.0932279108998323e-06, |
|
"logits/chosen": -1.9593979120254517, |
|
"logits/rejected": -1.9633464813232422, |
|
"logps/chosen": -33.067623138427734, |
|
"logps/rejected": -31.64206886291504, |
|
"loss": 0.9956, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": 0.003498472273349762, |
|
"rewards/margins": 0.004426136147230864, |
|
"rewards/rejected": -0.0009276636992581189, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 1.8712423238279358e-06, |
|
"logits/chosen": -1.9572632312774658, |
|
"logits/rejected": -1.9354908466339111, |
|
"logps/chosen": -33.843727111816406, |
|
"logps/rejected": -35.1453742980957, |
|
"loss": 0.995, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.0031002266332507133, |
|
"rewards/margins": 0.0050460235215723515, |
|
"rewards/rejected": -0.0019457967719063163, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 1.6544367689701824e-06, |
|
"logits/chosen": -1.9997854232788086, |
|
"logits/rejected": -1.9964803457260132, |
|
"logps/chosen": -32.75019454956055, |
|
"logps/rejected": -36.28661346435547, |
|
"loss": 0.9976, |
|
"rewards/accuracies": 0.637499988079071, |
|
"rewards/chosen": 0.0017487213481217623, |
|
"rewards/margins": 0.0023655896075069904, |
|
"rewards/rejected": -0.0006168682011775672, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 1.4445974030621963e-06, |
|
"logits/chosen": -1.8673791885375977, |
|
"logits/rejected": -1.8649587631225586, |
|
"logps/chosen": -34.018226623535156, |
|
"logps/rejected": -35.539276123046875, |
|
"loss": 0.998, |
|
"rewards/accuracies": 0.612500011920929, |
|
"rewards/chosen": 0.001701725646853447, |
|
"rewards/margins": 0.002048287307843566, |
|
"rewards/rejected": -0.0003465614281594753, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 1.243452991757889e-06, |
|
"logits/chosen": -1.8522275686264038, |
|
"logits/rejected": -1.849872350692749, |
|
"logps/chosen": -34.16339874267578, |
|
"logps/rejected": -31.845317840576172, |
|
"loss": 0.9969, |
|
"rewards/accuracies": 0.612500011920929, |
|
"rewards/chosen": 0.0023420110810548067, |
|
"rewards/margins": 0.0030819105450063944, |
|
"rewards/rejected": -0.0007398994639515877, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 1.0526606671603523e-06, |
|
"logits/chosen": -1.9549518823623657, |
|
"logits/rejected": -1.94447922706604, |
|
"logps/chosen": -35.027687072753906, |
|
"logps/rejected": -31.895471572875977, |
|
"loss": 0.9962, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.003289591521024704, |
|
"rewards/margins": 0.003834384260699153, |
|
"rewards/rejected": -0.0005447928560897708, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 8.737922755071455e-07, |
|
"logits/chosen": -2.0498766899108887, |
|
"logits/rejected": -2.034980297088623, |
|
"logps/chosen": -30.72440528869629, |
|
"logps/rejected": -32.658695220947266, |
|
"loss": 0.9979, |
|
"rewards/accuracies": 0.637499988079071, |
|
"rewards/chosen": 0.0019971781875938177, |
|
"rewards/margins": 0.002065772656351328, |
|
"rewards/rejected": -6.859400309622288e-05, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 7.08321427484816e-07, |
|
"logits/chosen": -1.9201946258544922, |
|
"logits/rejected": -1.9177051782608032, |
|
"logps/chosen": -32.3183479309082, |
|
"logps/rejected": -30.95510482788086, |
|
"loss": 0.9925, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": 0.006087628658860922, |
|
"rewards/margins": 0.007470599375665188, |
|
"rewards/rejected": -0.0013829706003889441, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"eval_logits/chosen": -2.2213804721832275, |
|
"eval_logits/rejected": -2.216555118560791, |
|
"eval_logps/chosen": -34.083614349365234, |
|
"eval_logps/rejected": -37.60634994506836, |
|
"eval_loss": 0.999591052532196, |
|
"eval_rewards/accuracies": 0.559385359287262, |
|
"eval_rewards/chosen": -0.0004906260874122381, |
|
"eval_rewards/margins": 0.00040669209556654096, |
|
"eval_rewards/rejected": -0.0008973181829787791, |
|
"eval_runtime": 145.8707, |
|
"eval_samples_per_second": 2.351, |
|
"eval_steps_per_second": 0.295, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 4.84533120650964e-06, |
|
"logits/chosen": -1.9058300256729126, |
|
"logits/rejected": -1.902580976486206, |
|
"logps/chosen": -31.3160400390625, |
|
"logps/rejected": -33.86355972290039, |
|
"loss": 0.9959, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": 0.0028731029015034437, |
|
"rewards/margins": 0.004087135195732117, |
|
"rewards/rejected": -0.0012140319449827075, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 4.825108134172131e-06, |
|
"logits/chosen": -1.9523051977157593, |
|
"logits/rejected": -1.9401578903198242, |
|
"logps/chosen": -34.279632568359375, |
|
"logps/rejected": -33.72241973876953, |
|
"loss": 0.995, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": 0.003164437832310796, |
|
"rewards/margins": 0.005000208970159292, |
|
"rewards/rejected": -0.0018357712542638183, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 4.80369052967602e-06, |
|
"logits/chosen": -1.985467553138733, |
|
"logits/rejected": -1.9840377569198608, |
|
"logps/chosen": -33.06474685668945, |
|
"logps/rejected": -32.610504150390625, |
|
"loss": 0.9945, |
|
"rewards/accuracies": 0.737500011920929, |
|
"rewards/chosen": 0.004210514482110739, |
|
"rewards/margins": 0.005519717000424862, |
|
"rewards/rejected": -0.0013092018198221922, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 4.781089396387968e-06, |
|
"logits/chosen": -2.0711922645568848, |
|
"logits/rejected": -2.0555691719055176, |
|
"logps/chosen": -33.74380874633789, |
|
"logps/rejected": -33.10470199584961, |
|
"loss": 0.9958, |
|
"rewards/accuracies": 0.637499988079071, |
|
"rewards/chosen": 0.0043558296747505665, |
|
"rewards/margins": 0.004222516901791096, |
|
"rewards/rejected": 0.00013331293303053826, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 4.757316345716554e-06, |
|
"logits/chosen": -1.94492506980896, |
|
"logits/rejected": -1.9441289901733398, |
|
"logps/chosen": -32.790000915527344, |
|
"logps/rejected": -32.51346969604492, |
|
"loss": 0.9946, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": 0.004920940846204758, |
|
"rewards/margins": 0.005404182709753513, |
|
"rewards/rejected": -0.0004832421545870602, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 4.73238359114687e-06, |
|
"logits/chosen": -1.8960033655166626, |
|
"logits/rejected": -1.9061963558197021, |
|
"logps/chosen": -31.744653701782227, |
|
"logps/rejected": -35.42363739013672, |
|
"loss": 0.9941, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.004600368440151215, |
|
"rewards/margins": 0.0059491791762411594, |
|
"rewards/rejected": -0.0013488102704286575, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 4.706303941965804e-06, |
|
"logits/chosen": -2.029266834259033, |
|
"logits/rejected": -2.0229573249816895, |
|
"logps/chosen": -33.29435729980469, |
|
"logps/rejected": -29.330352783203125, |
|
"loss": 0.9954, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": 0.003576122224330902, |
|
"rewards/margins": 0.004615402314811945, |
|
"rewards/rejected": -0.0010392797412350774, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 4.679090796681225e-06, |
|
"logits/chosen": -1.8853864669799805, |
|
"logits/rejected": -1.8876174688339233, |
|
"logps/chosen": -33.62360382080078, |
|
"logps/rejected": -31.0107479095459, |
|
"loss": 0.9925, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": 0.00615445151925087, |
|
"rewards/margins": 0.007535641081631184, |
|
"rewards/rejected": -0.0013811895623803139, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.650758136138454e-06, |
|
"logits/chosen": -1.9119482040405273, |
|
"logits/rejected": -1.910658597946167, |
|
"logps/chosen": -33.74114227294922, |
|
"logps/rejected": -36.10308074951172, |
|
"loss": 0.9891, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": 0.006436466239392757, |
|
"rewards/margins": 0.010907178744673729, |
|
"rewards/rejected": -0.004470714367926121, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 4.621320516337559e-06, |
|
"logits/chosen": -1.845229148864746, |
|
"logits/rejected": -1.8368085622787476, |
|
"logps/chosen": -30.943897247314453, |
|
"logps/rejected": -36.51512908935547, |
|
"loss": 0.986, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.00830076914280653, |
|
"rewards/margins": 0.014046055264770985, |
|
"rewards/rejected": -0.005745284259319305, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"eval_logits/chosen": -2.1925649642944336, |
|
"eval_logits/rejected": -2.1877267360687256, |
|
"eval_logps/chosen": -34.17170333862305, |
|
"eval_logps/rejected": -37.71051788330078, |
|
"eval_loss": 0.9994248747825623, |
|
"eval_rewards/accuracies": 0.5211793780326843, |
|
"eval_rewards/chosen": -0.0013715263921767473, |
|
"eval_rewards/margins": 0.0005674380226992071, |
|
"eval_rewards/rejected": -0.0019389643566682935, |
|
"eval_runtime": 146.3059, |
|
"eval_samples_per_second": 2.344, |
|
"eval_steps_per_second": 0.294, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 4.590793060955158e-06, |
|
"logits/chosen": -2.012714385986328, |
|
"logits/rejected": -2.015584945678711, |
|
"logps/chosen": -32.184898376464844, |
|
"logps/rejected": -35.35422897338867, |
|
"loss": 0.9857, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": 0.00869043916463852, |
|
"rewards/margins": 0.014304302632808685, |
|
"rewards/rejected": -0.005613864399492741, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 4.559191453574582e-06, |
|
"logits/chosen": -1.8472423553466797, |
|
"logits/rejected": -1.8458467721939087, |
|
"logps/chosen": -28.332477569580078, |
|
"logps/rejected": -32.81929397583008, |
|
"loss": 0.9876, |
|
"rewards/accuracies": 0.9125000238418579, |
|
"rewards/chosen": 0.0073507861234247684, |
|
"rewards/margins": 0.012367108836770058, |
|
"rewards/rejected": -0.005016324110329151, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 4.52653192962838e-06, |
|
"logits/chosen": -1.8031494617462158, |
|
"logits/rejected": -1.796241044998169, |
|
"logps/chosen": -33.12567901611328, |
|
"logps/rejected": -34.58681106567383, |
|
"loss": 0.9871, |
|
"rewards/accuracies": 0.9125000238418579, |
|
"rewards/chosen": 0.0094840619713068, |
|
"rewards/margins": 0.012899428606033325, |
|
"rewards/rejected": -0.003415366169065237, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 4.492831268057307e-06, |
|
"logits/chosen": -1.9694416522979736, |
|
"logits/rejected": -1.9643234014511108, |
|
"logps/chosen": -30.745025634765625, |
|
"logps/rejected": -32.631690979003906, |
|
"loss": 0.983, |
|
"rewards/accuracies": 0.9125000238418579, |
|
"rewards/chosen": 0.010382669046521187, |
|
"rewards/margins": 0.017023462802171707, |
|
"rewards/rejected": -0.0066407956182956696, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 4.458106782690094e-06, |
|
"logits/chosen": -1.8479034900665283, |
|
"logits/rejected": -1.8521947860717773, |
|
"logps/chosen": -33.47832107543945, |
|
"logps/rejected": -33.297340393066406, |
|
"loss": 0.9825, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": 0.010375277139246464, |
|
"rewards/margins": 0.017462292686104774, |
|
"rewards/rejected": -0.007087015546858311, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 4.422376313348405e-06, |
|
"logits/chosen": -1.848722219467163, |
|
"logits/rejected": -1.8431631326675415, |
|
"logps/chosen": -34.28997039794922, |
|
"logps/rejected": -35.908599853515625, |
|
"loss": 0.9791, |
|
"rewards/accuracies": 0.925000011920929, |
|
"rewards/chosen": 0.011295564472675323, |
|
"rewards/margins": 0.0208906102925539, |
|
"rewards/rejected": -0.009595044888556004, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 4.3856582166815696e-06, |
|
"logits/chosen": -1.8663924932479858, |
|
"logits/rejected": -1.8662691116333008, |
|
"logps/chosen": -33.07716369628906, |
|
"logps/rejected": -34.7927131652832, |
|
"loss": 0.983, |
|
"rewards/accuracies": 0.875, |
|
"rewards/chosen": 0.010911665856838226, |
|
"rewards/margins": 0.016983961686491966, |
|
"rewards/rejected": -0.006072297692298889, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 4.347971356735789e-06, |
|
"logits/chosen": -1.9087026119232178, |
|
"logits/rejected": -1.8900855779647827, |
|
"logps/chosen": -32.9819450378418, |
|
"logps/rejected": -33.98347854614258, |
|
"loss": 0.9791, |
|
"rewards/accuracies": 0.9125000238418579, |
|
"rewards/chosen": 0.011788198724389076, |
|
"rewards/margins": 0.02092314139008522, |
|
"rewards/rejected": -0.009134944528341293, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 4.309335095262675e-06, |
|
"logits/chosen": -1.8713194131851196, |
|
"logits/rejected": -1.870599389076233, |
|
"logps/chosen": -30.50040054321289, |
|
"logps/rejected": -31.814233779907227, |
|
"loss": 0.9832, |
|
"rewards/accuracies": 0.875, |
|
"rewards/chosen": 0.011109036393463612, |
|
"rewards/margins": 0.016790907829999924, |
|
"rewards/rejected": -0.005681873299181461, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 4.269769281772082e-06, |
|
"logits/chosen": -1.8269096612930298, |
|
"logits/rejected": -1.820021629333496, |
|
"logps/chosen": -31.442739486694336, |
|
"logps/rejected": -35.589012145996094, |
|
"loss": 0.9781, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": 0.012273869477212429, |
|
"rewards/margins": 0.02189657650887966, |
|
"rewards/rejected": -0.009622708894312382, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"eval_logits/chosen": -2.118459939956665, |
|
"eval_logits/rejected": -2.113690137863159, |
|
"eval_logps/chosen": -34.34184646606445, |
|
"eval_logps/rejected": -37.955074310302734, |
|
"eval_loss": 0.9986757040023804, |
|
"eval_rewards/accuracies": 0.5855481624603271, |
|
"eval_rewards/chosen": -0.0030729183927178383, |
|
"eval_rewards/margins": 0.0013115763431414962, |
|
"eval_rewards/rejected": -0.004384494852274656, |
|
"eval_runtime": 145.8957, |
|
"eval_samples_per_second": 2.351, |
|
"eval_steps_per_second": 0.295, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 4.22929424333435e-06, |
|
"logits/chosen": -1.818110466003418, |
|
"logits/rejected": -1.821658730506897, |
|
"logps/chosen": -28.320688247680664, |
|
"logps/rejected": -33.88779830932617, |
|
"loss": 0.9812, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.008541365154087543, |
|
"rewards/margins": 0.01882646046578884, |
|
"rewards/rejected": -0.010285094380378723, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 4.1879307741372085e-06, |
|
"logits/chosen": -1.8071190118789673, |
|
"logits/rejected": -1.817929983139038, |
|
"logps/chosen": -32.183380126953125, |
|
"logps/rejected": -31.759241104125977, |
|
"loss": 0.9781, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": 0.010789523832499981, |
|
"rewards/margins": 0.021865110844373703, |
|
"rewards/rejected": -0.011075586080551147, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 4.145700124802693e-06, |
|
"logits/chosen": -1.7446613311767578, |
|
"logits/rejected": -1.74239182472229, |
|
"logps/chosen": -30.616100311279297, |
|
"logps/rejected": -31.322246551513672, |
|
"loss": 0.978, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": 0.010509822517633438, |
|
"rewards/margins": 0.02196466363966465, |
|
"rewards/rejected": -0.011454842053353786, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 4.102623991469562e-06, |
|
"logits/chosen": -1.8112165927886963, |
|
"logits/rejected": -1.8044904470443726, |
|
"logps/chosen": -33.239593505859375, |
|
"logps/rejected": -34.19357681274414, |
|
"loss": 0.9783, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.009932359680533409, |
|
"rewards/margins": 0.021693259477615356, |
|
"rewards/rejected": -0.011760897003114223, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 4.058724504646834e-06, |
|
"logits/chosen": -1.77358078956604, |
|
"logits/rejected": -1.7800006866455078, |
|
"logps/chosen": -30.977636337280273, |
|
"logps/rejected": -33.71123123168945, |
|
"loss": 0.9827, |
|
"rewards/accuracies": 0.762499988079071, |
|
"rewards/chosen": 0.007755486760288477, |
|
"rewards/margins": 0.01728045754134655, |
|
"rewards/rejected": -0.00952497310936451, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 4.014024217844167e-06, |
|
"logits/chosen": -1.8399150371551514, |
|
"logits/rejected": -1.8170970678329468, |
|
"logps/chosen": -30.506546020507812, |
|
"logps/rejected": -33.8461799621582, |
|
"loss": 0.9822, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": 0.009310225024819374, |
|
"rewards/margins": 0.01781143620610237, |
|
"rewards/rejected": -0.008501212112605572, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 3.968546095984911e-06, |
|
"logits/chosen": -1.768343210220337, |
|
"logits/rejected": -1.7633851766586304, |
|
"logps/chosen": -31.467700958251953, |
|
"logps/rejected": -33.090362548828125, |
|
"loss": 0.9808, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": 0.009490304626524448, |
|
"rewards/margins": 0.019184768199920654, |
|
"rewards/rejected": -0.009694463573396206, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"learning_rate": 3.922313503607806e-06, |
|
"logits/chosen": -1.7968637943267822, |
|
"logits/rejected": -1.798661231994629, |
|
"logps/chosen": -33.596519470214844, |
|
"logps/rejected": -36.36685562133789, |
|
"loss": 0.9765, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": 0.007792733609676361, |
|
"rewards/margins": 0.02345210127532482, |
|
"rewards/rejected": -0.01565936766564846, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 3.875350192863368e-06, |
|
"logits/chosen": -1.7735790014266968, |
|
"logits/rejected": -1.773097038269043, |
|
"logps/chosen": -29.556873321533203, |
|
"logps/rejected": -32.799781799316406, |
|
"loss": 0.9763, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": 0.01035183947533369, |
|
"rewards/margins": 0.02373330667614937, |
|
"rewards/rejected": -0.013381466269493103, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 3.8276802913111436e-06, |
|
"logits/chosen": -1.77323317527771, |
|
"logits/rejected": -1.770957589149475, |
|
"logps/chosen": -32.05171585083008, |
|
"logps/rejected": -33.61422348022461, |
|
"loss": 0.9774, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": 0.009465712122619152, |
|
"rewards/margins": 0.02260836958885193, |
|
"rewards/rejected": -0.013142657466232777, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"eval_logits/chosen": -2.0208382606506348, |
|
"eval_logits/rejected": -2.0161898136138916, |
|
"eval_logps/chosen": -34.76713562011719, |
|
"eval_logps/rejected": -38.52279281616211, |
|
"eval_loss": 0.997253954410553, |
|
"eval_rewards/accuracies": 0.574335515499115, |
|
"eval_rewards/chosen": -0.007325790822505951, |
|
"eval_rewards/margins": 0.002735937014222145, |
|
"eval_rewards/rejected": -0.01006172876805067, |
|
"eval_runtime": 145.8309, |
|
"eval_samples_per_second": 2.352, |
|
"eval_steps_per_second": 0.295, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 3.7793282895240927e-06, |
|
"logits/chosen": -1.7979720830917358, |
|
"logits/rejected": -1.8042287826538086, |
|
"logps/chosen": -31.683053970336914, |
|
"logps/rejected": -33.631996154785156, |
|
"loss": 0.9788, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": 0.005543296225368977, |
|
"rewards/margins": 0.02123137377202511, |
|
"rewards/rejected": -0.015688076615333557, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 3.730319028506478e-06, |
|
"logits/chosen": -1.741254210472107, |
|
"logits/rejected": -1.7390741109848022, |
|
"logps/chosen": -33.84577178955078, |
|
"logps/rejected": -32.38778305053711, |
|
"loss": 0.9769, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.009101121686398983, |
|
"rewards/margins": 0.023123882710933685, |
|
"rewards/rejected": -0.014022761955857277, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 3.6806776869317074e-06, |
|
"logits/chosen": -1.681199312210083, |
|
"logits/rejected": -1.6748110055923462, |
|
"logps/chosen": -34.46784210205078, |
|
"logps/rejected": -34.00514221191406, |
|
"loss": 0.9736, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.00996384583413601, |
|
"rewards/margins": 0.02644318714737892, |
|
"rewards/rejected": -0.016479339450597763, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 3.6304297682067146e-06, |
|
"logits/chosen": -1.6963602304458618, |
|
"logits/rejected": -1.7026523351669312, |
|
"logps/chosen": -33.33148193359375, |
|
"logps/rejected": -34.69775390625, |
|
"loss": 0.9789, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": 0.00647822767496109, |
|
"rewards/margins": 0.021053150296211243, |
|
"rewards/rejected": -0.014574920758605003, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 3.579601087369492e-06, |
|
"logits/chosen": -1.767584204673767, |
|
"logits/rejected": -1.7815086841583252, |
|
"logps/chosen": -31.229949951171875, |
|
"logps/rejected": -33.55590057373047, |
|
"loss": 0.9782, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": 0.006225164048373699, |
|
"rewards/margins": 0.02180730551481247, |
|
"rewards/rejected": -0.015582139603793621, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 3.5282177578265295e-06, |
|
"logits/chosen": -1.6283632516860962, |
|
"logits/rejected": -1.6250991821289062, |
|
"logps/chosen": -32.852622985839844, |
|
"logps/rejected": -36.96779251098633, |
|
"loss": 0.9666, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.01070336066186428, |
|
"rewards/margins": 0.03337875381112099, |
|
"rewards/rejected": -0.022675391286611557, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 3.476306177936961e-06, |
|
"logits/chosen": -1.7106907367706299, |
|
"logits/rejected": -1.710681676864624, |
|
"logps/chosen": -30.867549896240234, |
|
"logps/rejected": -36.25127410888672, |
|
"loss": 0.9716, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": 0.00418041180819273, |
|
"rewards/margins": 0.028429334983229637, |
|
"rewards/rejected": -0.024248924106359482, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 3.423893017450324e-06, |
|
"logits/chosen": -1.6477956771850586, |
|
"logits/rejected": -1.6444308757781982, |
|
"logps/chosen": -30.36174964904785, |
|
"logps/rejected": -34.880332946777344, |
|
"loss": 0.9743, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": 0.005105969496071339, |
|
"rewards/margins": 0.025731250643730164, |
|
"rewards/rejected": -0.0206252820789814, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"learning_rate": 3.3710052038048794e-06, |
|
"logits/chosen": -1.6614688634872437, |
|
"logits/rejected": -1.6616519689559937, |
|
"logps/chosen": -29.40378189086914, |
|
"logps/rejected": -32.798709869384766, |
|
"loss": 0.9711, |
|
"rewards/accuracies": 0.824999988079071, |
|
"rewards/chosen": 0.007593109272420406, |
|
"rewards/margins": 0.028875216841697693, |
|
"rewards/rejected": -0.021282104775309563, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 3.3176699082935546e-06, |
|
"logits/chosen": -1.579671025276184, |
|
"logits/rejected": -1.583088994026184, |
|
"logps/chosen": -33.53199005126953, |
|
"logps/rejected": -33.63783264160156, |
|
"loss": 0.9688, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": 0.009464827366173267, |
|
"rewards/margins": 0.031196285039186478, |
|
"rewards/rejected": -0.021731454879045486, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"eval_logits/chosen": -1.9279817342758179, |
|
"eval_logits/rejected": -1.923504114151001, |
|
"eval_logps/chosen": -35.46812438964844, |
|
"eval_logps/rejected": -39.259788513183594, |
|
"eval_loss": 0.9969056844711304, |
|
"eval_rewards/accuracies": 0.5481727719306946, |
|
"eval_rewards/chosen": -0.014335726387798786, |
|
"eval_rewards/margins": 0.0030959637369960546, |
|
"eval_rewards/rejected": -0.01743169128894806, |
|
"eval_runtime": 145.8801, |
|
"eval_samples_per_second": 2.351, |
|
"eval_steps_per_second": 0.295, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 3.2639145321045933e-06, |
|
"logits/chosen": -1.6492347717285156, |
|
"logits/rejected": -1.6412155628204346, |
|
"logps/chosen": -36.10542297363281, |
|
"logps/rejected": -34.03607940673828, |
|
"loss": 0.9746, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": 0.004065921995788813, |
|
"rewards/margins": 0.025368575006723404, |
|
"rewards/rejected": -0.02130264975130558, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 3.2097666922441107e-06, |
|
"logits/chosen": -1.649939775466919, |
|
"logits/rejected": -1.651545524597168, |
|
"logps/chosen": -36.18703842163086, |
|
"logps/rejected": -35.8452262878418, |
|
"loss": 0.9698, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": 0.002483480144292116, |
|
"rewards/margins": 0.03017548657953739, |
|
"rewards/rejected": -0.027692005038261414, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 3.1552542073477554e-06, |
|
"logits/chosen": -1.6672731637954712, |
|
"logits/rejected": -1.6649030447006226, |
|
"logps/chosen": -31.656982421875, |
|
"logps/rejected": -35.12821578979492, |
|
"loss": 0.9696, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": 0.009063473902642727, |
|
"rewards/margins": 0.030376678332686424, |
|
"rewards/rejected": -0.02131320908665657, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 3.100405083388799e-06, |
|
"logits/chosen": -1.6315555572509766, |
|
"logits/rejected": -1.6366676092147827, |
|
"logps/chosen": -30.99042320251465, |
|
"logps/rejected": -35.70110321044922, |
|
"loss": 0.9666, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": 0.007953016087412834, |
|
"rewards/margins": 0.03341587260365486, |
|
"rewards/rejected": -0.025462854653596878, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 3.0452474992899645e-06, |
|
"logits/chosen": -1.5864739418029785, |
|
"logits/rejected": -1.5849554538726807, |
|
"logps/chosen": -32.77734375, |
|
"logps/rejected": -37.71215057373047, |
|
"loss": 0.9663, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.0034763626754283905, |
|
"rewards/margins": 0.03373824805021286, |
|
"rewards/rejected": -0.03026188537478447, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 2.989809792446417e-06, |
|
"logits/chosen": -1.453434705734253, |
|
"logits/rejected": -1.448988676071167, |
|
"logps/chosen": -35.56426239013672, |
|
"logps/rejected": -38.61144256591797, |
|
"loss": 0.9616, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": 0.004341802094131708, |
|
"rewards/margins": 0.03837207704782486, |
|
"rewards/rejected": -0.03403027355670929, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 2.9341204441673267e-06, |
|
"logits/chosen": -1.5692594051361084, |
|
"logits/rejected": -1.5738866329193115, |
|
"logps/chosen": -35.17799758911133, |
|
"logps/rejected": -36.24814224243164, |
|
"loss": 0.9731, |
|
"rewards/accuracies": 0.7208333611488342, |
|
"rewards/chosen": 0.0003566344385035336, |
|
"rewards/margins": 0.02688964083790779, |
|
"rewards/rejected": -0.026533011347055435, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 2.878208065043501e-06, |
|
"logits/chosen": -1.5177220106124878, |
|
"logits/rejected": -1.5161765813827515, |
|
"logps/chosen": -32.8176155090332, |
|
"logps/rejected": -39.03325653076172, |
|
"loss": 0.9429, |
|
"rewards/accuracies": 0.887499988079071, |
|
"rewards/chosen": 0.010839789174497128, |
|
"rewards/margins": 0.0571257583796978, |
|
"rewards/rejected": -0.0462859682738781, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 2.05, |
|
"learning_rate": 2.8221013802485974e-06, |
|
"logits/chosen": -1.5499645471572876, |
|
"logits/rejected": -1.5473253726959229, |
|
"logps/chosen": -32.2547607421875, |
|
"logps/rejected": -36.87131881713867, |
|
"loss": 0.9509, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.011209758929908276, |
|
"rewards/margins": 0.04905728995800018, |
|
"rewards/rejected": -0.037847526371479034, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 2.76582921478147e-06, |
|
"logits/chosen": -1.4648706912994385, |
|
"logits/rejected": -1.459632396697998, |
|
"logps/chosen": -33.919517517089844, |
|
"logps/rejected": -35.301055908203125, |
|
"loss": 0.957, |
|
"rewards/accuracies": 0.875, |
|
"rewards/chosen": 0.004635663237422705, |
|
"rewards/margins": 0.04300360754132271, |
|
"rewards/rejected": -0.038367945700883865, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"eval_logits/chosen": -1.840654969215393, |
|
"eval_logits/rejected": -1.8362884521484375, |
|
"eval_logps/chosen": -36.17332458496094, |
|
"eval_logps/rejected": -40.119380950927734, |
|
"eval_loss": 0.9953566193580627, |
|
"eval_rewards/accuracies": 0.5539867281913757, |
|
"eval_rewards/chosen": -0.021387748420238495, |
|
"eval_rewards/margins": 0.004639865830540657, |
|
"eval_rewards/rejected": -0.0260276161134243, |
|
"eval_runtime": 145.904, |
|
"eval_samples_per_second": 2.351, |
|
"eval_steps_per_second": 0.295, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 2.7094204786572254e-06, |
|
"logits/chosen": -1.547839641571045, |
|
"logits/rejected": -1.5548707246780396, |
|
"logps/chosen": -31.346927642822266, |
|
"logps/rejected": -39.03870391845703, |
|
"loss": 0.9444, |
|
"rewards/accuracies": 0.875, |
|
"rewards/chosen": 0.007813268341124058, |
|
"rewards/margins": 0.05564187094569206, |
|
"rewards/rejected": -0.04782859981060028, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 2.6529041520546072e-06, |
|
"logits/chosen": -1.514336347579956, |
|
"logits/rejected": -1.5161477327346802, |
|
"logps/chosen": -32.147666931152344, |
|
"logps/rejected": -36.56987762451172, |
|
"loss": 0.9624, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": 0.004234147723764181, |
|
"rewards/margins": 0.03763490542769432, |
|
"rewards/rejected": -0.0334007553756237, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 2.5963092704273302e-06, |
|
"logits/chosen": -1.3985604047775269, |
|
"logits/rejected": -1.4026976823806763, |
|
"logps/chosen": -32.4110221862793, |
|
"logps/rejected": -40.09288787841797, |
|
"loss": 0.9464, |
|
"rewards/accuracies": 0.875, |
|
"rewards/chosen": 0.0018043376039713621, |
|
"rewards/margins": 0.053631413727998734, |
|
"rewards/rejected": -0.051827073097229004, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.53966490958702e-06, |
|
"logits/chosen": -1.4642579555511475, |
|
"logits/rejected": -1.4605650901794434, |
|
"logps/chosen": -32.868621826171875, |
|
"logps/rejected": -37.319786071777344, |
|
"loss": 0.9557, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.0010002745548263192, |
|
"rewards/margins": 0.04425168037414551, |
|
"rewards/rejected": -0.0432514064013958, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 2.4830001707654135e-06, |
|
"logits/chosen": -1.5074342489242554, |
|
"logits/rejected": -1.5097802877426147, |
|
"logps/chosen": -32.264610290527344, |
|
"logps/rejected": -41.21857833862305, |
|
"loss": 0.9393, |
|
"rewards/accuracies": 0.8999999761581421, |
|
"rewards/chosen": 0.005107272416353226, |
|
"rewards/margins": 0.06067372113466263, |
|
"rewards/rejected": -0.0555664524435997, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 2.4263441656635054e-06, |
|
"logits/chosen": -1.3380223512649536, |
|
"logits/rejected": -1.3343451023101807, |
|
"logps/chosen": -37.09096908569336, |
|
"logps/rejected": -38.383792877197266, |
|
"loss": 0.9532, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -0.008092949166893959, |
|
"rewards/margins": 0.04682592675089836, |
|
"rewards/rejected": -0.05491887778043747, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 2.3697260014953107e-06, |
|
"logits/chosen": -1.360643744468689, |
|
"logits/rejected": -1.3605453968048096, |
|
"logps/chosen": -36.00204086303711, |
|
"logps/rejected": -40.74755096435547, |
|
"loss": 0.9426, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.0015923639293760061, |
|
"rewards/margins": 0.057420529425144196, |
|
"rewards/rejected": -0.059012897312641144, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 2.29, |
|
"learning_rate": 2.3131747660339396e-06, |
|
"logits/chosen": -1.3869389295578003, |
|
"logits/rejected": -1.3750654458999634, |
|
"logps/chosen": -34.1107292175293, |
|
"logps/rejected": -38.999202728271484, |
|
"loss": 0.9393, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": -4.148450534557924e-05, |
|
"rewards/margins": 0.06073024123907089, |
|
"rewards/rejected": -0.06077173352241516, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 2.31, |
|
"learning_rate": 2.256719512667651e-06, |
|
"logits/chosen": -1.476839303970337, |
|
"logits/rejected": -1.4818614721298218, |
|
"logps/chosen": -34.29611587524414, |
|
"logps/rejected": -39.490264892578125, |
|
"loss": 0.9391, |
|
"rewards/accuracies": 0.762499988079071, |
|
"rewards/chosen": -0.009661799296736717, |
|
"rewards/margins": 0.06087601184844971, |
|
"rewards/rejected": -0.07053781300783157, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 2.2003892454735786e-06, |
|
"logits/chosen": -1.4013174772262573, |
|
"logits/rejected": -1.3942521810531616, |
|
"logps/chosen": -35.21047592163086, |
|
"logps/rejected": -39.179725646972656, |
|
"loss": 0.9358, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": -0.0035639957059174776, |
|
"rewards/margins": 0.064211905002594, |
|
"rewards/rejected": -0.0677758976817131, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"eval_logits/chosen": -1.7029222249984741, |
|
"eval_logits/rejected": -1.6987833976745605, |
|
"eval_logps/chosen": -37.65321350097656, |
|
"eval_logps/rejected": -41.74834060668945, |
|
"eval_loss": 0.9938713312149048, |
|
"eval_rewards/accuracies": 0.5365448594093323, |
|
"eval_rewards/chosen": -0.03618660196661949, |
|
"eval_rewards/margins": 0.00613059988245368, |
|
"eval_rewards/rejected": -0.04231720417737961, |
|
"eval_runtime": 145.9089, |
|
"eval_samples_per_second": 2.351, |
|
"eval_steps_per_second": 0.295, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 2.36, |
|
"learning_rate": 2.1442129043167877e-06, |
|
"logits/chosen": -1.3909523487091064, |
|
"logits/rejected": -1.3909533023834229, |
|
"logps/chosen": -31.57651138305664, |
|
"logps/rejected": -42.50416564941406, |
|
"loss": 0.9259, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": -0.004093508701771498, |
|
"rewards/margins": 0.07405127584934235, |
|
"rewards/rejected": -0.07814478874206543, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 2.39, |
|
"learning_rate": 2.088219349982323e-06, |
|
"logits/chosen": -1.3331390619277954, |
|
"logits/rejected": -1.3248008489608765, |
|
"logps/chosen": -33.280738830566406, |
|
"logps/rejected": -41.13130569458008, |
|
"loss": 0.9388, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.014417916536331177, |
|
"rewards/margins": 0.06118142604827881, |
|
"rewards/rejected": -0.07559934258460999, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 2.42, |
|
"learning_rate": 2.0324373493478803e-06, |
|
"logits/chosen": -1.474799394607544, |
|
"logits/rejected": -1.473085641860962, |
|
"logps/chosen": -30.739898681640625, |
|
"logps/rejected": -40.02002716064453, |
|
"loss": 0.9359, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -0.007428646087646484, |
|
"rewards/margins": 0.06412303447723389, |
|
"rewards/rejected": -0.07155168056488037, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"learning_rate": 1.976895560604729e-06, |
|
"logits/chosen": -1.3616539239883423, |
|
"logits/rejected": -1.372312307357788, |
|
"logps/chosen": -36.05510330200195, |
|
"logps/rejected": -41.12104034423828, |
|
"loss": 0.935, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": -0.016300354152917862, |
|
"rewards/margins": 0.06501360237598419, |
|
"rewards/rejected": -0.08131395280361176, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 2.47, |
|
"learning_rate": 1.921622518534466e-06, |
|
"logits/chosen": -1.3951683044433594, |
|
"logits/rejected": -1.3979324102401733, |
|
"logps/chosen": -32.248931884765625, |
|
"logps/rejected": -39.3216552734375, |
|
"loss": 0.9402, |
|
"rewards/accuracies": 0.7124999761581421, |
|
"rewards/chosen": -0.016012068837881088, |
|
"rewards/margins": 0.05984366685152054, |
|
"rewards/rejected": -0.07585573941469193, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 2.49, |
|
"learning_rate": 1.8666466198491794e-06, |
|
"logits/chosen": -1.3717939853668213, |
|
"logits/rejected": -1.3667659759521484, |
|
"logps/chosen": -35.57506561279297, |
|
"logps/rejected": -42.31549835205078, |
|
"loss": 0.9309, |
|
"rewards/accuracies": 0.762499988079071, |
|
"rewards/chosen": -0.014998522587120533, |
|
"rewards/margins": 0.06909255683422089, |
|
"rewards/rejected": -0.08409108221530914, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 2.52, |
|
"learning_rate": 1.8119961086025376e-06, |
|
"logits/chosen": -1.294325590133667, |
|
"logits/rejected": -1.2963002920150757, |
|
"logps/chosen": -34.213260650634766, |
|
"logps/rejected": -43.49393081665039, |
|
"loss": 0.9309, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -0.014005112461745739, |
|
"rewards/margins": 0.06911824643611908, |
|
"rewards/rejected": -0.0831233561038971, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 2.55, |
|
"learning_rate": 1.7576990616793139e-06, |
|
"logits/chosen": -1.3165125846862793, |
|
"logits/rejected": -1.310859203338623, |
|
"logps/chosen": -38.791927337646484, |
|
"logps/rejected": -45.34046936035156, |
|
"loss": 0.9431, |
|
"rewards/accuracies": 0.762499988079071, |
|
"rewards/chosen": -0.03386862203478813, |
|
"rewards/margins": 0.05691651254892349, |
|
"rewards/rejected": -0.09078512340784073, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 2.57, |
|
"learning_rate": 1.7037833743707892e-06, |
|
"logits/chosen": -1.3041751384735107, |
|
"logits/rejected": -1.2975685596466064, |
|
"logps/chosen": -32.80427551269531, |
|
"logps/rejected": -45.00722885131836, |
|
"loss": 0.9278, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.018013231456279755, |
|
"rewards/margins": 0.07216015458106995, |
|
"rewards/rejected": -0.0901733785867691, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 2.6, |
|
"learning_rate": 1.6502767460434588e-06, |
|
"logits/chosen": -1.2888100147247314, |
|
"logits/rejected": -1.2791659832000732, |
|
"logps/chosen": -34.17406463623047, |
|
"logps/rejected": -36.677513122558594, |
|
"loss": 0.9535, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -0.024253757670521736, |
|
"rewards/margins": 0.04652407765388489, |
|
"rewards/rejected": -0.07077784091234207, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 2.6, |
|
"eval_logits/chosen": -1.6182585954666138, |
|
"eval_logits/rejected": -1.614295244216919, |
|
"eval_logps/chosen": -39.14792251586914, |
|
"eval_logps/rejected": -43.423744201660156, |
|
"eval_loss": 0.9920738935470581, |
|
"eval_rewards/accuracies": 0.545265793800354, |
|
"eval_rewards/chosen": -0.05113373324275017, |
|
"eval_rewards/margins": 0.007937532849609852, |
|
"eval_rewards/rejected": -0.059071265161037445, |
|
"eval_runtime": 145.6856, |
|
"eval_samples_per_second": 2.354, |
|
"eval_steps_per_second": 0.295, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 2.62, |
|
"learning_rate": 1.5972066659083796e-06, |
|
"logits/chosen": -1.353121042251587, |
|
"logits/rejected": -1.35282301902771, |
|
"logps/chosen": -34.075523376464844, |
|
"logps/rejected": -38.43246078491211, |
|
"loss": 0.9408, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -0.022513313218951225, |
|
"rewards/margins": 0.059226490557193756, |
|
"rewards/rejected": -0.08173979073762894, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 2.65, |
|
"learning_rate": 1.5446003988985041e-06, |
|
"logits/chosen": -1.403580904006958, |
|
"logits/rejected": -1.4037662744522095, |
|
"logps/chosen": -33.98898696899414, |
|
"logps/rejected": -39.30225372314453, |
|
"loss": 0.9391, |
|
"rewards/accuracies": 0.762499988079071, |
|
"rewards/chosen": -0.020454101264476776, |
|
"rewards/margins": 0.06091039255261421, |
|
"rewards/rejected": -0.08136448264122009, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 2.68, |
|
"learning_rate": 1.4924849716612211e-06, |
|
"logits/chosen": -1.3622567653656006, |
|
"logits/rejected": -1.3672521114349365, |
|
"logps/chosen": -35.09298324584961, |
|
"logps/rejected": -35.331912994384766, |
|
"loss": 0.9546, |
|
"rewards/accuracies": 0.6875, |
|
"rewards/chosen": -0.02812912128865719, |
|
"rewards/margins": 0.04535282403230667, |
|
"rewards/rejected": -0.07348194718360901, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 2.7, |
|
"learning_rate": 1.440887158673332e-06, |
|
"logits/chosen": -1.3483327627182007, |
|
"logits/rejected": -1.3397239446640015, |
|
"logps/chosen": -33.612525939941406, |
|
"logps/rejected": -43.52740478515625, |
|
"loss": 0.9241, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -0.024346303194761276, |
|
"rewards/margins": 0.07592549920082092, |
|
"rewards/rejected": -0.1002717837691307, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 2.73, |
|
"learning_rate": 1.3898334684855647e-06, |
|
"logits/chosen": -1.3182249069213867, |
|
"logits/rejected": -1.3285300731658936, |
|
"logps/chosen": -36.65705108642578, |
|
"logps/rejected": -41.1858024597168, |
|
"loss": 0.9416, |
|
"rewards/accuracies": 0.824999988079071, |
|
"rewards/chosen": -0.0310394074767828, |
|
"rewards/margins": 0.05836848169565201, |
|
"rewards/rejected": -0.08940788358449936, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 2.75, |
|
"learning_rate": 1.3393501301037245e-06, |
|
"logits/chosen": -1.3787955045700073, |
|
"logits/rejected": -1.3705497980117798, |
|
"logps/chosen": -35.9283561706543, |
|
"logps/rejected": -47.05866241455078, |
|
"loss": 0.917, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -0.024275777861475945, |
|
"rewards/margins": 0.08299694955348969, |
|
"rewards/rejected": -0.10727272927761078, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 2.78, |
|
"learning_rate": 1.2894630795134454e-06, |
|
"logits/chosen": -1.2876031398773193, |
|
"logits/rejected": -1.2903258800506592, |
|
"logps/chosen": -38.158668518066406, |
|
"logps/rejected": -41.497459411621094, |
|
"loss": 0.9337, |
|
"rewards/accuracies": 0.762499988079071, |
|
"rewards/chosen": -0.021693874150514603, |
|
"rewards/margins": 0.06630205363035202, |
|
"rewards/rejected": -0.08799593150615692, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 2.81, |
|
"learning_rate": 1.2401979463554984e-06, |
|
"logits/chosen": -1.3835922479629517, |
|
"logits/rejected": -1.38229238986969, |
|
"logps/chosen": -35.45813751220703, |
|
"logps/rejected": -44.54060363769531, |
|
"loss": 0.9198, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -0.024004962295293808, |
|
"rewards/margins": 0.08024422824382782, |
|
"rewards/rejected": -0.10424919426441193, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 2.83, |
|
"learning_rate": 1.1915800407584705e-06, |
|
"logits/chosen": -1.3808025121688843, |
|
"logits/rejected": -1.3838340044021606, |
|
"logps/chosen": -33.26921463012695, |
|
"logps/rejected": -42.89200210571289, |
|
"loss": 0.9303, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -0.023758988827466965, |
|
"rewards/margins": 0.06968275457620621, |
|
"rewards/rejected": -0.09344174712896347, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 2.86, |
|
"learning_rate": 1.1436343403356019e-06, |
|
"logits/chosen": -1.3661361932754517, |
|
"logits/rejected": -1.3705623149871826, |
|
"logps/chosen": -36.603370666503906, |
|
"logps/rejected": -37.271080017089844, |
|
"loss": 0.9616, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -0.030343661084771156, |
|
"rewards/margins": 0.03835037350654602, |
|
"rewards/rejected": -0.06869403272867203, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 2.86, |
|
"eval_logits/chosen": -1.5919773578643799, |
|
"eval_logits/rejected": -1.5880482196807861, |
|
"eval_logps/chosen": -39.65049743652344, |
|
"eval_logps/rejected": -43.97543716430664, |
|
"eval_loss": 0.9915903806686401, |
|
"eval_rewards/accuracies": 0.545265793800354, |
|
"eval_rewards/chosen": -0.05615944042801857, |
|
"eval_rewards/margins": 0.008428744971752167, |
|
"eval_rewards/rejected": -0.06458818912506104, |
|
"eval_runtime": 145.863, |
|
"eval_samples_per_second": 2.352, |
|
"eval_steps_per_second": 0.295, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 2.88, |
|
"learning_rate": 1.0963854773524548e-06, |
|
"logits/chosen": -1.355208158493042, |
|
"logits/rejected": -1.3564389944076538, |
|
"logps/chosen": -34.99750900268555, |
|
"logps/rejected": -39.34065628051758, |
|
"loss": 0.9373, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -0.020573578774929047, |
|
"rewards/margins": 0.06267382204532623, |
|
"rewards/rejected": -0.08324739336967468, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 2.91, |
|
"learning_rate": 1.049857726072005e-06, |
|
"logits/chosen": -1.227752923965454, |
|
"logits/rejected": -1.2306647300720215, |
|
"logps/chosen": -37.04521942138672, |
|
"logps/rejected": -41.808509826660156, |
|
"loss": 0.9371, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -0.02480238489806652, |
|
"rewards/margins": 0.06287000328302383, |
|
"rewards/rejected": -0.0876723974943161, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 2.94, |
|
"learning_rate": 1.0040749902836508e-06, |
|
"logits/chosen": -1.250138521194458, |
|
"logits/rejected": -1.2487337589263916, |
|
"logps/chosen": -34.29961013793945, |
|
"logps/rejected": -39.423301696777344, |
|
"loss": 0.9503, |
|
"rewards/accuracies": 0.7124999761581421, |
|
"rewards/chosen": -0.032878875732421875, |
|
"rewards/margins": 0.04969673603773117, |
|
"rewards/rejected": -0.08257561177015305, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 2.96, |
|
"learning_rate": 9.59060791022566e-07, |
|
"logits/chosen": -1.363937258720398, |
|
"logits/rejected": -1.360128402709961, |
|
"logps/chosen": -34.89466094970703, |
|
"logps/rejected": -41.99665832519531, |
|
"loss": 0.9281, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.018548697233200073, |
|
"rewards/margins": 0.07187779247760773, |
|
"rewards/rejected": -0.0904264897108078, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 2.99, |
|
"learning_rate": 9.148382544856885e-07, |
|
"logits/chosen": -1.2763797044754028, |
|
"logits/rejected": -1.2683088779449463, |
|
"logps/chosen": -36.54337692260742, |
|
"logps/rejected": -40.25540542602539, |
|
"loss": 0.9413, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -0.030916398391127586, |
|
"rewards/margins": 0.05871362239122391, |
|
"rewards/rejected": -0.08963000774383545, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 3.01, |
|
"learning_rate": 8.714301001505568e-07, |
|
"logits/chosen": -1.301361083984375, |
|
"logits/rejected": -1.3027942180633545, |
|
"logps/chosen": -36.6050910949707, |
|
"logps/rejected": -39.68968963623047, |
|
"loss": 0.9444, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -0.029233410954475403, |
|
"rewards/margins": 0.05564901977777481, |
|
"rewards/rejected": -0.08488242328166962, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 3.04, |
|
"learning_rate": 8.288586291031025e-07, |
|
"logits/chosen": -1.3797643184661865, |
|
"logits/rejected": -1.3745005130767822, |
|
"logps/chosen": -36.120018005371094, |
|
"logps/rejected": -41.199337005615234, |
|
"loss": 0.9434, |
|
"rewards/accuracies": 0.7124999761581421, |
|
"rewards/chosen": -0.02594051882624626, |
|
"rewards/margins": 0.05664067342877388, |
|
"rewards/rejected": -0.08258119225502014, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 3.06, |
|
"learning_rate": 7.871457125803897e-07, |
|
"logits/chosen": -1.2806499004364014, |
|
"logits/rejected": -1.289930820465088, |
|
"logps/chosen": -36.737998962402344, |
|
"logps/rejected": -41.14373016357422, |
|
"loss": 0.9462, |
|
"rewards/accuracies": 0.7124999761581421, |
|
"rewards/chosen": -0.03404999524354935, |
|
"rewards/margins": 0.053772974759340286, |
|
"rewards/rejected": -0.08782295882701874, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 3.09, |
|
"learning_rate": 7.463127807341966e-07, |
|
"logits/chosen": -1.2992810010910034, |
|
"logits/rejected": -1.2938917875289917, |
|
"logps/chosen": -33.985870361328125, |
|
"logps/rejected": -42.187801361083984, |
|
"loss": 0.9291, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.018122535198926926, |
|
"rewards/margins": 0.07092103362083435, |
|
"rewards/rejected": -0.08904357254505157, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 3.12, |
|
"learning_rate": 7.063808116212021e-07, |
|
"logits/chosen": -1.261099934577942, |
|
"logits/rejected": -1.262424349784851, |
|
"logps/chosen": -35.9713249206543, |
|
"logps/rejected": -43.56898880004883, |
|
"loss": 0.9167, |
|
"rewards/accuracies": 0.7124999761581421, |
|
"rewards/chosen": -0.02480882592499256, |
|
"rewards/margins": 0.08328056335449219, |
|
"rewards/rejected": -0.1080893874168396, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 3.12, |
|
"eval_logits/chosen": -1.5890045166015625, |
|
"eval_logits/rejected": -1.585091471672058, |
|
"eval_logps/chosen": -39.66655349731445, |
|
"eval_logps/rejected": -44.02885818481445, |
|
"eval_loss": 0.9912154078483582, |
|
"eval_rewards/accuracies": 0.5481727719306946, |
|
"eval_rewards/chosen": -0.056320033967494965, |
|
"eval_rewards/margins": 0.00880234595388174, |
|
"eval_rewards/rejected": -0.06512238085269928, |
|
"eval_runtime": 145.8882, |
|
"eval_samples_per_second": 2.351, |
|
"eval_steps_per_second": 0.295, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 3.14, |
|
"learning_rate": 6.673703204254348e-07, |
|
"logits/chosen": -1.2175233364105225, |
|
"logits/rejected": -1.2172149419784546, |
|
"logps/chosen": -37.891151428222656, |
|
"logps/rejected": -42.90294647216797, |
|
"loss": 0.9165, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": -0.01838667131960392, |
|
"rewards/margins": 0.08354915678501129, |
|
"rewards/rejected": -0.10193584114313126, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 3.17, |
|
"learning_rate": 6.293013489185315e-07, |
|
"logits/chosen": -1.3444710969924927, |
|
"logits/rejected": -1.3372966051101685, |
|
"logps/chosen": -34.07860565185547, |
|
"logps/rejected": -43.33241653442383, |
|
"loss": 0.9192, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.02374918945133686, |
|
"rewards/margins": 0.08078955113887787, |
|
"rewards/rejected": -0.10453873872756958, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 3.19, |
|
"learning_rate": 5.921934551632086e-07, |
|
"logits/chosen": -1.2269846200942993, |
|
"logits/rejected": -1.2168152332305908, |
|
"logps/chosen": -36.28278350830078, |
|
"logps/rejected": -42.86933517456055, |
|
"loss": 0.9192, |
|
"rewards/accuracies": 0.824999988079071, |
|
"rewards/chosen": -0.01897098869085312, |
|
"rewards/margins": 0.08078411966562271, |
|
"rewards/rejected": -0.09975510835647583, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 3.22, |
|
"learning_rate": 5.560657034652405e-07, |
|
"logits/chosen": -1.3140621185302734, |
|
"logits/rejected": -1.308857798576355, |
|
"logps/chosen": -33.93239212036133, |
|
"logps/rejected": -37.61920928955078, |
|
"loss": 0.9448, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": -0.03232160210609436, |
|
"rewards/margins": 0.0551949217915535, |
|
"rewards/rejected": -0.08751652389764786, |
|
"step": 1240 |
|
}, |
|
{ |
|
"epoch": 3.25, |
|
"learning_rate": 5.2093665457911e-07, |
|
"logits/chosen": -1.3194572925567627, |
|
"logits/rejected": -1.3273426294326782, |
|
"logps/chosen": -38.09012985229492, |
|
"logps/rejected": -40.47280502319336, |
|
"loss": 0.9374, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -0.026947390288114548, |
|
"rewards/margins": 0.06259752810001373, |
|
"rewards/rejected": -0.08954491466283798, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 3.27, |
|
"learning_rate": 4.868243561723535e-07, |
|
"logits/chosen": -1.3223885297775269, |
|
"logits/rejected": -1.3221849203109741, |
|
"logps/chosen": -36.7206916809082, |
|
"logps/rejected": -43.311153411865234, |
|
"loss": 0.932, |
|
"rewards/accuracies": 0.8374999761581421, |
|
"rewards/chosen": -0.03692134469747543, |
|
"rewards/margins": 0.06804148107767105, |
|
"rewards/rejected": -0.10496282577514648, |
|
"step": 1260 |
|
}, |
|
{ |
|
"epoch": 3.3, |
|
"learning_rate": 4.537463335535161e-07, |
|
"logits/chosen": -1.2348535060882568, |
|
"logits/rejected": -1.2325520515441895, |
|
"logps/chosen": -34.88422393798828, |
|
"logps/rejected": -44.12788009643555, |
|
"loss": 0.9119, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.01793385110795498, |
|
"rewards/margins": 0.08813884109258652, |
|
"rewards/rejected": -0.10607268661260605, |
|
"step": 1270 |
|
}, |
|
{ |
|
"epoch": 3.32, |
|
"learning_rate": 4.217195806684629e-07, |
|
"logits/chosen": -1.1531862020492554, |
|
"logits/rejected": -1.1495788097381592, |
|
"logps/chosen": -37.46773147583008, |
|
"logps/rejected": -39.666324615478516, |
|
"loss": 0.9349, |
|
"rewards/accuracies": 0.737500011920929, |
|
"rewards/chosen": -0.022302847355604172, |
|
"rewards/margins": 0.06509546935558319, |
|
"rewards/rejected": -0.08739831298589706, |
|
"step": 1280 |
|
}, |
|
{ |
|
"epoch": 3.35, |
|
"learning_rate": 3.907605513696808e-07, |
|
"logits/chosen": -1.3214530944824219, |
|
"logits/rejected": -1.3079793453216553, |
|
"logps/chosen": -37.78839111328125, |
|
"logps/rejected": -46.277854919433594, |
|
"loss": 0.9193, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -0.03440812602639198, |
|
"rewards/margins": 0.0807172879576683, |
|
"rewards/rejected": -0.11512543261051178, |
|
"step": 1290 |
|
}, |
|
{ |
|
"epoch": 3.38, |
|
"learning_rate": 3.6088515096305675e-07, |
|
"logits/chosen": -1.274196982383728, |
|
"logits/rejected": -1.2778257131576538, |
|
"logps/chosen": -36.133689880371094, |
|
"logps/rejected": -48.38513946533203, |
|
"loss": 0.9033, |
|
"rewards/accuracies": 0.875, |
|
"rewards/chosen": -0.026160676032304764, |
|
"rewards/margins": 0.09665848314762115, |
|
"rewards/rejected": -0.12281917035579681, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 3.38, |
|
"eval_logits/chosen": -1.5856225490570068, |
|
"eval_logits/rejected": -1.5817017555236816, |
|
"eval_logps/chosen": -39.73164367675781, |
|
"eval_logps/rejected": -44.086830139160156, |
|
"eval_loss": 0.9912906289100647, |
|
"eval_rewards/accuracies": 0.545265793800354, |
|
"eval_rewards/chosen": -0.056970901787281036, |
|
"eval_rewards/margins": 0.00873124971985817, |
|
"eval_rewards/rejected": -0.06570214778184891, |
|
"eval_runtime": 145.8872, |
|
"eval_samples_per_second": 2.351, |
|
"eval_steps_per_second": 0.295, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 3.4, |
|
"learning_rate": 3.321087280364757e-07, |
|
"logits/chosen": -1.2609808444976807, |
|
"logits/rejected": -1.2618134021759033, |
|
"logps/chosen": -38.979774475097656, |
|
"logps/rejected": -48.166404724121094, |
|
"loss": 0.9165, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": -0.02735757827758789, |
|
"rewards/margins": 0.0834653228521347, |
|
"rewards/rejected": -0.11082291603088379, |
|
"step": 1310 |
|
}, |
|
{ |
|
"epoch": 3.43, |
|
"learning_rate": 3.044460665744284e-07, |
|
"logits/chosen": -1.3282499313354492, |
|
"logits/rejected": -1.3269433975219727, |
|
"logps/chosen": -34.594669342041016, |
|
"logps/rejected": -40.66201400756836, |
|
"loss": 0.9279, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -0.02500814199447632, |
|
"rewards/margins": 0.07207842171192169, |
|
"rewards/rejected": -0.09708657115697861, |
|
"step": 1320 |
|
}, |
|
{ |
|
"epoch": 3.45, |
|
"learning_rate": 2.779113783626916e-07, |
|
"logits/chosen": -1.2696067094802856, |
|
"logits/rejected": -1.2711015939712524, |
|
"logps/chosen": -36.31203079223633, |
|
"logps/rejected": -43.704010009765625, |
|
"loss": 0.9188, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.019377777352929115, |
|
"rewards/margins": 0.08117558807134628, |
|
"rewards/rejected": -0.10055337101221085, |
|
"step": 1330 |
|
}, |
|
{ |
|
"epoch": 3.48, |
|
"learning_rate": 2.5251829568697204e-07, |
|
"logits/chosen": -1.312336802482605, |
|
"logits/rejected": -1.3112585544586182, |
|
"logps/chosen": -33.062835693359375, |
|
"logps/rejected": -41.20447540283203, |
|
"loss": 0.9265, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -0.018641695380210876, |
|
"rewards/margins": 0.07348335534334183, |
|
"rewards/rejected": -0.0921250432729721, |
|
"step": 1340 |
|
}, |
|
{ |
|
"epoch": 3.51, |
|
"learning_rate": 2.2827986432927774e-07, |
|
"logits/chosen": -1.3330471515655518, |
|
"logits/rejected": -1.318232774734497, |
|
"logps/chosen": -37.70689392089844, |
|
"logps/rejected": -48.62843322753906, |
|
"loss": 0.9103, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": -0.033902253955602646, |
|
"rewards/margins": 0.08973778784275055, |
|
"rewards/rejected": -0.1236400380730629, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 3.53, |
|
"learning_rate": 2.0520853686560177e-07, |
|
"logits/chosen": -1.2979753017425537, |
|
"logits/rejected": -1.3076903820037842, |
|
"logps/chosen": -33.92186737060547, |
|
"logps/rejected": -42.06645965576172, |
|
"loss": 0.9203, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -0.018533948808908463, |
|
"rewards/margins": 0.07973320782184601, |
|
"rewards/rejected": -0.09826715290546417, |
|
"step": 1360 |
|
}, |
|
{ |
|
"epoch": 3.56, |
|
"learning_rate": 1.833161662683672e-07, |
|
"logits/chosen": -1.4156075716018677, |
|
"logits/rejected": -1.4154436588287354, |
|
"logps/chosen": -33.43849563598633, |
|
"logps/rejected": -48.292179107666016, |
|
"loss": 0.8895, |
|
"rewards/accuracies": 0.887499988079071, |
|
"rewards/chosen": -0.014080649241805077, |
|
"rewards/margins": 0.11053232103586197, |
|
"rewards/rejected": -0.1246129646897316, |
|
"step": 1370 |
|
}, |
|
{ |
|
"epoch": 3.58, |
|
"learning_rate": 1.626139998169246e-07, |
|
"logits/chosen": -1.2881263494491577, |
|
"logits/rejected": -1.295348882675171, |
|
"logps/chosen": -36.35157775878906, |
|
"logps/rejected": -48.889339447021484, |
|
"loss": 0.9073, |
|
"rewards/accuracies": 0.862500011920929, |
|
"rewards/chosen": -0.021710969507694244, |
|
"rewards/margins": 0.09268596768379211, |
|
"rewards/rejected": -0.11439694464206696, |
|
"step": 1380 |
|
}, |
|
{ |
|
"epoch": 3.61, |
|
"learning_rate": 1.4311267331922535e-07, |
|
"logits/chosen": -1.2490915060043335, |
|
"logits/rejected": -1.2452831268310547, |
|
"logps/chosen": -36.50374984741211, |
|
"logps/rejected": -40.645694732666016, |
|
"loss": 0.9269, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.01736537739634514, |
|
"rewards/margins": 0.07311762124300003, |
|
"rewards/rejected": -0.09048298746347427, |
|
"step": 1390 |
|
}, |
|
{ |
|
"epoch": 3.64, |
|
"learning_rate": 1.2482220564763669e-07, |
|
"logits/chosen": -1.37045419216156, |
|
"logits/rejected": -1.3693218231201172, |
|
"logps/chosen": -33.22849655151367, |
|
"logps/rejected": -41.22184371948242, |
|
"loss": 0.9285, |
|
"rewards/accuracies": 0.762499988079071, |
|
"rewards/chosen": -0.01931949146091938, |
|
"rewards/margins": 0.07145430892705917, |
|
"rewards/rejected": -0.090773805975914, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 3.64, |
|
"eval_logits/chosen": -1.586363673210144, |
|
"eval_logits/rejected": -1.5824512243270874, |
|
"eval_logps/chosen": -39.72161102294922, |
|
"eval_logps/rejected": -44.085166931152344, |
|
"eval_loss": 0.9912044405937195, |
|
"eval_rewards/accuracies": 0.5394518375396729, |
|
"eval_rewards/chosen": -0.056870583444833755, |
|
"eval_rewards/margins": 0.0088148582726717, |
|
"eval_rewards/rejected": -0.06568543612957001, |
|
"eval_runtime": 145.8196, |
|
"eval_samples_per_second": 2.352, |
|
"eval_steps_per_second": 0.295, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 3.66, |
|
"learning_rate": 1.0775199359171346e-07, |
|
"logits/chosen": -1.3426971435546875, |
|
"logits/rejected": -1.337646484375, |
|
"logps/chosen": -36.17827224731445, |
|
"logps/rejected": -37.915504455566406, |
|
"loss": 0.9417, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": -0.025357365608215332, |
|
"rewards/margins": 0.05825306102633476, |
|
"rewards/rejected": -0.0836104229092598, |
|
"step": 1410 |
|
}, |
|
{ |
|
"epoch": 3.69, |
|
"learning_rate": 9.191080703056604e-08, |
|
"logits/chosen": -1.2913800477981567, |
|
"logits/rejected": -1.2924379110336304, |
|
"logps/chosen": -35.52798843383789, |
|
"logps/rejected": -44.0432243347168, |
|
"loss": 0.9304, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.021631117910146713, |
|
"rewards/margins": 0.06956102699041367, |
|
"rewards/rejected": -0.09119214862585068, |
|
"step": 1420 |
|
}, |
|
{ |
|
"epoch": 3.71, |
|
"learning_rate": 7.730678442730539e-08, |
|
"logits/chosen": -1.241626501083374, |
|
"logits/rejected": -1.2360563278198242, |
|
"logps/chosen": -35.85869216918945, |
|
"logps/rejected": -48.056610107421875, |
|
"loss": 0.9035, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.017169218510389328, |
|
"rewards/margins": 0.09650282561779022, |
|
"rewards/rejected": -0.11367203295230865, |
|
"step": 1430 |
|
}, |
|
{ |
|
"epoch": 3.74, |
|
"learning_rate": 6.394742864787806e-08, |
|
"logits/chosen": -1.2561070919036865, |
|
"logits/rejected": -1.2504056692123413, |
|
"logps/chosen": -31.515949249267578, |
|
"logps/rejected": -41.77183151245117, |
|
"loss": 0.9216, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -0.02202703058719635, |
|
"rewards/margins": 0.07838472723960876, |
|
"rewards/rejected": -0.10041175782680511, |
|
"step": 1440 |
|
}, |
|
{ |
|
"epoch": 3.77, |
|
"learning_rate": 5.183960310644748e-08, |
|
"logits/chosen": -1.3071388006210327, |
|
"logits/rejected": -1.2970083951950073, |
|
"logps/chosen": -35.57398986816406, |
|
"logps/rejected": -45.840171813964844, |
|
"loss": 0.9254, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": -0.030924629420042038, |
|
"rewards/margins": 0.0746234804391861, |
|
"rewards/rejected": -0.10554809868335724, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 3.79, |
|
"learning_rate": 4.098952823928693e-08, |
|
"logits/chosen": -1.2653499841690063, |
|
"logits/rejected": -1.2617781162261963, |
|
"logps/chosen": -35.99557876586914, |
|
"logps/rejected": -39.989891052246094, |
|
"loss": 0.9359, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -0.026857968419790268, |
|
"rewards/margins": 0.06413265317678452, |
|
"rewards/rejected": -0.09099061787128448, |
|
"step": 1460 |
|
}, |
|
{ |
|
"epoch": 3.82, |
|
"learning_rate": 3.1402778309014284e-08, |
|
"logits/chosen": -1.3204753398895264, |
|
"logits/rejected": -1.3249455690383911, |
|
"logps/chosen": -33.896812438964844, |
|
"logps/rejected": -42.878475189208984, |
|
"loss": 0.9191, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.018276356160640717, |
|
"rewards/margins": 0.08088420331478119, |
|
"rewards/rejected": -0.0991605594754219, |
|
"step": 1470 |
|
}, |
|
{ |
|
"epoch": 3.84, |
|
"learning_rate": 2.3084278540791427e-08, |
|
"logits/chosen": -1.3220521211624146, |
|
"logits/rejected": -1.3311010599136353, |
|
"logps/chosen": -33.49136734008789, |
|
"logps/rejected": -38.53743362426758, |
|
"loss": 0.9289, |
|
"rewards/accuracies": 0.824999988079071, |
|
"rewards/chosen": -0.017961082980036736, |
|
"rewards/margins": 0.07112761586904526, |
|
"rewards/rejected": -0.08908869326114655, |
|
"step": 1480 |
|
}, |
|
{ |
|
"epoch": 3.87, |
|
"learning_rate": 1.6038302591975807e-08, |
|
"logits/chosen": -1.258320689201355, |
|
"logits/rejected": -1.2530874013900757, |
|
"logps/chosen": -36.1824951171875, |
|
"logps/rejected": -41.13605880737305, |
|
"loss": 0.936, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -0.02526010572910309, |
|
"rewards/margins": 0.06399567425251007, |
|
"rewards/rejected": -0.08925577253103256, |
|
"step": 1490 |
|
}, |
|
{ |
|
"epoch": 3.9, |
|
"learning_rate": 1.0268470356514237e-08, |
|
"logits/chosen": -1.3191821575164795, |
|
"logits/rejected": -1.3164719343185425, |
|
"logps/chosen": -36.40666961669922, |
|
"logps/rejected": -44.079254150390625, |
|
"loss": 0.9196, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -0.02781098522245884, |
|
"rewards/margins": 0.08037930727005005, |
|
"rewards/rejected": -0.10819028317928314, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 3.9, |
|
"eval_logits/chosen": -1.587085485458374, |
|
"eval_logits/rejected": -1.5831539630889893, |
|
"eval_logps/chosen": -39.70748519897461, |
|
"eval_logps/rejected": -44.0648307800293, |
|
"eval_loss": 0.9912670850753784, |
|
"eval_rewards/accuracies": 0.5423588156700134, |
|
"eval_rewards/chosen": -0.0567292720079422, |
|
"eval_rewards/margins": 0.008752820082008839, |
|
"eval_rewards/rejected": -0.06548209488391876, |
|
"eval_runtime": 145.8973, |
|
"eval_samples_per_second": 2.351, |
|
"eval_steps_per_second": 0.295, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 3.92, |
|
"learning_rate": 5.777746105209147e-09, |
|
"logits/chosen": -1.3802618980407715, |
|
"logits/rejected": -1.3801379203796387, |
|
"logps/chosen": -31.544485092163086, |
|
"logps/rejected": -42.76384735107422, |
|
"loss": 0.917, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -0.017282087355852127, |
|
"rewards/margins": 0.08295184373855591, |
|
"rewards/rejected": -0.10023393481969833, |
|
"step": 1510 |
|
}, |
|
{ |
|
"epoch": 3.95, |
|
"learning_rate": 2.5684369628148352e-09, |
|
"logits/chosen": -1.2462044954299927, |
|
"logits/rejected": -1.2448112964630127, |
|
"logps/chosen": -34.93328857421875, |
|
"logps/rejected": -43.264244079589844, |
|
"loss": 0.925, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -0.01891419291496277, |
|
"rewards/margins": 0.07499094307422638, |
|
"rewards/rejected": -0.09390512108802795, |
|
"step": 1520 |
|
}, |
|
{ |
|
"epoch": 3.97, |
|
"learning_rate": 6.421917227455999e-10, |
|
"logits/chosen": -1.3795892000198364, |
|
"logits/rejected": -1.3771770000457764, |
|
"logps/chosen": -33.931644439697266, |
|
"logps/rejected": -40.94723892211914, |
|
"loss": 0.9299, |
|
"rewards/accuracies": 0.8125, |
|
"rewards/chosen": -0.024575814604759216, |
|
"rewards/margins": 0.07006003707647324, |
|
"rewards/rejected": -0.09463585913181305, |
|
"step": 1530 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"learning_rate": 0.0, |
|
"logits/chosen": -1.3671410083770752, |
|
"logits/rejected": -1.3676393032073975, |
|
"logps/chosen": -33.82143020629883, |
|
"logps/rejected": -37.72132873535156, |
|
"loss": 0.9527, |
|
"rewards/accuracies": 0.762499988079071, |
|
"rewards/chosen": -0.0320148840546608, |
|
"rewards/margins": 0.04729658365249634, |
|
"rewards/rejected": -0.07931147515773773, |
|
"step": 1540 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"step": 1540, |
|
"total_flos": 0.0, |
|
"train_loss": 0.7654441864459546, |
|
"train_runtime": 10797.8934, |
|
"train_samples_per_second": 1.141, |
|
"train_steps_per_second": 0.143 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 1540, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 4, |
|
"save_steps": 100, |
|
"total_flos": 0.0, |
|
"train_batch_size": 4, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|