{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9982631930527722, "eval_steps": 400, "global_step": 467, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01068804275217101, "grad_norm": 56.89954388226368, "learning_rate": 1.0638297872340425e-07, "logits/chosen": -1.0110526084899902, "logits/rejected": -0.9819751977920532, "logps/chosen": -0.27428290247917175, "logps/rejected": -0.27152150869369507, "loss": 3.074, "rewards/accuracies": 0.4312500059604645, "rewards/chosen": -2.7428293228149414, "rewards/margins": -0.027614299207925797, "rewards/rejected": -2.7152152061462402, "step": 5 }, { "epoch": 0.02137608550434202, "grad_norm": 41.09804027603237, "learning_rate": 2.127659574468085e-07, "logits/chosen": -1.0431499481201172, "logits/rejected": -0.9762164354324341, "logps/chosen": -0.29434844851493835, "logps/rejected": -0.29961150884628296, "loss": 3.0104, "rewards/accuracies": 0.5, "rewards/chosen": -2.943484306335449, "rewards/margins": 0.052630893886089325, "rewards/rejected": -2.9961154460906982, "step": 10 }, { "epoch": 0.03206412825651302, "grad_norm": 54.12410975393932, "learning_rate": 3.1914893617021275e-07, "logits/chosen": -0.9705549478530884, "logits/rejected": -0.9907904863357544, "logps/chosen": -0.2644530236721039, "logps/rejected": -0.300653874874115, "loss": 2.9255, "rewards/accuracies": 0.59375, "rewards/chosen": -2.6445302963256836, "rewards/margins": 0.36200839281082153, "rewards/rejected": -3.0065386295318604, "step": 15 }, { "epoch": 0.04275217100868404, "grad_norm": 70.58531376681509, "learning_rate": 4.25531914893617e-07, "logits/chosen": -0.9636842608451843, "logits/rejected": -0.9370673298835754, "logps/chosen": -0.27763131260871887, "logps/rejected": -0.29128938913345337, "loss": 2.9498, "rewards/accuracies": 0.512499988079071, "rewards/chosen": -2.776313304901123, "rewards/margins": 0.13658040761947632, "rewards/rejected": -2.912893772125244, "step": 20 }, { "epoch": 0.053440213760855046, "grad_norm": 56.29890733120682, "learning_rate": 5.319148936170212e-07, "logits/chosen": -1.021801233291626, "logits/rejected": -0.9922389984130859, "logps/chosen": -0.2716570198535919, "logps/rejected": -0.27810558676719666, "loss": 2.996, "rewards/accuracies": 0.4937500059604645, "rewards/chosen": -2.7165703773498535, "rewards/margins": 0.06448549777269363, "rewards/rejected": -2.7810559272766113, "step": 25 }, { "epoch": 0.06412825651302605, "grad_norm": 43.82713664972921, "learning_rate": 6.382978723404255e-07, "logits/chosen": -0.9963313937187195, "logits/rejected": -0.9516829252243042, "logps/chosen": -0.2735390067100525, "logps/rejected": -0.2788829803466797, "loss": 2.9153, "rewards/accuracies": 0.4375, "rewards/chosen": -2.7353897094726562, "rewards/margins": 0.0534403920173645, "rewards/rejected": -2.788830041885376, "step": 30 }, { "epoch": 0.07481629926519706, "grad_norm": 70.24698865073822, "learning_rate": 7.446808510638297e-07, "logits/chosen": -1.052671194076538, "logits/rejected": -0.9770715832710266, "logps/chosen": -0.2941210865974426, "logps/rejected": -0.32225456833839417, "loss": 2.901, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -2.941210985183716, "rewards/margins": 0.2813349962234497, "rewards/rejected": -3.222545623779297, "step": 35 }, { "epoch": 0.08550434201736808, "grad_norm": 55.30141312601411, "learning_rate": 8.51063829787234e-07, "logits/chosen": -1.0156581401824951, "logits/rejected": -0.9718509912490845, "logps/chosen": -0.28176331520080566, "logps/rejected": -0.3227621614933014, "loss": 2.9376, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -2.8176331520080566, "rewards/margins": 0.40998831391334534, "rewards/rejected": -3.227621555328369, "step": 40 }, { "epoch": 0.09619238476953908, "grad_norm": 38.45710685921155, "learning_rate": 9.574468085106384e-07, "logits/chosen": -1.0545786619186401, "logits/rejected": -1.0107605457305908, "logps/chosen": -0.32836729288101196, "logps/rejected": -0.37582582235336304, "loss": 2.944, "rewards/accuracies": 0.53125, "rewards/chosen": -3.28367280960083, "rewards/margins": 0.4745853841304779, "rewards/rejected": -3.75825834274292, "step": 45 }, { "epoch": 0.10688042752171009, "grad_norm": 127.64245160013748, "learning_rate": 9.998741174712533e-07, "logits/chosen": -1.040578842163086, "logits/rejected": -0.9890502095222473, "logps/chosen": -0.3216663599014282, "logps/rejected": -0.3576571047306061, "loss": 3.0077, "rewards/accuracies": 0.48750001192092896, "rewards/chosen": -3.216662883758545, "rewards/margins": 0.35990777611732483, "rewards/rejected": -3.576570987701416, "step": 50 }, { "epoch": 0.11756847027388109, "grad_norm": 52.19635359720767, "learning_rate": 9.991050648838675e-07, "logits/chosen": -1.0747789144515991, "logits/rejected": -1.0378679037094116, "logps/chosen": -0.28550446033477783, "logps/rejected": -0.3416834771633148, "loss": 2.8502, "rewards/accuracies": 0.5874999761581421, "rewards/chosen": -2.8550446033477783, "rewards/margins": 0.5617905855178833, "rewards/rejected": -3.416834592819214, "step": 55 }, { "epoch": 0.1282565130260521, "grad_norm": 45.58873626472617, "learning_rate": 9.97637968732563e-07, "logits/chosen": -1.0949262380599976, "logits/rejected": -1.0604766607284546, "logps/chosen": -0.3153451085090637, "logps/rejected": -0.33443158864974976, "loss": 2.8532, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -3.1534507274627686, "rewards/margins": 0.19086508452892303, "rewards/rejected": -3.344316005706787, "step": 60 }, { "epoch": 0.13894455577822312, "grad_norm": 57.26986612669826, "learning_rate": 9.954748808839674e-07, "logits/chosen": -1.0162328481674194, "logits/rejected": -0.9866952896118164, "logps/chosen": -0.35168641805648804, "logps/rejected": -0.39575284719467163, "loss": 2.8058, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -3.516864061355591, "rewards/margins": 0.4406643509864807, "rewards/rejected": -3.9575283527374268, "step": 65 }, { "epoch": 0.14963259853039412, "grad_norm": 36.80891193149694, "learning_rate": 9.926188266120295e-07, "logits/chosen": -1.032679796218872, "logits/rejected": -1.009883999824524, "logps/chosen": -0.34069326519966125, "logps/rejected": -0.4097404479980469, "loss": 2.8492, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -3.406932830810547, "rewards/margins": 0.690471887588501, "rewards/rejected": -4.097404479980469, "step": 70 }, { "epoch": 0.16032064128256512, "grad_norm": 52.89939132813969, "learning_rate": 9.890738003669027e-07, "logits/chosen": -0.9840335845947266, "logits/rejected": -0.9154711961746216, "logps/chosen": -0.3410232663154602, "logps/rejected": -0.39265531301498413, "loss": 2.8212, "rewards/accuracies": 0.543749988079071, "rewards/chosen": -3.4102330207824707, "rewards/margins": 0.5163205862045288, "rewards/rejected": -3.926553249359131, "step": 75 }, { "epoch": 0.17100868403473615, "grad_norm": 43.121051189441694, "learning_rate": 9.848447601883433e-07, "logits/chosen": -0.9788318872451782, "logits/rejected": -0.9655908346176147, "logps/chosen": -0.3363918364048004, "logps/rejected": -0.4268081784248352, "loss": 2.7186, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": -3.3639183044433594, "rewards/margins": 0.9041633605957031, "rewards/rejected": -4.2680816650390625, "step": 80 }, { "epoch": 0.18169672678690715, "grad_norm": 54.440571510963004, "learning_rate": 9.799376207714444e-07, "logits/chosen": -1.003329873085022, "logits/rejected": -0.9817102551460266, "logps/chosen": -0.3231724500656128, "logps/rejected": -0.37237557768821716, "loss": 2.7361, "rewards/accuracies": 0.6312500238418579, "rewards/chosen": -3.231724262237549, "rewards/margins": 0.4920312464237213, "rewards/rejected": -3.7237555980682373, "step": 85 }, { "epoch": 0.19238476953907815, "grad_norm": 68.19418423227036, "learning_rate": 9.743592451943998e-07, "logits/chosen": -1.0398868322372437, "logits/rejected": -1.0045406818389893, "logps/chosen": -0.39259663224220276, "logps/rejected": -0.46695399284362793, "loss": 2.8464, "rewards/accuracies": 0.574999988079071, "rewards/chosen": -3.925966262817383, "rewards/margins": 0.7435733675956726, "rewards/rejected": -4.669539451599121, "step": 90 }, { "epoch": 0.20307281229124916, "grad_norm": 48.310300313424634, "learning_rate": 9.681174353198686e-07, "logits/chosen": -1.1072306632995605, "logits/rejected": -1.0259342193603516, "logps/chosen": -0.4255724549293518, "logps/rejected": -0.46118488907814026, "loss": 2.7441, "rewards/accuracies": 0.5562499761581421, "rewards/chosen": -4.2557244300842285, "rewards/margins": 0.3561245799064636, "rewards/rejected": -4.611849308013916, "step": 95 }, { "epoch": 0.21376085504342018, "grad_norm": 65.13984774258292, "learning_rate": 9.612209208833646e-07, "logits/chosen": -1.0061118602752686, "logits/rejected": -0.9820231199264526, "logps/chosen": -0.4049428403377533, "logps/rejected": -0.4770785868167877, "loss": 2.7252, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -4.049428462982178, "rewards/margins": 0.7213573455810547, "rewards/rejected": -4.770786285400391, "step": 100 }, { "epoch": 0.22444889779559118, "grad_norm": 67.14694271980007, "learning_rate": 9.536793472839324e-07, "logits/chosen": -1.0186512470245361, "logits/rejected": -0.9668880701065063, "logps/chosen": -0.40378451347351074, "logps/rejected": -0.4971798360347748, "loss": 2.7255, "rewards/accuracies": 0.6875, "rewards/chosen": -4.037845134735107, "rewards/margins": 0.9339532852172852, "rewards/rejected": -4.971798419952393, "step": 105 }, { "epoch": 0.23513694054776219, "grad_norm": 66.18323566645698, "learning_rate": 9.455032620941839e-07, "logits/chosen": -0.9757324457168579, "logits/rejected": -0.9178048372268677, "logps/chosen": -0.45775121450424194, "logps/rejected": -0.5744206309318542, "loss": 2.6384, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -4.577511787414551, "rewards/margins": 1.1666945219039917, "rewards/rejected": -5.744206428527832, "step": 110 }, { "epoch": 0.2458249832999332, "grad_norm": 59.4672474909725, "learning_rate": 9.367041003085648e-07, "logits/chosen": -1.0457837581634521, "logits/rejected": -0.9855262041091919, "logps/chosen": -0.5011107921600342, "logps/rejected": -0.5669585466384888, "loss": 2.5014, "rewards/accuracies": 0.643750011920929, "rewards/chosen": -5.011107444763184, "rewards/margins": 0.6584769487380981, "rewards/rejected": -5.66958475112915, "step": 115 }, { "epoch": 0.2565130260521042, "grad_norm": 66.03731752964929, "learning_rate": 9.272941683504808e-07, "logits/chosen": -0.9971206784248352, "logits/rejected": -0.9091175198554993, "logps/chosen": -0.5187912583351135, "logps/rejected": -0.7068670392036438, "loss": 2.3834, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -5.187912464141846, "rewards/margins": 1.8807573318481445, "rewards/rejected": -7.06866979598999, "step": 120 }, { "epoch": 0.26720106880427524, "grad_norm": 71.68229077695405, "learning_rate": 9.172866268606513e-07, "logits/chosen": -1.0721819400787354, "logits/rejected": -1.030027151107788, "logps/chosen": -0.5926575660705566, "logps/rejected": -0.6795297861099243, "loss": 2.3213, "rewards/accuracies": 0.6875, "rewards/chosen": -5.926576137542725, "rewards/margins": 0.8687216639518738, "rewards/rejected": -6.795297145843506, "step": 125 }, { "epoch": 0.27788911155644624, "grad_norm": 64.96923591468992, "learning_rate": 9.066954722907638e-07, "logits/chosen": -1.1085387468338013, "logits/rejected": -1.1034091711044312, "logps/chosen": -0.5947206020355225, "logps/rejected": -0.8544057607650757, "loss": 2.1425, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -5.947205543518066, "rewards/margins": 2.596850872039795, "rewards/rejected": -8.54405689239502, "step": 130 }, { "epoch": 0.28857715430861725, "grad_norm": 66.68661326155792, "learning_rate": 8.955355173281707e-07, "logits/chosen": -1.0726548433303833, "logits/rejected": -1.0262948274612427, "logps/chosen": -0.708818256855011, "logps/rejected": -0.8760965466499329, "loss": 2.0661, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -7.0881829261779785, "rewards/margins": 1.67278254032135, "rewards/rejected": -8.760965347290039, "step": 135 }, { "epoch": 0.29926519706078825, "grad_norm": 88.39987976267922, "learning_rate": 8.838223701790055e-07, "logits/chosen": -1.144708275794983, "logits/rejected": -1.1237266063690186, "logps/chosen": -0.8580458760261536, "logps/rejected": -1.0160259008407593, "loss": 2.0786, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -8.580458641052246, "rewards/margins": 1.5798008441925049, "rewards/rejected": -10.160260200500488, "step": 140 }, { "epoch": 0.30995323981295925, "grad_norm": 90.83612810193658, "learning_rate": 8.71572412738697e-07, "logits/chosen": -1.0413812398910522, "logits/rejected": -1.0159306526184082, "logps/chosen": -0.9142902493476868, "logps/rejected": -1.1587440967559814, "loss": 2.0446, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -9.142904281616211, "rewards/margins": 2.444538116455078, "rewards/rejected": -11.587442398071289, "step": 145 }, { "epoch": 0.32064128256513025, "grad_norm": 84.07324057563632, "learning_rate": 8.588027776804058e-07, "logits/chosen": -1.06027352809906, "logits/rejected": -1.0439517498016357, "logps/chosen": -1.0236120223999023, "logps/rejected": -1.2801861763000488, "loss": 1.9933, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -10.236120223999023, "rewards/margins": 2.565741777420044, "rewards/rejected": -12.801861763000488, "step": 150 }, { "epoch": 0.33132932531730125, "grad_norm": 83.2873729325751, "learning_rate": 8.455313244934324e-07, "logits/chosen": -1.0788214206695557, "logits/rejected": -1.059463620185852, "logps/chosen": -1.1024755239486694, "logps/rejected": -1.424791932106018, "loss": 2.0534, "rewards/accuracies": 0.71875, "rewards/chosen": -11.024755477905273, "rewards/margins": 3.2231624126434326, "rewards/rejected": -14.247919082641602, "step": 155 }, { "epoch": 0.3420173680694723, "grad_norm": 91.00920515798947, "learning_rate": 8.317766145051057e-07, "logits/chosen": -1.115094780921936, "logits/rejected": -1.0990221500396729, "logps/chosen": -1.2077633142471313, "logps/rejected": -1.5987141132354736, "loss": 1.9827, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -12.077634811401367, "rewards/margins": 3.909506320953369, "rewards/rejected": -15.987141609191895, "step": 160 }, { "epoch": 0.3527054108216433, "grad_norm": 65.16453388698586, "learning_rate": 8.175578849210894e-07, "logits/chosen": -1.1276443004608154, "logits/rejected": -1.106072187423706, "logps/chosen": -1.156090259552002, "logps/rejected": -1.5160917043685913, "loss": 1.8258, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -11.560903549194336, "rewards/margins": 3.6000137329101562, "rewards/rejected": -15.160917282104492, "step": 165 }, { "epoch": 0.3633934535738143, "grad_norm": 89.17851329246218, "learning_rate": 8.028950219204099e-07, "logits/chosen": -1.133327841758728, "logits/rejected": -1.1123793125152588, "logps/chosen": -1.1282349824905396, "logps/rejected": -1.5152461528778076, "loss": 1.805, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -11.282350540161133, "rewards/margins": 3.8701133728027344, "rewards/rejected": -15.15246295928955, "step": 170 }, { "epoch": 0.3740814963259853, "grad_norm": 90.91620530583262, "learning_rate": 7.878085328428368e-07, "logits/chosen": -1.1684499979019165, "logits/rejected": -1.1230958700180054, "logps/chosen": -1.199479341506958, "logps/rejected": -1.480486273765564, "loss": 1.7009, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -11.994794845581055, "rewards/margins": 2.8100688457489014, "rewards/rejected": -14.804861068725586, "step": 175 }, { "epoch": 0.3847695390781563, "grad_norm": 85.1133679017557, "learning_rate": 7.723195175075135e-07, "logits/chosen": -1.125327706336975, "logits/rejected": -1.107810139656067, "logps/chosen": -1.1546721458435059, "logps/rejected": -1.5301127433776855, "loss": 1.5941, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -11.546720504760742, "rewards/margins": 3.7544052600860596, "rewards/rejected": -15.301126480102539, "step": 180 }, { "epoch": 0.3954575818303273, "grad_norm": 80.80074018938376, "learning_rate": 7.564496387029531e-07, "logits/chosen": -1.1530572175979614, "logits/rejected": -1.1004054546356201, "logps/chosen": -1.1977171897888184, "logps/rejected": -1.6391410827636719, "loss": 1.6323, "rewards/accuracies": 0.84375, "rewards/chosen": -11.977171897888184, "rewards/margins": 4.414237976074219, "rewards/rejected": -16.39141273498535, "step": 185 }, { "epoch": 0.4061456245824983, "grad_norm": 87.8130211657885, "learning_rate": 7.402210918896689e-07, "logits/chosen": -1.1491271257400513, "logits/rejected": -1.159403681755066, "logps/chosen": -1.3192346096038818, "logps/rejected": -1.8000110387802124, "loss": 1.5253, "rewards/accuracies": 0.84375, "rewards/chosen": -13.192344665527344, "rewards/margins": 4.807766437530518, "rewards/rejected": -18.000110626220703, "step": 190 }, { "epoch": 0.4168336673346693, "grad_norm": 84.55372178403601, "learning_rate": 7.236565741578162e-07, "logits/chosen": -1.1186103820800781, "logits/rejected": -1.1020419597625732, "logps/chosen": -1.367811918258667, "logps/rejected": -1.7383407354354858, "loss": 1.6656, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -13.678120613098145, "rewards/margins": 3.705289363861084, "rewards/rejected": -17.38340950012207, "step": 195 }, { "epoch": 0.42752171008684037, "grad_norm": 84.20111385217919, "learning_rate": 7.067792524832603e-07, "logits/chosen": -1.1123974323272705, "logits/rejected": -1.1035972833633423, "logps/chosen": -1.3991031646728516, "logps/rejected": -1.8270728588104248, "loss": 1.514, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -13.991032600402832, "rewards/margins": 4.279696464538574, "rewards/rejected": -18.27073097229004, "step": 200 }, { "epoch": 0.43820975283901137, "grad_norm": 95.99111629240849, "learning_rate": 6.896127313264642e-07, "logits/chosen": -1.174342393875122, "logits/rejected": -1.1257025003433228, "logps/chosen": -1.463107705116272, "logps/rejected": -1.938598871231079, "loss": 1.6893, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -14.631075859069824, "rewards/margins": 4.754912853240967, "rewards/rejected": -19.385990142822266, "step": 205 }, { "epoch": 0.44889779559118237, "grad_norm": 100.92269633403131, "learning_rate": 6.721810196195174e-07, "logits/chosen": -1.2214075326919556, "logits/rejected": -1.2106014490127563, "logps/chosen": -1.468076467514038, "logps/rejected": -1.8905394077301025, "loss": 1.573, "rewards/accuracies": 0.8125, "rewards/chosen": -14.680766105651855, "rewards/margins": 4.224628925323486, "rewards/rejected": -18.905391693115234, "step": 210 }, { "epoch": 0.45958583834335337, "grad_norm": 98.80302089659659, "learning_rate": 6.545084971874736e-07, "logits/chosen": -1.1770942211151123, "logits/rejected": -1.161029577255249, "logps/chosen": -1.4916527271270752, "logps/rejected": -1.9983189105987549, "loss": 1.4988, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -14.916528701782227, "rewards/margins": 5.0666608810424805, "rewards/rejected": -19.98318862915039, "step": 215 }, { "epoch": 0.47027388109552437, "grad_norm": 94.29735384618333, "learning_rate": 6.3661988065096e-07, "logits/chosen": -1.2350765466690063, "logits/rejected": -1.2198480367660522, "logps/chosen": -1.5811980962753296, "logps/rejected": -2.1048731803894043, "loss": 1.3966, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -15.811983108520508, "rewards/margins": 5.236747741699219, "rewards/rejected": -21.048730850219727, "step": 220 }, { "epoch": 0.48096192384769537, "grad_norm": 73.78349743811617, "learning_rate": 6.185401888577487e-07, "logits/chosen": -1.2252886295318604, "logits/rejected": -1.195698618888855, "logps/chosen": -1.6302112340927124, "logps/rejected": -2.095083475112915, "loss": 1.4949, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -16.302112579345703, "rewards/margins": 4.6487226486206055, "rewards/rejected": -20.950834274291992, "step": 225 }, { "epoch": 0.4916499665998664, "grad_norm": 72.86824044049841, "learning_rate": 6.002947078916364e-07, "logits/chosen": -1.2797940969467163, "logits/rejected": -1.2363423109054565, "logps/chosen": -1.5519064664840698, "logps/rejected": -2.0577468872070312, "loss": 1.3568, "rewards/accuracies": 0.8125, "rewards/chosen": -15.519063949584961, "rewards/margins": 5.05840539932251, "rewards/rejected": -20.577470779418945, "step": 230 }, { "epoch": 0.5023380093520374, "grad_norm": 87.56760475487422, "learning_rate": 5.819089557075688e-07, "logits/chosen": -1.3070341348648071, "logits/rejected": -1.2829915285110474, "logps/chosen": -1.5769996643066406, "logps/rejected": -2.101868152618408, "loss": 1.4121, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -15.769996643066406, "rewards/margins": 5.248685359954834, "rewards/rejected": -21.0186824798584, "step": 235 }, { "epoch": 0.5130260521042084, "grad_norm": 97.95206336790469, "learning_rate": 5.634086464424742e-07, "logits/chosen": -1.2722411155700684, "logits/rejected": -1.2728557586669922, "logps/chosen": -1.5129470825195312, "logps/rejected": -2.0472776889801025, "loss": 1.4543, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -15.12946891784668, "rewards/margins": 5.343307018280029, "rewards/rejected": -20.4727783203125, "step": 240 }, { "epoch": 0.5237140948563794, "grad_norm": 91.68979328640371, "learning_rate": 5.448196544517167e-07, "logits/chosen": -1.3419743776321411, "logits/rejected": -1.2915674448013306, "logps/chosen": -1.5888983011245728, "logps/rejected": -2.2375972270965576, "loss": 1.3296, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -15.888982772827148, "rewards/margins": 6.486991882324219, "rewards/rejected": -22.375974655151367, "step": 245 }, { "epoch": 0.5344021376085505, "grad_norm": 90.04751296677803, "learning_rate": 5.26167978121472e-07, "logits/chosen": -1.3019921779632568, "logits/rejected": -1.287535309791565, "logps/chosen": -1.632799744606018, "logps/rejected": -2.2480623722076416, "loss": 1.3388, "rewards/accuracies": 0.8687499761581421, "rewards/chosen": -16.3279972076416, "rewards/margins": 6.1526265144348145, "rewards/rejected": -22.48062515258789, "step": 250 }, { "epoch": 0.5450901803607214, "grad_norm": 72.22778279348458, "learning_rate": 5.074797035076318e-07, "logits/chosen": -1.3449289798736572, "logits/rejected": -1.321571707725525, "logps/chosen": -1.6949834823608398, "logps/rejected": -2.1791787147521973, "loss": 1.4266, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -16.949832916259766, "rewards/margins": 4.84195613861084, "rewards/rejected": -21.791790008544922, "step": 255 }, { "epoch": 0.5557782231128925, "grad_norm": 82.32130834680915, "learning_rate": 4.887809678520975e-07, "logits/chosen": -1.3022774457931519, "logits/rejected": -1.2787460088729858, "logps/chosen": -1.6087223291397095, "logps/rejected": -2.112210750579834, "loss": 1.4418, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -16.087223052978516, "rewards/margins": 5.034885883331299, "rewards/rejected": -21.122108459472656, "step": 260 }, { "epoch": 0.5664662658650634, "grad_norm": 82.10988858763459, "learning_rate": 4.700979230274829e-07, "logits/chosen": -1.263526201248169, "logits/rejected": -1.2527527809143066, "logps/chosen": -1.7088844776153564, "logps/rejected": -2.2491986751556396, "loss": 1.2866, "rewards/accuracies": 0.8187500238418579, "rewards/chosen": -17.08884620666504, "rewards/margins": 5.40314245223999, "rewards/rejected": -22.491985321044922, "step": 265 }, { "epoch": 0.5771543086172345, "grad_norm": 74.22531378642971, "learning_rate": 4.514566989613559e-07, "logits/chosen": -1.276107907295227, "logits/rejected": -1.2517645359039307, "logps/chosen": -1.5033190250396729, "logps/rejected": -2.076201915740967, "loss": 1.2308, "rewards/accuracies": 0.84375, "rewards/chosen": -15.033190727233887, "rewards/margins": 5.728825569152832, "rewards/rejected": -20.762014389038086, "step": 270 }, { "epoch": 0.5878423513694054, "grad_norm": 66.56842436857252, "learning_rate": 4.328833670911724e-07, "logits/chosen": -1.2568957805633545, "logits/rejected": -1.2262030839920044, "logps/chosen": -1.562673568725586, "logps/rejected": -2.0573482513427734, "loss": 1.4699, "rewards/accuracies": 0.8062499761581421, "rewards/chosen": -15.626736640930176, "rewards/margins": 4.946745872497559, "rewards/rejected": -20.573482513427734, "step": 275 }, { "epoch": 0.5985303941215765, "grad_norm": 71.08794378986906, "learning_rate": 4.144039039010124e-07, "logits/chosen": -1.3348051309585571, "logits/rejected": -1.3145146369934082, "logps/chosen": -1.6333341598510742, "logps/rejected": -2.22784686088562, "loss": 1.3131, "rewards/accuracies": 0.8125, "rewards/chosen": -16.33333969116211, "rewards/margins": 5.945128440856934, "rewards/rejected": -22.27846908569336, "step": 280 }, { "epoch": 0.6092184368737475, "grad_norm": 93.72278509593903, "learning_rate": 3.960441545911204e-07, "logits/chosen": -1.297483205795288, "logits/rejected": -1.2733924388885498, "logps/chosen": -1.650708794593811, "logps/rejected": -2.298219680786133, "loss": 1.0059, "rewards/accuracies": 0.875, "rewards/chosen": -16.50708770751953, "rewards/margins": 6.4751081466674805, "rewards/rejected": -22.982196807861328, "step": 285 }, { "epoch": 0.6199064796259185, "grad_norm": 76.68999166591557, "learning_rate": 3.778297969310529e-07, "logits/chosen": -1.32032310962677, "logits/rejected": -1.2831968069076538, "logps/chosen": -1.706647276878357, "logps/rejected": -2.2490804195404053, "loss": 1.3171, "rewards/accuracies": 0.8125, "rewards/chosen": -17.066471099853516, "rewards/margins": 5.424333572387695, "rewards/rejected": -22.490802764892578, "step": 290 }, { "epoch": 0.6305945223780896, "grad_norm": 87.09453084365835, "learning_rate": 3.5978630534699865e-07, "logits/chosen": -1.2636520862579346, "logits/rejected": -1.2527822256088257, "logps/chosen": -1.6968777179718018, "logps/rejected": -2.257763624191284, "loss": 1.1554, "rewards/accuracies": 0.8187500238418579, "rewards/chosen": -16.96877670288086, "rewards/margins": 5.608861446380615, "rewards/rejected": -22.57763671875, "step": 295 }, { "epoch": 0.6412825651302605, "grad_norm": 82.76223672337171, "learning_rate": 3.4193891529348795e-07, "logits/chosen": -1.206903100013733, "logits/rejected": -1.1841256618499756, "logps/chosen": -1.7743552923202515, "logps/rejected": -2.2721633911132812, "loss": 1.5577, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -17.74355125427246, "rewards/margins": 4.978079795837402, "rewards/rejected": -22.72163200378418, "step": 300 }, { "epoch": 0.6519706078824316, "grad_norm": 78.16246422530884, "learning_rate": 3.243125879593286e-07, "logits/chosen": -1.3001482486724854, "logits/rejected": -1.2663486003875732, "logps/chosen": -1.6906330585479736, "logps/rejected": -2.1780858039855957, "loss": 1.3471, "rewards/accuracies": 0.8187500238418579, "rewards/chosen": -16.906330108642578, "rewards/margins": 4.874526023864746, "rewards/rejected": -21.78085708618164, "step": 305 }, { "epoch": 0.6626586506346025, "grad_norm": 91.76909016032246, "learning_rate": 3.069319753571269e-07, "logits/chosen": -1.3407663106918335, "logits/rejected": -1.3254361152648926, "logps/chosen": -1.6954444646835327, "logps/rejected": -2.250342845916748, "loss": 1.3722, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -16.95444679260254, "rewards/margins": 5.548983573913574, "rewards/rejected": -22.503429412841797, "step": 310 }, { "epoch": 0.6733466933867736, "grad_norm": 83.8485405296481, "learning_rate": 2.898213858452173e-07, "logits/chosen": -1.3287131786346436, "logits/rejected": -1.2830489873886108, "logps/chosen": -1.7108901739120483, "logps/rejected": -2.263986587524414, "loss": 1.2997, "rewards/accuracies": 0.8500000238418579, "rewards/chosen": -17.108901977539062, "rewards/margins": 5.530962944030762, "rewards/rejected": -22.63986587524414, "step": 315 }, { "epoch": 0.6840347361389446, "grad_norm": 110.4735398683659, "learning_rate": 2.730047501302266e-07, "logits/chosen": -1.3247960805892944, "logits/rejected": -1.3241022825241089, "logps/chosen": -1.7461349964141846, "logps/rejected": -2.435873508453369, "loss": 1.1823, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -17.461349487304688, "rewards/margins": 6.8973846435546875, "rewards/rejected": -24.358732223510742, "step": 320 }, { "epoch": 0.6947227788911156, "grad_norm": 83.494848255409, "learning_rate": 2.5650558779781635e-07, "logits/chosen": -1.332975149154663, "logits/rejected": -1.2938920259475708, "logps/chosen": -1.795236349105835, "logps/rejected": -2.5173287391662598, "loss": 1.2571, "rewards/accuracies": 0.856249988079071, "rewards/chosen": -17.952363967895508, "rewards/margins": 7.220925331115723, "rewards/rejected": -25.173290252685547, "step": 325 }, { "epoch": 0.7054108216432866, "grad_norm": 67.71202935932357, "learning_rate": 2.403469744184154e-07, "logits/chosen": -1.2552909851074219, "logits/rejected": -1.2208917140960693, "logps/chosen": -1.8006776571273804, "logps/rejected": -2.3320443630218506, "loss": 1.3218, "rewards/accuracies": 0.8500000238418579, "rewards/chosen": -18.006778717041016, "rewards/margins": 5.3136701583862305, "rewards/rejected": -23.32044792175293, "step": 330 }, { "epoch": 0.7160988643954576, "grad_norm": 100.05067628008052, "learning_rate": 2.2455150927394878e-07, "logits/chosen": -1.2909272909164429, "logits/rejected": -1.279365062713623, "logps/chosen": -1.7781559228897095, "logps/rejected": -2.3740553855895996, "loss": 1.2001, "rewards/accuracies": 0.8374999761581421, "rewards/chosen": -17.781558990478516, "rewards/margins": 5.9589948654174805, "rewards/rejected": -23.740550994873047, "step": 335 }, { "epoch": 0.7267869071476286, "grad_norm": 97.4036448779786, "learning_rate": 2.0914128375069722e-07, "logits/chosen": -1.2993793487548828, "logits/rejected": -1.2727620601654053, "logps/chosen": -1.6889365911483765, "logps/rejected": -2.2718539237976074, "loss": 1.3272, "rewards/accuracies": 0.8687499761581421, "rewards/chosen": -16.889366149902344, "rewards/margins": 5.829171180725098, "rewards/rejected": -22.718538284301758, "step": 340 }, { "epoch": 0.7374749498997996, "grad_norm": 80.17854676377132, "learning_rate": 1.9413785044249676e-07, "logits/chosen": -1.3267228603363037, "logits/rejected": -1.307962417602539, "logps/chosen": -1.7601877450942993, "logps/rejected": -2.4487035274505615, "loss": 1.268, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -17.601877212524414, "rewards/margins": 6.885159492492676, "rewards/rejected": -24.487037658691406, "step": 345 }, { "epoch": 0.7481629926519706, "grad_norm": 93.74425146775721, "learning_rate": 1.7956219300748792e-07, "logits/chosen": -1.301977515220642, "logits/rejected": -1.3074698448181152, "logps/chosen": -1.6333377361297607, "logps/rejected": -2.2102160453796387, "loss": 1.3055, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -16.333377838134766, "rewards/margins": 5.768781661987305, "rewards/rejected": -22.102161407470703, "step": 350 }, { "epoch": 0.7588510354041417, "grad_norm": 81.82295078846326, "learning_rate": 1.6543469682057104e-07, "logits/chosen": -1.231096625328064, "logits/rejected": -1.248560905456543, "logps/chosen": -1.6608985662460327, "logps/rejected": -2.275078535079956, "loss": 1.1623, "rewards/accuracies": 0.875, "rewards/chosen": -16.608983993530273, "rewards/margins": 6.141798496246338, "rewards/rejected": -22.750782012939453, "step": 355 }, { "epoch": 0.7695390781563126, "grad_norm": 83.05095937518779, "learning_rate": 1.5177512046261666e-07, "logits/chosen": -1.2908953428268433, "logits/rejected": -1.2959532737731934, "logps/chosen": -1.669966459274292, "logps/rejected": -2.3578410148620605, "loss": 1.2917, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -16.69966697692871, "rewards/margins": 6.878741264343262, "rewards/rejected": -23.578407287597656, "step": 360 }, { "epoch": 0.7802271209084837, "grad_norm": 74.89454351181128, "learning_rate": 1.3860256808630427e-07, "logits/chosen": -1.3286449909210205, "logits/rejected": -1.2749364376068115, "logps/chosen": -1.7136112451553345, "logps/rejected": -2.3888180255889893, "loss": 1.2062, "rewards/accuracies": 0.8500000238418579, "rewards/chosen": -17.136112213134766, "rewards/margins": 6.752066135406494, "rewards/rejected": -23.888179779052734, "step": 365 }, { "epoch": 0.7909151636606546, "grad_norm": 86.53704054424027, "learning_rate": 1.2593546269723647e-07, "logits/chosen": -1.251564860343933, "logits/rejected": -1.2445452213287354, "logps/chosen": -1.689417839050293, "logps/rejected": -2.242030620574951, "loss": 1.2397, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -16.89417839050293, "rewards/margins": 5.526127338409424, "rewards/rejected": -22.420307159423828, "step": 370 }, { "epoch": 0.8016032064128257, "grad_norm": 67.08154175834267, "learning_rate": 1.1379152038770029e-07, "logits/chosen": -1.2969611883163452, "logits/rejected": -1.3048449754714966, "logps/chosen": -1.8122236728668213, "logps/rejected": -2.4301235675811768, "loss": 1.1934, "rewards/accuracies": 0.875, "rewards/chosen": -18.122234344482422, "rewards/margins": 6.178996562957764, "rewards/rejected": -24.301233291625977, "step": 375 }, { "epoch": 0.8122912491649966, "grad_norm": 117.54208827116818, "learning_rate": 1.0218772555910954e-07, "logits/chosen": -1.2999813556671143, "logits/rejected": -1.2861607074737549, "logps/chosen": -1.7071046829223633, "logps/rejected": -2.316257953643799, "loss": 1.3229, "rewards/accuracies": 0.875, "rewards/chosen": -17.071046829223633, "rewards/margins": 6.091533184051514, "rewards/rejected": -23.162578582763672, "step": 380 }, { "epoch": 0.8229792919171677, "grad_norm": 81.44656810755579, "learning_rate": 9.114030716778432e-08, "logits/chosen": -1.2965797185897827, "logits/rejected": -1.2802777290344238, "logps/chosen": -1.7108449935913086, "logps/rejected": -2.4812264442443848, "loss": 1.0893, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -17.108448028564453, "rewards/margins": 7.703813076019287, "rewards/rejected": -24.8122615814209, "step": 385 }, { "epoch": 0.8336673346693386, "grad_norm": 77.70445335852864, "learning_rate": 8.066471602728803e-08, "logits/chosen": -1.3073532581329346, "logits/rejected": -1.298214316368103, "logps/chosen": -1.7747161388397217, "logps/rejected": -2.438636302947998, "loss": 1.243, "rewards/accuracies": 0.8687499761581421, "rewards/chosen": -17.747161865234375, "rewards/margins": 6.639204502105713, "rewards/rejected": -24.38636589050293, "step": 390 }, { "epoch": 0.8443553774215097, "grad_norm": 83.85991252299944, "learning_rate": 7.077560319906694e-08, "logits/chosen": -1.3133912086486816, "logits/rejected": -1.2960683107376099, "logps/chosen": -1.6985034942626953, "logps/rejected": -2.307638645172119, "loss": 1.1851, "rewards/accuracies": 0.84375, "rewards/chosen": -16.985034942626953, "rewards/margins": 6.091352462768555, "rewards/rejected": -23.07638931274414, "step": 395 }, { "epoch": 0.8550434201736807, "grad_norm": 71.13795001226813, "learning_rate": 6.148679950161672e-08, "logits/chosen": -1.3109194040298462, "logits/rejected": -1.2997634410858154, "logps/chosen": -1.7445151805877686, "logps/rejected": -2.31221604347229, "loss": 1.1882, "rewards/accuracies": 0.831250011920929, "rewards/chosen": -17.445152282714844, "rewards/margins": 5.677010536193848, "rewards/rejected": -23.122159957885742, "step": 400 }, { "epoch": 0.8550434201736807, "eval_logits/chosen": -1.4831929206848145, "eval_logits/rejected": -1.4942151308059692, "eval_logps/chosen": -1.7345198392868042, "eval_logps/rejected": -2.3351681232452393, "eval_loss": 1.17947518825531, "eval_rewards/accuracies": 0.8434959053993225, "eval_rewards/chosen": -17.345199584960938, "eval_rewards/margins": 6.006482124328613, "eval_rewards/rejected": -23.351680755615234, "eval_runtime": 66.8816, "eval_samples_per_second": 29.32, "eval_steps_per_second": 1.839, "step": 400 }, { "epoch": 0.8657314629258517, "grad_norm": 93.04526178302538, "learning_rate": 5.2811296166831666e-08, "logits/chosen": -1.2737401723861694, "logits/rejected": -1.2966753244400024, "logps/chosen": -1.8380101919174194, "logps/rejected": -2.4519333839416504, "loss": 1.1965, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -18.380102157592773, "rewards/margins": 6.139228820800781, "rewards/rejected": -24.51933479309082, "step": 405 }, { "epoch": 0.8764195056780227, "grad_norm": 146.600711895034, "learning_rate": 4.4761226670592066e-08, "logits/chosen": -1.29901921749115, "logits/rejected": -1.287117838859558, "logps/chosen": -1.744970679283142, "logps/rejected": -2.3512680530548096, "loss": 1.2833, "rewards/accuracies": 0.8500000238418579, "rewards/chosen": -17.44970703125, "rewards/margins": 6.062973499298096, "rewards/rejected": -23.512680053710938, "step": 410 }, { "epoch": 0.8871075484301937, "grad_norm": 75.88048221665215, "learning_rate": 3.734784976300165e-08, "logits/chosen": -1.2868596315383911, "logits/rejected": -1.2446833848953247, "logps/chosen": -1.6970937252044678, "logps/rejected": -2.4172720909118652, "loss": 1.2743, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -16.970937728881836, "rewards/margins": 7.201780796051025, "rewards/rejected": -24.172718048095703, "step": 415 }, { "epoch": 0.8977955911823647, "grad_norm": 89.93720385918785, "learning_rate": 3.058153372200695e-08, "logits/chosen": -1.322579026222229, "logits/rejected": -1.2816030979156494, "logps/chosen": -1.6371269226074219, "logps/rejected": -2.3092269897460938, "loss": 1.1839, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -16.371267318725586, "rewards/margins": 6.720999240875244, "rewards/rejected": -23.092267990112305, "step": 420 }, { "epoch": 0.9084836339345357, "grad_norm": 87.9502722305557, "learning_rate": 2.4471741852423233e-08, "logits/chosen": -1.320749044418335, "logits/rejected": -1.3164869546890259, "logps/chosen": -1.8138576745986938, "logps/rejected": -2.4155681133270264, "loss": 1.3872, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -18.13857650756836, "rewards/margins": 6.0171027183532715, "rewards/rejected": -24.15567970275879, "step": 425 }, { "epoch": 0.9191716766867067, "grad_norm": 86.56225821481218, "learning_rate": 1.9027019250647036e-08, "logits/chosen": -1.2986432313919067, "logits/rejected": -1.2903183698654175, "logps/chosen": -1.7905200719833374, "logps/rejected": -2.436361789703369, "loss": 1.2511, "rewards/accuracies": 0.8812500238418579, "rewards/chosen": -17.905202865600586, "rewards/margins": 6.458415985107422, "rewards/rejected": -24.363616943359375, "step": 430 }, { "epoch": 0.9298597194388778, "grad_norm": 72.5185778330974, "learning_rate": 1.4254980853566246e-08, "logits/chosen": -1.2492395639419556, "logits/rejected": -1.2206861972808838, "logps/chosen": -1.657965064048767, "logps/rejected": -2.2997374534606934, "loss": 1.1762, "rewards/accuracies": 0.875, "rewards/chosen": -16.57965087890625, "rewards/margins": 6.417726039886475, "rewards/rejected": -22.99737548828125, "step": 435 }, { "epoch": 0.9405477621910487, "grad_norm": 93.87912624876707, "learning_rate": 1.016230078838226e-08, "logits/chosen": -1.2905051708221436, "logits/rejected": -1.2409080266952515, "logps/chosen": -1.7906429767608643, "logps/rejected": -2.375685453414917, "loss": 1.2152, "rewards/accuracies": 0.875, "rewards/chosen": -17.906429290771484, "rewards/margins": 5.850423812866211, "rewards/rejected": -23.756855010986328, "step": 440 }, { "epoch": 0.9512358049432198, "grad_norm": 68.35754825310542, "learning_rate": 6.754703038239329e-09, "logits/chosen": -1.238844633102417, "logits/rejected": -1.2277412414550781, "logps/chosen": -1.769052267074585, "logps/rejected": -2.5054378509521484, "loss": 1.1183, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -17.690523147583008, "rewards/margins": 7.363855838775635, "rewards/rejected": -25.054378509521484, "step": 445 }, { "epoch": 0.9619238476953907, "grad_norm": 83.54543382327525, "learning_rate": 4.036953436716895e-09, "logits/chosen": -1.346346139907837, "logits/rejected": -1.3324317932128906, "logps/chosen": -1.7270358800888062, "logps/rejected": -2.3151395320892334, "loss": 1.242, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -17.27035903930664, "rewards/margins": 5.881035804748535, "rewards/rejected": -23.15139389038086, "step": 450 }, { "epoch": 0.9726118904475618, "grad_norm": 96.76907591863952, "learning_rate": 2.0128530023804656e-09, "logits/chosen": -1.3069255352020264, "logits/rejected": -1.277972936630249, "logps/chosen": -1.7244638204574585, "logps/rejected": -2.4305450916290283, "loss": 1.1012, "rewards/accuracies": 0.9312499761581421, "rewards/chosen": -17.244640350341797, "rewards/margins": 7.060812473297119, "rewards/rejected": -24.305450439453125, "step": 455 }, { "epoch": 0.9832999331997327, "grad_norm": 95.7750256826231, "learning_rate": 6.852326227130833e-10, "logits/chosen": -1.3161975145339966, "logits/rejected": -1.3086355924606323, "logps/chosen": -1.768693208694458, "logps/rejected": -2.425410747528076, "loss": 1.1545, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -17.686931610107422, "rewards/margins": 6.567179203033447, "rewards/rejected": -24.25411033630371, "step": 460 }, { "epoch": 0.9939879759519038, "grad_norm": 90.4800906374845, "learning_rate": 5.594909486328348e-11, "logits/chosen": -1.2925775051116943, "logits/rejected": -1.3013789653778076, "logps/chosen": -1.8294912576675415, "logps/rejected": -2.505312204360962, "loss": 1.2602, "rewards/accuracies": 0.84375, "rewards/chosen": -18.294912338256836, "rewards/margins": 6.758206844329834, "rewards/rejected": -25.053119659423828, "step": 465 }, { "epoch": 0.9982631930527722, "step": 467, "total_flos": 0.0, "train_loss": 1.7882541670789045, "train_runtime": 8154.3088, "train_samples_per_second": 7.343, "train_steps_per_second": 0.057 } ], "logging_steps": 5, "max_steps": 467, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 1000000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 2, "trial_name": null, "trial_params": null }