| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.10976647183117917, | |
| "eval_steps": 500, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.999970269475589e-06, | |
| "logits/chosen": -2.748720645904541, | |
| "logits/rejected": -3.5176024436950684, | |
| "logps/chosen": -165.87062072753906, | |
| "logps/rejected": -23.556869506835938, | |
| "loss": 1.5949, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 11.860421180725098, | |
| "rewards/margins": 2.147804021835327, | |
| "rewards/rejected": 9.712616920471191, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.999881078255916e-06, | |
| "logits/chosen": -3.261986255645752, | |
| "logits/rejected": -4.041715145111084, | |
| "logps/chosen": -153.11898803710938, | |
| "logps/rejected": -24.845947265625, | |
| "loss": 1.6208, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 12.161478042602539, | |
| "rewards/margins": 2.482882499694824, | |
| "rewards/rejected": 9.678595542907715, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.99973242740166e-06, | |
| "logits/chosen": -2.687460422515869, | |
| "logits/rejected": -3.837062358856201, | |
| "logps/chosen": -148.68341064453125, | |
| "logps/rejected": -34.57600402832031, | |
| "loss": 1.497, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 12.840588569641113, | |
| "rewards/margins": 2.2458953857421875, | |
| "rewards/rejected": 10.59469223022461, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.99954780509173e-06, | |
| "logits/chosen": -2.9765782356262207, | |
| "logits/rejected": -4.164349555969238, | |
| "logps/chosen": -201.6632537841797, | |
| "logps/rejected": -29.7652645111084, | |
| "loss": 1.8816, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 12.654102325439453, | |
| "rewards/margins": 2.569594383239746, | |
| "rewards/rejected": 10.084507942199707, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.999286186386431e-06, | |
| "logits/chosen": -3.153574228286743, | |
| "logits/rejected": -3.934051036834717, | |
| "logps/chosen": -138.03778076171875, | |
| "logps/rejected": -30.519153594970703, | |
| "loss": 1.9103, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 11.428102493286133, | |
| "rewards/margins": 1.7032209634780884, | |
| "rewards/rejected": 9.724882125854492, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.998965115121132e-06, | |
| "logits/chosen": -3.201958417892456, | |
| "logits/rejected": -4.274596691131592, | |
| "logps/chosen": -166.93260192871094, | |
| "logps/rejected": -35.59614944458008, | |
| "loss": 2.0164, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 11.013410568237305, | |
| "rewards/margins": 1.4899368286132812, | |
| "rewards/rejected": 9.523473739624023, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.998584595114074e-06, | |
| "logits/chosen": -3.6508994102478027, | |
| "logits/rejected": -4.988117694854736, | |
| "logps/chosen": -192.28738403320312, | |
| "logps/rejected": -39.88438415527344, | |
| "loss": 0.7081, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 12.313508987426758, | |
| "rewards/margins": 4.361058712005615, | |
| "rewards/rejected": 7.952448844909668, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.998144630890485e-06, | |
| "logits/chosen": -3.2785956859588623, | |
| "logits/rejected": -4.609368801116943, | |
| "logps/chosen": -177.5666961669922, | |
| "logps/rejected": -48.05603790283203, | |
| "loss": 1.1081, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 12.321931838989258, | |
| "rewards/margins": 4.14510440826416, | |
| "rewards/rejected": 8.176828384399414, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.997645227682513e-06, | |
| "logits/chosen": -3.4411368370056152, | |
| "logits/rejected": -4.750131607055664, | |
| "logps/chosen": -163.21450805664062, | |
| "logps/rejected": -53.2952880859375, | |
| "loss": 1.1417, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 11.179890632629395, | |
| "rewards/margins": 3.490216016769409, | |
| "rewards/rejected": 7.689675331115723, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.997086391429161e-06, | |
| "logits/chosen": -3.429443836212158, | |
| "logits/rejected": -4.5238776206970215, | |
| "logps/chosen": -176.43289184570312, | |
| "logps/rejected": -65.92268371582031, | |
| "loss": 0.9445, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 10.390297889709473, | |
| "rewards/margins": 3.6286094188690186, | |
| "rewards/rejected": 6.761689186096191, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.99646812877623e-06, | |
| "logits/chosen": -3.4572105407714844, | |
| "logits/rejected": -4.460841655731201, | |
| "logps/chosen": -170.5079803466797, | |
| "logps/rejected": -68.17359161376953, | |
| "loss": 0.6313, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 10.23057746887207, | |
| "rewards/margins": 4.633530616760254, | |
| "rewards/rejected": 5.5970458984375, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.995790447076232e-06, | |
| "logits/chosen": -3.5251846313476562, | |
| "logits/rejected": -4.237984657287598, | |
| "logps/chosen": -163.4308624267578, | |
| "logps/rejected": -69.18915557861328, | |
| "loss": 0.8116, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": 10.12580394744873, | |
| "rewards/margins": 4.935164451599121, | |
| "rewards/rejected": 5.190640449523926, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.995053354388295e-06, | |
| "logits/chosen": -3.2857518196105957, | |
| "logits/rejected": -4.0882792472839355, | |
| "logps/chosen": -211.8911590576172, | |
| "logps/rejected": -76.1227035522461, | |
| "loss": 0.1832, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 10.317492485046387, | |
| "rewards/margins": 5.812389373779297, | |
| "rewards/rejected": 4.505103588104248, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.994256859478082e-06, | |
| "logits/chosen": -3.0444934368133545, | |
| "logits/rejected": -3.676633358001709, | |
| "logps/chosen": -229.4088134765625, | |
| "logps/rejected": -82.24313354492188, | |
| "loss": 0.3931, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 10.056417465209961, | |
| "rewards/margins": 5.8769307136535645, | |
| "rewards/rejected": 4.17948579788208, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.993489232972927e-06, | |
| "logits/chosen": -3.0874276161193848, | |
| "logits/rejected": -3.6212425231933594, | |
| "logps/chosen": -163.163330078125, | |
| "logps/rejected": -76.16889953613281, | |
| "loss": 0.7483, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 9.392515182495117, | |
| "rewards/margins": 5.074815273284912, | |
| "rewards/rejected": 4.3177008628845215, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.992579900519751e-06, | |
| "logits/chosen": -2.9442169666290283, | |
| "logits/rejected": -3.488318681716919, | |
| "logps/chosen": -202.21347045898438, | |
| "logps/rejected": -84.56549072265625, | |
| "loss": 0.3173, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 9.955198287963867, | |
| "rewards/margins": 6.719505310058594, | |
| "rewards/rejected": 3.2356934547424316, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.99161119525913e-06, | |
| "logits/chosen": -2.819887161254883, | |
| "logits/rejected": -3.2742202281951904, | |
| "logps/chosen": -212.2724151611328, | |
| "logps/rejected": -90.73429870605469, | |
| "loss": 0.3464, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": 9.708670616149902, | |
| "rewards/margins": 5.832002639770508, | |
| "rewards/rejected": 3.8766684532165527, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.990688606280938e-06, | |
| "logits/chosen": -2.7344448566436768, | |
| "logits/rejected": -2.890084743499756, | |
| "logps/chosen": -184.69085693359375, | |
| "logps/rejected": -89.32521057128906, | |
| "loss": 0.6789, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": 9.630158424377441, | |
| "rewards/margins": 5.920450687408447, | |
| "rewards/rejected": 3.7097079753875732, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.989607125007361e-06, | |
| "logits/chosen": -2.5483996868133545, | |
| "logits/rejected": -2.572554588317871, | |
| "logps/chosen": -172.837158203125, | |
| "logps/rejected": -92.3455581665039, | |
| "loss": 0.4471, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": 8.161615371704102, | |
| "rewards/margins": 4.883408069610596, | |
| "rewards/rejected": 3.2782065868377686, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.98846630627921e-06, | |
| "logits/chosen": -2.7279348373413086, | |
| "logits/rejected": -2.8250045776367188, | |
| "logps/chosen": -216.2030487060547, | |
| "logps/rejected": -94.36392211914062, | |
| "loss": 0.1686, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 9.2470703125, | |
| "rewards/margins": 6.067578315734863, | |
| "rewards/rejected": 3.1794912815093994, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.98726616366334e-06, | |
| "logits/chosen": -2.476562261581421, | |
| "logits/rejected": -2.579338550567627, | |
| "logps/chosen": -169.55120849609375, | |
| "logps/rejected": -101.41313171386719, | |
| "loss": 0.3145, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": 8.830245018005371, | |
| "rewards/margins": 5.8404154777526855, | |
| "rewards/rejected": 2.9898295402526855, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.986006711432099e-06, | |
| "logits/chosen": -2.6179986000061035, | |
| "logits/rejected": -2.938459873199463, | |
| "logps/chosen": -239.5026397705078, | |
| "logps/rejected": -93.572998046875, | |
| "loss": 0.136, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 10.406990051269531, | |
| "rewards/margins": 7.794189453125, | |
| "rewards/rejected": 2.612799882888794, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.984687964563159e-06, | |
| "logits/chosen": -2.4739646911621094, | |
| "logits/rejected": -2.5754473209381104, | |
| "logps/chosen": -178.31881713867188, | |
| "logps/rejected": -95.78547668457031, | |
| "loss": 0.2827, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 8.740915298461914, | |
| "rewards/margins": 6.064324855804443, | |
| "rewards/rejected": 2.676591396331787, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.983309938739332e-06, | |
| "logits/chosen": -2.6331839561462402, | |
| "logits/rejected": -2.8605971336364746, | |
| "logps/chosen": -229.9063720703125, | |
| "logps/rejected": -90.2779541015625, | |
| "loss": 0.0817, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 10.207468032836914, | |
| "rewards/margins": 7.982739448547363, | |
| "rewards/rejected": 2.2247281074523926, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.98187265034839e-06, | |
| "logits/chosen": -2.3521595001220703, | |
| "logits/rejected": -2.5626819133758545, | |
| "logps/chosen": -190.8962860107422, | |
| "logps/rejected": -104.4652099609375, | |
| "loss": 0.2826, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 8.663321495056152, | |
| "rewards/margins": 6.1150994300842285, | |
| "rewards/rejected": 2.5482218265533447, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.980376116482871e-06, | |
| "logits/chosen": -2.4073798656463623, | |
| "logits/rejected": -2.531911849975586, | |
| "logps/chosen": -204.7512664794922, | |
| "logps/rejected": -108.78132629394531, | |
| "loss": 0.2884, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 9.377767562866211, | |
| "rewards/margins": 6.906778812408447, | |
| "rewards/rejected": 2.470989227294922, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.978820354939866e-06, | |
| "logits/chosen": -2.2742762565612793, | |
| "logits/rejected": -2.1533379554748535, | |
| "logps/chosen": -189.8150177001953, | |
| "logps/rejected": -112.7627944946289, | |
| "loss": 0.4643, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 9.204684257507324, | |
| "rewards/margins": 6.9484758377075195, | |
| "rewards/rejected": 2.2562096118927, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.97720538422082e-06, | |
| "logits/chosen": -2.1974568367004395, | |
| "logits/rejected": -2.233710765838623, | |
| "logps/chosen": -198.45361328125, | |
| "logps/rejected": -111.978515625, | |
| "loss": 0.1247, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 10.231064796447754, | |
| "rewards/margins": 8.133554458618164, | |
| "rewards/rejected": 2.0975089073181152, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.975531223531301e-06, | |
| "logits/chosen": -2.2490475177764893, | |
| "logits/rejected": -2.30739164352417, | |
| "logps/chosen": -187.04519653320312, | |
| "logps/rejected": -109.82148742675781, | |
| "loss": 0.1268, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 8.633919715881348, | |
| "rewards/margins": 6.911574363708496, | |
| "rewards/rejected": 1.7223446369171143, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.973797892780783e-06, | |
| "logits/chosen": -2.035712718963623, | |
| "logits/rejected": -1.961692452430725, | |
| "logps/chosen": -179.4069366455078, | |
| "logps/rejected": -116.109130859375, | |
| "loss": 0.0938, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 9.934889793395996, | |
| "rewards/margins": 8.173246383666992, | |
| "rewards/rejected": 1.7616443634033203, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.972005412582397e-06, | |
| "logits/chosen": -2.2436442375183105, | |
| "logits/rejected": -1.8246514797210693, | |
| "logps/chosen": -201.26980590820312, | |
| "logps/rejected": -117.66944885253906, | |
| "loss": 0.2073, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 8.626791954040527, | |
| "rewards/margins": 7.419760227203369, | |
| "rewards/rejected": 1.207031488418579, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.970153804252692e-06, | |
| "logits/chosen": -1.8984334468841553, | |
| "logits/rejected": -1.6126270294189453, | |
| "logps/chosen": -185.87290954589844, | |
| "logps/rejected": -117.4137954711914, | |
| "loss": 0.278, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 8.371435165405273, | |
| "rewards/margins": 7.151828765869141, | |
| "rewards/rejected": 1.2196067571640015, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.968243089811384e-06, | |
| "logits/chosen": -1.771867036819458, | |
| "logits/rejected": -1.5103999376296997, | |
| "logps/chosen": -192.2825469970703, | |
| "logps/rejected": -118.695556640625, | |
| "loss": 0.0729, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 8.544290542602539, | |
| "rewards/margins": 7.602395057678223, | |
| "rewards/rejected": 0.9418962597846985, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.96627329198109e-06, | |
| "logits/chosen": -1.924703598022461, | |
| "logits/rejected": -1.3986445665359497, | |
| "logps/chosen": -203.59628295898438, | |
| "logps/rejected": -128.64163208007812, | |
| "loss": 0.2888, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 8.449407577514648, | |
| "rewards/margins": 7.650280952453613, | |
| "rewards/rejected": 0.7991273999214172, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.96424443418706e-06, | |
| "logits/chosen": -1.8718311786651611, | |
| "logits/rejected": -1.552980661392212, | |
| "logps/chosen": -181.55087280273438, | |
| "logps/rejected": -120.4705581665039, | |
| "loss": 0.0762, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 8.385054588317871, | |
| "rewards/margins": 7.822892665863037, | |
| "rewards/rejected": 0.5621613264083862, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.962156540556895e-06, | |
| "logits/chosen": -1.7886946201324463, | |
| "logits/rejected": -1.5114587545394897, | |
| "logps/chosen": -183.34066772460938, | |
| "logps/rejected": -117.0723648071289, | |
| "loss": 0.1247, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 7.962329864501953, | |
| "rewards/margins": 7.366267204284668, | |
| "rewards/rejected": 0.5960620045661926, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.960009635920264e-06, | |
| "logits/chosen": -1.788691520690918, | |
| "logits/rejected": -1.3814653158187866, | |
| "logps/chosen": -197.45651245117188, | |
| "logps/rejected": -127.6612319946289, | |
| "loss": 0.076, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 9.019963264465332, | |
| "rewards/margins": 8.529821395874023, | |
| "rewards/rejected": 0.49014148116111755, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.95780374580861e-06, | |
| "logits/chosen": -1.5638988018035889, | |
| "logits/rejected": -1.1953327655792236, | |
| "logps/chosen": -189.52597045898438, | |
| "logps/rejected": -122.5257797241211, | |
| "loss": 0.2198, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 7.37133264541626, | |
| "rewards/margins": 6.876678466796875, | |
| "rewards/rejected": 0.4946538507938385, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.955538896454835e-06, | |
| "logits/chosen": -1.5664819478988647, | |
| "logits/rejected": -0.962997317314148, | |
| "logps/chosen": -204.67657470703125, | |
| "logps/rejected": -129.93968200683594, | |
| "loss": 0.0446, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 7.449831485748291, | |
| "rewards/margins": 7.785650730133057, | |
| "rewards/rejected": -0.33581945300102234, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.95321511479301e-06, | |
| "logits/chosen": -1.7048896551132202, | |
| "logits/rejected": -1.1939642429351807, | |
| "logps/chosen": -223.2345428466797, | |
| "logps/rejected": -130.3861846923828, | |
| "loss": 0.1745, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 8.562383651733398, | |
| "rewards/margins": 8.441825866699219, | |
| "rewards/rejected": 0.12055821716785431, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.950832428458025e-06, | |
| "logits/chosen": -1.2762507200241089, | |
| "logits/rejected": -0.9500716924667358, | |
| "logps/chosen": -195.8381805419922, | |
| "logps/rejected": -124.5715560913086, | |
| "loss": 0.1329, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 7.099652290344238, | |
| "rewards/margins": 7.511727809906006, | |
| "rewards/rejected": -0.41207608580589294, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.948390865785295e-06, | |
| "logits/chosen": -1.151587963104248, | |
| "logits/rejected": -0.925134539604187, | |
| "logps/chosen": -199.50967407226562, | |
| "logps/rejected": -137.6339111328125, | |
| "loss": 0.1983, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.476855278015137, | |
| "rewards/margins": 7.712057590484619, | |
| "rewards/rejected": -0.23520107567310333, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.94589045581039e-06, | |
| "logits/chosen": -1.0321991443634033, | |
| "logits/rejected": -0.8457671999931335, | |
| "logps/chosen": -206.13467407226562, | |
| "logps/rejected": -130.86544799804688, | |
| "loss": 0.2554, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 7.223177433013916, | |
| "rewards/margins": 7.446470737457275, | |
| "rewards/rejected": -0.2232942134141922, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.943331228268711e-06, | |
| "logits/chosen": -1.1853946447372437, | |
| "logits/rejected": -0.7296536564826965, | |
| "logps/chosen": -217.77304077148438, | |
| "logps/rejected": -127.19853210449219, | |
| "loss": 0.1477, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 7.660037040710449, | |
| "rewards/margins": 8.213974952697754, | |
| "rewards/rejected": -0.5539380311965942, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.94071321359513e-06, | |
| "logits/chosen": -0.9670246243476868, | |
| "logits/rejected": -0.314952552318573, | |
| "logps/chosen": -225.93533325195312, | |
| "logps/rejected": -154.71826171875, | |
| "loss": 0.172, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 7.689838409423828, | |
| "rewards/margins": 8.839930534362793, | |
| "rewards/rejected": -1.1500914096832275, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.938036442923624e-06, | |
| "logits/chosen": -0.9485570788383484, | |
| "logits/rejected": -0.33327311277389526, | |
| "logps/chosen": -225.99398803710938, | |
| "logps/rejected": -142.39173889160156, | |
| "loss": 0.0579, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.543333530426025, | |
| "rewards/margins": 8.586603164672852, | |
| "rewards/rejected": -1.0432696342468262, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.935300948086915e-06, | |
| "logits/chosen": -0.8986530303955078, | |
| "logits/rejected": -0.400226354598999, | |
| "logps/chosen": -209.4599609375, | |
| "logps/rejected": -130.23941040039062, | |
| "loss": 0.1114, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.468968391418457, | |
| "rewards/margins": 8.008504867553711, | |
| "rewards/rejected": -0.5395378470420837, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.932506761616078e-06, | |
| "logits/chosen": -0.7118775248527527, | |
| "logits/rejected": -0.03559774532914162, | |
| "logps/chosen": -193.45834350585938, | |
| "logps/rejected": -131.99844360351562, | |
| "loss": 0.0508, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 7.474764823913574, | |
| "rewards/margins": 8.308823585510254, | |
| "rewards/rejected": -0.8340586423873901, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.929653916740168e-06, | |
| "logits/chosen": -1.1100711822509766, | |
| "logits/rejected": -0.40355220437049866, | |
| "logps/chosen": -242.05697631835938, | |
| "logps/rejected": -139.188232421875, | |
| "loss": 0.1127, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 8.726511001586914, | |
| "rewards/margins": 9.716850280761719, | |
| "rewards/rejected": -0.9903383255004883, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.92674244738581e-06, | |
| "logits/chosen": -0.8486649394035339, | |
| "logits/rejected": 0.05283903330564499, | |
| "logps/chosen": -155.493408203125, | |
| "logps/rejected": -141.4785919189453, | |
| "loss": 0.0901, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 7.535587310791016, | |
| "rewards/margins": 8.244181632995605, | |
| "rewards/rejected": -0.7085943818092346, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.923772388176812e-06, | |
| "logits/chosen": -0.907477080821991, | |
| "logits/rejected": -0.171137273311615, | |
| "logps/chosen": -236.19198608398438, | |
| "logps/rejected": -133.06643676757812, | |
| "loss": 0.0178, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 8.850728034973145, | |
| "rewards/margins": 10.053266525268555, | |
| "rewards/rejected": -1.202539086341858, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.92074377443374e-06, | |
| "logits/chosen": -1.0009920597076416, | |
| "logits/rejected": -0.20153608918190002, | |
| "logps/chosen": -178.60147094726562, | |
| "logps/rejected": -143.53695678710938, | |
| "loss": 0.1976, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 7.2967023849487305, | |
| "rewards/margins": 8.339073181152344, | |
| "rewards/rejected": -1.0423698425292969, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.917656642173505e-06, | |
| "logits/chosen": -1.0008546113967896, | |
| "logits/rejected": -0.01453827042132616, | |
| "logps/chosen": -206.38461303710938, | |
| "logps/rejected": -144.1584014892578, | |
| "loss": 0.1575, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 8.916428565979004, | |
| "rewards/margins": 9.663307189941406, | |
| "rewards/rejected": -0.7468785047531128, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.91451102810893e-06, | |
| "logits/chosen": -1.326416015625, | |
| "logits/rejected": -0.38943323493003845, | |
| "logps/chosen": -220.2677764892578, | |
| "logps/rejected": -130.0069122314453, | |
| "loss": 0.054, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 8.959368705749512, | |
| "rewards/margins": 9.750789642333984, | |
| "rewards/rejected": -0.7914207577705383, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.911306969648318e-06, | |
| "logits/chosen": -0.932703971862793, | |
| "logits/rejected": -0.0587722584605217, | |
| "logps/chosen": -205.8779296875, | |
| "logps/rejected": -135.040283203125, | |
| "loss": 0.1965, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 8.557300567626953, | |
| "rewards/margins": 9.26187515258789, | |
| "rewards/rejected": -0.7045742869377136, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.908044504895004e-06, | |
| "logits/chosen": -0.9326754808425903, | |
| "logits/rejected": -0.19441956281661987, | |
| "logps/chosen": -195.15328979492188, | |
| "logps/rejected": -131.31869506835938, | |
| "loss": 0.0943, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 8.366579055786133, | |
| "rewards/margins": 9.570085525512695, | |
| "rewards/rejected": -1.203505277633667, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.904723672646903e-06, | |
| "logits/chosen": -0.9648317098617554, | |
| "logits/rejected": -0.06580810248851776, | |
| "logps/chosen": -211.77633666992188, | |
| "logps/rejected": -136.65756225585938, | |
| "loss": 0.2817, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 8.721487045288086, | |
| "rewards/margins": 9.342923164367676, | |
| "rewards/rejected": -0.6214349865913391, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.901344512396049e-06, | |
| "logits/chosen": -1.1457253694534302, | |
| "logits/rejected": -0.2043023556470871, | |
| "logps/chosen": -210.77450561523438, | |
| "logps/rejected": -128.11520385742188, | |
| "loss": 0.1145, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 8.162195205688477, | |
| "rewards/margins": 9.146950721740723, | |
| "rewards/rejected": -0.9847553968429565, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.897907064328125e-06, | |
| "logits/chosen": -0.8582237362861633, | |
| "logits/rejected": -0.14161944389343262, | |
| "logps/chosen": -202.4379425048828, | |
| "logps/rejected": -136.9176483154297, | |
| "loss": 0.2544, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 8.603880882263184, | |
| "rewards/margins": 9.390780448913574, | |
| "rewards/rejected": -0.7868987321853638, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.894411369321984e-06, | |
| "logits/chosen": -0.9913557767868042, | |
| "logits/rejected": -0.0038151517510414124, | |
| "logps/chosen": -182.00146484375, | |
| "logps/rejected": -135.45352172851562, | |
| "loss": 0.0794, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 7.4787445068359375, | |
| "rewards/margins": 8.599211692810059, | |
| "rewards/rejected": -1.120466947555542, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.890857468949164e-06, | |
| "logits/chosen": -0.9070969820022583, | |
| "logits/rejected": 0.03064725361764431, | |
| "logps/chosen": -223.1108856201172, | |
| "logps/rejected": -145.6663055419922, | |
| "loss": 0.0421, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 8.449647903442383, | |
| "rewards/margins": 10.343244552612305, | |
| "rewards/rejected": -1.8935970067977905, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.887245405473393e-06, | |
| "logits/chosen": -0.7473093867301941, | |
| "logits/rejected": 0.2375013381242752, | |
| "logps/chosen": -218.11428833007812, | |
| "logps/rejected": -137.17213439941406, | |
| "loss": 0.0778, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 7.949044704437256, | |
| "rewards/margins": 8.823598861694336, | |
| "rewards/rejected": -0.8745536804199219, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.88357522185009e-06, | |
| "logits/chosen": -0.913166344165802, | |
| "logits/rejected": -0.011747682467103004, | |
| "logps/chosen": -220.34603881835938, | |
| "logps/rejected": -141.53561401367188, | |
| "loss": 0.0287, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 8.446084976196289, | |
| "rewards/margins": 9.644007682800293, | |
| "rewards/rejected": -1.1979233026504517, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.879846961725846e-06, | |
| "logits/chosen": -0.8051003217697144, | |
| "logits/rejected": 0.030248349532485008, | |
| "logps/chosen": -199.5961456298828, | |
| "logps/rejected": -155.56056213378906, | |
| "loss": 0.0765, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 8.313097953796387, | |
| "rewards/margins": 9.846205711364746, | |
| "rewards/rejected": -1.5331077575683594, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.876060669437913e-06, | |
| "logits/chosen": -0.616194486618042, | |
| "logits/rejected": 0.2745420038700104, | |
| "logps/chosen": -215.4296417236328, | |
| "logps/rejected": -142.76492309570312, | |
| "loss": 0.0848, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 7.84356164932251, | |
| "rewards/margins": 9.544157981872559, | |
| "rewards/rejected": -1.7005951404571533, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.872216390013675e-06, | |
| "logits/chosen": -0.7901605367660522, | |
| "logits/rejected": 0.41072148084640503, | |
| "logps/chosen": -185.0421142578125, | |
| "logps/rejected": -144.97398376464844, | |
| "loss": 0.1577, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.704103946685791, | |
| "rewards/margins": 9.240386009216309, | |
| "rewards/rejected": -1.5362818241119385, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.868314169170108e-06, | |
| "logits/chosen": -0.5289187431335449, | |
| "logits/rejected": 0.397865355014801, | |
| "logps/chosen": -219.9363555908203, | |
| "logps/rejected": -142.4893798828125, | |
| "loss": 0.1996, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 7.231544494628906, | |
| "rewards/margins": 8.826276779174805, | |
| "rewards/rejected": -1.5947315692901611, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.86435405331324e-06, | |
| "logits/chosen": -0.3374841809272766, | |
| "logits/rejected": 0.4616989195346832, | |
| "logps/chosen": -197.3474578857422, | |
| "logps/rejected": -141.45425415039062, | |
| "loss": 0.0638, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 7.858680725097656, | |
| "rewards/margins": 9.586578369140625, | |
| "rewards/rejected": -1.7278966903686523, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.860336089537601e-06, | |
| "logits/chosen": -0.4742531180381775, | |
| "logits/rejected": 0.3263815939426422, | |
| "logps/chosen": -197.79933166503906, | |
| "logps/rejected": -145.15786743164062, | |
| "loss": 0.1294, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 7.016458988189697, | |
| "rewards/margins": 8.728564262390137, | |
| "rewards/rejected": -1.7121059894561768, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.856260325625657e-06, | |
| "logits/chosen": -0.6684874892234802, | |
| "logits/rejected": 0.5160067677497864, | |
| "logps/chosen": -215.1957550048828, | |
| "logps/rejected": -157.3702392578125, | |
| "loss": 0.0588, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 6.882597923278809, | |
| "rewards/margins": 9.006999969482422, | |
| "rewards/rejected": -2.1244006156921387, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.85212681004725e-06, | |
| "logits/chosen": -0.7230426073074341, | |
| "logits/rejected": 0.48965024948120117, | |
| "logps/chosen": -238.30581665039062, | |
| "logps/rejected": -151.21774291992188, | |
| "loss": 0.0211, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 8.455827713012695, | |
| "rewards/margins": 10.841313362121582, | |
| "rewards/rejected": -2.385484457015991, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.847935591959011e-06, | |
| "logits/chosen": -0.2026168555021286, | |
| "logits/rejected": 0.6747661828994751, | |
| "logps/chosen": -206.3802490234375, | |
| "logps/rejected": -141.45103454589844, | |
| "loss": 0.0742, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 7.532872676849365, | |
| "rewards/margins": 9.587210655212402, | |
| "rewards/rejected": -2.0543372631073, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.84368672120379e-06, | |
| "logits/chosen": -0.3312041461467743, | |
| "logits/rejected": 0.7338079214096069, | |
| "logps/chosen": -200.46630859375, | |
| "logps/rejected": -147.34877014160156, | |
| "loss": 0.1475, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 7.3677077293396, | |
| "rewards/margins": 9.639891624450684, | |
| "rewards/rejected": -2.272183656692505, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.839380248310043e-06, | |
| "logits/chosen": -0.16066689789295197, | |
| "logits/rejected": 0.8108816146850586, | |
| "logps/chosen": -175.85104370117188, | |
| "logps/rejected": -157.10450744628906, | |
| "loss": 0.0537, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 7.471951961517334, | |
| "rewards/margins": 9.829510688781738, | |
| "rewards/rejected": -2.3575589656829834, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.835016224491251e-06, | |
| "logits/chosen": -0.3715950846672058, | |
| "logits/rejected": 0.7917531728744507, | |
| "logps/chosen": -183.4439697265625, | |
| "logps/rejected": -141.3053436279297, | |
| "loss": 0.0628, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.702794551849365, | |
| "rewards/margins": 9.382607460021973, | |
| "rewards/rejected": -1.6798127889633179, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.830594701645304e-06, | |
| "logits/chosen": -0.3017290234565735, | |
| "logits/rejected": 0.6022787094116211, | |
| "logps/chosen": -198.8234100341797, | |
| "logps/rejected": -144.2804412841797, | |
| "loss": 0.0508, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.64525842666626, | |
| "rewards/margins": 9.416332244873047, | |
| "rewards/rejected": -1.7710736989974976, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.826115732353877e-06, | |
| "logits/chosen": -0.36430418491363525, | |
| "logits/rejected": 0.5193418860435486, | |
| "logps/chosen": -256.91583251953125, | |
| "logps/rejected": -153.57138061523438, | |
| "loss": 0.1406, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 7.545983791351318, | |
| "rewards/margins": 10.080999374389648, | |
| "rewards/rejected": -2.535015821456909, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.82157936988181e-06, | |
| "logits/chosen": -0.4545150697231293, | |
| "logits/rejected": 0.928754985332489, | |
| "logps/chosen": -220.9812469482422, | |
| "logps/rejected": -150.5502166748047, | |
| "loss": 0.0332, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 7.531121730804443, | |
| "rewards/margins": 9.915712356567383, | |
| "rewards/rejected": -2.3845906257629395, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.816985668176482e-06, | |
| "logits/chosen": -0.7030404806137085, | |
| "logits/rejected": 0.5952990055084229, | |
| "logps/chosen": -231.7360382080078, | |
| "logps/rejected": -148.20033264160156, | |
| "loss": 0.152, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 8.060033798217773, | |
| "rewards/margins": 9.900753021240234, | |
| "rewards/rejected": -1.8407185077667236, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.812334681867152e-06, | |
| "logits/chosen": -0.7694862484931946, | |
| "logits/rejected": 0.30579477548599243, | |
| "logps/chosen": -182.74615478515625, | |
| "logps/rejected": -146.8044891357422, | |
| "loss": 0.106, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 7.797210693359375, | |
| "rewards/margins": 9.230527877807617, | |
| "rewards/rejected": -1.4333173036575317, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.807626466264328e-06, | |
| "logits/chosen": -0.6776883006095886, | |
| "logits/rejected": 0.4075050354003906, | |
| "logps/chosen": -188.23257446289062, | |
| "logps/rejected": -147.1197052001953, | |
| "loss": 0.0302, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 8.79939079284668, | |
| "rewards/margins": 10.872949600219727, | |
| "rewards/rejected": -2.073558807373047, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.802861077359098e-06, | |
| "logits/chosen": -0.6511849164962769, | |
| "logits/rejected": 0.4911740720272064, | |
| "logps/chosen": -197.52383422851562, | |
| "logps/rejected": -153.37123107910156, | |
| "loss": 0.1162, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.9149274826049805, | |
| "rewards/margins": 10.323246955871582, | |
| "rewards/rejected": -2.4083173274993896, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.798038571822463e-06, | |
| "logits/chosen": -0.8261006474494934, | |
| "logits/rejected": 0.3168261647224426, | |
| "logps/chosen": -200.90821838378906, | |
| "logps/rejected": -147.97134399414062, | |
| "loss": 0.0839, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 7.613385200500488, | |
| "rewards/margins": 9.651713371276855, | |
| "rewards/rejected": -2.038329601287842, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.793159007004675e-06, | |
| "logits/chosen": -0.9683536291122437, | |
| "logits/rejected": -0.2409909963607788, | |
| "logps/chosen": -231.24169921875, | |
| "logps/rejected": -154.0245361328125, | |
| "loss": 0.0287, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 8.987038612365723, | |
| "rewards/margins": 10.66617202758789, | |
| "rewards/rejected": -1.6791340112686157, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.78822244093454e-06, | |
| "logits/chosen": -0.7381896376609802, | |
| "logits/rejected": 0.2965128421783447, | |
| "logps/chosen": -174.1750946044922, | |
| "logps/rejected": -151.99537658691406, | |
| "loss": 0.0919, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 7.649115085601807, | |
| "rewards/margins": 9.841694831848145, | |
| "rewards/rejected": -2.192580223083496, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.783228932318736e-06, | |
| "logits/chosen": -0.6059221625328064, | |
| "logits/rejected": 0.08314159512519836, | |
| "logps/chosen": -210.4627227783203, | |
| "logps/rejected": -152.2406768798828, | |
| "loss": 0.0535, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.937192440032959, | |
| "rewards/margins": 9.917118072509766, | |
| "rewards/rejected": -1.979925513267517, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.778178540541116e-06, | |
| "logits/chosen": -0.730563759803772, | |
| "logits/rejected": 0.34863799810409546, | |
| "logps/chosen": -200.68209838867188, | |
| "logps/rejected": -151.97323608398438, | |
| "loss": 0.0397, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 7.986664772033691, | |
| "rewards/margins": 10.529966354370117, | |
| "rewards/rejected": -2.543302059173584, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.773071325662e-06, | |
| "logits/chosen": -0.8053274154663086, | |
| "logits/rejected": 0.12549547851085663, | |
| "logps/chosen": -192.92202758789062, | |
| "logps/rejected": -146.79693603515625, | |
| "loss": 0.1526, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 8.616433143615723, | |
| "rewards/margins": 10.316951751708984, | |
| "rewards/rejected": -1.7005198001861572, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.767907348417458e-06, | |
| "logits/chosen": -0.6939282417297363, | |
| "logits/rejected": 0.5801184773445129, | |
| "logps/chosen": -237.7701873779297, | |
| "logps/rejected": -150.30313110351562, | |
| "loss": 0.2363, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 8.603517532348633, | |
| "rewards/margins": 10.700518608093262, | |
| "rewards/rejected": -2.0970005989074707, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.762686670218588e-06, | |
| "logits/chosen": -0.5126532912254333, | |
| "logits/rejected": 0.5091619491577148, | |
| "logps/chosen": -213.9130401611328, | |
| "logps/rejected": -155.66085815429688, | |
| "loss": 0.0143, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 8.648975372314453, | |
| "rewards/margins": 11.388581275939941, | |
| "rewards/rejected": -2.739605665206909, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.757409353150793e-06, | |
| "logits/chosen": -0.4479140341281891, | |
| "logits/rejected": 0.5997704267501831, | |
| "logps/chosen": -206.91928100585938, | |
| "logps/rejected": -153.65902709960938, | |
| "loss": 0.0504, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 6.941058158874512, | |
| "rewards/margins": 9.638825416564941, | |
| "rewards/rejected": -2.6977667808532715, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.752075459973034e-06, | |
| "logits/chosen": -0.48702773451805115, | |
| "logits/rejected": 0.5097901225090027, | |
| "logps/chosen": -253.9662322998047, | |
| "logps/rejected": -161.4595947265625, | |
| "loss": 0.0841, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 8.159878730773926, | |
| "rewards/margins": 10.821348190307617, | |
| "rewards/rejected": -2.661470890045166, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.746685054117086e-06, | |
| "logits/chosen": -0.4953214228153229, | |
| "logits/rejected": 0.5581221580505371, | |
| "logps/chosen": -248.16482543945312, | |
| "logps/rejected": -150.33505249023438, | |
| "loss": 0.0869, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.8457465171813965, | |
| "rewards/margins": 10.702252388000488, | |
| "rewards/rejected": -2.856506109237671, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.741238199686789e-06, | |
| "logits/chosen": -0.5145618915557861, | |
| "logits/rejected": 0.7150206565856934, | |
| "logps/chosen": -236.22683715820312, | |
| "logps/rejected": -149.36941528320312, | |
| "loss": 0.0362, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 7.276123046875, | |
| "rewards/margins": 10.051583290100098, | |
| "rewards/rejected": -2.775459051132202, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.735734961457277e-06, | |
| "logits/chosen": -0.4139340817928314, | |
| "logits/rejected": 0.7916821241378784, | |
| "logps/chosen": -179.39382934570312, | |
| "logps/rejected": -153.31869506835938, | |
| "loss": 0.0393, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.780765533447266, | |
| "rewards/margins": 10.50986099243164, | |
| "rewards/rejected": -2.729094982147217, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.730175404874212e-06, | |
| "logits/chosen": -0.3699759542942047, | |
| "logits/rejected": 0.7061504125595093, | |
| "logps/chosen": -173.6160888671875, | |
| "logps/rejected": -157.6312713623047, | |
| "loss": 0.15, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 8.363532066345215, | |
| "rewards/margins": 10.816950798034668, | |
| "rewards/rejected": -2.4534192085266113, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.724559596053009e-06, | |
| "logits/chosen": -0.6346268057823181, | |
| "logits/rejected": 0.48736244440078735, | |
| "logps/chosen": -197.9464111328125, | |
| "logps/rejected": -163.78814697265625, | |
| "loss": 0.0861, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 8.08003044128418, | |
| "rewards/margins": 10.763525009155273, | |
| "rewards/rejected": -2.683495044708252, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.718887601778044e-06, | |
| "logits/chosen": -0.4699893891811371, | |
| "logits/rejected": 0.5934313535690308, | |
| "logps/chosen": -172.5706787109375, | |
| "logps/rejected": -147.60012817382812, | |
| "loss": 0.0598, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 7.218860626220703, | |
| "rewards/margins": 9.611445426940918, | |
| "rewards/rejected": -2.3925843238830566, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.713159489501863e-06, | |
| "logits/chosen": -0.7212098836898804, | |
| "logits/rejected": 0.6830633282661438, | |
| "logps/chosen": -195.26705932617188, | |
| "logps/rejected": -144.74806213378906, | |
| "loss": 0.0301, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": 7.268791198730469, | |
| "rewards/margins": 9.785104751586914, | |
| "rewards/rejected": -2.5163135528564453, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.707956263849591e-06, | |
| "logits/chosen": -0.653510570526123, | |
| "logits/rejected": 0.6825904250144958, | |
| "logps/chosen": -214.06729125976562, | |
| "logps/rejected": -151.3599395751953, | |
| "loss": 0.1073, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 8.91938591003418, | |
| "rewards/margins": 11.606144905090332, | |
| "rewards/rejected": -2.6867592334747314, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 9110, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "total_flos": 1.2405468070721618e+18, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |