| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1274, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0007849293563579278, | |
| "grad_norm": 6.033449115554074, | |
| "learning_rate": 2.34375e-09, | |
| "logits/chosen": 5914.86328125, | |
| "logits/rejected": 2785.05078125, | |
| "logps/chosen": -212.39956665039062, | |
| "logps/rejected": -98.58319854736328, | |
| "loss": 1.6265, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.007849293563579277, | |
| "grad_norm": 6.334644894564701, | |
| "learning_rate": 2.34375e-08, | |
| "logits/chosen": 4973.83203125, | |
| "logits/rejected": 4328.20947265625, | |
| "logps/chosen": -204.1883544921875, | |
| "logps/rejected": -179.72076416015625, | |
| "loss": 1.6265, | |
| "rewards/accuracies": 0.5277777910232544, | |
| "rewards/chosen": 0.01907867193222046, | |
| "rewards/margins": 0.04872569814324379, | |
| "rewards/rejected": -0.029647022485733032, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.015698587127158554, | |
| "grad_norm": 5.9916217113774, | |
| "learning_rate": 4.6875e-08, | |
| "logits/chosen": 6084.1591796875, | |
| "logits/rejected": 4833.9228515625, | |
| "logps/chosen": -217.2135467529297, | |
| "logps/rejected": -196.72286987304688, | |
| "loss": 1.6265, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.02391243539750576, | |
| "rewards/margins": -0.033898498862981796, | |
| "rewards/rejected": 0.009986063465476036, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.023547880690737835, | |
| "grad_norm": 5.7641173333433775, | |
| "learning_rate": 7.03125e-08, | |
| "logits/chosen": 6082.0751953125, | |
| "logits/rejected": 5103.20458984375, | |
| "logps/chosen": -250.56283569335938, | |
| "logps/rejected": -209.3272705078125, | |
| "loss": 1.6265, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 0.023494163528084755, | |
| "rewards/margins": 0.016794539988040924, | |
| "rewards/rejected": 0.006699626334011555, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03139717425431711, | |
| "grad_norm": 5.454897887476028, | |
| "learning_rate": 9.375e-08, | |
| "logits/chosen": 5302.84423828125, | |
| "logits/rejected": 4339.72900390625, | |
| "logps/chosen": -212.10537719726562, | |
| "logps/rejected": -181.6745147705078, | |
| "loss": 1.6265, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.0172278955578804, | |
| "rewards/margins": -0.01973465457558632, | |
| "rewards/rejected": 0.0025067573878914118, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03924646781789639, | |
| "grad_norm": 6.38829428743751, | |
| "learning_rate": 1.1718749999999999e-07, | |
| "logits/chosen": 6392.1474609375, | |
| "logits/rejected": 5017.2919921875, | |
| "logps/chosen": -265.69232177734375, | |
| "logps/rejected": -206.754150390625, | |
| "loss": 1.6261, | |
| "rewards/accuracies": 0.4583333432674408, | |
| "rewards/chosen": -0.07548192888498306, | |
| "rewards/margins": -0.040568120777606964, | |
| "rewards/rejected": -0.03491383045911789, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04709576138147567, | |
| "grad_norm": 6.31748832974153, | |
| "learning_rate": 1.40625e-07, | |
| "logits/chosen": 5422.49609375, | |
| "logits/rejected": 4506.4208984375, | |
| "logps/chosen": -214.3790740966797, | |
| "logps/rejected": -209.01828002929688, | |
| "loss": 1.6253, | |
| "rewards/accuracies": 0.3166666626930237, | |
| "rewards/chosen": -0.40180811285972595, | |
| "rewards/margins": -0.24679434299468994, | |
| "rewards/rejected": -0.1550137847661972, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.054945054945054944, | |
| "grad_norm": 7.1862223915388315, | |
| "learning_rate": 1.640625e-07, | |
| "logits/chosen": 5049.93408203125, | |
| "logits/rejected": 4787.21630859375, | |
| "logps/chosen": -179.75225830078125, | |
| "logps/rejected": -177.83863830566406, | |
| "loss": 1.6216, | |
| "rewards/accuracies": 0.44166669249534607, | |
| "rewards/chosen": -1.2514512538909912, | |
| "rewards/margins": -0.1759890913963318, | |
| "rewards/rejected": -1.0754622220993042, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06279434850863422, | |
| "grad_norm": 8.327706520679039, | |
| "learning_rate": 1.875e-07, | |
| "logits/chosen": 5466.95947265625, | |
| "logits/rejected": 5001.8310546875, | |
| "logps/chosen": -199.88381958007812, | |
| "logps/rejected": -184.80918884277344, | |
| "loss": 1.6118, | |
| "rewards/accuracies": 0.4583333432674408, | |
| "rewards/chosen": -3.306250810623169, | |
| "rewards/margins": -0.05246015638113022, | |
| "rewards/rejected": -3.2537906169891357, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0706436420722135, | |
| "grad_norm": 12.704545270148827, | |
| "learning_rate": 2.109375e-07, | |
| "logits/chosen": 5456.98583984375, | |
| "logits/rejected": 4672.60986328125, | |
| "logps/chosen": -231.42489624023438, | |
| "logps/rejected": -196.01947021484375, | |
| "loss": 1.5778, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -12.56345272064209, | |
| "rewards/margins": -0.9916127324104309, | |
| "rewards/rejected": -11.571840286254883, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07849293563579278, | |
| "grad_norm": 7.59113886837309, | |
| "learning_rate": 2.3437499999999998e-07, | |
| "logits/chosen": 4993.03955078125, | |
| "logits/rejected": 4812.00146484375, | |
| "logps/chosen": -246.10147094726562, | |
| "logps/rejected": -233.54025268554688, | |
| "loss": 1.5199, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -35.86996078491211, | |
| "rewards/margins": -1.7471039295196533, | |
| "rewards/rejected": -34.122859954833984, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08634222919937205, | |
| "grad_norm": 9.015849618304676, | |
| "learning_rate": 2.5781249999999997e-07, | |
| "logits/chosen": 4971.4384765625, | |
| "logits/rejected": 4100.7958984375, | |
| "logps/chosen": -258.1769104003906, | |
| "logps/rejected": -242.14089965820312, | |
| "loss": 1.5138, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -75.07661437988281, | |
| "rewards/margins": -11.86035442352295, | |
| "rewards/rejected": -63.21626663208008, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09419152276295134, | |
| "grad_norm": 10.274748882232315, | |
| "learning_rate": 2.8125e-07, | |
| "logits/chosen": 4943.21923828125, | |
| "logits/rejected": 4107.5322265625, | |
| "logps/chosen": -274.8851318359375, | |
| "logps/rejected": -264.3096008300781, | |
| "loss": 1.4822, | |
| "rewards/accuracies": 0.5250000357627869, | |
| "rewards/chosen": -72.01099395751953, | |
| "rewards/margins": -5.851285934448242, | |
| "rewards/rejected": -66.15970611572266, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.10204081632653061, | |
| "grad_norm": 7.144257110565484, | |
| "learning_rate": 2.999977454977299e-07, | |
| "logits/chosen": 4948.9169921875, | |
| "logits/rejected": 4537.37646484375, | |
| "logps/chosen": -287.4667663574219, | |
| "logps/rejected": -281.4789123535156, | |
| "loss": 1.4274, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -77.65235137939453, | |
| "rewards/margins": 6.16226863861084, | |
| "rewards/rejected": -83.81462097167969, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.10989010989010989, | |
| "grad_norm": 8.226551521487808, | |
| "learning_rate": 2.999188450339425e-07, | |
| "logits/chosen": 4828.34375, | |
| "logits/rejected": 4192.6845703125, | |
| "logps/chosen": -281.3506164550781, | |
| "logps/rejected": -279.585205078125, | |
| "loss": 1.3961, | |
| "rewards/accuracies": 0.5666666626930237, | |
| "rewards/chosen": -95.67347717285156, | |
| "rewards/margins": 6.334682464599609, | |
| "rewards/rejected": -102.00816345214844, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11773940345368916, | |
| "grad_norm": 11.507860546274475, | |
| "learning_rate": 2.997272872179022e-07, | |
| "logits/chosen": 4970.33837890625, | |
| "logits/rejected": 4065.40673828125, | |
| "logps/chosen": -291.5596923828125, | |
| "logps/rejected": -275.5636291503906, | |
| "loss": 1.3885, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -89.9407730102539, | |
| "rewards/margins": 13.317327499389648, | |
| "rewards/rejected": -103.25809478759766, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12558869701726844, | |
| "grad_norm": 10.994369595848676, | |
| "learning_rate": 2.994232159967983e-07, | |
| "logits/chosen": 4982.4130859375, | |
| "logits/rejected": 4372.6259765625, | |
| "logps/chosen": -284.61468505859375, | |
| "logps/rejected": -296.10565185546875, | |
| "loss": 1.3818, | |
| "rewards/accuracies": 0.6333333849906921, | |
| "rewards/chosen": -87.11192321777344, | |
| "rewards/margins": 11.931414604187012, | |
| "rewards/rejected": -99.0433349609375, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13343799058084774, | |
| "grad_norm": 9.481017294519704, | |
| "learning_rate": 2.990068598666458e-07, | |
| "logits/chosen": 5221.8046875, | |
| "logits/rejected": 4557.0537109375, | |
| "logps/chosen": -301.30633544921875, | |
| "logps/rejected": -305.81646728515625, | |
| "loss": 1.3726, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": -91.6949234008789, | |
| "rewards/margins": 17.794422149658203, | |
| "rewards/rejected": -109.48934173583984, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.141287284144427, | |
| "grad_norm": 14.020275412135652, | |
| "learning_rate": 2.984785317005808e-07, | |
| "logits/chosen": 4500.76220703125, | |
| "logits/rejected": 3867.770751953125, | |
| "logps/chosen": -281.3055114746094, | |
| "logps/rejected": -287.490478515625, | |
| "loss": 1.3686, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": -92.7432861328125, | |
| "rewards/margins": 16.572912216186523, | |
| "rewards/rejected": -109.31620025634766, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14913657770800628, | |
| "grad_norm": 15.646225937667412, | |
| "learning_rate": 2.978386285137506e-07, | |
| "logits/chosen": 4642.57958984375, | |
| "logits/rejected": 3846.667236328125, | |
| "logps/chosen": -290.1266174316406, | |
| "logps/rejected": -312.9205627441406, | |
| "loss": 1.355, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -86.35670471191406, | |
| "rewards/margins": 35.351932525634766, | |
| "rewards/rejected": -121.70863342285156, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15698587127158556, | |
| "grad_norm": 11.466454800803227, | |
| "learning_rate": 2.9708763116497365e-07, | |
| "logits/chosen": 4456.6640625, | |
| "logits/rejected": 4046.647705078125, | |
| "logps/chosen": -272.8510437011719, | |
| "logps/rejected": -312.5729675292969, | |
| "loss": 1.3576, | |
| "rewards/accuracies": 0.7166666388511658, | |
| "rewards/chosen": -90.31404876708984, | |
| "rewards/margins": 25.549606323242188, | |
| "rewards/rejected": -115.86366271972656, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16483516483516483, | |
| "grad_norm": 18.323993013074592, | |
| "learning_rate": 2.962261039953964e-07, | |
| "logits/chosen": 3986.946044921875, | |
| "logits/rejected": 3482.71142578125, | |
| "logps/chosen": -242.8981170654297, | |
| "logps/rejected": -276.468994140625, | |
| "loss": 1.3525, | |
| "rewards/accuracies": 0.7166666984558105, | |
| "rewards/chosen": -88.96837615966797, | |
| "rewards/margins": 30.285064697265625, | |
| "rewards/rejected": -119.2534408569336, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.1726844583987441, | |
| "grad_norm": 14.15120837947915, | |
| "learning_rate": 2.952546944044155e-07, | |
| "logits/chosen": 4956.119140625, | |
| "logits/rejected": 4112.83984375, | |
| "logps/chosen": -283.2933349609375, | |
| "logps/rejected": -285.0084533691406, | |
| "loss": 1.3478, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -87.01918029785156, | |
| "rewards/margins": 34.038204193115234, | |
| "rewards/rejected": -121.057373046875, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.18053375196232338, | |
| "grad_norm": 13.56003679991926, | |
| "learning_rate": 2.941741323631859e-07, | |
| "logits/chosen": 4587.6279296875, | |
| "logits/rejected": 4040.880126953125, | |
| "logps/chosen": -276.8036804199219, | |
| "logps/rejected": -302.5361328125, | |
| "loss": 1.3477, | |
| "rewards/accuracies": 0.7166666984558105, | |
| "rewards/chosen": -89.46044921875, | |
| "rewards/margins": 33.30833435058594, | |
| "rewards/rejected": -122.76878356933594, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18838304552590268, | |
| "grad_norm": 19.695765717669524, | |
| "learning_rate": 2.929852298660803e-07, | |
| "logits/chosen": 4106.96533203125, | |
| "logits/rejected": 3192.676513671875, | |
| "logps/chosen": -258.5592346191406, | |
| "logps/rejected": -276.740966796875, | |
| "loss": 1.3478, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -90.93244934082031, | |
| "rewards/margins": 31.089950561523438, | |
| "rewards/rejected": -122.02239990234375, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19623233908948196, | |
| "grad_norm": 17.837485568500572, | |
| "learning_rate": 2.9168888032051097e-07, | |
| "logits/chosen": 4755.09423828125, | |
| "logits/rejected": 4138.0048828125, | |
| "logps/chosen": -256.274169921875, | |
| "logps/rejected": -310.71746826171875, | |
| "loss": 1.3551, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": -89.11972045898438, | |
| "rewards/margins": 37.10137176513672, | |
| "rewards/rejected": -126.2210922241211, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.20408163265306123, | |
| "grad_norm": 15.060829491384922, | |
| "learning_rate": 2.9028605787557423e-07, | |
| "logits/chosen": 4961.2783203125, | |
| "logits/rejected": 3883.23583984375, | |
| "logps/chosen": -295.96575927734375, | |
| "logps/rejected": -291.8458251953125, | |
| "loss": 1.362, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -96.2462158203125, | |
| "rewards/margins": 29.351673126220703, | |
| "rewards/rejected": -125.59788513183594, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2119309262166405, | |
| "grad_norm": 21.725472315572194, | |
| "learning_rate": 2.8877781669002034e-07, | |
| "logits/chosen": 4907.73681640625, | |
| "logits/rejected": 4685.39306640625, | |
| "logps/chosen": -290.1836242675781, | |
| "logps/rejected": -312.2330627441406, | |
| "loss": 1.3387, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -93.42694854736328, | |
| "rewards/margins": 30.104965209960938, | |
| "rewards/rejected": -123.53192138671875, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.21978021978021978, | |
| "grad_norm": 22.380743969527877, | |
| "learning_rate": 2.871652901401002e-07, | |
| "logits/chosen": 4144.8955078125, | |
| "logits/rejected": 3516.05615234375, | |
| "logps/chosen": -250.620361328125, | |
| "logps/rejected": -303.0557861328125, | |
| "loss": 1.3452, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": -87.56240844726562, | |
| "rewards/margins": 46.16553497314453, | |
| "rewards/rejected": -133.7279510498047, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22762951334379905, | |
| "grad_norm": 19.023537926104407, | |
| "learning_rate": 2.854496899678834e-07, | |
| "logits/chosen": 5847.7109375, | |
| "logits/rejected": 4171.96630859375, | |
| "logps/chosen": -327.1379699707031, | |
| "logps/rejected": -330.09600830078125, | |
| "loss": 1.3403, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -90.06501007080078, | |
| "rewards/margins": 57.50885772705078, | |
| "rewards/rejected": -147.57388305664062, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.23547880690737832, | |
| "grad_norm": 17.58097454746967, | |
| "learning_rate": 2.8363230537068793e-07, | |
| "logits/chosen": 4251.00634765625, | |
| "logits/rejected": 4112.849609375, | |
| "logps/chosen": -255.7366485595703, | |
| "logps/rejected": -293.42156982421875, | |
| "loss": 1.3468, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": -88.24583435058594, | |
| "rewards/margins": 34.95277404785156, | |
| "rewards/rejected": -123.19859313964844, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.24332810047095763, | |
| "grad_norm": 14.70842299502561, | |
| "learning_rate": 2.817145020323055e-07, | |
| "logits/chosen": 4780.62255859375, | |
| "logits/rejected": 4147.02685546875, | |
| "logps/chosen": -269.6231384277344, | |
| "logps/rejected": -292.3729553222656, | |
| "loss": 1.3542, | |
| "rewards/accuracies": 0.6833333373069763, | |
| "rewards/chosen": -92.05260467529297, | |
| "rewards/margins": 24.870826721191406, | |
| "rewards/rejected": -116.9234390258789, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.25117739403453687, | |
| "grad_norm": 13.938859657249182, | |
| "learning_rate": 2.7969772109675106e-07, | |
| "logits/chosen": 4895.505859375, | |
| "logits/rejected": 4120.4462890625, | |
| "logps/chosen": -288.72723388671875, | |
| "logps/rejected": -322.5893859863281, | |
| "loss": 1.35, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": -87.61097717285156, | |
| "rewards/margins": 32.92052459716797, | |
| "rewards/rejected": -120.53150939941406, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25902668759811615, | |
| "grad_norm": 11.637579462880986, | |
| "learning_rate": 2.775834780853069e-07, | |
| "logits/chosen": 4723.48291015625, | |
| "logits/rejected": 4224.01220703125, | |
| "logps/chosen": -292.0691833496094, | |
| "logps/rejected": -297.52789306640625, | |
| "loss": 1.357, | |
| "rewards/accuracies": 0.6500000357627869, | |
| "rewards/chosen": -97.65233612060547, | |
| "rewards/margins": 22.543180465698242, | |
| "rewards/rejected": -120.19551849365234, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2668759811616955, | |
| "grad_norm": 11.50915278056431, | |
| "learning_rate": 2.75373361757676e-07, | |
| "logits/chosen": 4986.0654296875, | |
| "logits/rejected": 4922.3876953125, | |
| "logps/chosen": -282.6824035644531, | |
| "logps/rejected": -336.1164855957031, | |
| "loss": 1.3449, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -88.81416320800781, | |
| "rewards/margins": 26.942358016967773, | |
| "rewards/rejected": -115.75653076171875, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.27472527472527475, | |
| "grad_norm": 19.664421369914834, | |
| "learning_rate": 2.730690329180992e-07, | |
| "logits/chosen": 5255.826171875, | |
| "logits/rejected": 4648.89111328125, | |
| "logps/chosen": -287.81707763671875, | |
| "logps/rejected": -344.7186279296875, | |
| "loss": 1.3353, | |
| "rewards/accuracies": 0.6916666626930237, | |
| "rewards/chosen": -92.95384216308594, | |
| "rewards/margins": 44.985557556152344, | |
| "rewards/rejected": -137.93942260742188, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.282574568288854, | |
| "grad_norm": 14.042295784485253, | |
| "learning_rate": 2.7067222316733527e-07, | |
| "logits/chosen": 4749.18994140625, | |
| "logits/rejected": 4491.1953125, | |
| "logps/chosen": -266.18914794921875, | |
| "logps/rejected": -342.7701416015625, | |
| "loss": 1.3367, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -88.32334899902344, | |
| "rewards/margins": 55.10559844970703, | |
| "rewards/rejected": -143.42892456054688, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2904238618524333, | |
| "grad_norm": 15.2758713919473, | |
| "learning_rate": 2.6818473360143916e-07, | |
| "logits/chosen": 4407.02783203125, | |
| "logits/rejected": 4003.607421875, | |
| "logps/chosen": -270.6219787597656, | |
| "logps/rejected": -287.3055419921875, | |
| "loss": 1.3537, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -86.1965560913086, | |
| "rewards/margins": 33.04121780395508, | |
| "rewards/rejected": -119.2377700805664, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.29827315541601257, | |
| "grad_norm": 13.766452688703493, | |
| "learning_rate": 2.6560843345831893e-07, | |
| "logits/chosen": 5333.09375, | |
| "logits/rejected": 4200.95947265625, | |
| "logps/chosen": -317.053955078125, | |
| "logps/rejected": -308.80609130859375, | |
| "loss": 1.3395, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -88.68672943115234, | |
| "rewards/margins": 37.15703582763672, | |
| "rewards/rejected": -125.84376525878906, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.30612244897959184, | |
| "grad_norm": 36.85374368236901, | |
| "learning_rate": 2.62945258713086e-07, | |
| "logits/chosen": 5441.1943359375, | |
| "logits/rejected": 3993.84423828125, | |
| "logps/chosen": -302.8955993652344, | |
| "logps/rejected": -322.1986999511719, | |
| "loss": 1.3343, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -87.42176818847656, | |
| "rewards/margins": 54.94853591918945, | |
| "rewards/rejected": -142.3703155517578, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3139717425431711, | |
| "grad_norm": 14.099963149615544, | |
| "learning_rate": 2.601972106232561e-07, | |
| "logits/chosen": 4693.27587890625, | |
| "logits/rejected": 4468.32373046875, | |
| "logps/chosen": -273.7616882324219, | |
| "logps/rejected": -314.75164794921875, | |
| "loss": 1.3598, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -82.94029235839844, | |
| "rewards/margins": 43.07162857055664, | |
| "rewards/rejected": -126.01191711425781, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3218210361067504, | |
| "grad_norm": 22.82272878384856, | |
| "learning_rate": 2.5736635422489343e-07, | |
| "logits/chosen": 5083.9404296875, | |
| "logits/rejected": 4400.0224609375, | |
| "logps/chosen": -309.4847717285156, | |
| "logps/rejected": -357.3506774902344, | |
| "loss": 1.3187, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -97.01141357421875, | |
| "rewards/margins": 51.86890411376953, | |
| "rewards/rejected": -148.88031005859375, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.32967032967032966, | |
| "grad_norm": 13.413614592026427, | |
| "learning_rate": 2.5445481678082706e-07, | |
| "logits/chosen": 4738.033203125, | |
| "logits/rejected": 4165.1875, | |
| "logps/chosen": -292.02459716796875, | |
| "logps/rejected": -313.284912109375, | |
| "loss": 1.3412, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -87.97795104980469, | |
| "rewards/margins": 44.012813568115234, | |
| "rewards/rejected": -131.9907684326172, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.33751962323390894, | |
| "grad_norm": 16.35310936737267, | |
| "learning_rate": 2.514647861821081e-07, | |
| "logits/chosen": 4596.02392578125, | |
| "logits/rejected": 3646.234375, | |
| "logps/chosen": -269.6007080078125, | |
| "logps/rejected": -309.58538818359375, | |
| "loss": 1.3277, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -89.86016845703125, | |
| "rewards/margins": 58.46223831176758, | |
| "rewards/rejected": -148.32241821289062, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3453689167974882, | |
| "grad_norm": 13.924787630126689, | |
| "learning_rate": 2.483985093039065e-07, | |
| "logits/chosen": 4906.8564453125, | |
| "logits/rejected": 4671.38818359375, | |
| "logps/chosen": -275.3404846191406, | |
| "logps/rejected": -325.6501770019531, | |
| "loss": 1.3556, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -92.95171356201172, | |
| "rewards/margins": 41.062828063964844, | |
| "rewards/rejected": -134.01454162597656, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3532182103610675, | |
| "grad_norm": 12.974922593697578, | |
| "learning_rate": 2.4525829031708404e-07, | |
| "logits/chosen": 5700.68115234375, | |
| "logits/rejected": 3948.375732421875, | |
| "logps/chosen": -295.39019775390625, | |
| "logps/rejected": -297.334228515625, | |
| "loss": 1.3449, | |
| "rewards/accuracies": 0.7416667342185974, | |
| "rewards/chosen": -86.63603973388672, | |
| "rewards/margins": 47.681427001953125, | |
| "rewards/rejected": -134.3174591064453, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.36106750392464676, | |
| "grad_norm": 13.7663993271018, | |
| "learning_rate": 2.420464889567126e-07, | |
| "logits/chosen": 5984.6328125, | |
| "logits/rejected": 4481.2509765625, | |
| "logps/chosen": -346.4881286621094, | |
| "logps/rejected": -334.8788146972656, | |
| "loss": 1.3296, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -96.00253295898438, | |
| "rewards/margins": 48.739898681640625, | |
| "rewards/rejected": -144.74244689941406, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.36891679748822603, | |
| "grad_norm": 16.285406224013258, | |
| "learning_rate": 2.3876551874883775e-07, | |
| "logits/chosen": 5035.26953125, | |
| "logits/rejected": 4289.298828125, | |
| "logps/chosen": -301.6008605957031, | |
| "logps/rejected": -343.3747253417969, | |
| "loss": 1.3369, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -89.3971939086914, | |
| "rewards/margins": 51.47382736206055, | |
| "rewards/rejected": -140.8710174560547, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.37676609105180536, | |
| "grad_norm": 12.840284775101187, | |
| "learning_rate": 2.354178451968214e-07, | |
| "logits/chosen": 5432.525390625, | |
| "logits/rejected": 4493.26953125, | |
| "logps/chosen": -295.23321533203125, | |
| "logps/rejected": -313.70111083984375, | |
| "loss": 1.3372, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -87.82811737060547, | |
| "rewards/margins": 51.47626876831055, | |
| "rewards/rejected": -139.30441284179688, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 13.012539888722776, | |
| "learning_rate": 2.3200598392862542e-07, | |
| "logits/chosen": 5000.28125, | |
| "logits/rejected": 4610.8955078125, | |
| "logps/chosen": -264.51531982421875, | |
| "logps/rejected": -339.9250183105469, | |
| "loss": 1.3246, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -87.90812683105469, | |
| "rewards/margins": 49.94524383544922, | |
| "rewards/rejected": -137.85336303710938, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3924646781789639, | |
| "grad_norm": 18.56442430897208, | |
| "learning_rate": 2.2853249880642897e-07, | |
| "logits/chosen": 5522.9404296875, | |
| "logits/rejected": 4092.22265625, | |
| "logps/chosen": -323.59771728515625, | |
| "logps/rejected": -357.1122131347656, | |
| "loss": 1.3349, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -101.6022720336914, | |
| "rewards/margins": 52.95317459106445, | |
| "rewards/rejected": -154.55545043945312, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4003139717425432, | |
| "grad_norm": 14.082902463369917, | |
| "learning_rate": 2.25e-07, | |
| "logits/chosen": 4505.0869140625, | |
| "logits/rejected": 3814.422607421875, | |
| "logps/chosen": -273.663330078125, | |
| "logps/rejected": -298.57855224609375, | |
| "loss": 1.328, | |
| "rewards/accuracies": 0.7916668057441711, | |
| "rewards/chosen": -84.23956298828125, | |
| "rewards/margins": 48.00691223144531, | |
| "rewards/rejected": -132.2464599609375, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 12.873522668189558, | |
| "learning_rate": 2.2141114202526867e-07, | |
| "logits/chosen": 4780.2392578125, | |
| "logits/rejected": 4262.83203125, | |
| "logps/chosen": -296.81536865234375, | |
| "logps/rejected": -335.5638122558594, | |
| "loss": 1.365, | |
| "rewards/accuracies": 0.6666666269302368, | |
| "rewards/chosen": -98.39509582519531, | |
| "rewards/margins": 41.327919006347656, | |
| "rewards/rejected": -139.7230224609375, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.41601255886970173, | |
| "grad_norm": 12.684426666795588, | |
| "learning_rate": 2.1776862174957646e-07, | |
| "logits/chosen": 5397.5029296875, | |
| "logits/rejected": 4275.8896484375, | |
| "logps/chosen": -295.6401672363281, | |
| "logps/rejected": -316.23468017578125, | |
| "loss": 1.3303, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -82.70533752441406, | |
| "rewards/margins": 34.76225662231445, | |
| "rewards/rejected": -117.46760559082031, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.423861852433281, | |
| "grad_norm": 22.323475438849997, | |
| "learning_rate": 2.1407517636510023e-07, | |
| "logits/chosen": 5051.541015625, | |
| "logits/rejected": 4335.1279296875, | |
| "logps/chosen": -312.322265625, | |
| "logps/rejected": -339.65802001953125, | |
| "loss": 1.3352, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -101.51814270019531, | |
| "rewards/margins": 49.8724479675293, | |
| "rewards/rejected": -151.39059448242188, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4317111459968603, | |
| "grad_norm": 21.570737156393047, | |
| "learning_rate": 2.1033358133197376e-07, | |
| "logits/chosen": 4971.50927734375, | |
| "logits/rejected": 4415.0771484375, | |
| "logps/chosen": -298.01519775390625, | |
| "logps/rejected": -359.4136962890625, | |
| "loss": 1.3335, | |
| "rewards/accuracies": 0.7500000596046448, | |
| "rewards/chosen": -83.20237731933594, | |
| "rewards/margins": 51.15970230102539, | |
| "rewards/rejected": -134.36209106445312, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.43956043956043955, | |
| "grad_norm": 12.415018699318011, | |
| "learning_rate": 2.0654664829265295e-07, | |
| "logits/chosen": 4912.80859375, | |
| "logits/rejected": 4474.53515625, | |
| "logps/chosen": -297.15765380859375, | |
| "logps/rejected": -323.38116455078125, | |
| "loss": 1.3455, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -93.88143157958984, | |
| "rewards/margins": 36.26217269897461, | |
| "rewards/rejected": -130.1436004638672, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4474097331240188, | |
| "grad_norm": 20.61913444595419, | |
| "learning_rate": 2.027172229590914e-07, | |
| "logits/chosen": 5299.6044921875, | |
| "logits/rejected": 4737.7197265625, | |
| "logps/chosen": -336.34356689453125, | |
| "logps/rejected": -368.6340026855469, | |
| "loss": 1.3334, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": -93.41209411621094, | |
| "rewards/margins": 51.496124267578125, | |
| "rewards/rejected": -144.90821838378906, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.4552590266875981, | |
| "grad_norm": 31.600089809367816, | |
| "learning_rate": 1.9884818297431403e-07, | |
| "logits/chosen": 5278.46240234375, | |
| "logits/rejected": 4592.17578125, | |
| "logps/chosen": -296.7460632324219, | |
| "logps/rejected": -345.90313720703125, | |
| "loss": 1.3445, | |
| "rewards/accuracies": 0.7583333849906921, | |
| "rewards/chosen": -84.3451156616211, | |
| "rewards/margins": 48.27422332763672, | |
| "rewards/rejected": -132.6193389892578, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4631083202511774, | |
| "grad_norm": 15.260249014970388, | |
| "learning_rate": 1.9494243574999645e-07, | |
| "logits/chosen": 5775.82470703125, | |
| "logits/rejected": 4653.576171875, | |
| "logps/chosen": -324.98199462890625, | |
| "logps/rejected": -319.8547668457031, | |
| "loss": 1.3314, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -93.31060028076172, | |
| "rewards/margins": 47.719669342041016, | |
| "rewards/rejected": -141.0302734375, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.47095761381475665, | |
| "grad_norm": 15.88452181021661, | |
| "learning_rate": 1.9100291628167416e-07, | |
| "logits/chosen": 5445.01904296875, | |
| "logits/rejected": 4488.2998046875, | |
| "logps/chosen": -265.55645751953125, | |
| "logps/rejected": -320.50567626953125, | |
| "loss": 1.3278, | |
| "rewards/accuracies": 0.7500000596046448, | |
| "rewards/chosen": -88.00885009765625, | |
| "rewards/margins": 42.79914093017578, | |
| "rewards/rejected": -130.8079833984375, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.478806907378336, | |
| "grad_norm": 20.55547909163982, | |
| "learning_rate": 1.870325849432238e-07, | |
| "logits/chosen": 5234.193359375, | |
| "logits/rejected": 4424.484375, | |
| "logps/chosen": -293.6166076660156, | |
| "logps/rejected": -331.92974853515625, | |
| "loss": 1.3129, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -87.48274993896484, | |
| "rewards/margins": 59.040428161621094, | |
| "rewards/rejected": -146.52317810058594, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.48665620094191525, | |
| "grad_norm": 17.927137367169394, | |
| "learning_rate": 1.8303442526227364e-07, | |
| "logits/chosen": 4438.71240234375, | |
| "logits/rejected": 3490.155029296875, | |
| "logps/chosen": -275.29241943359375, | |
| "logps/rejected": -313.7520751953125, | |
| "loss": 1.3229, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -91.52749633789062, | |
| "rewards/margins": 61.12894821166992, | |
| "rewards/rejected": -152.6564178466797, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.4945054945054945, | |
| "grad_norm": 20.192182222445886, | |
| "learning_rate": 1.7901144167821507e-07, | |
| "logits/chosen": 5873.173828125, | |
| "logits/rejected": 4774.25537109375, | |
| "logps/chosen": -323.35992431640625, | |
| "logps/rejected": -342.5325012207031, | |
| "loss": 1.3214, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -80.52757263183594, | |
| "rewards/margins": 67.7401123046875, | |
| "rewards/rejected": -148.26768493652344, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.5023547880690737, | |
| "grad_norm": 13.166592462596656, | |
| "learning_rate": 1.749666572845e-07, | |
| "logits/chosen": 5518.23046875, | |
| "logits/rejected": 4573.6103515625, | |
| "logps/chosen": -308.5555419921875, | |
| "logps/rejected": -345.7183837890625, | |
| "loss": 1.3288, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -97.77859497070312, | |
| "rewards/margins": 33.44771194458008, | |
| "rewards/rejected": -131.226318359375, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5102040816326531, | |
| "grad_norm": 13.697385802349345, | |
| "learning_rate": 1.7090311155692034e-07, | |
| "logits/chosen": 4693.83349609375, | |
| "logits/rejected": 3723.333984375, | |
| "logps/chosen": -293.64495849609375, | |
| "logps/rejected": -310.81787109375, | |
| "loss": 1.3321, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -97.3224105834961, | |
| "rewards/margins": 44.76685333251953, | |
| "rewards/rejected": -142.08926391601562, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5180533751962323, | |
| "grad_norm": 25.46823282035701, | |
| "learning_rate": 1.6682385806957695e-07, | |
| "logits/chosen": 5368.3603515625, | |
| "logits/rejected": 4091.384765625, | |
| "logps/chosen": -303.13446044921875, | |
| "logps/rejected": -332.54107666015625, | |
| "loss": 1.3345, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -85.75367736816406, | |
| "rewards/margins": 58.79213333129883, | |
| "rewards/rejected": -144.5458221435547, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5259026687598116, | |
| "grad_norm": 14.259535892393998, | |
| "learning_rate": 1.627319622002544e-07, | |
| "logits/chosen": 5289.35595703125, | |
| "logits/rejected": 4184.24462890625, | |
| "logps/chosen": -298.56573486328125, | |
| "logps/rejected": -323.38275146484375, | |
| "loss": 1.3366, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -91.83341979980469, | |
| "rewards/margins": 42.687252044677734, | |
| "rewards/rejected": -134.52066040039062, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.533751962323391, | |
| "grad_norm": 13.615931001877374, | |
| "learning_rate": 1.5863049882692565e-07, | |
| "logits/chosen": 5200.74267578125, | |
| "logits/rejected": 4202.2919921875, | |
| "logps/chosen": -306.4901123046875, | |
| "logps/rejected": -322.50390625, | |
| "loss": 1.3358, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -92.8583984375, | |
| "rewards/margins": 50.42744445800781, | |
| "rewards/rejected": -143.28582763671875, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5416012558869702, | |
| "grad_norm": 18.445984962657718, | |
| "learning_rate": 1.5452255001711795e-07, | |
| "logits/chosen": 5152.58740234375, | |
| "logits/rejected": 4540.1044921875, | |
| "logps/chosen": -297.11669921875, | |
| "logps/rejected": -357.1095275878906, | |
| "loss": 1.3182, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -87.97896575927734, | |
| "rewards/margins": 52.92973709106445, | |
| "rewards/rejected": -140.90872192382812, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5494505494505495, | |
| "grad_norm": 22.733099236513056, | |
| "learning_rate": 1.50411202711876e-07, | |
| "logits/chosen": 4857.0771484375, | |
| "logits/rejected": 4486.8955078125, | |
| "logps/chosen": -285.587646484375, | |
| "logps/rejected": -335.50311279296875, | |
| "loss": 1.3496, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -96.36004638671875, | |
| "rewards/margins": 52.195396423339844, | |
| "rewards/rejected": -148.55545043945312, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5572998430141287, | |
| "grad_norm": 13.753296852144786, | |
| "learning_rate": 1.4629954640606275e-07, | |
| "logits/chosen": 4837.16650390625, | |
| "logits/rejected": 4699.06884765625, | |
| "logps/chosen": -282.49725341796875, | |
| "logps/rejected": -339.46014404296875, | |
| "loss": 1.3241, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -88.5594253540039, | |
| "rewards/margins": 48.25908660888672, | |
| "rewards/rejected": -136.81851196289062, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.565149136577708, | |
| "grad_norm": 23.74236196605101, | |
| "learning_rate": 1.4219067082674153e-07, | |
| "logits/chosen": 5229.64599609375, | |
| "logits/rejected": 4363.564453125, | |
| "logps/chosen": -307.42694091796875, | |
| "logps/rejected": -333.94476318359375, | |
| "loss": 1.3345, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -88.42107391357422, | |
| "rewards/margins": 45.73460006713867, | |
| "rewards/rejected": -134.1556854248047, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5729984301412873, | |
| "grad_norm": 14.581185643023554, | |
| "learning_rate": 1.3808766361138294e-07, | |
| "logits/chosen": 5660.58642578125, | |
| "logits/rejected": 4180.7021484375, | |
| "logps/chosen": -318.3136901855469, | |
| "logps/rejected": -321.05255126953125, | |
| "loss": 1.3196, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -93.13568115234375, | |
| "rewards/margins": 52.86260223388672, | |
| "rewards/rejected": -145.99827575683594, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5808477237048666, | |
| "grad_norm": 20.680345594369143, | |
| "learning_rate": 1.3399360798764307e-07, | |
| "logits/chosen": 4665.7041015625, | |
| "logits/rejected": 3810.73828125, | |
| "logps/chosen": -277.14202880859375, | |
| "logps/rejected": -301.589111328125, | |
| "loss": 1.3096, | |
| "rewards/accuracies": 0.7916666269302368, | |
| "rewards/chosen": -84.7000732421875, | |
| "rewards/margins": 54.24741744995117, | |
| "rewards/rejected": -138.94747924804688, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5886970172684458, | |
| "grad_norm": 21.101874083978554, | |
| "learning_rate": 1.2991158045645435e-07, | |
| "logits/chosen": 5032.05322265625, | |
| "logits/rejected": 4430.30322265625, | |
| "logps/chosen": -268.8847961425781, | |
| "logps/rejected": -301.6675109863281, | |
| "loss": 1.3294, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -88.50262451171875, | |
| "rewards/margins": 46.05238723754883, | |
| "rewards/rejected": -134.55502319335938, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5965463108320251, | |
| "grad_norm": 16.80574616569174, | |
| "learning_rate": 1.2584464848017214e-07, | |
| "logits/chosen": 5412.44775390625, | |
| "logits/rejected": 4117.3740234375, | |
| "logps/chosen": -287.7677001953125, | |
| "logps/rejected": -306.7326354980469, | |
| "loss": 1.3314, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -84.49854278564453, | |
| "rewards/margins": 47.73679733276367, | |
| "rewards/rejected": -132.23533630371094, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.6043956043956044, | |
| "grad_norm": 26.26369158658654, | |
| "learning_rate": 1.2179586817751263e-07, | |
| "logits/chosen": 5509.0380859375, | |
| "logits/rejected": 4625.6474609375, | |
| "logps/chosen": -320.41656494140625, | |
| "logps/rejected": -366.6982421875, | |
| "loss": 1.3377, | |
| "rewards/accuracies": 0.7166667580604553, | |
| "rewards/chosen": -90.7030258178711, | |
| "rewards/margins": 52.11555099487305, | |
| "rewards/rejected": -142.81857299804688, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6122448979591837, | |
| "grad_norm": 16.19764581259732, | |
| "learning_rate": 1.177682820270159e-07, | |
| "logits/chosen": 5451.91162109375, | |
| "logits/rejected": 5010.38232421875, | |
| "logps/chosen": -305.00799560546875, | |
| "logps/rejected": -377.30596923828125, | |
| "loss": 1.3328, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -88.89636993408203, | |
| "rewards/margins": 51.006568908691406, | |
| "rewards/rejected": -139.9029541015625, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6200941915227629, | |
| "grad_norm": 16.344358737664763, | |
| "learning_rate": 1.13764916580758e-07, | |
| "logits/chosen": 4994.599609375, | |
| "logits/rejected": 4273.96875, | |
| "logps/chosen": -293.0992126464844, | |
| "logps/rejected": -346.97467041015625, | |
| "loss": 1.3234, | |
| "rewards/accuracies": 0.7916667461395264, | |
| "rewards/chosen": -94.93836975097656, | |
| "rewards/margins": 63.40740966796875, | |
| "rewards/rejected": -158.3457794189453, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6279434850863422, | |
| "grad_norm": 18.968431335137478, | |
| "learning_rate": 1.0978878019003183e-07, | |
| "logits/chosen": 5044.96875, | |
| "logits/rejected": 4262.8154296875, | |
| "logps/chosen": -310.9752502441406, | |
| "logps/rejected": -353.9684143066406, | |
| "loss": 1.3377, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -94.22391510009766, | |
| "rewards/margins": 47.09145736694336, | |
| "rewards/rejected": -141.31536865234375, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6357927786499215, | |
| "grad_norm": 13.606253137803622, | |
| "learning_rate": 1.0584286074470446e-07, | |
| "logits/chosen": 4622.6064453125, | |
| "logits/rejected": 3968.47412109375, | |
| "logps/chosen": -293.10089111328125, | |
| "logps/rejected": -363.0287170410156, | |
| "loss": 1.3322, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -90.6499252319336, | |
| "rewards/margins": 51.11803436279297, | |
| "rewards/rejected": -141.7679443359375, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6436420722135008, | |
| "grad_norm": 38.62289956267384, | |
| "learning_rate": 1.0193012342795095e-07, | |
| "logits/chosen": 5154.56494140625, | |
| "logits/rejected": 3781.949951171875, | |
| "logps/chosen": -306.31817626953125, | |
| "logps/rejected": -307.9537048339844, | |
| "loss": 1.32, | |
| "rewards/accuracies": 0.8333333730697632, | |
| "rewards/chosen": -85.11204528808594, | |
| "rewards/margins": 54.1730842590332, | |
| "rewards/rejected": -139.28512573242188, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6514913657770801, | |
| "grad_norm": 22.79781152184824, | |
| "learning_rate": 9.805350848805078e-08, | |
| "logits/chosen": 5019.9873046875, | |
| "logits/rejected": 4204.40673828125, | |
| "logps/chosen": -296.72509765625, | |
| "logps/rejected": -338.36236572265625, | |
| "loss": 1.3248, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -92.53424072265625, | |
| "rewards/margins": 53.35478973388672, | |
| "rewards/rejected": -145.88902282714844, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6593406593406593, | |
| "grad_norm": 14.988614689944379, | |
| "learning_rate": 9.421592902892184e-08, | |
| "logits/chosen": 5479.388671875, | |
| "logits/rejected": 4344.3359375, | |
| "logps/chosen": -319.1064147949219, | |
| "logps/rejected": -352.5812683105469, | |
| "loss": 1.3196, | |
| "rewards/accuracies": 0.6833332777023315, | |
| "rewards/chosen": -91.74089050292969, | |
| "rewards/margins": 48.036460876464844, | |
| "rewards/rejected": -139.77737426757812, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6671899529042387, | |
| "grad_norm": 16.378454591394703, | |
| "learning_rate": 9.042026882105215e-08, | |
| "logits/chosen": 5068.8427734375, | |
| "logits/rejected": 4380.408203125, | |
| "logps/chosen": -304.6463928222656, | |
| "logps/rejected": -351.93988037109375, | |
| "loss": 1.311, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -89.17027282714844, | |
| "rewards/margins": 62.253204345703125, | |
| "rewards/rejected": -151.42349243164062, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6750392464678179, | |
| "grad_norm": 33.867305610136256, | |
| "learning_rate": 8.666938013447484e-08, | |
| "logits/chosen": 5499.62353515625, | |
| "logits/rejected": 4524.25390625, | |
| "logps/chosen": -333.63714599609375, | |
| "logps/rejected": -339.46429443359375, | |
| "loss": 1.3436, | |
| "rewards/accuracies": 0.7416666150093079, | |
| "rewards/chosen": -85.83646392822266, | |
| "rewards/margins": 50.95392990112305, | |
| "rewards/rejected": -136.79037475585938, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6828885400313972, | |
| "grad_norm": 16.295742425420926, | |
| "learning_rate": 8.29660815954134e-08, | |
| "logits/chosen": 5463.05712890625, | |
| "logits/rejected": 4347.86669921875, | |
| "logps/chosen": -314.76629638671875, | |
| "logps/rejected": -340.78515625, | |
| "loss": 1.3232, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -84.63117980957031, | |
| "rewards/margins": 55.43559646606445, | |
| "rewards/rejected": -140.0667724609375, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6907378335949764, | |
| "grad_norm": 35.343044136950866, | |
| "learning_rate": 7.931315606820979e-08, | |
| "logits/chosen": 4798.94140625, | |
| "logits/rejected": 4442.7099609375, | |
| "logps/chosen": -299.4132080078125, | |
| "logps/rejected": -370.7325744628906, | |
| "loss": 1.3375, | |
| "rewards/accuracies": 0.7500000596046448, | |
| "rewards/chosen": -88.87822723388672, | |
| "rewards/margins": 50.25794219970703, | |
| "rewards/rejected": -139.13616943359375, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6985871271585558, | |
| "grad_norm": 15.65569216217348, | |
| "learning_rate": 7.571334856412503e-08, | |
| "logits/chosen": 5700.92724609375, | |
| "logits/rejected": 4396.40234375, | |
| "logps/chosen": -342.78717041015625, | |
| "logps/rejected": -354.91058349609375, | |
| "loss": 1.337, | |
| "rewards/accuracies": 0.8083333969116211, | |
| "rewards/chosen": -85.16032409667969, | |
| "rewards/margins": 57.98085403442383, | |
| "rewards/rejected": -143.14117431640625, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.706436420722135, | |
| "grad_norm": 18.88682608723191, | |
| "learning_rate": 7.216936417858549e-08, | |
| "logits/chosen": 4812.2412109375, | |
| "logits/rejected": 4236.50439453125, | |
| "logps/chosen": -288.49560546875, | |
| "logps/rejected": -342.0299987792969, | |
| "loss": 1.3348, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -84.1703872680664, | |
| "rewards/margins": 47.417232513427734, | |
| "rewards/rejected": -131.58761596679688, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 16.830087200212606, | |
| "learning_rate": 6.868386605842379e-08, | |
| "logits/chosen": 4705.6962890625, | |
| "logits/rejected": 4288.7353515625, | |
| "logps/chosen": -298.02349853515625, | |
| "logps/rejected": -354.6422119140625, | |
| "loss": 1.3302, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -89.75446319580078, | |
| "rewards/margins": 56.94415283203125, | |
| "rewards/rejected": -146.6986083984375, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.7221350078492935, | |
| "grad_norm": 15.005561893820257, | |
| "learning_rate": 6.525947340064232e-08, | |
| "logits/chosen": 5283.912109375, | |
| "logits/rejected": 4178.82958984375, | |
| "logps/chosen": -321.24029541015625, | |
| "logps/rejected": -337.23944091796875, | |
| "loss": 1.3283, | |
| "rewards/accuracies": 0.7666667103767395, | |
| "rewards/chosen": -88.7941665649414, | |
| "rewards/margins": 50.259727478027344, | |
| "rewards/rejected": -139.0539093017578, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.7299843014128728, | |
| "grad_norm": 15.702994402687256, | |
| "learning_rate": 6.189875948420276e-08, | |
| "logits/chosen": 5172.5234375, | |
| "logits/rejected": 3729.58544921875, | |
| "logps/chosen": -328.5713806152344, | |
| "logps/rejected": -337.1180114746094, | |
| "loss": 1.324, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": -92.56468200683594, | |
| "rewards/margins": 58.63731002807617, | |
| "rewards/rejected": -151.20199584960938, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7378335949764521, | |
| "grad_norm": 28.79822002326502, | |
| "learning_rate": 5.86042497363215e-08, | |
| "logits/chosen": 5023.8671875, | |
| "logits/rejected": 4205.1982421875, | |
| "logps/chosen": -307.396240234375, | |
| "logps/rejected": -361.35791015625, | |
| "loss": 1.3224, | |
| "rewards/accuracies": 0.8083333969116211, | |
| "rewards/chosen": -89.06114959716797, | |
| "rewards/margins": 67.33191680908203, | |
| "rewards/rejected": -156.39306640625, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7456828885400314, | |
| "grad_norm": 29.49903030029792, | |
| "learning_rate": 5.537841983472337e-08, | |
| "logits/chosen": 5192.91015625, | |
| "logits/rejected": 4614.15087890625, | |
| "logps/chosen": -308.1656188964844, | |
| "logps/rejected": -341.88336181640625, | |
| "loss": 1.3437, | |
| "rewards/accuracies": 0.7166666388511658, | |
| "rewards/chosen": -90.37879943847656, | |
| "rewards/margins": 37.416908264160156, | |
| "rewards/rejected": -127.79571533203125, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7535321821036107, | |
| "grad_norm": 19.260341973231338, | |
| "learning_rate": 5.222369384727954e-08, | |
| "logits/chosen": 5086.7275390625, | |
| "logits/rejected": 4373.75390625, | |
| "logps/chosen": -302.8155212402344, | |
| "logps/rejected": -341.2423095703125, | |
| "loss": 1.3244, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": -84.31675720214844, | |
| "rewards/margins": 54.4622802734375, | |
| "rewards/rejected": -138.77902221679688, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7613814756671899, | |
| "grad_norm": 21.497254099738637, | |
| "learning_rate": 4.914244241042832e-08, | |
| "logits/chosen": 5195.4404296875, | |
| "logits/rejected": 4174.501953125, | |
| "logps/chosen": -328.4187316894531, | |
| "logps/rejected": -357.6769714355469, | |
| "loss": 1.3155, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -88.17876434326172, | |
| "rewards/margins": 62.826393127441406, | |
| "rewards/rejected": -151.00515747070312, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 14.487995995435664, | |
| "learning_rate": 4.613698094774715e-08, | |
| "logits/chosen": 4949.87158203125, | |
| "logits/rejected": 3807.67724609375, | |
| "logps/chosen": -312.44573974609375, | |
| "logps/rejected": -349.9386291503906, | |
| "loss": 1.3152, | |
| "rewards/accuracies": 0.8333333134651184, | |
| "rewards/chosen": -83.61039733886719, | |
| "rewards/margins": 75.85111236572266, | |
| "rewards/rejected": -159.46151733398438, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7770800627943485, | |
| "grad_norm": 19.108590316137075, | |
| "learning_rate": 4.3209567930014746e-08, | |
| "logits/chosen": 5185.36865234375, | |
| "logits/rejected": 4195.95068359375, | |
| "logps/chosen": -315.67938232421875, | |
| "logps/rejected": -348.85284423828125, | |
| "loss": 1.3174, | |
| "rewards/accuracies": 0.7500000596046448, | |
| "rewards/chosen": -89.78607177734375, | |
| "rewards/margins": 64.9991683959961, | |
| "rewards/rejected": -154.7852325439453, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7849293563579278, | |
| "grad_norm": 22.072531979816144, | |
| "learning_rate": 4.036240317807059e-08, | |
| "logits/chosen": 4905.2353515625, | |
| "logits/rejected": 3987.02978515625, | |
| "logps/chosen": -303.3591003417969, | |
| "logps/rejected": -321.552978515625, | |
| "loss": 1.3349, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -89.55894470214844, | |
| "rewards/margins": 51.63554000854492, | |
| "rewards/rejected": -141.1945037841797, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.792778649921507, | |
| "grad_norm": 21.75070665962702, | |
| "learning_rate": 3.759762620974716e-08, | |
| "logits/chosen": 5060.6669921875, | |
| "logits/rejected": 4109.7626953125, | |
| "logps/chosen": -304.08740234375, | |
| "logps/rejected": -337.9571838378906, | |
| "loss": 1.334, | |
| "rewards/accuracies": 0.7083333134651184, | |
| "rewards/chosen": -89.3779296875, | |
| "rewards/margins": 57.18464279174805, | |
| "rewards/rejected": -146.56259155273438, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.8006279434850864, | |
| "grad_norm": 17.10653047682008, | |
| "learning_rate": 3.4917314632117423e-08, | |
| "logits/chosen": 5245.35009765625, | |
| "logits/rejected": 4436.5830078125, | |
| "logps/chosen": -326.85821533203125, | |
| "logps/rejected": -367.40740966796875, | |
| "loss": 1.3302, | |
| "rewards/accuracies": 0.8333333730697632, | |
| "rewards/chosen": -78.67982482910156, | |
| "rewards/margins": 62.575775146484375, | |
| "rewards/rejected": -141.25559997558594, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.8084772370486656, | |
| "grad_norm": 16.25681330236911, | |
| "learning_rate": 3.232348258026552e-08, | |
| "logits/chosen": 5550.1083984375, | |
| "logits/rejected": 5013.5693359375, | |
| "logps/chosen": -318.5128479003906, | |
| "logps/rejected": -371.39556884765625, | |
| "loss": 1.329, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": -87.0693130493164, | |
| "rewards/margins": 51.2072868347168, | |
| "rewards/rejected": -138.276611328125, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 15.78931624178539, | |
| "learning_rate": 2.9818079203753654e-08, | |
| "logits/chosen": 4891.5419921875, | |
| "logits/rejected": 4282.7998046875, | |
| "logps/chosen": -314.73773193359375, | |
| "logps/rejected": -368.1146240234375, | |
| "loss": 1.3238, | |
| "rewards/accuracies": 0.7250000834465027, | |
| "rewards/chosen": -94.64768981933594, | |
| "rewards/margins": 64.27806854248047, | |
| "rewards/rejected": -158.92575073242188, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.8241758241758241, | |
| "grad_norm": 16.54458435339151, | |
| "learning_rate": 2.740298720192245e-08, | |
| "logits/chosen": 4927.3154296875, | |
| "logits/rejected": 4254.25341796875, | |
| "logps/chosen": -328.06463623046875, | |
| "logps/rejected": -370.9690856933594, | |
| "loss": 1.3151, | |
| "rewards/accuracies": 0.8083332777023315, | |
| "rewards/chosen": -90.51560974121094, | |
| "rewards/margins": 64.61587524414062, | |
| "rewards/rejected": -155.13148498535156, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.8320251177394035, | |
| "grad_norm": 26.179961673866252, | |
| "learning_rate": 2.5080021409126655e-08, | |
| "logits/chosen": 6100.56396484375, | |
| "logits/rejected": 4631.39599609375, | |
| "logps/chosen": -352.55450439453125, | |
| "logps/rejected": -360.7041015625, | |
| "loss": 1.3358, | |
| "rewards/accuracies": 0.7333332896232605, | |
| "rewards/chosen": -91.52915954589844, | |
| "rewards/margins": 49.11104202270508, | |
| "rewards/rejected": -140.6402130126953, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.8398744113029827, | |
| "grad_norm": 17.503149002663815, | |
| "learning_rate": 2.2850927430967287e-08, | |
| "logits/chosen": 5088.08935546875, | |
| "logits/rejected": 4477.36572265625, | |
| "logps/chosen": -312.17938232421875, | |
| "logps/rejected": -340.80975341796875, | |
| "loss": 1.3326, | |
| "rewards/accuracies": 0.73333340883255, | |
| "rewards/chosen": -86.90293884277344, | |
| "rewards/margins": 60.76922607421875, | |
| "rewards/rejected": -147.67214965820312, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.847723704866562, | |
| "grad_norm": 16.608334704247707, | |
| "learning_rate": 2.0717380332546768e-08, | |
| "logits/chosen": 5636.8876953125, | |
| "logits/rejected": 4137.8525390625, | |
| "logps/chosen": -351.24298095703125, | |
| "logps/rejected": -354.1505126953125, | |
| "loss": 1.3203, | |
| "rewards/accuracies": 0.8333333730697632, | |
| "rewards/chosen": -84.64915466308594, | |
| "rewards/margins": 68.51604461669922, | |
| "rewards/rejected": -153.1652069091797, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8555729984301413, | |
| "grad_norm": 16.306285487869843, | |
| "learning_rate": 1.8680983379731618e-08, | |
| "logits/chosen": 5661.28173828125, | |
| "logits/rejected": 4081.420654296875, | |
| "logps/chosen": -317.64776611328125, | |
| "logps/rejected": -356.91375732421875, | |
| "loss": 1.3259, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -90.13359069824219, | |
| "rewards/margins": 53.870460510253906, | |
| "rewards/rejected": -144.0040283203125, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8634222919937206, | |
| "grad_norm": 17.07320149919412, | |
| "learning_rate": 1.6743266834369856e-08, | |
| "logits/chosen": 5442.111328125, | |
| "logits/rejected": 4267.1455078125, | |
| "logps/chosen": -330.0649108886719, | |
| "logps/rejected": -343.8323669433594, | |
| "loss": 1.3308, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -91.5087661743164, | |
| "rewards/margins": 62.82781982421875, | |
| "rewards/rejected": -154.3365936279297, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8712715855572999, | |
| "grad_norm": 13.747396863762589, | |
| "learning_rate": 1.4905686804366853e-08, | |
| "logits/chosen": 5771.70458984375, | |
| "logits/rejected": 4442.76708984375, | |
| "logps/chosen": -355.2906799316406, | |
| "logps/rejected": -374.15447998046875, | |
| "loss": 1.3248, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -90.69126892089844, | |
| "rewards/margins": 61.19647216796875, | |
| "rewards/rejected": -151.8877410888672, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8791208791208791, | |
| "grad_norm": 27.204997822783117, | |
| "learning_rate": 1.3169624149485037e-08, | |
| "logits/chosen": 5658.005859375, | |
| "logits/rejected": 4368.58642578125, | |
| "logps/chosen": -337.535888671875, | |
| "logps/rejected": -371.7395324707031, | |
| "loss": 1.3197, | |
| "rewards/accuracies": 0.7500000596046448, | |
| "rewards/chosen": -92.40483093261719, | |
| "rewards/margins": 62.56023025512695, | |
| "rewards/rejected": -154.96505737304688, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8869701726844584, | |
| "grad_norm": 20.723713223509073, | |
| "learning_rate": 1.1536383443688968e-08, | |
| "logits/chosen": 5141.8701171875, | |
| "logits/rejected": 3853.640625, | |
| "logps/chosen": -318.36944580078125, | |
| "logps/rejected": -326.81610107421875, | |
| "loss": 1.3163, | |
| "rewards/accuracies": 0.7916666865348816, | |
| "rewards/chosen": -83.58235168457031, | |
| "rewards/margins": 63.00525665283203, | |
| "rewards/rejected": -146.58761596679688, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8948194662480377, | |
| "grad_norm": 15.383586929186926, | |
| "learning_rate": 1.0007191994816272e-08, | |
| "logits/chosen": 5663.7626953125, | |
| "logits/rejected": 4409.96728515625, | |
| "logps/chosen": -343.25408935546875, | |
| "logps/rejected": -359.54864501953125, | |
| "loss": 1.3158, | |
| "rewards/accuracies": 0.7166666388511658, | |
| "rewards/chosen": -85.61146545410156, | |
| "rewards/margins": 53.071754455566406, | |
| "rewards/rejected": -138.68319702148438, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.902668759811617, | |
| "grad_norm": 15.9052219677393, | |
| "learning_rate": 8.583198922310031e-09, | |
| "logits/chosen": 4978.74658203125, | |
| "logits/rejected": 4447.34521484375, | |
| "logps/chosen": -312.7223815917969, | |
| "logps/rejected": -353.9768981933594, | |
| "loss": 1.3155, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": -85.73565673828125, | |
| "rewards/margins": 48.706329345703125, | |
| "rewards/rejected": -134.44198608398438, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.9105180533751962, | |
| "grad_norm": 19.701859698836042, | |
| "learning_rate": 7.2654742937067916e-09, | |
| "logits/chosen": 5107.5390625, | |
| "logits/rejected": 3909.550048828125, | |
| "logps/chosen": -320.367431640625, | |
| "logps/rejected": -338.51812744140625, | |
| "loss": 1.3246, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -91.79454803466797, | |
| "rewards/margins": 50.79388427734375, | |
| "rewards/rejected": -142.5884246826172, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.9183673469387755, | |
| "grad_norm": 32.20474236778808, | |
| "learning_rate": 6.0550083205280325e-09, | |
| "logits/chosen": 5575.20849609375, | |
| "logits/rejected": 5152.0107421875, | |
| "logps/chosen": -332.0797424316406, | |
| "logps/rejected": -392.1387023925781, | |
| "loss": 1.3201, | |
| "rewards/accuracies": 0.73333340883255, | |
| "rewards/chosen": -89.12452697753906, | |
| "rewards/margins": 56.22435760498047, | |
| "rewards/rejected": -145.34889221191406, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.9262166405023547, | |
| "grad_norm": 19.875692057488717, | |
| "learning_rate": 4.952710614180567e-09, | |
| "logits/chosen": 4974.9326171875, | |
| "logits/rejected": 4310.21337890625, | |
| "logps/chosen": -328.6920166015625, | |
| "logps/rejected": -362.4015197753906, | |
| "loss": 1.3248, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -88.78321075439453, | |
| "rewards/margins": 56.8275032043457, | |
| "rewards/rejected": -145.61070251464844, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.9340659340659341, | |
| "grad_norm": 18.2157711653637, | |
| "learning_rate": 3.959409502423744e-09, | |
| "logits/chosen": 5420.66015625, | |
| "logits/rejected": 4498.2080078125, | |
| "logps/chosen": -345.4786071777344, | |
| "logps/rejected": -369.42694091796875, | |
| "loss": 1.3206, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -96.6148452758789, | |
| "rewards/margins": 55.66218185424805, | |
| "rewards/rejected": -152.2770233154297, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.9419152276295133, | |
| "grad_norm": 19.275111817920287, | |
| "learning_rate": 3.075851406917995e-09, | |
| "logits/chosen": 5607.0615234375, | |
| "logits/rejected": 4485.181640625, | |
| "logps/chosen": -340.6480407714844, | |
| "logps/rejected": -372.07196044921875, | |
| "loss": 1.3207, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -85.0034408569336, | |
| "rewards/margins": 68.38439178466797, | |
| "rewards/rejected": -153.38784790039062, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9497645211930926, | |
| "grad_norm": 47.451083320046365, | |
| "learning_rate": 2.3027002823221064e-09, | |
| "logits/chosen": 5242.3955078125, | |
| "logits/rejected": 4309.052734375, | |
| "logps/chosen": -328.3695983886719, | |
| "logps/rejected": -372.03692626953125, | |
| "loss": 1.3288, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -88.16230773925781, | |
| "rewards/margins": 65.71019744873047, | |
| "rewards/rejected": -153.87249755859375, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.957613814756672, | |
| "grad_norm": 17.39982846549713, | |
| "learning_rate": 1.6405371173608073e-09, | |
| "logits/chosen": 6537.92041015625, | |
| "logits/rejected": 4597.3408203125, | |
| "logps/chosen": -389.57415771484375, | |
| "logps/rejected": -396.27960205078125, | |
| "loss": 1.3285, | |
| "rewards/accuracies": 0.7500000596046448, | |
| "rewards/chosen": -92.74442291259766, | |
| "rewards/margins": 58.356590270996094, | |
| "rewards/rejected": -151.10101318359375, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9654631083202512, | |
| "grad_norm": 17.667026886080837, | |
| "learning_rate": 1.0898594982375387e-09, | |
| "logits/chosen": 5799.66064453125, | |
| "logits/rejected": 4413.10302734375, | |
| "logps/chosen": -349.25537109375, | |
| "logps/rejected": -362.83990478515625, | |
| "loss": 1.3307, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -88.6881332397461, | |
| "rewards/margins": 61.15995407104492, | |
| "rewards/rejected": -149.8480987548828, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9733124018838305, | |
| "grad_norm": 13.882974741420169, | |
| "learning_rate": 6.510812347205185e-10, | |
| "logits/chosen": 4966.72021484375, | |
| "logits/rejected": 4371.02392578125, | |
| "logps/chosen": -324.99078369140625, | |
| "logps/rejected": -384.681640625, | |
| "loss": 1.324, | |
| "rewards/accuracies": 0.7083333134651184, | |
| "rewards/chosen": -93.39674377441406, | |
| "rewards/margins": 56.08783721923828, | |
| "rewards/rejected": -149.48458862304688, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9811616954474097, | |
| "grad_norm": 15.154457626536647, | |
| "learning_rate": 3.245320491830683e-10, | |
| "logits/chosen": 5032.3193359375, | |
| "logits/rejected": 4055.299560546875, | |
| "logps/chosen": -315.40277099609375, | |
| "logps/rejected": -327.2320251464844, | |
| "loss": 1.3286, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -85.07127380371094, | |
| "rewards/margins": 52.71143341064453, | |
| "rewards/rejected": -137.78269958496094, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.989010989010989, | |
| "grad_norm": 15.979946674969982, | |
| "learning_rate": 1.1045732883197545e-10, | |
| "logits/chosen": 4809.5751953125, | |
| "logits/rejected": 4305.59765625, | |
| "logps/chosen": -319.3553771972656, | |
| "logps/rejected": -360.33917236328125, | |
| "loss": 1.3166, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -87.20452117919922, | |
| "rewards/margins": 56.80842208862305, | |
| "rewards/rejected": -144.01295471191406, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9968602825745683, | |
| "grad_norm": 27.24614835648412, | |
| "learning_rate": 9.017941309813482e-12, | |
| "logits/chosen": 4971.7294921875, | |
| "logits/rejected": 3906.59228515625, | |
| "logps/chosen": -316.85064697265625, | |
| "logps/rejected": -327.2884521484375, | |
| "loss": 1.3334, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -92.36187744140625, | |
| "rewards/margins": 48.6181755065918, | |
| "rewards/rejected": -140.9800567626953, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1274, | |
| "total_flos": 0.0, | |
| "train_loss": 1.3597402943172485, | |
| "train_runtime": 13015.2109, | |
| "train_samples_per_second": 4.697, | |
| "train_steps_per_second": 0.098 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1274, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 2000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": false, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 6, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |