| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 400, | |
| "global_step": 564, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.008865248226950355, | |
| "grad_norm": 149.09850698280093, | |
| "learning_rate": 4.385964912280701e-09, | |
| "logps/chosen": -2.8927114009857178, | |
| "logps/rejected": -0.7169164419174194, | |
| "loss": 25.3611, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.927114486694336, | |
| "rewards/margins": -21.757949829101562, | |
| "rewards/rejected": -7.169164180755615, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01773049645390071, | |
| "grad_norm": 133.9237477557748, | |
| "learning_rate": 8.771929824561403e-09, | |
| "logps/chosen": -2.677772045135498, | |
| "logps/rejected": -0.7764161825180054, | |
| "loss": 24.583, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.777721405029297, | |
| "rewards/margins": -19.013559341430664, | |
| "rewards/rejected": -7.764161109924316, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.026595744680851064, | |
| "grad_norm": 152.65228499963413, | |
| "learning_rate": 1.3157894736842104e-08, | |
| "logps/chosen": -2.491425037384033, | |
| "logps/rejected": -0.7799981832504272, | |
| "loss": 24.3721, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -24.914249420166016, | |
| "rewards/margins": -17.114269256591797, | |
| "rewards/rejected": -7.79998254776001, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03546099290780142, | |
| "grad_norm": 167.82984889261684, | |
| "learning_rate": 1.7543859649122805e-08, | |
| "logps/chosen": -2.818477153778076, | |
| "logps/rejected": -0.8185766935348511, | |
| "loss": 24.5782, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.184768676757812, | |
| "rewards/margins": -19.999004364013672, | |
| "rewards/rejected": -8.185766220092773, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.044326241134751775, | |
| "grad_norm": 158.80212467372465, | |
| "learning_rate": 2.1929824561403507e-08, | |
| "logps/chosen": -3.06237530708313, | |
| "logps/rejected": -0.685140073299408, | |
| "loss": 24.7471, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -30.62375259399414, | |
| "rewards/margins": -23.77235221862793, | |
| "rewards/rejected": -6.851400852203369, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05319148936170213, | |
| "grad_norm": 146.70835684303188, | |
| "learning_rate": 2.6315789473684208e-08, | |
| "logps/chosen": -2.6460320949554443, | |
| "logps/rejected": -0.7569402456283569, | |
| "loss": 24.7665, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.4603214263916, | |
| "rewards/margins": -18.890918731689453, | |
| "rewards/rejected": -7.569401741027832, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06205673758865248, | |
| "grad_norm": 155.08422103006285, | |
| "learning_rate": 3.070175438596491e-08, | |
| "logps/chosen": -2.725968837738037, | |
| "logps/rejected": -0.7673903703689575, | |
| "loss": 24.0323, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.259693145751953, | |
| "rewards/margins": -19.58578872680664, | |
| "rewards/rejected": -7.673903465270996, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.07092198581560284, | |
| "grad_norm": 144.68445966753478, | |
| "learning_rate": 3.508771929824561e-08, | |
| "logps/chosen": -2.8730432987213135, | |
| "logps/rejected": -0.7648515105247498, | |
| "loss": 24.9421, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.73043441772461, | |
| "rewards/margins": -21.081920623779297, | |
| "rewards/rejected": -7.648515224456787, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0797872340425532, | |
| "grad_norm": 136.66852055782576, | |
| "learning_rate": 3.947368421052631e-08, | |
| "logps/chosen": -2.648432970046997, | |
| "logps/rejected": -0.7622749209403992, | |
| "loss": 24.9349, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.484333038330078, | |
| "rewards/margins": -18.861581802368164, | |
| "rewards/rejected": -7.622749328613281, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.08865248226950355, | |
| "grad_norm": 137.8970167243902, | |
| "learning_rate": 4.385964912280701e-08, | |
| "logps/chosen": -2.7805895805358887, | |
| "logps/rejected": -0.7085495591163635, | |
| "loss": 23.7959, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.805896759033203, | |
| "rewards/margins": -20.72039794921875, | |
| "rewards/rejected": -7.085495948791504, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0975177304964539, | |
| "grad_norm": 144.53363403919823, | |
| "learning_rate": 4.8245614035087715e-08, | |
| "logps/chosen": -2.7633585929870605, | |
| "logps/rejected": -0.766043484210968, | |
| "loss": 24.5894, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.63358497619629, | |
| "rewards/margins": -19.973148345947266, | |
| "rewards/rejected": -7.660434722900391, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.10638297872340426, | |
| "grad_norm": 186.10146168517474, | |
| "learning_rate": 4.999568059583401e-08, | |
| "logps/chosen": -2.605541706085205, | |
| "logps/rejected": -0.7882484197616577, | |
| "loss": 24.1209, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.0554141998291, | |
| "rewards/margins": -18.172931671142578, | |
| "rewards/rejected": -7.88248348236084, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11524822695035461, | |
| "grad_norm": 144.73338067190508, | |
| "learning_rate": 4.9969289642076847e-08, | |
| "logps/chosen": -2.691685199737549, | |
| "logps/rejected": -0.7605332136154175, | |
| "loss": 24.5817, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.916851043701172, | |
| "rewards/margins": -19.311519622802734, | |
| "rewards/rejected": -7.6053314208984375, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.12411347517730496, | |
| "grad_norm": 142.10076756137335, | |
| "learning_rate": 4.991893270335525e-08, | |
| "logps/chosen": -2.887664318084717, | |
| "logps/rejected": -0.7798042297363281, | |
| "loss": 25.4368, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.876644134521484, | |
| "rewards/margins": -21.078603744506836, | |
| "rewards/rejected": -7.798041343688965, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13297872340425532, | |
| "grad_norm": 136.39554296889824, | |
| "learning_rate": 4.9844658113188256e-08, | |
| "logps/chosen": -2.554642915725708, | |
| "logps/rejected": -0.741515576839447, | |
| "loss": 25.032, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.546428680419922, | |
| "rewards/margins": -18.131275177001953, | |
| "rewards/rejected": -7.41515588760376, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.14184397163120568, | |
| "grad_norm": 150.8035431184681, | |
| "learning_rate": 4.974653716169812e-08, | |
| "logps/chosen": -2.592639923095703, | |
| "logps/rejected": -0.7398035526275635, | |
| "loss": 24.5209, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.9263973236084, | |
| "rewards/margins": -18.528364181518555, | |
| "rewards/rejected": -7.398035526275635, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.15070921985815602, | |
| "grad_norm": 131.31880148234296, | |
| "learning_rate": 4.962466402718474e-08, | |
| "logps/chosen": -2.76173734664917, | |
| "logps/rejected": -0.7157766819000244, | |
| "loss": 24.4644, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.61737060546875, | |
| "rewards/margins": -20.459604263305664, | |
| "rewards/rejected": -7.157766819000244, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1595744680851064, | |
| "grad_norm": 134.79926804392827, | |
| "learning_rate": 4.9479155685731585e-08, | |
| "logps/chosen": -2.651050329208374, | |
| "logps/rejected": -0.8522893190383911, | |
| "loss": 23.8276, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.510501861572266, | |
| "rewards/margins": -17.98760986328125, | |
| "rewards/rejected": -8.522892951965332, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.16843971631205673, | |
| "grad_norm": 148.45315521916737, | |
| "learning_rate": 4.93101517989299e-08, | |
| "logps/chosen": -2.806105136871338, | |
| "logps/rejected": -0.8227446675300598, | |
| "loss": 24.8131, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.061050415039062, | |
| "rewards/margins": -19.833606719970703, | |
| "rewards/rejected": -8.227446556091309, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.1773049645390071, | |
| "grad_norm": 143.9707085484871, | |
| "learning_rate": 4.91178145798289e-08, | |
| "logps/chosen": -2.8190388679504395, | |
| "logps/rejected": -0.7673857808113098, | |
| "loss": 24.4349, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.19038963317871, | |
| "rewards/margins": -20.516530990600586, | |
| "rewards/rejected": -7.673857688903809, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18617021276595744, | |
| "grad_norm": 142.14896402628358, | |
| "learning_rate": 4.890232863724074e-08, | |
| "logps/chosen": -2.5820021629333496, | |
| "logps/rejected": -0.7970396876335144, | |
| "loss": 24.1908, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.820018768310547, | |
| "rewards/margins": -17.849624633789062, | |
| "rewards/rejected": -7.970396518707275, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.1950354609929078, | |
| "grad_norm": 146.1009589129858, | |
| "learning_rate": 4.8663900798549555e-08, | |
| "logps/chosen": -2.6352057456970215, | |
| "logps/rejected": -0.7787631154060364, | |
| "loss": 23.47, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.3520565032959, | |
| "rewards/margins": -18.564428329467773, | |
| "rewards/rejected": -7.787631034851074, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.20390070921985815, | |
| "grad_norm": 140.00337741613868, | |
| "learning_rate": 4.84027599111947e-08, | |
| "logps/chosen": -2.8055331707000732, | |
| "logps/rejected": -0.7519856095314026, | |
| "loss": 24.6991, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.055328369140625, | |
| "rewards/margins": -20.535472869873047, | |
| "rewards/rejected": -7.5198564529418945, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2127659574468085, | |
| "grad_norm": 141.35660015466556, | |
| "learning_rate": 4.8119156623018765e-08, | |
| "logps/chosen": -2.79301118850708, | |
| "logps/rejected": -0.7545806765556335, | |
| "loss": 24.7028, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.93011474609375, | |
| "rewards/margins": -20.384307861328125, | |
| "rewards/rejected": -7.545806884765625, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.22163120567375885, | |
| "grad_norm": 142.6726786477248, | |
| "learning_rate": 4.781336314169116e-08, | |
| "logps/chosen": -2.820788860321045, | |
| "logps/rejected": -0.8155437707901001, | |
| "loss": 24.0466, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.207889556884766, | |
| "rewards/margins": -20.05245018005371, | |
| "rewards/rejected": -8.155437469482422, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.23049645390070922, | |
| "grad_norm": 132.8640819326495, | |
| "learning_rate": 4.748567297343817e-08, | |
| "logps/chosen": -2.764223575592041, | |
| "logps/rejected": -0.7645096182823181, | |
| "loss": 24.7092, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.642236709594727, | |
| "rewards/margins": -19.997140884399414, | |
| "rewards/rejected": -7.6450958251953125, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2393617021276596, | |
| "grad_norm": 139.96100373862689, | |
| "learning_rate": 4.713640064133024e-08, | |
| "logps/chosen": -2.7403454780578613, | |
| "logps/rejected": -0.745998740196228, | |
| "loss": 23.5556, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.403453826904297, | |
| "rewards/margins": -19.943464279174805, | |
| "rewards/rejected": -7.459986686706543, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.24822695035460993, | |
| "grad_norm": 148.02227270943942, | |
| "learning_rate": 4.676588138339698e-08, | |
| "logps/chosen": -2.7284622192382812, | |
| "logps/rejected": -0.7950869798660278, | |
| "loss": 23.8238, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.284622192382812, | |
| "rewards/margins": -19.333749771118164, | |
| "rewards/rejected": -7.950869560241699, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.2570921985815603, | |
| "grad_norm": 176.18112226614372, | |
| "learning_rate": 4.6374470830859435e-08, | |
| "logps/chosen": -2.7229888439178467, | |
| "logps/rejected": -0.8719980120658875, | |
| "loss": 24.1807, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.229888916015625, | |
| "rewards/margins": -18.50990867614746, | |
| "rewards/rejected": -8.719980239868164, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.26595744680851063, | |
| "grad_norm": 135.96118644362534, | |
| "learning_rate": 4.596254466678876e-08, | |
| "logps/chosen": -2.56044340133667, | |
| "logps/rejected": -0.8695181012153625, | |
| "loss": 22.9327, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.60443687438965, | |
| "rewards/margins": -16.90925407409668, | |
| "rewards/rejected": -8.695180892944336, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.274822695035461, | |
| "grad_norm": 132.24528952538998, | |
| "learning_rate": 4.5530498265518635e-08, | |
| "logps/chosen": -2.813053607940674, | |
| "logps/rejected": -0.808445155620575, | |
| "loss": 23.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.130535125732422, | |
| "rewards/margins": -20.04608726501465, | |
| "rewards/rejected": -8.084451675415039, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.28368794326241137, | |
| "grad_norm": 127.79059567307556, | |
| "learning_rate": 4.507874631315768e-08, | |
| "logps/chosen": -2.66123628616333, | |
| "logps/rejected": -0.8127639889717102, | |
| "loss": 23.47, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.61236000061035, | |
| "rewards/margins": -18.48472023010254, | |
| "rewards/rejected": -8.127640724182129, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.2925531914893617, | |
| "grad_norm": 129.2315474298329, | |
| "learning_rate": 4.460772240956608e-08, | |
| "logps/chosen": -2.7900562286376953, | |
| "logps/rejected": -0.849897027015686, | |
| "loss": 23.8895, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.900564193725586, | |
| "rewards/margins": -19.401594161987305, | |
| "rewards/rejected": -8.498970985412598, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.30141843971631205, | |
| "grad_norm": 130.10241912996838, | |
| "learning_rate": 4.411787865217846e-08, | |
| "logps/chosen": -2.605811595916748, | |
| "logps/rejected": -0.7789794206619263, | |
| "loss": 24.0528, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.0581111907959, | |
| "rewards/margins": -18.268321990966797, | |
| "rewards/rejected": -7.789793491363525, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3102836879432624, | |
| "grad_norm": 125.41956154105759, | |
| "learning_rate": 4.36096852020724e-08, | |
| "logps/chosen": -2.852963924407959, | |
| "logps/rejected": -0.8932555913925171, | |
| "loss": 24.7971, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.529638290405273, | |
| "rewards/margins": -19.59708023071289, | |
| "rewards/rejected": -8.932558059692383, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3191489361702128, | |
| "grad_norm": 129.52798142464744, | |
| "learning_rate": 4.308362983269915e-08, | |
| "logps/chosen": -2.6496150493621826, | |
| "logps/rejected": -0.8426092267036438, | |
| "loss": 24.0708, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.496150970458984, | |
| "rewards/margins": -18.070056915283203, | |
| "rewards/rejected": -8.426092147827148, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3280141843971631, | |
| "grad_norm": 124.39272381055714, | |
| "learning_rate": 4.2540217461709714e-08, | |
| "logps/chosen": -2.640799045562744, | |
| "logps/rejected": -0.7842020988464355, | |
| "loss": 23.3526, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.407989501953125, | |
| "rewards/margins": -18.565967559814453, | |
| "rewards/rejected": -7.8420209884643555, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.33687943262411346, | |
| "grad_norm": 135.5931742125443, | |
| "learning_rate": 4.1979969666325505e-08, | |
| "logps/chosen": -2.4907784461975098, | |
| "logps/rejected": -0.8824084997177124, | |
| "loss": 23.1102, | |
| "rewards/accuracies": 0.10000000149011612, | |
| "rewards/chosen": -24.907785415649414, | |
| "rewards/margins": -16.083698272705078, | |
| "rewards/rejected": -8.824087142944336, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.34574468085106386, | |
| "grad_norm": 124.78318512461612, | |
| "learning_rate": 4.140342418271896e-08, | |
| "logps/chosen": -2.566643476486206, | |
| "logps/rejected": -0.8403929471969604, | |
| "loss": 23.9738, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.666433334350586, | |
| "rewards/margins": -17.262508392333984, | |
| "rewards/rejected": -8.403928756713867, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.3546099290780142, | |
| "grad_norm": 124.54166567693945, | |
| "learning_rate": 4.0811134389884425e-08, | |
| "logps/chosen": -2.662815570831299, | |
| "logps/rejected": -0.8375980257987976, | |
| "loss": 23.2471, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.628154754638672, | |
| "rewards/margins": -18.25217628479004, | |
| "rewards/rejected": -8.37597942352295, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.36347517730496454, | |
| "grad_norm": 129.10521131399426, | |
| "learning_rate": 4.020366877849477e-08, | |
| "logps/chosen": -2.899678945541382, | |
| "logps/rejected": -0.9157842397689819, | |
| "loss": 22.6395, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.99679183959961, | |
| "rewards/margins": -19.838947296142578, | |
| "rewards/rejected": -9.157841682434082, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.3723404255319149, | |
| "grad_norm": 118.9218343714434, | |
| "learning_rate": 3.958161040525354e-08, | |
| "logps/chosen": -2.7788329124450684, | |
| "logps/rejected": -0.7856232523918152, | |
| "loss": 24.5067, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.788330078125, | |
| "rewards/margins": -19.93209457397461, | |
| "rewards/rejected": -7.856232643127441, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.38120567375886527, | |
| "grad_norm": 124.20603604955339, | |
| "learning_rate": 3.894555633326642e-08, | |
| "logps/chosen": -2.766160011291504, | |
| "logps/rejected": -0.8535798192024231, | |
| "loss": 23.9262, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.66160011291504, | |
| "rewards/margins": -19.125802993774414, | |
| "rewards/rejected": -8.535799026489258, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.3900709219858156, | |
| "grad_norm": 122.4047572823285, | |
| "learning_rate": 3.829611705896899e-08, | |
| "logps/chosen": -2.758810520172119, | |
| "logps/rejected": -0.8758818507194519, | |
| "loss": 23.0933, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.58810806274414, | |
| "rewards/margins": -18.829286575317383, | |
| "rewards/rejected": -8.758818626403809, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.39893617021276595, | |
| "grad_norm": 136.57138875427492, | |
| "learning_rate": 3.763391592616104e-08, | |
| "logps/chosen": -2.629612445831299, | |
| "logps/rejected": -0.8354926109313965, | |
| "loss": 23.2906, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.296123504638672, | |
| "rewards/margins": -17.941198348999023, | |
| "rewards/rejected": -8.354926109313965, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.4078014184397163, | |
| "grad_norm": 125.42475135454727, | |
| "learning_rate": 3.695958852770963e-08, | |
| "logps/chosen": -2.629051685333252, | |
| "logps/rejected": -0.8992561101913452, | |
| "loss": 23.2326, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.290517807006836, | |
| "rewards/margins": -17.297956466674805, | |
| "rewards/rejected": -8.992559432983398, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.4166666666666667, | |
| "grad_norm": 130.81865526034508, | |
| "learning_rate": 3.627378209549536e-08, | |
| "logps/chosen": -2.713019847869873, | |
| "logps/rejected": -0.9180544018745422, | |
| "loss": 23.2666, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.130197525024414, | |
| "rewards/margins": -17.94965171813965, | |
| "rewards/rejected": -9.180543899536133, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.425531914893617, | |
| "grad_norm": 154.01811109390695, | |
| "learning_rate": 3.557715487918728e-08, | |
| "logps/chosen": -2.845736265182495, | |
| "logps/rejected": -0.8981745839118958, | |
| "loss": 22.8914, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.45736312866211, | |
| "rewards/margins": -19.475618362426758, | |
| "rewards/rejected": -8.981744766235352, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.43439716312056736, | |
| "grad_norm": 124.72008912272919, | |
| "learning_rate": 3.487037551444267e-08, | |
| "logps/chosen": -2.6402182579040527, | |
| "logps/rejected": -0.9148176908493042, | |
| "loss": 24.0945, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.40218162536621, | |
| "rewards/margins": -17.254005432128906, | |
| "rewards/rejected": -9.148177146911621, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.4432624113475177, | |
| "grad_norm": 126.67782694015744, | |
| "learning_rate": 3.4154122381138226e-08, | |
| "logps/chosen": -2.8172085285186768, | |
| "logps/rejected": -0.9476315379142761, | |
| "loss": 22.8683, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.17208480834961, | |
| "rewards/margins": -18.695770263671875, | |
| "rewards/rejected": -9.476313591003418, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.4521276595744681, | |
| "grad_norm": 128.76198121819777, | |
| "learning_rate": 3.3429082952248535e-08, | |
| "logps/chosen": -2.6602797508239746, | |
| "logps/rejected": -0.9177526235580444, | |
| "loss": 22.7192, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.602794647216797, | |
| "rewards/margins": -17.42526626586914, | |
| "rewards/rejected": -9.177526473999023, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.46099290780141844, | |
| "grad_norm": 120.09861391040339, | |
| "learning_rate": 3.2695953133996826e-08, | |
| "logps/chosen": -2.6814820766448975, | |
| "logps/rejected": -0.8409668803215027, | |
| "loss": 22.8322, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.8148193359375, | |
| "rewards/margins": -18.405153274536133, | |
| "rewards/rejected": -8.409668922424316, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.4698581560283688, | |
| "grad_norm": 149.0293076595361, | |
| "learning_rate": 3.195543659791132e-08, | |
| "logps/chosen": -2.6782455444335938, | |
| "logps/rejected": -0.988556981086731, | |
| "loss": 22.749, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.782455444335938, | |
| "rewards/margins": -16.896886825561523, | |
| "rewards/rejected": -9.88556957244873, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.4787234042553192, | |
| "grad_norm": 120.49824463691422, | |
| "learning_rate": 3.120824410542833e-08, | |
| "logps/chosen": -2.619314193725586, | |
| "logps/rejected": -0.9258754849433899, | |
| "loss": 22.1466, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.19314193725586, | |
| "rewards/margins": -16.93438720703125, | |
| "rewards/rejected": -9.25875473022461, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.4875886524822695, | |
| "grad_norm": 126.28759504359856, | |
| "learning_rate": 3.045509282569031e-08, | |
| "logps/chosen": -2.569866895675659, | |
| "logps/rejected": -0.843720555305481, | |
| "loss": 22.4486, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.69866943359375, | |
| "rewards/margins": -17.261463165283203, | |
| "rewards/rejected": -8.43720531463623, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.49645390070921985, | |
| "grad_norm": 133.61205279748194, | |
| "learning_rate": 2.969670564719369e-08, | |
| "logps/chosen": -2.705709457397461, | |
| "logps/rejected": -0.8904563784599304, | |
| "loss": 22.7856, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.057092666625977, | |
| "rewards/margins": -18.152530670166016, | |
| "rewards/rejected": -8.904562950134277, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5053191489361702, | |
| "grad_norm": 125.31007604414539, | |
| "learning_rate": 2.893381048394715e-08, | |
| "logps/chosen": -2.5439467430114746, | |
| "logps/rejected": -0.9845563173294067, | |
| "loss": 22.3419, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.43946647644043, | |
| "rewards/margins": -15.593902587890625, | |
| "rewards/rejected": -9.845562934875488, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.5141843971631206, | |
| "grad_norm": 133.3039752253482, | |
| "learning_rate": 2.8167139576806304e-08, | |
| "logps/chosen": -2.7466816902160645, | |
| "logps/rejected": -0.9437241554260254, | |
| "loss": 22.5713, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.466812133789062, | |
| "rewards/margins": -18.029569625854492, | |
| "rewards/rejected": -9.43724250793457, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.5230496453900709, | |
| "grad_norm": 125.8510714937569, | |
| "learning_rate": 2.739742879065544e-08, | |
| "logps/chosen": -2.6576900482177734, | |
| "logps/rejected": -0.9526101350784302, | |
| "loss": 22.5247, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.576900482177734, | |
| "rewards/margins": -17.050800323486328, | |
| "rewards/rejected": -9.526101112365723, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.5319148936170213, | |
| "grad_norm": 132.63761511287686, | |
| "learning_rate": 2.662541690811082e-08, | |
| "logps/chosen": -2.3779425621032715, | |
| "logps/rejected": -0.8925439715385437, | |
| "loss": 21.2831, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -23.779422760009766, | |
| "rewards/margins": -14.853982925415039, | |
| "rewards/rejected": -8.925439834594727, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5407801418439716, | |
| "grad_norm": 136.13215779336474, | |
| "learning_rate": 2.585184492042347e-08, | |
| "logps/chosen": -2.6994504928588867, | |
| "logps/rejected": -0.9614452123641968, | |
| "loss": 21.5851, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.9945068359375, | |
| "rewards/margins": -17.38005256652832, | |
| "rewards/rejected": -9.614453315734863, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.549645390070922, | |
| "grad_norm": 132.02028150495246, | |
| "learning_rate": 2.5077455316262147e-08, | |
| "logps/chosen": -2.6731841564178467, | |
| "logps/rejected": -0.9815571904182434, | |
| "loss": 21.7416, | |
| "rewards/accuracies": 0.10000000149011612, | |
| "rewards/chosen": -26.73184585571289, | |
| "rewards/margins": -16.916271209716797, | |
| "rewards/rejected": -9.815571784973145, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.5585106382978723, | |
| "grad_norm": 130.67893366658126, | |
| "learning_rate": 2.4302991369058958e-08, | |
| "logps/chosen": -2.51406192779541, | |
| "logps/rejected": -1.0564053058624268, | |
| "loss": 20.9766, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.1406192779541, | |
| "rewards/margins": -14.576566696166992, | |
| "rewards/rejected": -10.564051628112793, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.5673758865248227, | |
| "grad_norm": 131.46127648897104, | |
| "learning_rate": 2.3529196423601876e-08, | |
| "logps/chosen": -2.443092107772827, | |
| "logps/rejected": -0.9888316988945007, | |
| "loss": 22.5219, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -24.43092155456543, | |
| "rewards/margins": -14.54260540008545, | |
| "rewards/rejected": -9.888317108154297, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.5762411347517731, | |
| "grad_norm": 134.82856786462523, | |
| "learning_rate": 2.27568131825586e-08, | |
| "logps/chosen": -2.7351233959198, | |
| "logps/rejected": -1.0152182579040527, | |
| "loss": 21.4016, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.35123634338379, | |
| "rewards/margins": -17.199050903320312, | |
| "rewards/rejected": -10.152182579040527, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.5851063829787234, | |
| "grad_norm": 138.32340221747404, | |
| "learning_rate": 2.1986582993616925e-08, | |
| "logps/chosen": -2.5989935398101807, | |
| "logps/rejected": -0.9776498079299927, | |
| "loss": 21.6639, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.989938735961914, | |
| "rewards/margins": -16.213438034057617, | |
| "rewards/rejected": -9.776496887207031, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.5939716312056738, | |
| "grad_norm": 144.27555433213394, | |
| "learning_rate": 2.121924513792548e-08, | |
| "logps/chosen": -2.7326598167419434, | |
| "logps/rejected": -0.993288516998291, | |
| "loss": 22.0246, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.326595306396484, | |
| "rewards/margins": -17.39371109008789, | |
| "rewards/rejected": -9.93288516998291, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.6028368794326241, | |
| "grad_norm": 131.66189209302496, | |
| "learning_rate": 2.0455536120518093e-08, | |
| "logps/chosen": -2.7750537395477295, | |
| "logps/rejected": -1.0254470109939575, | |
| "loss": 21.3832, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.750534057617188, | |
| "rewards/margins": -17.496063232421875, | |
| "rewards/rejected": -10.254469871520996, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.6117021276595744, | |
| "grad_norm": 133.29472374802276, | |
| "learning_rate": 1.969618896340261e-08, | |
| "logps/chosen": -2.791393756866455, | |
| "logps/rejected": -0.9777949452400208, | |
| "loss": 21.8087, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.9139404296875, | |
| "rewards/margins": -18.1359920501709, | |
| "rewards/rejected": -9.777949333190918, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.6205673758865248, | |
| "grad_norm": 132.7071056456499, | |
| "learning_rate": 1.8941932501992915e-08, | |
| "logps/chosen": -2.528127670288086, | |
| "logps/rejected": -0.9332917928695679, | |
| "loss": 22.4318, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.281272888183594, | |
| "rewards/margins": -15.948354721069336, | |
| "rewards/rejected": -9.332918167114258, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.6294326241134752, | |
| "grad_norm": 133.23550694363342, | |
| "learning_rate": 1.8193490685559176e-08, | |
| "logps/chosen": -2.6291937828063965, | |
| "logps/rejected": -1.0154783725738525, | |
| "loss": 21.633, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.291940689086914, | |
| "rewards/margins": -16.137157440185547, | |
| "rewards/rejected": -10.154783248901367, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.6382978723404256, | |
| "grad_norm": 131.9862073113527, | |
| "learning_rate": 1.745158188236805e-08, | |
| "logps/chosen": -2.7716784477233887, | |
| "logps/rejected": -0.9892207980155945, | |
| "loss": 21.8071, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.716785430908203, | |
| "rewards/margins": -17.824573516845703, | |
| "rewards/rejected": -9.89220905303955, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6471631205673759, | |
| "grad_norm": 133.59469928327707, | |
| "learning_rate": 1.6716918190179507e-08, | |
| "logps/chosen": -2.824233055114746, | |
| "logps/rejected": -1.0325210094451904, | |
| "loss": 21.8658, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.24233055114746, | |
| "rewards/margins": -17.91712188720703, | |
| "rewards/rejected": -10.325210571289062, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.6560283687943262, | |
| "grad_norm": 137.91947812898098, | |
| "learning_rate": 1.599020475276227e-08, | |
| "logps/chosen": -2.6540284156799316, | |
| "logps/rejected": -1.0216987133026123, | |
| "loss": 21.8092, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.540287017822266, | |
| "rewards/margins": -16.323299407958984, | |
| "rewards/rejected": -10.216985702514648, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.6648936170212766, | |
| "grad_norm": 139.4931571924634, | |
| "learning_rate": 1.527213908308386e-08, | |
| "logps/chosen": -2.6086392402648926, | |
| "logps/rejected": -1.0575668811798096, | |
| "loss": 21.1347, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.086395263671875, | |
| "rewards/margins": -15.510726928710938, | |
| "rewards/rejected": -10.575668334960938, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.6737588652482269, | |
| "grad_norm": 140.7825368139536, | |
| "learning_rate": 1.45634103938247e-08, | |
| "logps/chosen": -2.7265937328338623, | |
| "logps/rejected": -1.1351561546325684, | |
| "loss": 21.6724, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.265939712524414, | |
| "rewards/margins": -15.914377212524414, | |
| "rewards/rejected": -11.3515625, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.6826241134751773, | |
| "grad_norm": 155.36198841615501, | |
| "learning_rate": 1.3864698935859152e-08, | |
| "logps/chosen": -2.785165309906006, | |
| "logps/rejected": -1.016316533088684, | |
| "loss": 21.7173, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.851654052734375, | |
| "rewards/margins": -17.688488006591797, | |
| "rewards/rejected": -10.163165092468262, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.6914893617021277, | |
| "grad_norm": 142.46678106209737, | |
| "learning_rate": 1.3176675345338084e-08, | |
| "logps/chosen": -2.720693349838257, | |
| "logps/rejected": -1.0303252935409546, | |
| "loss": 21.5356, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.206933975219727, | |
| "rewards/margins": -16.903682708740234, | |
| "rewards/rejected": -10.303252220153809, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.700354609929078, | |
| "grad_norm": 145.09087804013313, | |
| "learning_rate": 1.2500000000000004e-08, | |
| "logps/chosen": -2.8458621501922607, | |
| "logps/rejected": -1.0235410928726196, | |
| "loss": 21.0196, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.4586181640625, | |
| "rewards/margins": -18.22321128845215, | |
| "rewards/rejected": -10.2354097366333, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.7092198581560284, | |
| "grad_norm": 141.31236991069352, | |
| "learning_rate": 1.1835322385328258e-08, | |
| "logps/chosen": -2.525390625, | |
| "logps/rejected": -1.0862586498260498, | |
| "loss": 21.4429, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.25390625, | |
| "rewards/margins": -14.391319274902344, | |
| "rewards/rejected": -10.862588882446289, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.7092198581560284, | |
| "eval_logps/chosen": -2.753817319869995, | |
| "eval_logps/rejected": -1.0216267108917236, | |
| "eval_loss": 21.640687942504883, | |
| "eval_rewards/accuracies": 0.0, | |
| "eval_rewards/chosen": -27.538171768188477, | |
| "eval_rewards/margins": -17.3219051361084, | |
| "eval_rewards/rejected": -10.216267585754395, | |
| "eval_runtime": 4.3362, | |
| "eval_samples_per_second": 2.767, | |
| "eval_steps_per_second": 0.692, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.7180851063829787, | |
| "grad_norm": 138.30907528011366, | |
| "learning_rate": 1.1183280471162916e-08, | |
| "logps/chosen": -2.6900100708007812, | |
| "logps/rejected": -1.054898977279663, | |
| "loss": 21.6015, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.900100708007812, | |
| "rewards/margins": -16.351110458374023, | |
| "rewards/rejected": -10.548989295959473, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.7269503546099291, | |
| "grad_norm": 140.60691958793103, | |
| "learning_rate": 1.0544500099365513e-08, | |
| "logps/chosen": -2.5441603660583496, | |
| "logps/rejected": -1.0485824346542358, | |
| "loss": 20.2114, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.441608428955078, | |
| "rewards/margins": -14.955782890319824, | |
| "rewards/rejected": -10.485824584960938, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.7358156028368794, | |
| "grad_norm": 158.54062894661507, | |
| "learning_rate": 9.91959438312451e-09, | |
| "logps/chosen": -2.586798667907715, | |
| "logps/rejected": -1.0556939840316772, | |
| "loss": 20.7519, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.867984771728516, | |
| "rewards/margins": -15.31104564666748, | |
| "rewards/rejected": -10.556941032409668, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.7446808510638298, | |
| "grad_norm": 145.7088174939863, | |
| "learning_rate": 9.309163118477952e-09, | |
| "logps/chosen": -2.7213327884674072, | |
| "logps/rejected": -1.10366690158844, | |
| "loss": 20.9045, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.213327407836914, | |
| "rewards/margins": -16.176658630371094, | |
| "rewards/rejected": -11.036666870117188, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.7535460992907801, | |
| "grad_norm": 140.87803118012943, | |
| "learning_rate": 8.713792208618095e-09, | |
| "logps/chosen": -2.7333548069000244, | |
| "logps/rejected": -1.0871623754501343, | |
| "loss": 21.3148, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.333545684814453, | |
| "rewards/margins": -16.461925506591797, | |
| "rewards/rejected": -10.871623039245605, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.7624113475177305, | |
| "grad_norm": 148.2931552966102, | |
| "learning_rate": 8.134053101530814e-09, | |
| "logps/chosen": -2.6974215507507324, | |
| "logps/rejected": -1.0488805770874023, | |
| "loss": 22.011, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.974212646484375, | |
| "rewards/margins": -16.485408782958984, | |
| "rewards/rejected": -10.488804817199707, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.7712765957446809, | |
| "grad_norm": 156.37941932984688, | |
| "learning_rate": 7.570502241509161e-09, | |
| "logps/chosen": -2.6490612030029297, | |
| "logps/rejected": -1.0499160289764404, | |
| "loss": 20.7465, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.490612030029297, | |
| "rewards/margins": -15.991450309753418, | |
| "rewards/rejected": -10.499159812927246, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.7801418439716312, | |
| "grad_norm": 142.1179108847554, | |
| "learning_rate": 7.023680535067997e-09, | |
| "logps/chosen": -2.540945053100586, | |
| "logps/rejected": -1.1362855434417725, | |
| "loss": 20.9313, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.40945053100586, | |
| "rewards/margins": -14.046595573425293, | |
| "rewards/rejected": -11.362855911254883, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.7890070921985816, | |
| "grad_norm": 142.7591601522917, | |
| "learning_rate": 6.4941128317718e-09, | |
| "logps/chosen": -2.71937894821167, | |
| "logps/rejected": -1.0216131210327148, | |
| "loss": 21.3747, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.19378662109375, | |
| "rewards/margins": -16.977657318115234, | |
| "rewards/rejected": -10.216130256652832, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.7978723404255319, | |
| "grad_norm": 146.19832437425603, | |
| "learning_rate": 5.9823074204745e-09, | |
| "logps/chosen": -2.6064822673797607, | |
| "logps/rejected": -1.1812522411346436, | |
| "loss": 20.7936, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.0648250579834, | |
| "rewards/margins": -14.252301216125488, | |
| "rewards/rejected": -11.812520980834961, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.8067375886524822, | |
| "grad_norm": 138.88047068502686, | |
| "learning_rate": 5.4887555414543344e-09, | |
| "logps/chosen": -2.6501286029815674, | |
| "logps/rejected": -1.0125081539154053, | |
| "loss": 20.8237, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.50128746032715, | |
| "rewards/margins": -16.376205444335938, | |
| "rewards/rejected": -10.125081062316895, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.8156028368794326, | |
| "grad_norm": 136.88698229561803, | |
| "learning_rate": 5.0139309149124755e-09, | |
| "logps/chosen": -2.5912132263183594, | |
| "logps/rejected": -1.1049864292144775, | |
| "loss": 20.3607, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.91213035583496, | |
| "rewards/margins": -14.862268447875977, | |
| "rewards/rejected": -11.049863815307617, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.824468085106383, | |
| "grad_norm": 150.56544212976812, | |
| "learning_rate": 4.558289286287545e-09, | |
| "logps/chosen": -2.563302516937256, | |
| "logps/rejected": -1.1222257614135742, | |
| "loss": 20.5554, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.63302993774414, | |
| "rewards/margins": -14.41076946258545, | |
| "rewards/rejected": -11.222257614135742, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 151.83025187850424, | |
| "learning_rate": 4.122267988822792e-09, | |
| "logps/chosen": -2.8155665397644043, | |
| "logps/rejected": -1.1182525157928467, | |
| "loss": 20.6342, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.15566635131836, | |
| "rewards/margins": -16.973140716552734, | |
| "rewards/rejected": -11.182525634765625, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.8421985815602837, | |
| "grad_norm": 143.90652908649776, | |
| "learning_rate": 3.7062855238055775e-09, | |
| "logps/chosen": -2.8534111976623535, | |
| "logps/rejected": -1.079416275024414, | |
| "loss": 21.6896, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.534109115600586, | |
| "rewards/margins": -17.739948272705078, | |
| "rewards/rejected": -10.794163703918457, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.851063829787234, | |
| "grad_norm": 146.70532693647684, | |
| "learning_rate": 3.310741158882052e-09, | |
| "logps/chosen": -2.5945401191711426, | |
| "logps/rejected": -1.1171306371688843, | |
| "loss": 20.8278, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.945398330688477, | |
| "rewards/margins": -14.774093627929688, | |
| "rewards/rejected": -11.171305656433105, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.8599290780141844, | |
| "grad_norm": 152.1225967764815, | |
| "learning_rate": 2.9360145448327933e-09, | |
| "logps/chosen": -2.7132070064544678, | |
| "logps/rejected": -1.10798978805542, | |
| "loss": 20.2016, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.132068634033203, | |
| "rewards/margins": -16.052169799804688, | |
| "rewards/rejected": -11.079896926879883, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.8687943262411347, | |
| "grad_norm": 148.4298648799797, | |
| "learning_rate": 2.5824653511768905e-09, | |
| "logps/chosen": -2.7444825172424316, | |
| "logps/rejected": -1.0877773761749268, | |
| "loss": 20.6239, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.44482421875, | |
| "rewards/margins": -16.56705093383789, | |
| "rewards/rejected": -10.877774238586426, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.8776595744680851, | |
| "grad_norm": 149.3789958144051, | |
| "learning_rate": 2.250432920954584e-09, | |
| "logps/chosen": -2.625457286834717, | |
| "logps/rejected": -1.1413992643356323, | |
| "loss": 20.8323, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.25457191467285, | |
| "rewards/margins": -14.840580940246582, | |
| "rewards/rejected": -11.413991928100586, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.8865248226950354, | |
| "grad_norm": 149.24437602693683, | |
| "learning_rate": 1.9402359450194834e-09, | |
| "logps/chosen": -2.633744716644287, | |
| "logps/rejected": -1.0395913124084473, | |
| "loss": 21.1488, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.337448120117188, | |
| "rewards/margins": -15.941534042358398, | |
| "rewards/rejected": -10.395914077758789, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8953900709219859, | |
| "grad_norm": 148.69800996723336, | |
| "learning_rate": 1.6521721561532642e-09, | |
| "logps/chosen": -2.753237724304199, | |
| "logps/rejected": -1.091736078262329, | |
| "loss": 21.1955, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.53237533569336, | |
| "rewards/margins": -16.61501693725586, | |
| "rewards/rejected": -10.91736125946045, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.9042553191489362, | |
| "grad_norm": 140.23345832085906, | |
| "learning_rate": 1.3865180432961975e-09, | |
| "logps/chosen": -2.7156546115875244, | |
| "logps/rejected": -1.033905029296875, | |
| "loss": 20.9517, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.156545639038086, | |
| "rewards/margins": -16.8174991607666, | |
| "rewards/rejected": -10.33905029296875, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.9131205673758865, | |
| "grad_norm": 135.7495898891431, | |
| "learning_rate": 1.1435285861680105e-09, | |
| "logps/chosen": -2.6717796325683594, | |
| "logps/rejected": -1.1057870388031006, | |
| "loss": 20.2493, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.71779441833496, | |
| "rewards/margins": -15.65992546081543, | |
| "rewards/rejected": -11.057870864868164, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.9219858156028369, | |
| "grad_norm": 146.54140042392083, | |
| "learning_rate": 9.234370105336037e-10, | |
| "logps/chosen": -2.7197585105895996, | |
| "logps/rejected": -1.0512771606445312, | |
| "loss": 20.9221, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.197586059570312, | |
| "rewards/margins": -16.684812545776367, | |
| "rewards/rejected": -10.512772560119629, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9308510638297872, | |
| "grad_norm": 154.92783543731412, | |
| "learning_rate": 7.264545643486997e-10, | |
| "logps/chosen": -2.7849745750427246, | |
| "logps/rejected": -1.0905792713165283, | |
| "loss": 21.5713, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.849742889404297, | |
| "rewards/margins": -16.943950653076172, | |
| "rewards/rejected": -10.905792236328125, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.9397163120567376, | |
| "grad_norm": 151.56459387858007, | |
| "learning_rate": 5.527703150001173e-10, | |
| "logps/chosen": -2.8318445682525635, | |
| "logps/rejected": -1.042283535003662, | |
| "loss": 21.4995, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -28.31844711303711, | |
| "rewards/margins": -17.895610809326172, | |
| "rewards/rejected": -10.422834396362305, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.9485815602836879, | |
| "grad_norm": 163.48745110663447, | |
| "learning_rate": 4.0255096783542995e-10, | |
| "logps/chosen": -2.5114052295684814, | |
| "logps/rejected": -1.1133381128311157, | |
| "loss": 20.1608, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.114055633544922, | |
| "rewards/margins": -13.980671882629395, | |
| "rewards/rejected": -11.133380889892578, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.9574468085106383, | |
| "grad_norm": 149.97953788101626, | |
| "learning_rate": 2.759407061560942e-10, | |
| "logps/chosen": -2.641373872756958, | |
| "logps/rejected": -1.0878345966339111, | |
| "loss": 20.6395, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -26.41373634338379, | |
| "rewards/margins": -15.535390853881836, | |
| "rewards/rejected": -10.878347396850586, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.9663120567375887, | |
| "grad_norm": 146.82377405274374, | |
| "learning_rate": 1.7306105282764162e-10, | |
| "logps/chosen": -2.7007031440734863, | |
| "logps/rejected": -1.1214299201965332, | |
| "loss": 21.5189, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -27.007030487060547, | |
| "rewards/margins": -15.792730331420898, | |
| "rewards/rejected": -11.214300155639648, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.975177304964539, | |
| "grad_norm": 146.05557779309504, | |
| "learning_rate": 9.401075363981436e-11, | |
| "logps/chosen": -2.5641696453094482, | |
| "logps/rejected": -1.0827744007110596, | |
| "loss": 20.7057, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.641698837280273, | |
| "rewards/margins": -14.813952445983887, | |
| "rewards/rejected": -10.827742576599121, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.9840425531914894, | |
| "grad_norm": 141.1483903438878, | |
| "learning_rate": 3.886568252850497e-11, | |
| "logps/chosen": -2.5521302223205566, | |
| "logps/rejected": -1.1713207960128784, | |
| "loss": 20.7053, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": -25.521303176879883, | |
| "rewards/margins": -13.808095932006836, | |
| "rewards/rejected": -11.713208198547363, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.9929078014184397, | |
| "grad_norm": 150.04371038548183, | |
| "learning_rate": 7.678768750579711e-12, | |
| "logps/chosen": -2.5215792655944824, | |
| "logps/rejected": -1.0503337383270264, | |
| "loss": 20.6892, | |
| "rewards/accuracies": 0.10000000149011612, | |
| "rewards/chosen": -25.215795516967773, | |
| "rewards/margins": -14.712457656860352, | |
| "rewards/rejected": -10.503335952758789, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 564, | |
| "total_flos": 0.0, | |
| "train_loss": 22.58866818745931, | |
| "train_runtime": 4347.6828, | |
| "train_samples_per_second": 1.297, | |
| "train_steps_per_second": 0.13 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 564, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |