| { | |
| "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": 7.607854381527219, | |
| "learning_rate": 3.90625e-09, | |
| "logits/chosen": 5881.4375, | |
| "logits/rejected": 2834.66162109375, | |
| "logps/chosen": -257.5969543457031, | |
| "logps/rejected": -120.09489440917969, | |
| "loss": 1.3863, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.007849293563579277, | |
| "grad_norm": 8.72860536992559, | |
| "learning_rate": 3.9062499999999997e-08, | |
| "logits/chosen": 4945.994140625, | |
| "logits/rejected": 4332.44775390625, | |
| "logps/chosen": -248.813232421875, | |
| "logps/rejected": -214.26055908203125, | |
| "loss": 1.3863, | |
| "rewards/accuracies": 0.4166666567325592, | |
| "rewards/chosen": 0.0029773255810141563, | |
| "rewards/margins": -0.0285730492323637, | |
| "rewards/rejected": 0.03155038133263588, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.015698587127158554, | |
| "grad_norm": 6.983503663611063, | |
| "learning_rate": 7.812499999999999e-08, | |
| "logits/chosen": 6023.13232421875, | |
| "logits/rejected": 4842.7783203125, | |
| "logps/chosen": -266.24188232421875, | |
| "logps/rejected": -232.4458465576172, | |
| "loss": 1.3863, | |
| "rewards/accuracies": 0.5416666865348816, | |
| "rewards/chosen": 0.07147032022476196, | |
| "rewards/margins": 0.06520499289035797, | |
| "rewards/rejected": 0.0062653301283717155, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.023547880690737835, | |
| "grad_norm": 6.8914768137357285, | |
| "learning_rate": 1.1718749999999999e-07, | |
| "logits/chosen": 6032.46484375, | |
| "logits/rejected": 5038.45751953125, | |
| "logps/chosen": -310.44793701171875, | |
| "logps/rejected": -262.9197998046875, | |
| "loss": 1.3861, | |
| "rewards/accuracies": 0.5333333015441895, | |
| "rewards/chosen": 0.16140249371528625, | |
| "rewards/margins": 0.04461865872144699, | |
| "rewards/rejected": 0.11678384244441986, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03139717425431711, | |
| "grad_norm": 7.344666737423356, | |
| "learning_rate": 1.5624999999999999e-07, | |
| "logits/chosen": 5242.39111328125, | |
| "logits/rejected": 4344.1669921875, | |
| "logps/chosen": -263.51727294921875, | |
| "logps/rejected": -224.874267578125, | |
| "loss": 1.3853, | |
| "rewards/accuracies": 0.6666666269302368, | |
| "rewards/chosen": 0.5029774904251099, | |
| "rewards/margins": 0.3132503926753998, | |
| "rewards/rejected": 0.18972712755203247, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03924646781789639, | |
| "grad_norm": 7.642832099780162, | |
| "learning_rate": 1.9531249999999998e-07, | |
| "logits/chosen": 6364.5029296875, | |
| "logits/rejected": 4975.6201171875, | |
| "logps/chosen": -324.6713562011719, | |
| "logps/rejected": -258.1625061035156, | |
| "loss": 1.3845, | |
| "rewards/accuracies": 0.6999999284744263, | |
| "rewards/chosen": 1.4382458925247192, | |
| "rewards/margins": 1.0509032011032104, | |
| "rewards/rejected": 0.3873426020145416, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04709576138147567, | |
| "grad_norm": 7.604175877836808, | |
| "learning_rate": 2.3437499999999998e-07, | |
| "logits/chosen": 5383.66748046875, | |
| "logits/rejected": 4505.18896484375, | |
| "logps/chosen": -269.0326232910156, | |
| "logps/rejected": -262.60595703125, | |
| "loss": 1.3827, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": 0.8944981694221497, | |
| "rewards/margins": 1.3096297979354858, | |
| "rewards/rejected": -0.4151315689086914, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.054945054945054944, | |
| "grad_norm": 6.208056202858818, | |
| "learning_rate": 2.734375e-07, | |
| "logits/chosen": 5107.27587890625, | |
| "logits/rejected": 4841.3271484375, | |
| "logps/chosen": -233.1592254638672, | |
| "logps/rejected": -234.7693328857422, | |
| "loss": 1.3809, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": 0.09875397384166718, | |
| "rewards/margins": 2.2150731086730957, | |
| "rewards/rejected": -2.116318941116333, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06279434850863422, | |
| "grad_norm": 6.428456558078536, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": 5695.2880859375, | |
| "logits/rejected": 5164.94677734375, | |
| "logps/chosen": -256.11041259765625, | |
| "logps/rejected": -246.0963134765625, | |
| "loss": 1.3773, | |
| "rewards/accuracies": 0.6666666269302368, | |
| "rewards/chosen": -0.131959468126297, | |
| "rewards/margins": 3.6519851684570312, | |
| "rewards/rejected": -3.783945083618164, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0706436420722135, | |
| "grad_norm": 7.066386348445373, | |
| "learning_rate": 3.5156249999999997e-07, | |
| "logits/chosen": 5876.4501953125, | |
| "logits/rejected": 5056.3515625, | |
| "logps/chosen": -282.83660888671875, | |
| "logps/rejected": -248.5631561279297, | |
| "loss": 1.3727, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": 2.3544561862945557, | |
| "rewards/margins": 7.451170921325684, | |
| "rewards/rejected": -5.096714973449707, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07849293563579278, | |
| "grad_norm": 7.392556025808417, | |
| "learning_rate": 3.9062499999999997e-07, | |
| "logits/chosen": 5770.21875, | |
| "logits/rejected": 5554.04052734375, | |
| "logps/chosen": -277.7173156738281, | |
| "logps/rejected": -277.0956726074219, | |
| "loss": 1.3715, | |
| "rewards/accuracies": 0.6833333373069763, | |
| "rewards/chosen": 0.5577901601791382, | |
| "rewards/margins": 9.465188026428223, | |
| "rewards/rejected": -8.907397270202637, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08634222919937205, | |
| "grad_norm": 7.851432814181509, | |
| "learning_rate": 4.2968749999999996e-07, | |
| "logits/chosen": 6210.16064453125, | |
| "logits/rejected": 5102.63427734375, | |
| "logps/chosen": -256.07000732421875, | |
| "logps/rejected": -254.1890869140625, | |
| "loss": 1.3644, | |
| "rewards/accuracies": 0.7250000834465027, | |
| "rewards/chosen": 1.0678449869155884, | |
| "rewards/margins": 11.644010543823242, | |
| "rewards/rejected": -10.576166152954102, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09419152276295134, | |
| "grad_norm": 12.300002755478918, | |
| "learning_rate": 4.6874999999999996e-07, | |
| "logits/chosen": 6063.0439453125, | |
| "logits/rejected": 5074.71826171875, | |
| "logps/chosen": -279.06317138671875, | |
| "logps/rejected": -286.5476379394531, | |
| "loss": 1.3566, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.5236625671386719, | |
| "rewards/margins": 17.469036102294922, | |
| "rewards/rejected": -16.94537353515625, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.10204081632653061, | |
| "grad_norm": 7.110167356695566, | |
| "learning_rate": 4.999962424962166e-07, | |
| "logits/chosen": 6032.046875, | |
| "logits/rejected": 5570.3583984375, | |
| "logps/chosen": -286.85101318359375, | |
| "logps/rejected": -281.3219299316406, | |
| "loss": 1.3643, | |
| "rewards/accuracies": 0.6416667103767395, | |
| "rewards/chosen": 3.256235122680664, | |
| "rewards/margins": 11.147451400756836, | |
| "rewards/rejected": -7.8912153244018555, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.10989010989010989, | |
| "grad_norm": 9.933114313171957, | |
| "learning_rate": 4.998647417232375e-07, | |
| "logits/chosen": 5896.25390625, | |
| "logits/rejected": 5167.8720703125, | |
| "logps/chosen": -264.71954345703125, | |
| "logps/rejected": -270.3382263183594, | |
| "loss": 1.3659, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -4.4773268699646, | |
| "rewards/margins": 15.50507926940918, | |
| "rewards/rejected": -19.98240852355957, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11773940345368916, | |
| "grad_norm": 18.02912449650011, | |
| "learning_rate": 4.995454786965036e-07, | |
| "logits/chosen": 5985.2431640625, | |
| "logits/rejected": 4986.9541015625, | |
| "logps/chosen": -277.7461853027344, | |
| "logps/rejected": -262.54913330078125, | |
| "loss": 1.3621, | |
| "rewards/accuracies": 0.7000000476837158, | |
| "rewards/chosen": 0.12583641707897186, | |
| "rewards/margins": 20.78557014465332, | |
| "rewards/rejected": -20.659732818603516, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12558869701726844, | |
| "grad_norm": 10.259154769962691, | |
| "learning_rate": 4.990386933279972e-07, | |
| "logits/chosen": 5920.44921875, | |
| "logits/rejected": 5265.1396484375, | |
| "logps/chosen": -279.40008544921875, | |
| "logps/rejected": -288.27239990234375, | |
| "loss": 1.3602, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -5.8358893394470215, | |
| "rewards/margins": 16.303333282470703, | |
| "rewards/rejected": -22.139223098754883, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13343799058084774, | |
| "grad_norm": 10.334859382833892, | |
| "learning_rate": 4.983447664444096e-07, | |
| "logits/chosen": 6151.115234375, | |
| "logits/rejected": 5482.80712890625, | |
| "logps/chosen": -289.50921630859375, | |
| "logps/rejected": -283.6671142578125, | |
| "loss": 1.3577, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": 0.28831419348716736, | |
| "rewards/margins": 14.94935417175293, | |
| "rewards/rejected": -14.661041259765625, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.141287284144427, | |
| "grad_norm": 10.89182659405604, | |
| "learning_rate": 4.97464219500968e-07, | |
| "logits/chosen": 5435.22216796875, | |
| "logits/rejected": 4739.29638671875, | |
| "logps/chosen": -264.37957763671875, | |
| "logps/rejected": -265.5071716308594, | |
| "loss": 1.3546, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": -0.935233473777771, | |
| "rewards/margins": 17.30495834350586, | |
| "rewards/rejected": -18.24019432067871, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14913657770800628, | |
| "grad_norm": 10.587152467990663, | |
| "learning_rate": 4.963977141895843e-07, | |
| "logits/chosen": 5499.7109375, | |
| "logits/rejected": 4725.96484375, | |
| "logps/chosen": -288.96429443359375, | |
| "logps/rejected": -293.37158203125, | |
| "loss": 1.3504, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.6796563863754272, | |
| "rewards/margins": 30.230484008789062, | |
| "rewards/rejected": -31.910140991210938, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15698587127158556, | |
| "grad_norm": 12.9585002988099, | |
| "learning_rate": 4.951460519416227e-07, | |
| "logits/chosen": 5437.3857421875, | |
| "logits/rejected": 5018.8994140625, | |
| "logps/chosen": -249.23568725585938, | |
| "logps/rejected": -282.5234680175781, | |
| "loss": 1.3535, | |
| "rewards/accuracies": 0.6583333611488342, | |
| "rewards/chosen": 2.014281749725342, | |
| "rewards/margins": 21.064096450805664, | |
| "rewards/rejected": -19.049816131591797, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16483516483516483, | |
| "grad_norm": 13.988339309057054, | |
| "learning_rate": 4.937101733256606e-07, | |
| "logits/chosen": 4930.84326171875, | |
| "logits/rejected": 4372.6025390625, | |
| "logps/chosen": -228.283203125, | |
| "logps/rejected": -246.59848022460938, | |
| "loss": 1.3514, | |
| "rewards/accuracies": 0.6416667699813843, | |
| "rewards/chosen": -3.5702528953552246, | |
| "rewards/margins": 20.567468643188477, | |
| "rewards/rejected": -24.137720108032227, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.1726844583987441, | |
| "grad_norm": 13.390095593837465, | |
| "learning_rate": 4.920911573406924e-07, | |
| "logits/chosen": 6006.0947265625, | |
| "logits/rejected": 5091.2421875, | |
| "logps/chosen": -286.8468017578125, | |
| "logps/rejected": -264.01007080078125, | |
| "loss": 1.3475, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -4.213888645172119, | |
| "rewards/margins": 26.255664825439453, | |
| "rewards/rejected": -30.469554901123047, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.18053375196232338, | |
| "grad_norm": 8.835282950532482, | |
| "learning_rate": 4.902902206053098e-07, | |
| "logits/chosen": 5451.99462890625, | |
| "logits/rejected": 4929.2568359375, | |
| "logps/chosen": -274.7848815917969, | |
| "logps/rejected": -285.9403076171875, | |
| "loss": 1.3613, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": -4.50766658782959, | |
| "rewards/margins": 23.41042709350586, | |
| "rewards/rejected": -27.918094635009766, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18838304552590268, | |
| "grad_norm": 11.386605076391048, | |
| "learning_rate": 4.883087164434672e-07, | |
| "logits/chosen": 4924.3642578125, | |
| "logits/rejected": 3931.905517578125, | |
| "logps/chosen": -235.32815551757812, | |
| "logps/rejected": -233.06259155273438, | |
| "loss": 1.3466, | |
| "rewards/accuracies": 0.7999999523162842, | |
| "rewards/chosen": 4.365412712097168, | |
| "rewards/margins": 24.50173568725586, | |
| "rewards/rejected": -20.136322021484375, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19623233908948196, | |
| "grad_norm": 14.852101177024576, | |
| "learning_rate": 4.861481338675183e-07, | |
| "logits/chosen": 5796.91650390625, | |
| "logits/rejected": 5151.455078125, | |
| "logps/chosen": -241.3012237548828, | |
| "logps/rejected": -289.451171875, | |
| "loss": 1.3523, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": -5.653787136077881, | |
| "rewards/margins": 29.40525245666504, | |
| "rewards/rejected": -35.05904006958008, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.20408163265306123, | |
| "grad_norm": 25.34165509811899, | |
| "learning_rate": 4.838100964592904e-07, | |
| "logits/chosen": 5968.93798828125, | |
| "logits/rejected": 4793.3125, | |
| "logps/chosen": -284.72705078125, | |
| "logps/rejected": -264.1895446777344, | |
| "loss": 1.3619, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": 0.0255860798060894, | |
| "rewards/margins": 25.690975189208984, | |
| "rewards/rejected": -25.665390014648438, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2119309262166405, | |
| "grad_norm": 19.332212090573815, | |
| "learning_rate": 4.812963611500339e-07, | |
| "logits/chosen": 5883.005859375, | |
| "logits/rejected": 5666.29296875, | |
| "logps/chosen": -276.0812683105469, | |
| "logps/rejected": -295.21917724609375, | |
| "loss": 1.3351, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.5726773738861084, | |
| "rewards/margins": 26.818140029907227, | |
| "rewards/rejected": -28.390817642211914, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.21978021978021978, | |
| "grad_norm": 14.608788344306173, | |
| "learning_rate": 4.786088169001671e-07, | |
| "logits/chosen": 5039.642578125, | |
| "logits/rejected": 4366.6572265625, | |
| "logps/chosen": -227.117919921875, | |
| "logps/rejected": -264.2989196777344, | |
| "loss": 1.349, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": 1.5974067449569702, | |
| "rewards/margins": 31.4863338470459, | |
| "rewards/rejected": -29.888927459716797, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22762951334379905, | |
| "grad_norm": 13.162440911557498, | |
| "learning_rate": 4.7574948327980567e-07, | |
| "logits/chosen": 6984.78271484375, | |
| "logits/rejected": 5142.07568359375, | |
| "logps/chosen": -320.518310546875, | |
| "logps/rejected": -291.1632385253906, | |
| "loss": 1.3413, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": 4.272213935852051, | |
| "rewards/margins": 37.669044494628906, | |
| "rewards/rejected": -33.39683151245117, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.23547880690737832, | |
| "grad_norm": 11.078754356565836, | |
| "learning_rate": 4.727205089511466e-07, | |
| "logits/chosen": 5084.3916015625, | |
| "logits/rejected": 5000.9306640625, | |
| "logps/chosen": -242.4727020263672, | |
| "logps/rejected": -278.3855895996094, | |
| "loss": 1.3511, | |
| "rewards/accuracies": 0.7249999642372131, | |
| "rewards/chosen": -5.732133388519287, | |
| "rewards/margins": 28.726451873779297, | |
| "rewards/rejected": -34.458587646484375, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.24332810047095763, | |
| "grad_norm": 10.77472855962222, | |
| "learning_rate": 4.6952417005384247e-07, | |
| "logits/chosen": 5697.62109375, | |
| "logits/rejected": 5098.0419921875, | |
| "logps/chosen": -252.02890014648438, | |
| "logps/rejected": -257.55767822265625, | |
| "loss": 1.3556, | |
| "rewards/accuracies": 0.6833333373069763, | |
| "rewards/chosen": 3.639686107635498, | |
| "rewards/margins": 19.53166961669922, | |
| "rewards/rejected": -15.89198112487793, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.25117739403453687, | |
| "grad_norm": 17.11307388675024, | |
| "learning_rate": 4.661628684945851e-07, | |
| "logits/chosen": 5720.63427734375, | |
| "logits/rejected": 4948.8916015625, | |
| "logps/chosen": -271.60260009765625, | |
| "logps/rejected": -297.1649169921875, | |
| "loss": 1.3498, | |
| "rewards/accuracies": 0.7833333015441895, | |
| "rewards/chosen": 5.684656620025635, | |
| "rewards/margins": 30.92440414428711, | |
| "rewards/rejected": -25.239748001098633, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25902668759811615, | |
| "grad_norm": 12.367352919357952, | |
| "learning_rate": 4.626391301421782e-07, | |
| "logits/chosen": 5532.48876953125, | |
| "logits/rejected": 5025.5048828125, | |
| "logps/chosen": -269.034423828125, | |
| "logps/rejected": -272.80877685546875, | |
| "loss": 1.3575, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 1.0636268854141235, | |
| "rewards/margins": 24.480344772338867, | |
| "rewards/rejected": -23.41672134399414, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2668759811616955, | |
| "grad_norm": 8.994486651516668, | |
| "learning_rate": 4.5895560292945996e-07, | |
| "logits/chosen": 5816.6611328125, | |
| "logits/rejected": 5900.87353515625, | |
| "logps/chosen": -263.0292663574219, | |
| "logps/rejected": -319.2271423339844, | |
| "loss": 1.3448, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": 5.053846836090088, | |
| "rewards/margins": 22.39950180053711, | |
| "rewards/rejected": -17.345653533935547, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.27472527472527475, | |
| "grad_norm": 19.827599555279765, | |
| "learning_rate": 4.5511505486349865e-07, | |
| "logits/chosen": 6142.09912109375, | |
| "logits/rejected": 5550.3779296875, | |
| "logps/chosen": -271.07275390625, | |
| "logps/rejected": -319.1446533203125, | |
| "loss": 1.3436, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": 1.834998369216919, | |
| "rewards/margins": 37.48114013671875, | |
| "rewards/rejected": -35.646141052246094, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.282574568288854, | |
| "grad_norm": 13.444707696548154, | |
| "learning_rate": 4.5112037194555876e-07, | |
| "logits/chosen": 5679.64794921875, | |
| "logits/rejected": 5532.9794921875, | |
| "logps/chosen": -245.96444702148438, | |
| "logps/rejected": -313.7906494140625, | |
| "loss": 1.3421, | |
| "rewards/accuracies": 0.7166666984558105, | |
| "rewards/chosen": -2.279477596282959, | |
| "rewards/margins": 37.538108825683594, | |
| "rewards/rejected": -39.817588806152344, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2904238618524333, | |
| "grad_norm": 19.583734659018468, | |
| "learning_rate": 4.4697455600239863e-07, | |
| "logits/chosen": 5170.92236328125, | |
| "logits/rejected": 4834.27001953125, | |
| "logps/chosen": -258.157470703125, | |
| "logps/rejected": -263.2154846191406, | |
| "loss": 1.3526, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.7448989152908325, | |
| "rewards/margins": 25.542724609375, | |
| "rewards/rejected": -24.797826766967773, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.29827315541601257, | |
| "grad_norm": 8.08546602135427, | |
| "learning_rate": 4.426807224305315e-07, | |
| "logits/chosen": 6216.36181640625, | |
| "logits/rejected": 5109.68505859375, | |
| "logps/chosen": -301.31640625, | |
| "logps/rejected": -283.6510925292969, | |
| "loss": 1.3469, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": 6.660558223724365, | |
| "rewards/margins": 34.403236389160156, | |
| "rewards/rejected": -27.742671966552734, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.30612244897959184, | |
| "grad_norm": 27.74929847224248, | |
| "learning_rate": 4.3824209785514326e-07, | |
| "logits/chosen": 6327.078125, | |
| "logits/rejected": 4816.5087890625, | |
| "logps/chosen": -285.81243896484375, | |
| "logps/rejected": -281.25262451171875, | |
| "loss": 1.3417, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": 5.714006423950195, | |
| "rewards/margins": 41.474945068359375, | |
| "rewards/rejected": -35.76093673706055, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3139717425431711, | |
| "grad_norm": 9.104859038759448, | |
| "learning_rate": 4.3366201770542687e-07, | |
| "logits/chosen": 5376.41943359375, | |
| "logits/rejected": 5212.60009765625, | |
| "logps/chosen": -280.355712890625, | |
| "logps/rejected": -326.0173034667969, | |
| "loss": 1.3628, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -5.738234043121338, | |
| "rewards/margins": 33.6685791015625, | |
| "rewards/rejected": -39.40680694580078, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3218210361067504, | |
| "grad_norm": 16.18185965000689, | |
| "learning_rate": 4.2894392370815567e-07, | |
| "logits/chosen": 5832.12939453125, | |
| "logits/rejected": 5222.6982421875, | |
| "logps/chosen": -287.0929260253906, | |
| "logps/rejected": -319.9795227050781, | |
| "loss": 1.3244, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 4.7759108543396, | |
| "rewards/margins": 42.1308708190918, | |
| "rewards/rejected": -37.35496520996094, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.32967032967032966, | |
| "grad_norm": 18.594382630147045, | |
| "learning_rate": 4.2409136130137845e-07, | |
| "logits/chosen": 5506.63427734375, | |
| "logits/rejected": 4987.7587890625, | |
| "logps/chosen": -291.8835144042969, | |
| "logps/rejected": -295.0062561035156, | |
| "loss": 1.3448, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -5.990305423736572, | |
| "rewards/margins": 35.93355178833008, | |
| "rewards/rejected": -41.923858642578125, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.33751962323390894, | |
| "grad_norm": 16.377815387474406, | |
| "learning_rate": 4.1910797697018017e-07, | |
| "logits/chosen": 5318.04443359375, | |
| "logits/rejected": 4443.08154296875, | |
| "logps/chosen": -257.94244384765625, | |
| "logps/rejected": -276.1806335449219, | |
| "loss": 1.3303, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -1.1475229263305664, | |
| "rewards/margins": 50.02882766723633, | |
| "rewards/rejected": -51.17634963989258, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3453689167974882, | |
| "grad_norm": 13.298196752411583, | |
| "learning_rate": 4.1399751550651084e-07, | |
| "logits/chosen": 5691.7333984375, | |
| "logits/rejected": 5633.2880859375, | |
| "logps/chosen": -255.7645721435547, | |
| "logps/rejected": -298.67950439453125, | |
| "loss": 1.3412, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -4.996587753295898, | |
| "rewards/margins": 33.46734619140625, | |
| "rewards/rejected": -38.46393585205078, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3532182103610675, | |
| "grad_norm": 11.949337864294227, | |
| "learning_rate": 4.087638171951401e-07, | |
| "logits/chosen": 6557.4560546875, | |
| "logits/rejected": 4709.63916015625, | |
| "logps/chosen": -282.85821533203125, | |
| "logps/rejected": -264.0398254394531, | |
| "loss": 1.3394, | |
| "rewards/accuracies": 0.7666667103767395, | |
| "rewards/chosen": 4.198666572570801, | |
| "rewards/margins": 42.2392578125, | |
| "rewards/rejected": -38.04059600830078, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.36106750392464676, | |
| "grad_norm": 18.009080463194742, | |
| "learning_rate": 4.034108149278543e-07, | |
| "logits/chosen": 6794.8349609375, | |
| "logits/rejected": 5243.083984375, | |
| "logps/chosen": -328.0745849609375, | |
| "logps/rejected": -295.5428771972656, | |
| "loss": 1.3422, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": 4.63649320602417, | |
| "rewards/margins": 35.52788543701172, | |
| "rewards/rejected": -30.89139175415039, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.36891679748822603, | |
| "grad_norm": 15.242057503851473, | |
| "learning_rate": 3.979425312480629e-07, | |
| "logits/chosen": 5803.8564453125, | |
| "logits/rejected": 5091.99755859375, | |
| "logps/chosen": -288.36767578125, | |
| "logps/rejected": -314.0430908203125, | |
| "loss": 1.3403, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.5818802118301392, | |
| "rewards/margins": 35.362457275390625, | |
| "rewards/rejected": -35.9443359375, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.37676609105180536, | |
| "grad_norm": 8.425440898911905, | |
| "learning_rate": 3.923630753280357e-07, | |
| "logits/chosen": 6259.91064453125, | |
| "logits/rejected": 5369.96923828125, | |
| "logps/chosen": -282.23291015625, | |
| "logps/rejected": -289.3865051269531, | |
| "loss": 1.3425, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -2.4263367652893066, | |
| "rewards/margins": 40.16687774658203, | |
| "rewards/rejected": -42.59321594238281, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 11.00106948084356, | |
| "learning_rate": 3.866766398810424e-07, | |
| "logits/chosen": 5882.1298828125, | |
| "logits/rejected": 5595.41796875, | |
| "logps/chosen": -231.893798828125, | |
| "logps/rejected": -307.16790771484375, | |
| "loss": 1.3228, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": 9.04726791381836, | |
| "rewards/margins": 43.61574172973633, | |
| "rewards/rejected": -34.56847381591797, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3924646781789639, | |
| "grad_norm": 19.585809671843773, | |
| "learning_rate": 3.8088749801071496e-07, | |
| "logits/chosen": 6353.7998046875, | |
| "logits/rejected": 4928.9228515625, | |
| "logps/chosen": -306.9959716796875, | |
| "logps/rejected": -332.1584777832031, | |
| "loss": 1.3408, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": -8.50652027130127, | |
| "rewards/margins": 59.219215393066406, | |
| "rewards/rejected": -67.7257308959961, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4003139717425432, | |
| "grad_norm": 11.818213453951092, | |
| "learning_rate": 3.75e-07, | |
| "logits/chosen": 5125.52392578125, | |
| "logits/rejected": 4506.7783203125, | |
| "logps/chosen": -254.2230682373047, | |
| "logps/rejected": -276.11163330078125, | |
| "loss": 1.3461, | |
| "rewards/accuracies": 0.7583333849906921, | |
| "rewards/chosen": 2.636770725250244, | |
| "rewards/margins": 49.630699157714844, | |
| "rewards/rejected": -46.993927001953125, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 12.192569188299785, | |
| "learning_rate": 3.6901857004211443e-07, | |
| "logits/chosen": 5393.0078125, | |
| "logits/rejected": 4975.08251953125, | |
| "logps/chosen": -268.563232421875, | |
| "logps/rejected": -304.7891540527344, | |
| "loss": 1.3551, | |
| "rewards/accuracies": 0.6833332777023315, | |
| "rewards/chosen": -2.390415906906128, | |
| "rewards/margins": 33.272422790527344, | |
| "rewards/rejected": -35.662837982177734, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.41601255886970173, | |
| "grad_norm": 14.033496034633881, | |
| "learning_rate": 3.6294770291596076e-07, | |
| "logits/chosen": 6087.6201171875, | |
| "logits/rejected": 4996.3271484375, | |
| "logps/chosen": -278.80938720703125, | |
| "logps/rejected": -286.06890869140625, | |
| "loss": 1.3329, | |
| "rewards/accuracies": 0.7166666388511658, | |
| "rewards/chosen": 5.48272705078125, | |
| "rewards/margins": 27.480846405029297, | |
| "rewards/rejected": -21.998119354248047, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.423861852433281, | |
| "grad_norm": 14.81241119572004, | |
| "learning_rate": 3.5679196060850034e-07, | |
| "logits/chosen": 5809.55859375, | |
| "logits/rejected": 5171.65673828125, | |
| "logps/chosen": -277.66326904296875, | |
| "logps/rejected": -297.49688720703125, | |
| "loss": 1.3437, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -0.2796575725078583, | |
| "rewards/margins": 41.81981658935547, | |
| "rewards/rejected": -42.0994758605957, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4317111459968603, | |
| "grad_norm": 12.798701067262096, | |
| "learning_rate": 3.505559688866229e-07, | |
| "logits/chosen": 5618.95849609375, | |
| "logits/rejected": 5222.8896484375, | |
| "logps/chosen": -278.93939208984375, | |
| "logps/rejected": -324.45550537109375, | |
| "loss": 1.3355, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": 2.3972537517547607, | |
| "rewards/margins": 35.66948318481445, | |
| "rewards/rejected": -33.27222442626953, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.43956043956043955, | |
| "grad_norm": 13.21925433462796, | |
| "learning_rate": 3.4424441382108826e-07, | |
| "logits/chosen": 5613.3388671875, | |
| "logits/rejected": 5243.99072265625, | |
| "logps/chosen": -282.6348571777344, | |
| "logps/rejected": -307.70916748046875, | |
| "loss": 1.3454, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -6.507281303405762, | |
| "rewards/margins": 34.423736572265625, | |
| "rewards/rejected": -40.93102264404297, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4474097331240188, | |
| "grad_norm": 20.97640858075027, | |
| "learning_rate": 3.378620382651523e-07, | |
| "logits/chosen": 5962.16357421875, | |
| "logits/rejected": 5507.55810546875, | |
| "logps/chosen": -317.55755615234375, | |
| "logps/rejected": -332.68597412109375, | |
| "loss": 1.3433, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -2.992626190185547, | |
| "rewards/margins": 37.75238037109375, | |
| "rewards/rejected": -40.7450065612793, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.4552590266875981, | |
| "grad_norm": 14.558248023475134, | |
| "learning_rate": 3.314136382905234e-07, | |
| "logits/chosen": 5928.56640625, | |
| "logits/rejected": 5350.7138671875, | |
| "logps/chosen": -282.3759460449219, | |
| "logps/rejected": -322.36962890625, | |
| "loss": 1.3422, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": 4.87086296081543, | |
| "rewards/margins": 42.390098571777344, | |
| "rewards/rejected": -37.51923370361328, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4631083202511774, | |
| "grad_norm": 11.970316354669556, | |
| "learning_rate": 3.249040595833274e-07, | |
| "logits/chosen": 6434.35302734375, | |
| "logits/rejected": 5420.3125, | |
| "logps/chosen": -303.90411376953125, | |
| "logps/rejected": -283.4837341308594, | |
| "loss": 1.3363, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": 2.2493205070495605, | |
| "rewards/margins": 39.89289093017578, | |
| "rewards/rejected": -37.64357376098633, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.47095761381475665, | |
| "grad_norm": 18.089656280102407, | |
| "learning_rate": 3.1833819380279023e-07, | |
| "logits/chosen": 6090.2900390625, | |
| "logits/rejected": 5182.3095703125, | |
| "logps/chosen": -250.2621612548828, | |
| "logps/rejected": -292.43292236328125, | |
| "loss": 1.3306, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 2.228858709335327, | |
| "rewards/margins": 39.69800567626953, | |
| "rewards/rejected": -37.46915054321289, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.478806907378336, | |
| "grad_norm": 11.22555391650791, | |
| "learning_rate": 3.11720974905373e-07, | |
| "logits/chosen": 5875.5830078125, | |
| "logits/rejected": 5115.8701171875, | |
| "logps/chosen": -272.0517883300781, | |
| "logps/rejected": -296.71087646484375, | |
| "loss": 1.3224, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.7897025942802429, | |
| "rewards/margins": 47.00892639160156, | |
| "rewards/rejected": -46.21921920776367, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.48665620094191525, | |
| "grad_norm": 14.568276742620993, | |
| "learning_rate": 3.0505737543712275e-07, | |
| "logits/chosen": 4994.525390625, | |
| "logits/rejected": 4103.8837890625, | |
| "logps/chosen": -249.3625946044922, | |
| "logps/rejected": -255.7006378173828, | |
| "loss": 1.3334, | |
| "rewards/accuracies": 0.7583333849906921, | |
| "rewards/chosen": -1.5366663932800293, | |
| "rewards/margins": 40.58672332763672, | |
| "rewards/rejected": -42.12339401245117, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.4945054945054945, | |
| "grad_norm": 14.733773271335288, | |
| "learning_rate": 2.9835240279702513e-07, | |
| "logits/chosen": 6521.90283203125, | |
| "logits/rejected": 5524.4736328125, | |
| "logps/chosen": -310.3549499511719, | |
| "logps/rejected": -315.75042724609375, | |
| "loss": 1.3311, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": 6.356592178344727, | |
| "rewards/margins": 56.0721435546875, | |
| "rewards/rejected": -49.715553283691406, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.5023547880690737, | |
| "grad_norm": 14.093171590433894, | |
| "learning_rate": 2.9161109547416667e-07, | |
| "logits/chosen": 6216.2978515625, | |
| "logits/rejected": 5327.9072265625, | |
| "logps/chosen": -279.8126525878906, | |
| "logps/rejected": -307.3542175292969, | |
| "loss": 1.3315, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -1.1222234964370728, | |
| "rewards/margins": 28.391210556030273, | |
| "rewards/rejected": -29.5134334564209, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5102040816326531, | |
| "grad_norm": 13.6593664706907, | |
| "learning_rate": 2.848385192615339e-07, | |
| "logits/chosen": 5323.177734375, | |
| "logits/rejected": 4330.21826171875, | |
| "logps/chosen": -256.6112976074219, | |
| "logps/rejected": -267.56719970703125, | |
| "loss": 1.3318, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.9272038340568542, | |
| "rewards/margins": 41.3925666809082, | |
| "rewards/rejected": -40.465354919433594, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5180533751962323, | |
| "grad_norm": 20.41441904521654, | |
| "learning_rate": 2.780397634492949e-07, | |
| "logits/chosen": 5928.0263671875, | |
| "logits/rejected": 4747.13720703125, | |
| "logps/chosen": -286.15582275390625, | |
| "logps/rejected": -301.84051513671875, | |
| "loss": 1.3373, | |
| "rewards/accuracies": 0.7916667461395264, | |
| "rewards/chosen": 4.162437915802002, | |
| "rewards/margins": 53.73211669921875, | |
| "rewards/rejected": -49.569679260253906, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5259026687598116, | |
| "grad_norm": 13.818709240958135, | |
| "learning_rate": 2.71219937000424e-07, | |
| "logits/chosen": 5939.3681640625, | |
| "logits/rejected": 4897.11962890625, | |
| "logps/chosen": -273.56671142578125, | |
| "logps/rejected": -280.0821838378906, | |
| "loss": 1.3386, | |
| "rewards/accuracies": 0.7166666984558105, | |
| "rewards/chosen": 2.6437764167785645, | |
| "rewards/margins": 31.636972427368164, | |
| "rewards/rejected": -28.99319839477539, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.533751962323391, | |
| "grad_norm": 41.130057456711704, | |
| "learning_rate": 2.6438416471154273e-07, | |
| "logits/chosen": 5823.55078125, | |
| "logits/rejected": 4877.8759765625, | |
| "logps/chosen": -275.37005615234375, | |
| "logps/rejected": -281.0362854003906, | |
| "loss": 1.3356, | |
| "rewards/accuracies": 0.7666667103767395, | |
| "rewards/chosen": 2.457808256149292, | |
| "rewards/margins": 48.48851776123047, | |
| "rewards/rejected": -46.03070831298828, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5416012558869702, | |
| "grad_norm": 18.04694819817481, | |
| "learning_rate": 2.5753758336186326e-07, | |
| "logits/chosen": 5727.6728515625, | |
| "logits/rejected": 5275.44482421875, | |
| "logps/chosen": -274.7696228027344, | |
| "logps/rejected": -328.78533935546875, | |
| "loss": 1.3293, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.31948238611221313, | |
| "rewards/margins": 51.6459846496582, | |
| "rewards/rejected": -51.32649612426758, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5494505494505495, | |
| "grad_norm": 13.091605442505317, | |
| "learning_rate": 2.5068533785312666e-07, | |
| "logits/chosen": 5472.88671875, | |
| "logits/rejected": 5261.16552734375, | |
| "logps/chosen": -251.00350952148438, | |
| "logps/rejected": -291.24249267578125, | |
| "loss": 1.3473, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.8872472643852234, | |
| "rewards/margins": 44.903663635253906, | |
| "rewards/rejected": -45.79090881347656, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5572998430141287, | |
| "grad_norm": 14.635710242423256, | |
| "learning_rate": 2.4383257734343794e-07, | |
| "logits/chosen": 5453.39111328125, | |
| "logits/rejected": 5437.5087890625, | |
| "logps/chosen": -249.7123260498047, | |
| "logps/rejected": -304.453369140625, | |
| "loss": 1.333, | |
| "rewards/accuracies": 0.7166666984558105, | |
| "rewards/chosen": -0.4664379060268402, | |
| "rewards/margins": 39.897727966308594, | |
| "rewards/rejected": -40.364166259765625, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.565149136577708, | |
| "grad_norm": 11.3725426273517, | |
| "learning_rate": 2.3698445137790258e-07, | |
| "logits/chosen": 5855.744140625, | |
| "logits/rejected": 5054.41455078125, | |
| "logps/chosen": -278.29632568359375, | |
| "logps/rejected": -295.22430419921875, | |
| "loss": 1.3376, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": 6.468451499938965, | |
| "rewards/margins": 43.57428741455078, | |
| "rewards/rejected": -37.1058349609375, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5729984301412873, | |
| "grad_norm": 13.835969138728538, | |
| "learning_rate": 2.3014610601897157e-07, | |
| "logits/chosen": 6335.21630859375, | |
| "logits/rejected": 4871.65283203125, | |
| "logps/chosen": -292.9911193847656, | |
| "logps/rejected": -279.36297607421875, | |
| "loss": 1.329, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": 2.0679280757904053, | |
| "rewards/margins": 41.14617156982422, | |
| "rewards/rejected": -39.07823944091797, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5808477237048666, | |
| "grad_norm": 14.836930457465817, | |
| "learning_rate": 2.2332267997940513e-07, | |
| "logits/chosen": 5278.6005859375, | |
| "logits/rejected": 4482.12890625, | |
| "logps/chosen": -242.5696258544922, | |
| "logps/rejected": -257.9692077636719, | |
| "loss": 1.3248, | |
| "rewards/accuracies": 0.7916666269302368, | |
| "rewards/chosen": 5.438414573669434, | |
| "rewards/margins": 45.854732513427734, | |
| "rewards/rejected": -40.41631317138672, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5886970172684458, | |
| "grad_norm": 18.322582602210563, | |
| "learning_rate": 2.1651930076075723e-07, | |
| "logits/chosen": 5739.4140625, | |
| "logits/rejected": 5206.27783203125, | |
| "logps/chosen": -240.5334930419922, | |
| "logps/rejected": -252.7237548828125, | |
| "loss": 1.3372, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.27130642533302307, | |
| "rewards/margins": 29.90248680114746, | |
| "rewards/rejected": -30.173797607421875, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5965463108320251, | |
| "grad_norm": 16.105294021082695, | |
| "learning_rate": 2.0974108080028692e-07, | |
| "logits/chosen": 6041.63232421875, | |
| "logits/rejected": 4766.7265625, | |
| "logps/chosen": -258.1484069824219, | |
| "logps/rejected": -270.5816955566406, | |
| "loss": 1.3332, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 5.212155342102051, | |
| "rewards/margins": 37.87400436401367, | |
| "rewards/rejected": -32.6618537902832, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.6043956043956044, | |
| "grad_norm": 21.542454507655048, | |
| "learning_rate": 2.0299311362918773e-07, | |
| "logits/chosen": 6268.74267578125, | |
| "logits/rejected": 5378.11181640625, | |
| "logps/chosen": -292.4772644042969, | |
| "logps/rejected": -327.83941650390625, | |
| "loss": 1.3404, | |
| "rewards/accuracies": 0.7166666388511658, | |
| "rewards/chosen": 2.167691707611084, | |
| "rewards/margins": 45.20887756347656, | |
| "rewards/rejected": -43.04119110107422, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6122448979591837, | |
| "grad_norm": 14.859266227724207, | |
| "learning_rate": 1.962804700450265e-07, | |
| "logits/chosen": 6079.2265625, | |
| "logits/rejected": 5769.8720703125, | |
| "logps/chosen": -277.9334716796875, | |
| "logps/rejected": -348.91180419921875, | |
| "loss": 1.3422, | |
| "rewards/accuracies": 0.7416667342185974, | |
| "rewards/chosen": 5.1991095542907715, | |
| "rewards/margins": 39.5477180480957, | |
| "rewards/rejected": -34.348609924316406, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6200941915227629, | |
| "grad_norm": 17.289507430070785, | |
| "learning_rate": 1.8960819430126334e-07, | |
| "logits/chosen": 5685.2568359375, | |
| "logits/rejected": 5022.52587890625, | |
| "logps/chosen": -256.14813232421875, | |
| "logps/rejected": -313.5560302734375, | |
| "loss": 1.3308, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -1.6967357397079468, | |
| "rewards/margins": 56.51359939575195, | |
| "rewards/rejected": -58.21033477783203, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6279434850863422, | |
| "grad_norm": 20.11535614922616, | |
| "learning_rate": 1.8298130031671972e-07, | |
| "logits/chosen": 5686.21923828125, | |
| "logits/rejected": 4978.49072265625, | |
| "logps/chosen": -277.88140869140625, | |
| "logps/rejected": -307.9370422363281, | |
| "loss": 1.349, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -4.144988059997559, | |
| "rewards/margins": 37.27484893798828, | |
| "rewards/rejected": -41.419837951660156, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6357927786499215, | |
| "grad_norm": 14.165928356937492, | |
| "learning_rate": 1.7640476790784075e-07, | |
| "logits/chosen": 5179.6455078125, | |
| "logits/rejected": 4716.197265625, | |
| "logps/chosen": -275.0489807128906, | |
| "logps/rejected": -312.2055969238281, | |
| "loss": 1.3374, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -1.2190592288970947, | |
| "rewards/margins": 37.21118927001953, | |
| "rewards/rejected": -38.43024444580078, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6436420722135008, | |
| "grad_norm": 14.347930044372637, | |
| "learning_rate": 1.6988353904658492e-07, | |
| "logits/chosen": 5679.46044921875, | |
| "logits/rejected": 4414.927734375, | |
| "logps/chosen": -277.84991455078125, | |
| "logps/rejected": -252.51123046875, | |
| "loss": 1.3239, | |
| "rewards/accuracies": 0.8166666030883789, | |
| "rewards/chosen": 5.222801685333252, | |
| "rewards/margins": 41.908973693847656, | |
| "rewards/rejected": -36.68617248535156, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6514913657770801, | |
| "grad_norm": 19.898579796392823, | |
| "learning_rate": 1.634225141467513e-07, | |
| "logits/chosen": 5611.337890625, | |
| "logits/rejected": 5031.56640625, | |
| "logps/chosen": -265.5953369140625, | |
| "logps/rejected": -299.775146484375, | |
| "loss": 1.3353, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -2.1408400535583496, | |
| "rewards/margins": 49.84906005859375, | |
| "rewards/rejected": -51.989906311035156, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6593406593406593, | |
| "grad_norm": 11.367717976315854, | |
| "learning_rate": 1.570265483815364e-07, | |
| "logits/chosen": 6171.8974609375, | |
| "logits/rejected": 5058.6689453125, | |
| "logps/chosen": -285.18670654296875, | |
| "logps/rejected": -314.2545471191406, | |
| "loss": 1.3313, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.3465190827846527, | |
| "rewards/margins": 40.40586471557617, | |
| "rewards/rejected": -40.05934524536133, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6671899529042387, | |
| "grad_norm": 15.257111920071068, | |
| "learning_rate": 1.5070044803508691e-07, | |
| "logits/chosen": 5713.53955078125, | |
| "logits/rejected": 5173.865234375, | |
| "logps/chosen": -271.6879577636719, | |
| "logps/rejected": -294.98724365234375, | |
| "loss": 1.3144, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.8542098999023438, | |
| "rewards/margins": 49.883445739746094, | |
| "rewards/rejected": -49.02924346923828, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6750392464678179, | |
| "grad_norm": 24.549427196976698, | |
| "learning_rate": 1.444489668907914e-07, | |
| "logits/chosen": 6195.8759765625, | |
| "logits/rejected": 5269.583984375, | |
| "logps/chosen": -299.3851623535156, | |
| "logps/rejected": -303.155517578125, | |
| "loss": 1.3406, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 3.483236789703369, | |
| "rewards/margins": 52.33366012573242, | |
| "rewards/rejected": -48.850425720214844, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6828885400313972, | |
| "grad_norm": 15.338561329629815, | |
| "learning_rate": 1.3827680265902232e-07, | |
| "logits/chosen": 6130.16943359375, | |
| "logits/rejected": 5076.16650390625, | |
| "logps/chosen": -287.5099792480469, | |
| "logps/rejected": -301.68994140625, | |
| "loss": 1.3344, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 1.9024982452392578, | |
| "rewards/margins": 49.383323669433594, | |
| "rewards/rejected": -47.48082733154297, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6907378335949764, | |
| "grad_norm": 18.349687582837216, | |
| "learning_rate": 1.3218859344701632e-07, | |
| "logits/chosen": 5425.09814453125, | |
| "logits/rejected": 5145.62744140625, | |
| "logps/chosen": -254.3935546875, | |
| "logps/rejected": -331.16436767578125, | |
| "loss": 1.3298, | |
| "rewards/accuracies": 0.8166667222976685, | |
| "rewards/chosen": 4.359163284301758, | |
| "rewards/margins": 47.21587371826172, | |
| "rewards/rejected": -42.856712341308594, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6985871271585558, | |
| "grad_norm": 19.217261044107826, | |
| "learning_rate": 1.2618891427354172e-07, | |
| "logits/chosen": 6371.8828125, | |
| "logits/rejected": 5201.703125, | |
| "logps/chosen": -306.90191650390625, | |
| "logps/rejected": -306.8157958984375, | |
| "loss": 1.3413, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 5.626570224761963, | |
| "rewards/margins": 50.41263961791992, | |
| "rewards/rejected": -44.786067962646484, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.706436420722135, | |
| "grad_norm": 12.27601531904447, | |
| "learning_rate": 1.202822736309758e-07, | |
| "logits/chosen": 5404.2451171875, | |
| "logits/rejected": 4977.263671875, | |
| "logps/chosen": -253.3689422607422, | |
| "logps/rejected": -301.351806640625, | |
| "loss": 1.3373, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 4.672505855560303, | |
| "rewards/margins": 41.6027946472168, | |
| "rewards/rejected": -36.93029022216797, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 16.891852269390416, | |
| "learning_rate": 1.1447311009737299e-07, | |
| "logits/chosen": 5318.9541015625, | |
| "logits/rejected": 5051.8701171875, | |
| "logps/chosen": -257.08343505859375, | |
| "logps/rejected": -300.7640380859375, | |
| "loss": 1.3363, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.48396730422973633, | |
| "rewards/margins": 43.27261734008789, | |
| "rewards/rejected": -42.78865051269531, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.7221350078492935, | |
| "grad_norm": 13.331899210748277, | |
| "learning_rate": 1.0876578900107053e-07, | |
| "logits/chosen": 5862.32861328125, | |
| "logits/rejected": 4909.03173828125, | |
| "logps/chosen": -283.63812255859375, | |
| "logps/rejected": -280.3291320800781, | |
| "loss": 1.3348, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": 3.0835793018341064, | |
| "rewards/margins": 38.459007263183594, | |
| "rewards/rejected": -35.37542724609375, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.7299843014128728, | |
| "grad_norm": 12.315909203426335, | |
| "learning_rate": 1.0316459914033793e-07, | |
| "logits/chosen": 5814.38818359375, | |
| "logits/rejected": 4369.14013671875, | |
| "logps/chosen": -290.85711669921875, | |
| "logps/rejected": -276.1246643066406, | |
| "loss": 1.3357, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -2.9035708904266357, | |
| "rewards/margins": 40.27419662475586, | |
| "rewards/rejected": -43.17776870727539, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7378335949764521, | |
| "grad_norm": 12.761052610243647, | |
| "learning_rate": 9.767374956053584e-08, | |
| "logits/chosen": 5648.11865234375, | |
| "logits/rejected": 4961.22021484375, | |
| "logps/chosen": -262.9266662597656, | |
| "logps/rejected": -296.5054626464844, | |
| "loss": 1.3331, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": 1.9393322467803955, | |
| "rewards/margins": 51.28413009643555, | |
| "rewards/rejected": -49.34479904174805, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7456828885400314, | |
| "grad_norm": 15.961811740621364, | |
| "learning_rate": 9.229736639120561e-08, | |
| "logits/chosen": 5820.294921875, | |
| "logits/rejected": 5358.4501953125, | |
| "logps/chosen": -261.51019287109375, | |
| "logps/rejected": -291.1629638671875, | |
| "loss": 1.3374, | |
| "rewards/accuracies": 0.7166666984558105, | |
| "rewards/chosen": 2.049938201904297, | |
| "rewards/margins": 30.419775009155273, | |
| "rewards/rejected": -28.369836807250977, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7535321821036107, | |
| "grad_norm": 17.74540755293784, | |
| "learning_rate": 8.70394897454659e-08, | |
| "logits/chosen": 5693.8564453125, | |
| "logits/rejected": 5088.1923828125, | |
| "logps/chosen": -256.2207946777344, | |
| "logps/rejected": -285.6395568847656, | |
| "loss": 1.3262, | |
| "rewards/accuracies": 0.8333333730697632, | |
| "rewards/chosen": 7.792778968811035, | |
| "rewards/margins": 47.109642028808594, | |
| "rewards/rejected": -39.31686019897461, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7613814756671899, | |
| "grad_norm": 16.946371930331026, | |
| "learning_rate": 8.19040706840472e-08, | |
| "logits/chosen": 5793.71728515625, | |
| "logits/rejected": 4867.8349609375, | |
| "logps/chosen": -285.85101318359375, | |
| "logps/rejected": -297.75115966796875, | |
| "loss": 1.3191, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 3.2804324626922607, | |
| "rewards/margins": 54.697120666503906, | |
| "rewards/rejected": -51.41667938232422, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 14.121877040769963, | |
| "learning_rate": 7.689496824624525e-08, | |
| "logits/chosen": 5501.0263671875, | |
| "logits/rejected": 4394.03515625, | |
| "logps/chosen": -272.18084716796875, | |
| "logps/rejected": -304.90155029296875, | |
| "loss": 1.314, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 4.461153984069824, | |
| "rewards/margins": 70.67555236816406, | |
| "rewards/rejected": -66.21439361572266, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7770800627943485, | |
| "grad_norm": 26.732470735289972, | |
| "learning_rate": 7.201594655002458e-08, | |
| "logits/chosen": 5842.39208984375, | |
| "logits/rejected": 4866.20166015625, | |
| "logps/chosen": -270.3237609863281, | |
| "logps/rejected": -294.47418212890625, | |
| "loss": 1.3256, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": -2.2163407802581787, | |
| "rewards/margins": 55.973243713378906, | |
| "rewards/rejected": -58.1895866394043, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7849293563579278, | |
| "grad_norm": 18.77323467526337, | |
| "learning_rate": 6.727067196345099e-08, | |
| "logits/chosen": 5498.07763671875, | |
| "logits/rejected": 4639.53173828125, | |
| "logps/chosen": -261.0239562988281, | |
| "logps/rejected": -268.3658142089844, | |
| "loss": 1.3318, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.702414333820343, | |
| "rewards/margins": 41.98760986328125, | |
| "rewards/rejected": -42.690025329589844, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.792778649921507, | |
| "grad_norm": 17.592174205857383, | |
| "learning_rate": 6.26627103495786e-08, | |
| "logits/chosen": 5701.921875, | |
| "logits/rejected": 4740.48681640625, | |
| "logps/chosen": -254.7830047607422, | |
| "logps/rejected": -281.86212158203125, | |
| "loss": 1.3313, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 4.632332801818848, | |
| "rewards/margins": 50.97934341430664, | |
| "rewards/rejected": -46.347007751464844, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.8006279434850864, | |
| "grad_norm": 13.752063548346952, | |
| "learning_rate": 5.8195524386862374e-08, | |
| "logits/chosen": 5787.1845703125, | |
| "logits/rejected": 5125.1923828125, | |
| "logps/chosen": -290.4544372558594, | |
| "logps/rejected": -315.5395202636719, | |
| "loss": 1.3331, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 7.442157745361328, | |
| "rewards/margins": 49.381126403808594, | |
| "rewards/rejected": -41.938968658447266, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.8084772370486656, | |
| "grad_norm": 11.958970899521432, | |
| "learning_rate": 5.38724709671092e-08, | |
| "logits/chosen": 6183.4560546875, | |
| "logits/rejected": 5846.205078125, | |
| "logps/chosen": -272.84075927734375, | |
| "logps/rejected": -320.28857421875, | |
| "loss": 1.3268, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": 4.959445953369141, | |
| "rewards/margins": 50.137760162353516, | |
| "rewards/rejected": -45.178314208984375, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 15.636426596469342, | |
| "learning_rate": 4.969679867292276e-08, | |
| "logits/chosen": 5491.4345703125, | |
| "logits/rejected": 4991.15625, | |
| "logps/chosen": -267.9794921875, | |
| "logps/rejected": -309.1411437988281, | |
| "loss": 1.3281, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": -2.7362513542175293, | |
| "rewards/margins": 52.83111572265625, | |
| "rewards/rejected": -55.5673713684082, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.8241758241758241, | |
| "grad_norm": 16.08684334634875, | |
| "learning_rate": 4.5671645336537416e-08, | |
| "logits/chosen": 5548.25634765625, | |
| "logits/rejected": 5025.25390625, | |
| "logps/chosen": -280.34722900390625, | |
| "logps/rejected": -319.4068603515625, | |
| "loss": 1.3262, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": 1.185887336730957, | |
| "rewards/margins": 54.8686637878418, | |
| "rewards/rejected": -53.682777404785156, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.8320251177394035, | |
| "grad_norm": 25.21510864266244, | |
| "learning_rate": 4.180003568187776e-08, | |
| "logits/chosen": 6859.25634765625, | |
| "logits/rejected": 5394.68603515625, | |
| "logps/chosen": -308.07086181640625, | |
| "logps/rejected": -305.1750793457031, | |
| "loss": 1.3342, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.14819645881652832, | |
| "rewards/margins": 40.491920471191406, | |
| "rewards/rejected": -40.640113830566406, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.8398744113029827, | |
| "grad_norm": 23.166205130811267, | |
| "learning_rate": 3.8084879051612144e-08, | |
| "logits/chosen": 5705.005859375, | |
| "logits/rejected": 5211.6767578125, | |
| "logps/chosen": -268.3230285644531, | |
| "logps/rejected": -279.81500244140625, | |
| "loss": 1.3313, | |
| "rewards/accuracies": 0.7416667342185974, | |
| "rewards/chosen": 4.569327354431152, | |
| "rewards/margins": 47.14904022216797, | |
| "rewards/rejected": -42.5797119140625, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.847723704866562, | |
| "grad_norm": 14.145560157344223, | |
| "learning_rate": 3.452896722091128e-08, | |
| "logits/chosen": 6269.6328125, | |
| "logits/rejected": 4840.6416015625, | |
| "logps/chosen": -304.7834777832031, | |
| "logps/rejected": -297.2294921875, | |
| "loss": 1.3203, | |
| "rewards/accuracies": 0.8166667819023132, | |
| "rewards/chosen": 7.704338073730469, | |
| "rewards/margins": 63.69272994995117, | |
| "rewards/rejected": -55.98839569091797, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8555729984301413, | |
| "grad_norm": 22.06234881634468, | |
| "learning_rate": 3.11349722995527e-08, | |
| "logits/chosen": 6309.28466796875, | |
| "logits/rejected": 4754.0556640625, | |
| "logps/chosen": -271.60272216796875, | |
| "logps/rejected": -295.8482666015625, | |
| "loss": 1.3331, | |
| "rewards/accuracies": 0.7416667342185974, | |
| "rewards/chosen": 2.194936752319336, | |
| "rewards/margins": 48.374412536621094, | |
| "rewards/rejected": -46.17947769165039, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8634222919937206, | |
| "grad_norm": 16.765509803508074, | |
| "learning_rate": 2.7905444723949762e-08, | |
| "logits/chosen": 6098.36474609375, | |
| "logits/rejected": 5048.1748046875, | |
| "logps/chosen": -282.4183654785156, | |
| "logps/rejected": -289.3445129394531, | |
| "loss": 1.336, | |
| "rewards/accuracies": 0.7916666269302368, | |
| "rewards/chosen": 0.8044170141220093, | |
| "rewards/margins": 58.03986358642578, | |
| "rewards/rejected": -57.235450744628906, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8712715855572999, | |
| "grad_norm": 13.49722567059801, | |
| "learning_rate": 2.484281134061142e-08, | |
| "logits/chosen": 6470.77734375, | |
| "logits/rejected": 5227.4248046875, | |
| "logps/chosen": -308.28125, | |
| "logps/rejected": -317.0091857910156, | |
| "loss": 1.3261, | |
| "rewards/accuracies": 0.7916666865348816, | |
| "rewards/chosen": 0.9931009411811829, | |
| "rewards/margins": 55.281524658203125, | |
| "rewards/rejected": -54.288421630859375, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8791208791208791, | |
| "grad_norm": 22.162318246756865, | |
| "learning_rate": 2.194937358247506e-08, | |
| "logits/chosen": 6336.37744140625, | |
| "logits/rejected": 5156.53662109375, | |
| "logps/chosen": -290.1517028808594, | |
| "logps/rejected": -313.71923828125, | |
| "loss": 1.3272, | |
| "rewards/accuracies": 0.7833333015441895, | |
| "rewards/chosen": 1.1864078044891357, | |
| "rewards/margins": 55.42217254638672, | |
| "rewards/rejected": -54.23577117919922, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8869701726844584, | |
| "grad_norm": 18.39920794977249, | |
| "learning_rate": 1.9227305739481612e-08, | |
| "logits/chosen": 5778.82958984375, | |
| "logits/rejected": 4524.93798828125, | |
| "logps/chosen": -270.95416259765625, | |
| "logps/rejected": -272.36578369140625, | |
| "loss": 1.3283, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": 7.6158647537231445, | |
| "rewards/margins": 55.379432678222656, | |
| "rewards/rejected": -47.76355743408203, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8948194662480377, | |
| "grad_norm": 13.813322430164584, | |
| "learning_rate": 1.6678653324693787e-08, | |
| "logits/chosen": 6345.8369140625, | |
| "logits/rejected": 5146.07373046875, | |
| "logps/chosen": -302.70758056640625, | |
| "logps/rejected": -304.568115234375, | |
| "loss": 1.326, | |
| "rewards/accuracies": 0.7416667342185974, | |
| "rewards/chosen": 2.731863260269165, | |
| "rewards/margins": 45.693050384521484, | |
| "rewards/rejected": -42.96118927001953, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.902668759811617, | |
| "grad_norm": 13.562562133674499, | |
| "learning_rate": 1.4305331537183384e-08, | |
| "logits/chosen": 5643.845703125, | |
| "logits/rejected": 5149.87744140625, | |
| "logps/chosen": -265.35552978515625, | |
| "logps/rejected": -306.31292724609375, | |
| "loss": 1.3213, | |
| "rewards/accuracies": 0.6916666626930237, | |
| "rewards/chosen": 1.8398481607437134, | |
| "rewards/margins": 44.68225860595703, | |
| "rewards/rejected": -42.84241485595703, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.9105180533751962, | |
| "grad_norm": 13.602390776530422, | |
| "learning_rate": 1.2109123822844653e-08, | |
| "logits/chosen": 5800.7470703125, | |
| "logits/rejected": 4576.7470703125, | |
| "logps/chosen": -271.863037109375, | |
| "logps/rejected": -276.7607727050781, | |
| "loss": 1.3286, | |
| "rewards/accuracies": 0.7333332896232605, | |
| "rewards/chosen": -1.1851859092712402, | |
| "rewards/margins": 41.55708312988281, | |
| "rewards/rejected": -42.742271423339844, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.9183673469387755, | |
| "grad_norm": 22.181340292160385, | |
| "learning_rate": 1.0091680534213387e-08, | |
| "logits/chosen": 6310.87353515625, | |
| "logits/rejected": 6083.5595703125, | |
| "logps/chosen": -291.30230712890625, | |
| "logps/rejected": -343.88775634765625, | |
| "loss": 1.3346, | |
| "rewards/accuracies": 0.6833333969116211, | |
| "rewards/chosen": 0.8247451782226562, | |
| "rewards/margins": 40.90442657470703, | |
| "rewards/rejected": -40.07968521118164, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.9262166405023547, | |
| "grad_norm": 19.386965396369924, | |
| "learning_rate": 8.254517690300944e-09, | |
| "logits/chosen": 5587.4599609375, | |
| "logits/rejected": 5026.15380859375, | |
| "logps/chosen": -275.68817138671875, | |
| "logps/rejected": -304.5224914550781, | |
| "loss": 1.3333, | |
| "rewards/accuracies": 0.7666667103767395, | |
| "rewards/chosen": 3.26338267326355, | |
| "rewards/margins": 48.9727668762207, | |
| "rewards/rejected": -45.70938491821289, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.9340659340659341, | |
| "grad_norm": 18.166944148946065, | |
| "learning_rate": 6.599015837372907e-09, | |
| "logits/chosen": 6081.8134765625, | |
| "logits/rejected": 5268.125, | |
| "logps/chosen": -296.06268310546875, | |
| "logps/rejected": -314.55780029296875, | |
| "loss": 1.3257, | |
| "rewards/accuracies": 0.7166666388511658, | |
| "rewards/chosen": -3.7439887523651123, | |
| "rewards/margins": 46.977317810058594, | |
| "rewards/rejected": -50.721309661865234, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.9419152276295133, | |
| "grad_norm": 23.054621837524056, | |
| "learning_rate": 5.126419011529992e-09, | |
| "logits/chosen": 6278.8330078125, | |
| "logits/rejected": 5327.0478515625, | |
| "logps/chosen": -300.5643310546875, | |
| "logps/rejected": -308.908447265625, | |
| "loss": 1.3288, | |
| "rewards/accuracies": 0.841666579246521, | |
| "rewards/chosen": 4.18377685546875, | |
| "rewards/margins": 54.943260192871094, | |
| "rewards/rejected": -50.759490966796875, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9497645211930926, | |
| "grad_norm": 30.456570394001158, | |
| "learning_rate": 3.837833803870177e-09, | |
| "logits/chosen": 5893.2646484375, | |
| "logits/rejected": 5159.4443359375, | |
| "logps/chosen": -275.4104309082031, | |
| "logps/rejected": -302.34478759765625, | |
| "loss": 1.3282, | |
| "rewards/accuracies": 0.8000000715255737, | |
| "rewards/chosen": 4.3925042152404785, | |
| "rewards/margins": 54.420494079589844, | |
| "rewards/rejected": -50.027992248535156, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.957613814756672, | |
| "grad_norm": 17.154490819112805, | |
| "learning_rate": 2.734228528934679e-09, | |
| "logits/chosen": 7311.48974609375, | |
| "logits/rejected": 5375.12841796875, | |
| "logps/chosen": -347.67498779296875, | |
| "logps/rejected": -334.7488708496094, | |
| "loss": 1.3403, | |
| "rewards/accuracies": 0.7583333849906921, | |
| "rewards/chosen": -1.259454369544983, | |
| "rewards/margins": 48.19050598144531, | |
| "rewards/rejected": -49.4499626159668, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9654631083202512, | |
| "grad_norm": 21.407381285740787, | |
| "learning_rate": 1.8164324970625645e-09, | |
| "logits/chosen": 6486.4833984375, | |
| "logits/rejected": 5080.07470703125, | |
| "logps/chosen": -302.5705871582031, | |
| "logps/rejected": -313.39874267578125, | |
| "loss": 1.3316, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": 5.784834384918213, | |
| "rewards/margins": 54.12310791015625, | |
| "rewards/rejected": -48.3382682800293, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9733124018838305, | |
| "grad_norm": 12.65026863678564, | |
| "learning_rate": 1.0851353912008642e-09, | |
| "logits/chosen": 5584.55908203125, | |
| "logits/rejected": 5134.20703125, | |
| "logps/chosen": -274.4143371582031, | |
| "logps/rejected": -323.35443115234375, | |
| "loss": 1.3204, | |
| "rewards/accuracies": 0.7166666984558105, | |
| "rewards/chosen": -1.4607696533203125, | |
| "rewards/margins": 45.6775016784668, | |
| "rewards/rejected": -47.138275146484375, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9811616954474097, | |
| "grad_norm": 14.086764845840083, | |
| "learning_rate": 5.408867486384471e-10, | |
| "logits/chosen": 5697.59228515625, | |
| "logits/rejected": 4794.82666015625, | |
| "logps/chosen": -263.6590270996094, | |
| "logps/rejected": -265.0268859863281, | |
| "loss": 1.3285, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 6.779554843902588, | |
| "rewards/margins": 48.45376968383789, | |
| "rewards/rejected": -41.674217224121094, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.989010989010989, | |
| "grad_norm": 14.635164517100653, | |
| "learning_rate": 1.840955480532924e-10, | |
| "logits/chosen": 5442.82275390625, | |
| "logits/rejected": 5116.64990234375, | |
| "logps/chosen": -265.51348876953125, | |
| "logps/rejected": -295.3118896484375, | |
| "loss": 1.3256, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": 1.139338493347168, | |
| "rewards/margins": 38.744808197021484, | |
| "rewards/rejected": -37.60546875, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9968602825745683, | |
| "grad_norm": 16.61323875313643, | |
| "learning_rate": 1.502990218302247e-11, | |
| "logits/chosen": 5668.91748046875, | |
| "logits/rejected": 4585.67529296875, | |
| "logps/chosen": -261.23193359375, | |
| "logps/rejected": -272.4540710449219, | |
| "loss": 1.326, | |
| "rewards/accuracies": 0.7916666865348816, | |
| "rewards/chosen": 0.24156558513641357, | |
| "rewards/margins": 44.537353515625, | |
| "rewards/rejected": -44.2957878112793, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1274, | |
| "total_flos": 0.0, | |
| "train_loss": 1.3416044334040331, | |
| "train_runtime": 14497.0358, | |
| "train_samples_per_second": 4.217, | |
| "train_steps_per_second": 0.088 | |
| } | |
| ], | |
| "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 | |
| } | |