|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 1.0, |
|
"eval_steps": 500, |
|
"global_step": 1274, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0007849293563579278, |
|
"grad_norm": 6.093131966217363, |
|
"learning_rate": 3.90625e-09, |
|
"logits/chosen": 5914.52099609375, |
|
"logits/rejected": 2785.021484375, |
|
"logps/chosen": -212.45889282226562, |
|
"logps/rejected": -98.59669494628906, |
|
"loss": 0.6931, |
|
"rewards/accuracies": 0.0, |
|
"rewards/chosen": 0.0, |
|
"rewards/margins": 0.0, |
|
"rewards/rejected": 0.0, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.007849293563579277, |
|
"grad_norm": 6.063672072770918, |
|
"learning_rate": 3.9062499999999997e-08, |
|
"logits/chosen": 4973.8095703125, |
|
"logits/rejected": 4328.318359375, |
|
"logps/chosen": -204.25518798828125, |
|
"logps/rejected": -179.73281860351562, |
|
"loss": 0.6931, |
|
"rewards/accuracies": 0.5092592835426331, |
|
"rewards/chosen": 0.02867966890335083, |
|
"rewards/margins": 0.05586132034659386, |
|
"rewards/rejected": -0.02718164585530758, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.015698587127158554, |
|
"grad_norm": 6.188456944981756, |
|
"learning_rate": 7.812499999999999e-08, |
|
"logits/chosen": 6084.14111328125, |
|
"logits/rejected": 4834.31640625, |
|
"logps/chosen": -217.20156860351562, |
|
"logps/rejected": -196.7338104248047, |
|
"loss": 0.6932, |
|
"rewards/accuracies": 0.5416666865348816, |
|
"rewards/chosen": 0.009626108221709728, |
|
"rewards/margins": 0.029883313924074173, |
|
"rewards/rejected": -0.020257214084267616, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.023547880690737835, |
|
"grad_norm": 5.474904458795652, |
|
"learning_rate": 1.1718749999999999e-07, |
|
"logits/chosen": 6084.2373046875, |
|
"logits/rejected": 5105.03662109375, |
|
"logps/chosen": -250.5583953857422, |
|
"logps/rejected": -209.34817504882812, |
|
"loss": 0.693, |
|
"rewards/accuracies": 0.5083333253860474, |
|
"rewards/chosen": 0.013943061232566833, |
|
"rewards/margins": 0.0645134299993515, |
|
"rewards/rejected": -0.05057036876678467, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.03139717425431711, |
|
"grad_norm": 5.700132865060452, |
|
"learning_rate": 1.5624999999999999e-07, |
|
"logits/chosen": 5311.515625, |
|
"logits/rejected": 4346.8349609375, |
|
"logps/chosen": -211.9991455078125, |
|
"logps/rejected": -181.73727416992188, |
|
"loss": 0.6928, |
|
"rewards/accuracies": 0.6750000715255737, |
|
"rewards/chosen": 0.08009053766727448, |
|
"rewards/margins": 0.1303517371416092, |
|
"rewards/rejected": -0.05026119947433472, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.03924646781789639, |
|
"grad_norm": 5.75968341286992, |
|
"learning_rate": 1.9531249999999998e-07, |
|
"logits/chosen": 6424.73291015625, |
|
"logits/rejected": 5042.3193359375, |
|
"logps/chosen": -265.3091735839844, |
|
"logps/rejected": -206.7834014892578, |
|
"loss": 0.6924, |
|
"rewards/accuracies": 0.6750000715255737, |
|
"rewards/chosen": 0.3174227178096771, |
|
"rewards/margins": 0.4012085497379303, |
|
"rewards/rejected": -0.08378583192825317, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.04709576138147567, |
|
"grad_norm": 5.546891159199977, |
|
"learning_rate": 2.3437499999999998e-07, |
|
"logits/chosen": 5484.2392578125, |
|
"logits/rejected": 4559.8701171875, |
|
"logps/chosen": -213.73947143554688, |
|
"logps/rejected": -209.15982055664062, |
|
"loss": 0.6918, |
|
"rewards/accuracies": 0.6583333611488342, |
|
"rewards/chosen": 0.26449495553970337, |
|
"rewards/margins": 0.6242072582244873, |
|
"rewards/rejected": -0.3597122132778168, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.054945054945054944, |
|
"grad_norm": 5.331755041716918, |
|
"learning_rate": 2.734375e-07, |
|
"logits/chosen": 5194.4169921875, |
|
"logits/rejected": 4918.47998046875, |
|
"logps/chosen": -178.27944946289062, |
|
"logps/rejected": -177.4241485595703, |
|
"loss": 0.6911, |
|
"rewards/accuracies": 0.6666667461395264, |
|
"rewards/chosen": 0.2014751136302948, |
|
"rewards/margins": 0.8939019441604614, |
|
"rewards/rejected": -0.6924268007278442, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.06279434850863422, |
|
"grad_norm": 5.584999261034571, |
|
"learning_rate": 3.1249999999999997e-07, |
|
"logits/chosen": 5774.71337890625, |
|
"logits/rejected": 5269.7578125, |
|
"logps/chosen": -196.8366241455078, |
|
"logps/rejected": -183.1029815673828, |
|
"loss": 0.6888, |
|
"rewards/accuracies": 0.6583333015441895, |
|
"rewards/chosen": -0.26804548501968384, |
|
"rewards/margins": 1.244419813156128, |
|
"rewards/rejected": -1.5124653577804565, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.0706436420722135, |
|
"grad_norm": 6.357864437998398, |
|
"learning_rate": 3.5156249999999997e-07, |
|
"logits/chosen": 6036.94384765625, |
|
"logits/rejected": 5178.0302734375, |
|
"logps/chosen": -220.6177520751953, |
|
"logps/rejected": -191.21290588378906, |
|
"loss": 0.6847, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -1.735548973083496, |
|
"rewards/margins": 5.033526420593262, |
|
"rewards/rejected": -6.769075870513916, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.07849293563579278, |
|
"grad_norm": 7.040364327766666, |
|
"learning_rate": 3.9062499999999997e-07, |
|
"logits/chosen": 5949.79296875, |
|
"logits/rejected": 5727.5908203125, |
|
"logps/chosen": -216.8369140625, |
|
"logps/rejected": -212.58670043945312, |
|
"loss": 0.6839, |
|
"rewards/accuracies": 0.6583333611488342, |
|
"rewards/chosen": -6.606234073638916, |
|
"rewards/margins": 6.538748264312744, |
|
"rewards/rejected": -13.144981384277344, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.08634222919937205, |
|
"grad_norm": 6.763596378595654, |
|
"learning_rate": 4.2968749999999996e-07, |
|
"logits/chosen": 6423.328125, |
|
"logits/rejected": 5241.2958984375, |
|
"logps/chosen": -198.6194305419922, |
|
"logps/rejected": -202.97906494140625, |
|
"loss": 0.6778, |
|
"rewards/accuracies": 0.6583333611488342, |
|
"rewards/chosen": -15.504582405090332, |
|
"rewards/margins": 8.482271194458008, |
|
"rewards/rejected": -23.986852645874023, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.09419152276295134, |
|
"grad_norm": 10.481073772174538, |
|
"learning_rate": 4.6874999999999996e-07, |
|
"logits/chosen": 6349.3115234375, |
|
"logits/rejected": 5272.1162109375, |
|
"logps/chosen": -227.03030395507812, |
|
"logps/rejected": -234.8236846923828, |
|
"loss": 0.672, |
|
"rewards/accuracies": 0.625, |
|
"rewards/chosen": -24.193578720092773, |
|
"rewards/margins": 12.465006828308105, |
|
"rewards/rejected": -36.65858840942383, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.10204081632653061, |
|
"grad_norm": 8.025924997034352, |
|
"learning_rate": 4.999962424962166e-07, |
|
"logits/chosen": 6358.8583984375, |
|
"logits/rejected": 5857.8310546875, |
|
"logps/chosen": -239.5376434326172, |
|
"logps/rejected": -246.1665496826172, |
|
"loss": 0.6658, |
|
"rewards/accuracies": 0.6833333373069763, |
|
"rewards/chosen": -29.729150772094727, |
|
"rewards/margins": 18.7928466796875, |
|
"rewards/rejected": -48.521995544433594, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.10989010989010989, |
|
"grad_norm": 9.521563540986381, |
|
"learning_rate": 4.998647417232375e-07, |
|
"logits/chosen": 6323.96923828125, |
|
"logits/rejected": 5568.6796875, |
|
"logps/chosen": -233.1209259033203, |
|
"logps/rejected": -242.63882446289062, |
|
"loss": 0.6641, |
|
"rewards/accuracies": 0.6500000357627869, |
|
"rewards/chosen": -47.453453063964844, |
|
"rewards/margins": 17.59344482421875, |
|
"rewards/rejected": -65.04689025878906, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.11773940345368916, |
|
"grad_norm": 12.644608831856553, |
|
"learning_rate": 4.995454786965036e-07, |
|
"logits/chosen": 6541.3330078125, |
|
"logits/rejected": 5415.2822265625, |
|
"logps/chosen": -245.84255981445312, |
|
"logps/rejected": -239.1229705810547, |
|
"loss": 0.6607, |
|
"rewards/accuracies": 0.6916667222976685, |
|
"rewards/chosen": -44.15815734863281, |
|
"rewards/margins": 22.66707992553711, |
|
"rewards/rejected": -66.82524108886719, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.12558869701726844, |
|
"grad_norm": 8.265424491202015, |
|
"learning_rate": 4.990386933279972e-07, |
|
"logits/chosen": 6377.3125, |
|
"logits/rejected": 5663.359375, |
|
"logps/chosen": -241.92959594726562, |
|
"logps/rejected": -266.5514831542969, |
|
"loss": 0.6574, |
|
"rewards/accuracies": 0.6916667222976685, |
|
"rewards/chosen": -44.40141296386719, |
|
"rewards/margins": 25.092517852783203, |
|
"rewards/rejected": -69.4939193725586, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.13343799058084774, |
|
"grad_norm": 9.232273980396416, |
|
"learning_rate": 4.983447664444096e-07, |
|
"logits/chosen": 6607.70947265625, |
|
"logits/rejected": 5855.15625, |
|
"logps/chosen": -261.1861267089844, |
|
"logps/rejected": -268.4610900878906, |
|
"loss": 0.6564, |
|
"rewards/accuracies": 0.6666666865348816, |
|
"rewards/chosen": -51.56547927856445, |
|
"rewards/margins": 20.574033737182617, |
|
"rewards/rejected": -72.13951110839844, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.141287284144427, |
|
"grad_norm": 9.866448872437736, |
|
"learning_rate": 4.97464219500968e-07, |
|
"logits/chosen": 5737.3740234375, |
|
"logits/rejected": 4962.4228515625, |
|
"logps/chosen": -247.4786834716797, |
|
"logps/rejected": -266.4071960449219, |
|
"loss": 0.6521, |
|
"rewards/accuracies": 0.6916667222976685, |
|
"rewards/chosen": -58.9509162902832, |
|
"rewards/margins": 29.29502296447754, |
|
"rewards/rejected": -88.24593353271484, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.14913657770800628, |
|
"grad_norm": 8.7062689635517, |
|
"learning_rate": 4.963977141895843e-07, |
|
"logits/chosen": 5834.91796875, |
|
"logits/rejected": 4950.6669921875, |
|
"logps/chosen": -265.50799560546875, |
|
"logps/rejected": -303.29736328125, |
|
"loss": 0.6419, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -61.749046325683594, |
|
"rewards/margins": 50.30883026123047, |
|
"rewards/rejected": -112.05787658691406, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.15698587127158556, |
|
"grad_norm": 10.780388842430504, |
|
"learning_rate": 4.951460519416227e-07, |
|
"logits/chosen": 5741.23486328125, |
|
"logits/rejected": 5287.05322265625, |
|
"logps/chosen": -251.54641723632812, |
|
"logps/rejected": -304.95709228515625, |
|
"loss": 0.6438, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -69.00214385986328, |
|
"rewards/margins": 39.283905029296875, |
|
"rewards/rejected": -108.28604888916016, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.16483516483516483, |
|
"grad_norm": 17.073665708890324, |
|
"learning_rate": 4.937101733256606e-07, |
|
"logits/chosen": 5240.689453125, |
|
"logits/rejected": 4608.19482421875, |
|
"logps/chosen": -215.47265625, |
|
"logps/rejected": -251.48239135742188, |
|
"loss": 0.6445, |
|
"rewards/accuracies": 0.6833332777023315, |
|
"rewards/chosen": -61.55937957763672, |
|
"rewards/margins": 32.7420539855957, |
|
"rewards/rejected": -94.30143737792969, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.1726844583987441, |
|
"grad_norm": 12.221014555466162, |
|
"learning_rate": 4.920911573406924e-07, |
|
"logits/chosen": 6383.455078125, |
|
"logits/rejected": 5361.6767578125, |
|
"logps/chosen": -253.47500610351562, |
|
"logps/rejected": -262.1346435546875, |
|
"loss": 0.6407, |
|
"rewards/accuracies": 0.7083333730697632, |
|
"rewards/chosen": -57.13840866088867, |
|
"rewards/margins": 40.999515533447266, |
|
"rewards/rejected": -98.13792419433594, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.18053375196232338, |
|
"grad_norm": 13.329208345468095, |
|
"learning_rate": 4.902902206053098e-07, |
|
"logits/chosen": 5734.9677734375, |
|
"logits/rejected": 5143.1357421875, |
|
"logps/chosen": -279.87896728515625, |
|
"logps/rejected": -315.6455078125, |
|
"loss": 0.6407, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -92.54025268554688, |
|
"rewards/margins": 43.33777618408203, |
|
"rewards/rejected": -135.87803649902344, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.18838304552590268, |
|
"grad_norm": 10.653934386649361, |
|
"learning_rate": 4.883087164434672e-07, |
|
"logits/chosen": 5146.69921875, |
|
"logits/rejected": 4056.839111328125, |
|
"logps/chosen": -252.93594360351562, |
|
"logps/rejected": -288.0379943847656, |
|
"loss": 0.6314, |
|
"rewards/accuracies": 0.6916667222976685, |
|
"rewards/chosen": -85.30692291259766, |
|
"rewards/margins": 48.041629791259766, |
|
"rewards/rejected": -133.34857177734375, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.19623233908948196, |
|
"grad_norm": 13.800858939449958, |
|
"learning_rate": 4.861481338675183e-07, |
|
"logits/chosen": 6170.0908203125, |
|
"logits/rejected": 5447.45458984375, |
|
"logps/chosen": -256.0116271972656, |
|
"logps/rejected": -326.3557434082031, |
|
"loss": 0.6432, |
|
"rewards/accuracies": 0.6750000715255737, |
|
"rewards/chosen": -88.86774444580078, |
|
"rewards/margins": 52.98130416870117, |
|
"rewards/rejected": -141.8490447998047, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.20408163265306123, |
|
"grad_norm": 16.199410438264415, |
|
"learning_rate": 4.838100964592904e-07, |
|
"logits/chosen": 6424.16455078125, |
|
"logits/rejected": 5117.61083984375, |
|
"logps/chosen": -273.86700439453125, |
|
"logps/rejected": -284.8200378417969, |
|
"loss": 0.6465, |
|
"rewards/accuracies": 0.7583333849906921, |
|
"rewards/chosen": -74.1485824584961, |
|
"rewards/margins": 44.447505950927734, |
|
"rewards/rejected": -118.59608459472656, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.2119309262166405, |
|
"grad_norm": 11.286813732099095, |
|
"learning_rate": 4.812963611500339e-07, |
|
"logits/chosen": 6215.70947265625, |
|
"logits/rejected": 5984.4306640625, |
|
"logps/chosen": -285.27606201171875, |
|
"logps/rejected": -319.1112365722656, |
|
"loss": 0.6303, |
|
"rewards/accuracies": 0.6916666626930237, |
|
"rewards/chosen": -88.52018737792969, |
|
"rewards/margins": 41.961997985839844, |
|
"rewards/rejected": -130.48216247558594, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.21978021978021978, |
|
"grad_norm": 16.531534462677037, |
|
"learning_rate": 4.786088169001671e-07, |
|
"logits/chosen": 5287.22216796875, |
|
"logits/rejected": 4560.978515625, |
|
"logps/chosen": -243.2444305419922, |
|
"logps/rejected": -313.148193359375, |
|
"loss": 0.6282, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -80.0935287475586, |
|
"rewards/margins": 63.753204345703125, |
|
"rewards/rejected": -143.8467559814453, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.22762951334379905, |
|
"grad_norm": 16.90462731667777, |
|
"learning_rate": 4.7574948327980567e-07, |
|
"logits/chosen": 7445.8447265625, |
|
"logits/rejected": 5403.94482421875, |
|
"logps/chosen": -331.681884765625, |
|
"logps/rejected": -354.81011962890625, |
|
"loss": 0.6276, |
|
"rewards/accuracies": 0.783333420753479, |
|
"rewards/chosen": -94.65342712402344, |
|
"rewards/margins": 77.64093780517578, |
|
"rewards/rejected": -172.29437255859375, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.23547880690737832, |
|
"grad_norm": 13.245971629591574, |
|
"learning_rate": 4.727205089511466e-07, |
|
"logits/chosen": 5381.85302734375, |
|
"logits/rejected": 5297.11083984375, |
|
"logps/chosen": -255.672607421875, |
|
"logps/rejected": -314.2789611816406, |
|
"loss": 0.636, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -88.2001724243164, |
|
"rewards/margins": 55.892372131347656, |
|
"rewards/rejected": -144.09254455566406, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.24332810047095763, |
|
"grad_norm": 12.146215443307637, |
|
"learning_rate": 4.6952417005384247e-07, |
|
"logits/chosen": 6014.71875, |
|
"logits/rejected": 5335.70556640625, |
|
"logps/chosen": -260.8976745605469, |
|
"logps/rejected": -300.9578857421875, |
|
"loss": 0.6408, |
|
"rewards/accuracies": 0.6833333373069763, |
|
"rewards/chosen": -83.31695556640625, |
|
"rewards/margins": 42.17268753051758, |
|
"rewards/rejected": -125.4896469116211, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.25117739403453687, |
|
"grad_norm": 13.90650852576195, |
|
"learning_rate": 4.661628684945851e-07, |
|
"logits/chosen": 5960.9345703125, |
|
"logits/rejected": 5119.1533203125, |
|
"logps/chosen": -308.49700927734375, |
|
"logps/rejected": -376.7956848144531, |
|
"loss": 0.6322, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -107.44000244140625, |
|
"rewards/margins": 67.33781433105469, |
|
"rewards/rejected": -174.77781677246094, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.25902668759811615, |
|
"grad_norm": 18.153298332822626, |
|
"learning_rate": 4.626391301421782e-07, |
|
"logits/chosen": 5755.26318359375, |
|
"logits/rejected": 5219.13671875, |
|
"logps/chosen": -295.9870300292969, |
|
"logps/rejected": -321.1059265136719, |
|
"loss": 0.643, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -101.53385925292969, |
|
"rewards/margins": 42.24925994873047, |
|
"rewards/rejected": -143.78311157226562, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.2668759811616955, |
|
"grad_norm": 10.626065491733366, |
|
"learning_rate": 4.5895560292945996e-07, |
|
"logits/chosen": 6115.3408203125, |
|
"logits/rejected": 6242.7255859375, |
|
"logps/chosen": -266.7374267578125, |
|
"logps/rejected": -334.5285949707031, |
|
"loss": 0.6505, |
|
"rewards/accuracies": 0.7166666984558105, |
|
"rewards/chosen": -72.87915802001953, |
|
"rewards/margins": 41.2818717956543, |
|
"rewards/rejected": -114.16102600097656, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.27472527472527475, |
|
"grad_norm": 19.123518096220113, |
|
"learning_rate": 4.5511505486349865e-07, |
|
"logits/chosen": 6534.66162109375, |
|
"logits/rejected": 5885.20654296875, |
|
"logps/chosen": -251.6634979248047, |
|
"logps/rejected": -319.6004943847656, |
|
"loss": 0.628, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -56.837799072265625, |
|
"rewards/margins": 55.940704345703125, |
|
"rewards/rejected": -112.77848815917969, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.282574568288854, |
|
"grad_norm": 14.298240994707886, |
|
"learning_rate": 4.5112037194555876e-07, |
|
"logits/chosen": 6060.9228515625, |
|
"logits/rejected": 5880.5185546875, |
|
"logps/chosen": -276.15887451171875, |
|
"logps/rejected": -374.9804992675781, |
|
"loss": 0.626, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -98.23542785644531, |
|
"rewards/margins": 77.46324920654297, |
|
"rewards/rejected": -175.6986846923828, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.2904238618524333, |
|
"grad_norm": 14.585395330021852, |
|
"learning_rate": 4.4697455600239863e-07, |
|
"logits/chosen": 5437.0537109375, |
|
"logits/rejected": 5045.03271484375, |
|
"logps/chosen": -286.1153564453125, |
|
"logps/rejected": -318.96649169921875, |
|
"loss": 0.6403, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -101.69328308105469, |
|
"rewards/margins": 49.202125549316406, |
|
"rewards/rejected": -150.89541625976562, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.29827315541601257, |
|
"grad_norm": 12.011840689440003, |
|
"learning_rate": 4.426807224305315e-07, |
|
"logits/chosen": 6565.9169921875, |
|
"logits/rejected": 5335.61669921875, |
|
"logps/chosen": -314.1622009277344, |
|
"logps/rejected": -326.3265380859375, |
|
"loss": 0.6373, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -85.82119750976562, |
|
"rewards/margins": 57.495765686035156, |
|
"rewards/rejected": -143.3169708251953, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.30612244897959184, |
|
"grad_norm": 14.603970390337606, |
|
"learning_rate": 4.3824209785514326e-07, |
|
"logits/chosen": 6679.51708984375, |
|
"logits/rejected": 5036.5615234375, |
|
"logps/chosen": -299.2738342285156, |
|
"logps/rejected": -336.33966064453125, |
|
"loss": 0.6334, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -83.78034210205078, |
|
"rewards/margins": 72.7061767578125, |
|
"rewards/rejected": -156.4865264892578, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.3139717425431711, |
|
"grad_norm": 12.263570083382138, |
|
"learning_rate": 4.3366201770542687e-07, |
|
"logits/chosen": 5716.26025390625, |
|
"logits/rejected": 5553.00634765625, |
|
"logps/chosen": -288.0743713378906, |
|
"logps/rejected": -344.66259765625, |
|
"loss": 0.6377, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -97.28577423095703, |
|
"rewards/margins": 58.6373176574707, |
|
"rewards/rejected": -155.92308044433594, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.3218210361067504, |
|
"grad_norm": 30.10421699468723, |
|
"learning_rate": 4.2894392370815567e-07, |
|
"logits/chosen": 6109.4443359375, |
|
"logits/rejected": 5399.1669921875, |
|
"logps/chosen": -343.89434814453125, |
|
"logps/rejected": -421.060302734375, |
|
"loss": 0.6049, |
|
"rewards/accuracies": 0.7833333015441895, |
|
"rewards/chosen": -131.40748596191406, |
|
"rewards/margins": 81.200439453125, |
|
"rewards/rejected": -212.60791015625, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.32967032967032966, |
|
"grad_norm": 23.67201136623266, |
|
"learning_rate": 4.2409136130137845e-07, |
|
"logits/chosen": 5768.19677734375, |
|
"logits/rejected": 5116.79296875, |
|
"logps/chosen": -338.7046203613281, |
|
"logps/rejected": -397.24517822265625, |
|
"loss": 0.6238, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -134.6746826171875, |
|
"rewards/margins": 81.32905578613281, |
|
"rewards/rejected": -216.00375366210938, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.33751962323390894, |
|
"grad_norm": 17.987511578768668, |
|
"learning_rate": 4.1910797697018017e-07, |
|
"logits/chosen": 5607.255859375, |
|
"logits/rejected": 4614.8291015625, |
|
"logps/chosen": -280.25738525390625, |
|
"logps/rejected": -348.2909240722656, |
|
"loss": 0.6312, |
|
"rewards/accuracies": 0.8416666984558105, |
|
"rewards/chosen": -100.51243591308594, |
|
"rewards/margins": 86.54534912109375, |
|
"rewards/rejected": -187.0577850341797, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.3453689167974882, |
|
"grad_norm": 16.95800834501449, |
|
"learning_rate": 4.1399751550651084e-07, |
|
"logits/chosen": 5963.57470703125, |
|
"logits/rejected": 5890.51171875, |
|
"logps/chosen": -275.5081787109375, |
|
"logps/rejected": -343.4457702636719, |
|
"loss": 0.6238, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -93.11483764648438, |
|
"rewards/margins": 58.647361755371094, |
|
"rewards/rejected": -151.76220703125, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.3532182103610675, |
|
"grad_norm": 11.15998480161967, |
|
"learning_rate": 4.087638171951401e-07, |
|
"logits/chosen": 6931.64697265625, |
|
"logits/rejected": 4930.6923828125, |
|
"logps/chosen": -293.3556213378906, |
|
"logps/rejected": -322.02337646484375, |
|
"loss": 0.6253, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -84.58689880371094, |
|
"rewards/margins": 74.44380950927734, |
|
"rewards/rejected": -159.0307159423828, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.36106750392464676, |
|
"grad_norm": 11.8715155801604, |
|
"learning_rate": 4.034108149278543e-07, |
|
"logits/chosen": 7102.58349609375, |
|
"logits/rejected": 5428.9775390625, |
|
"logps/chosen": -346.68634033203125, |
|
"logps/rejected": -355.1321716308594, |
|
"loss": 0.6207, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -96.25406646728516, |
|
"rewards/margins": 68.67012786865234, |
|
"rewards/rejected": -164.9241943359375, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.36891679748822603, |
|
"grad_norm": 16.98432474874273, |
|
"learning_rate": 3.979425312480629e-07, |
|
"logits/chosen": 6031.1064453125, |
|
"logits/rejected": 5226.72216796875, |
|
"logps/chosen": -335.190673828125, |
|
"logps/rejected": -399.636962890625, |
|
"loss": 0.6369, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -122.92594146728516, |
|
"rewards/margins": 74.1447525024414, |
|
"rewards/rejected": -197.07069396972656, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.37676609105180536, |
|
"grad_norm": 10.201495638028796, |
|
"learning_rate": 3.923630753280357e-07, |
|
"logits/chosen": 6538.5703125, |
|
"logits/rejected": 5552.7998046875, |
|
"logps/chosen": -298.9499816894531, |
|
"logps/rejected": -345.0953369140625, |
|
"loss": 0.6213, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -91.55477142333984, |
|
"rewards/margins": 79.14476776123047, |
|
"rewards/rejected": -170.6995391845703, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.38461538461538464, |
|
"grad_norm": 17.170911712736423, |
|
"learning_rate": 3.866766398810424e-07, |
|
"logits/chosen": 6079.12646484375, |
|
"logits/rejected": 5729.90625, |
|
"logps/chosen": -269.0557556152344, |
|
"logps/rejected": -384.9722900390625, |
|
"loss": 0.6034, |
|
"rewards/accuracies": 0.824999988079071, |
|
"rewards/chosen": -92.447265625, |
|
"rewards/margins": 90.50070190429688, |
|
"rewards/rejected": -182.94796752929688, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.3924646781789639, |
|
"grad_norm": 17.90792915865348, |
|
"learning_rate": 3.8088749801071496e-07, |
|
"logits/chosen": 6501.1435546875, |
|
"logits/rejected": 4916.962890625, |
|
"logps/chosen": -370.4066467285156, |
|
"logps/rejected": -442.2255859375, |
|
"loss": 0.6376, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -148.45297241210938, |
|
"rewards/margins": 91.25639343261719, |
|
"rewards/rejected": -239.7093505859375, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.4003139717425432, |
|
"grad_norm": 11.093400815043566, |
|
"learning_rate": 3.75e-07, |
|
"logits/chosen": 5320.60302734375, |
|
"logits/rejected": 4625.2607421875, |
|
"logps/chosen": -280.2235107421875, |
|
"logps/rejected": -348.7869567871094, |
|
"loss": 0.613, |
|
"rewards/accuracies": 0.8250001072883606, |
|
"rewards/chosen": -90.74053955078125, |
|
"rewards/margins": 91.68112182617188, |
|
"rewards/rejected": -182.42164611816406, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.40816326530612246, |
|
"grad_norm": 16.873928239705048, |
|
"learning_rate": 3.6901857004211443e-07, |
|
"logits/chosen": 5609.70556640625, |
|
"logits/rejected": 5175.36328125, |
|
"logps/chosen": -295.0995178222656, |
|
"logps/rejected": -360.041015625, |
|
"loss": 0.6383, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -96.6335220336914, |
|
"rewards/margins": 67.6194839477539, |
|
"rewards/rejected": -164.2530059814453, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.41601255886970173, |
|
"grad_norm": 12.66871333637675, |
|
"learning_rate": 3.6294770291596076e-07, |
|
"logits/chosen": 6355.0244140625, |
|
"logits/rejected": 5158.5986328125, |
|
"logps/chosen": -314.4820251464844, |
|
"logps/rejected": -356.3719177246094, |
|
"loss": 0.6139, |
|
"rewards/accuracies": 0.6999999284744263, |
|
"rewards/chosen": -101.55928802490234, |
|
"rewards/margins": 56.05229949951172, |
|
"rewards/rejected": -157.611572265625, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.423861852433281, |
|
"grad_norm": 15.194295742706043, |
|
"learning_rate": 3.5679196060850034e-07, |
|
"logits/chosen": 6004.0322265625, |
|
"logits/rejected": 5311.98583984375, |
|
"logps/chosen": -328.72637939453125, |
|
"logps/rejected": -383.19464111328125, |
|
"loss": 0.6213, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -117.90850830078125, |
|
"rewards/margins": 77.04728698730469, |
|
"rewards/rejected": -194.95579528808594, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.4317111459968603, |
|
"grad_norm": 17.3949205074546, |
|
"learning_rate": 3.505559688866229e-07, |
|
"logits/chosen": 5810.20166015625, |
|
"logits/rejected": 5345.8447265625, |
|
"logps/chosen": -338.4986267089844, |
|
"logps/rejected": -431.74285888671875, |
|
"loss": 0.6214, |
|
"rewards/accuracies": 0.73333340883255, |
|
"rewards/chosen": -123.7582015991211, |
|
"rewards/margins": 82.92054748535156, |
|
"rewards/rejected": -206.6787567138672, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.43956043956043955, |
|
"grad_norm": 16.745493319885668, |
|
"learning_rate": 3.4424441382108826e-07, |
|
"logits/chosen": 5873.52099609375, |
|
"logits/rejected": 5474.826171875, |
|
"logps/chosen": -316.10479736328125, |
|
"logps/rejected": -369.5398864746094, |
|
"loss": 0.6278, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -112.82951354980469, |
|
"rewards/margins": 63.50117874145508, |
|
"rewards/rejected": -176.33071899414062, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.4474097331240188, |
|
"grad_norm": 20.733746374707195, |
|
"learning_rate": 3.378620382651523e-07, |
|
"logits/chosen": 6166.53125, |
|
"logits/rejected": 5672.7578125, |
|
"logps/chosen": -361.2176513671875, |
|
"logps/rejected": -411.9073791503906, |
|
"loss": 0.6119, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -118.25773620605469, |
|
"rewards/margins": 69.87909698486328, |
|
"rewards/rejected": -188.13682556152344, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.4552590266875981, |
|
"grad_norm": 39.46262291218352, |
|
"learning_rate": 3.314136382905234e-07, |
|
"logits/chosen": 6161.3798828125, |
|
"logits/rejected": 5444.43408203125, |
|
"logps/chosen": -331.9513854980469, |
|
"logps/rejected": -450.3265686035156, |
|
"loss": 0.623, |
|
"rewards/accuracies": 0.8333333134651184, |
|
"rewards/chosen": -119.53593444824219, |
|
"rewards/margins": 117.47932434082031, |
|
"rewards/rejected": -237.01522827148438, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.4631083202511774, |
|
"grad_norm": 16.486304218382795, |
|
"learning_rate": 3.249040595833274e-07, |
|
"logits/chosen": 6643.578125, |
|
"logits/rejected": 5511.3427734375, |
|
"logps/chosen": -372.27203369140625, |
|
"logps/rejected": -422.9007873535156, |
|
"loss": 0.6139, |
|
"rewards/accuracies": 0.7916666865348816, |
|
"rewards/chosen": -140.54617309570312, |
|
"rewards/margins": 103.49578857421875, |
|
"rewards/rejected": -244.04196166992188, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.47095761381475665, |
|
"grad_norm": 20.09270587846568, |
|
"learning_rate": 3.1833819380279023e-07, |
|
"logits/chosen": 6314.1455078125, |
|
"logits/rejected": 5327.32275390625, |
|
"logps/chosen": -312.32403564453125, |
|
"logps/rejected": -400.52081298828125, |
|
"loss": 0.6082, |
|
"rewards/accuracies": 0.783333420753479, |
|
"rewards/chosen": -134.77569580078125, |
|
"rewards/margins": 76.05888366699219, |
|
"rewards/rejected": -210.83456420898438, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.478806907378336, |
|
"grad_norm": 13.1370939820266, |
|
"learning_rate": 3.11720974905373e-07, |
|
"logits/chosen": 6039.34130859375, |
|
"logits/rejected": 5201.23388671875, |
|
"logps/chosen": -330.3988952636719, |
|
"logps/rejected": -401.22540283203125, |
|
"loss": 0.6108, |
|
"rewards/accuracies": 0.7916666865348816, |
|
"rewards/chosen": -124.28079986572266, |
|
"rewards/margins": 91.52204895019531, |
|
"rewards/rejected": -215.80282592773438, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.48665620094191525, |
|
"grad_norm": 13.32277955977118, |
|
"learning_rate": 3.0505737543712275e-07, |
|
"logits/chosen": 5050.4599609375, |
|
"logits/rejected": 4102.0205078125, |
|
"logps/chosen": -343.1001281738281, |
|
"logps/rejected": -393.78594970703125, |
|
"loss": 0.6182, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -159.34146118164062, |
|
"rewards/margins": 73.34559631347656, |
|
"rewards/rejected": -232.6870574951172, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.4945054945054945, |
|
"grad_norm": 20.488144807475322, |
|
"learning_rate": 2.9835240279702513e-07, |
|
"logits/chosen": 6585.72119140625, |
|
"logits/rejected": 5542.74755859375, |
|
"logps/chosen": -375.4349365234375, |
|
"logps/rejected": -429.8561096191406, |
|
"loss": 0.6128, |
|
"rewards/accuracies": 0.8083333969116211, |
|
"rewards/chosen": -132.61892700195312, |
|
"rewards/margins": 102.98915100097656, |
|
"rewards/rejected": -235.6080322265625, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.5023547880690737, |
|
"grad_norm": 15.285970338736266, |
|
"learning_rate": 2.9161109547416667e-07, |
|
"logits/chosen": 6281.32177734375, |
|
"logits/rejected": 5335.0029296875, |
|
"logps/chosen": -345.87286376953125, |
|
"logps/rejected": -405.34576416015625, |
|
"loss": 0.6137, |
|
"rewards/accuracies": 0.7000000476837158, |
|
"rewards/chosen": -135.1112823486328, |
|
"rewards/margins": 55.7255859375, |
|
"rewards/rejected": -190.83685302734375, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.5102040816326531, |
|
"grad_norm": 13.467647025010642, |
|
"learning_rate": 2.848385192615339e-07, |
|
"logits/chosen": 5367.36474609375, |
|
"logits/rejected": 4323.5107421875, |
|
"logps/chosen": -317.5534362792969, |
|
"logps/rejected": -361.20721435546875, |
|
"loss": 0.6173, |
|
"rewards/accuracies": 0.7750000357627869, |
|
"rewards/chosen": -121.21412658691406, |
|
"rewards/margins": 71.26534271240234, |
|
"rewards/rejected": -192.47947692871094, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.5180533751962323, |
|
"grad_norm": 16.80725232592256, |
|
"learning_rate": 2.780397634492949e-07, |
|
"logits/chosen": 6023.45703125, |
|
"logits/rejected": 4714.66015625, |
|
"logps/chosen": -344.6996154785156, |
|
"logps/rejected": -413.51885986328125, |
|
"loss": 0.6221, |
|
"rewards/accuracies": 0.841666579246521, |
|
"rewards/chosen": -127.31312561035156, |
|
"rewards/margins": 98.25833892822266, |
|
"rewards/rejected": -225.57144165039062, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.5259026687598116, |
|
"grad_norm": 25.673672433544294, |
|
"learning_rate": 2.71219937000424e-07, |
|
"logits/chosen": 6010.3349609375, |
|
"logits/rejected": 4891.86328125, |
|
"logps/chosen": -337.61956787109375, |
|
"logps/rejected": -385.28363037109375, |
|
"loss": 0.6274, |
|
"rewards/accuracies": 0.783333420753479, |
|
"rewards/chosen": -130.86009216308594, |
|
"rewards/margins": 65.56916809082031, |
|
"rewards/rejected": -196.42926025390625, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.533751962323391, |
|
"grad_norm": 18.590503250558747, |
|
"learning_rate": 2.6438416471154273e-07, |
|
"logits/chosen": 5842.89208984375, |
|
"logits/rejected": 4827.79345703125, |
|
"logps/chosen": -347.49615478515625, |
|
"logps/rejected": -392.47955322265625, |
|
"loss": 0.6184, |
|
"rewards/accuracies": 0.7916666269302368, |
|
"rewards/chosen": -133.83489990234375, |
|
"rewards/margins": 79.41798400878906, |
|
"rewards/rejected": -213.2528839111328, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.5416012558869702, |
|
"grad_norm": 17.15200634352273, |
|
"learning_rate": 2.5753758336186326e-07, |
|
"logits/chosen": 5809.44482421875, |
|
"logits/rejected": 5264.39306640625, |
|
"logps/chosen": -337.9293212890625, |
|
"logps/rejected": -433.0084533691406, |
|
"loss": 0.6031, |
|
"rewards/accuracies": 0.7666666507720947, |
|
"rewards/chosen": -128.82289123535156, |
|
"rewards/margins": 87.98890686035156, |
|
"rewards/rejected": -216.811767578125, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 0.5494505494505495, |
|
"grad_norm": 15.797111525049983, |
|
"learning_rate": 2.5068533785312666e-07, |
|
"logits/chosen": 5499.59619140625, |
|
"logits/rejected": 5237.15771484375, |
|
"logps/chosen": -330.9737854003906, |
|
"logps/rejected": -413.58660888671875, |
|
"loss": 0.6328, |
|
"rewards/accuracies": 0.7083333730697632, |
|
"rewards/chosen": -141.71347045898438, |
|
"rewards/margins": 84.89921569824219, |
|
"rewards/rejected": -226.61270141601562, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.5572998430141287, |
|
"grad_norm": 15.51787747131488, |
|
"learning_rate": 2.4383257734343794e-07, |
|
"logits/chosen": 5412.13134765625, |
|
"logits/rejected": 5404.90283203125, |
|
"logps/chosen": -334.6061706542969, |
|
"logps/rejected": -422.1012268066406, |
|
"loss": 0.6078, |
|
"rewards/accuracies": 0.7500000596046448, |
|
"rewards/chosen": -140.69480895996094, |
|
"rewards/margins": 78.7857666015625, |
|
"rewards/rejected": -219.48056030273438, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 0.565149136577708, |
|
"grad_norm": 19.301105032278493, |
|
"learning_rate": 2.3698445137790258e-07, |
|
"logits/chosen": 5852.1748046875, |
|
"logits/rejected": 4949.2099609375, |
|
"logps/chosen": -339.65594482421875, |
|
"logps/rejected": -409.58099365234375, |
|
"loss": 0.6125, |
|
"rewards/accuracies": 0.7500000596046448, |
|
"rewards/chosen": -120.65275573730469, |
|
"rewards/margins": 89.16310119628906, |
|
"rewards/rejected": -209.8158416748047, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 0.5729984301412873, |
|
"grad_norm": 17.264432929616067, |
|
"learning_rate": 2.3014610601897157e-07, |
|
"logits/chosen": 6345.07080078125, |
|
"logits/rejected": 4777.46826171875, |
|
"logps/chosen": -357.9107971191406, |
|
"logps/rejected": -393.86749267578125, |
|
"loss": 0.6193, |
|
"rewards/accuracies": 0.7666667699813843, |
|
"rewards/chosen": -132.76486206054688, |
|
"rewards/margins": 86.02892303466797, |
|
"rewards/rejected": -218.79379272460938, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 0.5808477237048666, |
|
"grad_norm": 16.99769270663946, |
|
"learning_rate": 2.2332267997940513e-07, |
|
"logits/chosen": 5260.65673828125, |
|
"logits/rejected": 4413.0966796875, |
|
"logps/chosen": -320.5991516113281, |
|
"logps/rejected": -370.4900817871094, |
|
"loss": 0.6068, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -128.17471313476562, |
|
"rewards/margins": 79.66935729980469, |
|
"rewards/rejected": -207.8440704345703, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 0.5886970172684458, |
|
"grad_norm": 19.890647703266293, |
|
"learning_rate": 2.1651930076075723e-07, |
|
"logits/chosen": 5752.275390625, |
|
"logits/rejected": 5127.4677734375, |
|
"logps/chosen": -327.98681640625, |
|
"logps/rejected": -381.2528381347656, |
|
"loss": 0.6251, |
|
"rewards/accuracies": 0.73333340883255, |
|
"rewards/chosen": -147.56088256835938, |
|
"rewards/margins": 66.59766387939453, |
|
"rewards/rejected": -214.1585693359375, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.5965463108320251, |
|
"grad_norm": 14.729863717931694, |
|
"learning_rate": 2.0974108080028692e-07, |
|
"logits/chosen": 6052.0751953125, |
|
"logits/rejected": 4696.7666015625, |
|
"logps/chosen": -337.3818664550781, |
|
"logps/rejected": -385.4659118652344, |
|
"loss": 0.6144, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -134.1309051513672, |
|
"rewards/margins": 76.8044662475586, |
|
"rewards/rejected": -210.9353790283203, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 0.6043956043956044, |
|
"grad_norm": 17.668089788843886, |
|
"learning_rate": 2.0299311362918773e-07, |
|
"logits/chosen": 6281.421875, |
|
"logits/rejected": 5333.64794921875, |
|
"logps/chosen": -369.94805908203125, |
|
"logps/rejected": -441.9542541503906, |
|
"loss": 0.6302, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -140.21438598632812, |
|
"rewards/margins": 77.87971496582031, |
|
"rewards/rejected": -218.09408569335938, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 0.6122448979591837, |
|
"grad_norm": 15.214960693707779, |
|
"learning_rate": 1.962804700450265e-07, |
|
"logits/chosen": 6088.87744140625, |
|
"logits/rejected": 5759.99560546875, |
|
"logps/chosen": -357.53375244140625, |
|
"logps/rejected": -451.697509765625, |
|
"loss": 0.6171, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -141.428466796875, |
|
"rewards/margins": 72.88134002685547, |
|
"rewards/rejected": -214.30978393554688, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 0.6200941915227629, |
|
"grad_norm": 32.04354354649456, |
|
"learning_rate": 1.8960819430126334e-07, |
|
"logits/chosen": 5662.08056640625, |
|
"logits/rejected": 4959.47998046875, |
|
"logps/chosen": -362.9315490722656, |
|
"logps/rejected": -448.6405334472656, |
|
"loss": 0.6126, |
|
"rewards/accuracies": 0.783333420753479, |
|
"rewards/chosen": -164.80050659179688, |
|
"rewards/margins": 95.24226379394531, |
|
"rewards/rejected": -260.04278564453125, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 0.6279434850863422, |
|
"grad_norm": 22.34135272884677, |
|
"learning_rate": 1.8298130031671972e-07, |
|
"logits/chosen": 5636.14599609375, |
|
"logits/rejected": 4883.76123046875, |
|
"logps/chosen": -372.5877380371094, |
|
"logps/rejected": -449.95355224609375, |
|
"loss": 0.6251, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -155.86485290527344, |
|
"rewards/margins": 81.3904800415039, |
|
"rewards/rejected": -237.25534057617188, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.6357927786499215, |
|
"grad_norm": 22.27973267170805, |
|
"learning_rate": 1.7640476790784075e-07, |
|
"logits/chosen": 5207.5654296875, |
|
"logits/rejected": 4584.2197265625, |
|
"logps/chosen": -348.1822204589844, |
|
"logps/rejected": -463.4768981933594, |
|
"loss": 0.6154, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -145.72683715820312, |
|
"rewards/margins": 96.45280456542969, |
|
"rewards/rejected": -242.1796417236328, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 0.6436420722135008, |
|
"grad_norm": 21.878335066672264, |
|
"learning_rate": 1.6988353904658492e-07, |
|
"logits/chosen": 5700.16650390625, |
|
"logits/rejected": 4322.6650390625, |
|
"logps/chosen": -365.1958923339844, |
|
"logps/rejected": -391.56353759765625, |
|
"loss": 0.6154, |
|
"rewards/accuracies": 0.7666666507720947, |
|
"rewards/chosen": -143.98133850097656, |
|
"rewards/margins": 78.89381408691406, |
|
"rewards/rejected": -222.8751983642578, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 0.6514913657770801, |
|
"grad_norm": 30.16348408203026, |
|
"learning_rate": 1.634225141467513e-07, |
|
"logits/chosen": 5625.34326171875, |
|
"logits/rejected": 4964.39306640625, |
|
"logps/chosen": -347.97393798828125, |
|
"logps/rejected": -428.678466796875, |
|
"loss": 0.613, |
|
"rewards/accuracies": 0.7750000357627869, |
|
"rewards/chosen": -143.7879180908203, |
|
"rewards/margins": 92.43571472167969, |
|
"rewards/rejected": -236.2236328125, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 0.6593406593406593, |
|
"grad_norm": 12.309600681345202, |
|
"learning_rate": 1.570265483815364e-07, |
|
"logits/chosen": 6213.0498046875, |
|
"logits/rejected": 5020.015625, |
|
"logps/chosen": -358.1226806640625, |
|
"logps/rejected": -418.3009338378906, |
|
"loss": 0.618, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -130.79283142089844, |
|
"rewards/margins": 74.71489715576172, |
|
"rewards/rejected": -205.5077362060547, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 0.6671899529042387, |
|
"grad_norm": 12.13079365289034, |
|
"learning_rate": 1.5070044803508691e-07, |
|
"logits/chosen": 5740.8427734375, |
|
"logits/rejected": 5088.7275390625, |
|
"logps/chosen": -333.6675720214844, |
|
"logps/rejected": -408.93280029296875, |
|
"loss": 0.5962, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -118.1900863647461, |
|
"rewards/margins": 90.28736877441406, |
|
"rewards/rejected": -208.4774627685547, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 0.6750392464678179, |
|
"grad_norm": 17.975318476548516, |
|
"learning_rate": 1.444489668907914e-07, |
|
"logits/chosen": 6176.43212890625, |
|
"logits/rejected": 5197.5263671875, |
|
"logps/chosen": -364.1601867675781, |
|
"logps/rejected": -395.86810302734375, |
|
"loss": 0.6293, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -116.38951110839844, |
|
"rewards/margins": 76.79676818847656, |
|
"rewards/rejected": -193.186279296875, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 0.6828885400313972, |
|
"grad_norm": 18.675632935473597, |
|
"learning_rate": 1.3827680265902232e-07, |
|
"logits/chosen": 6169.16015625, |
|
"logits/rejected": 5053.54638671875, |
|
"logps/chosen": -336.77423095703125, |
|
"logps/rejected": -384.07647705078125, |
|
"loss": 0.6181, |
|
"rewards/accuracies": 0.7833333611488342, |
|
"rewards/chosen": -106.67745208740234, |
|
"rewards/margins": 76.71006774902344, |
|
"rewards/rejected": -183.38754272460938, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 0.6907378335949764, |
|
"grad_norm": 15.556936298105454, |
|
"learning_rate": 1.3218859344701632e-07, |
|
"logits/chosen": 5415.5537109375, |
|
"logits/rejected": 5134.35888671875, |
|
"logps/chosen": -307.6910400390625, |
|
"logps/rejected": -403.2095947265625, |
|
"loss": 0.6154, |
|
"rewards/accuracies": 0.7583333849906921, |
|
"rewards/chosen": -97.1800765991211, |
|
"rewards/margins": 74.44038391113281, |
|
"rewards/rejected": -171.62045288085938, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 0.6985871271585558, |
|
"grad_norm": 14.598418420491413, |
|
"learning_rate": 1.2618891427354172e-07, |
|
"logits/chosen": 6405.8955078125, |
|
"logits/rejected": 5148.23828125, |
|
"logps/chosen": -354.01531982421875, |
|
"logps/rejected": -396.88348388671875, |
|
"loss": 0.6246, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -96.41930389404297, |
|
"rewards/margins": 88.68364715576172, |
|
"rewards/rejected": -185.1029510498047, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 0.706436420722135, |
|
"grad_norm": 17.248604251826578, |
|
"learning_rate": 1.202822736309758e-07, |
|
"logits/chosen": 5427.36181640625, |
|
"logits/rejected": 4946.220703125, |
|
"logps/chosen": -309.7834777832031, |
|
"logps/rejected": -402.7059326171875, |
|
"loss": 0.6169, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -105.43426513671875, |
|
"rewards/margins": 86.841552734375, |
|
"rewards/rejected": -192.27581787109375, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.7142857142857143, |
|
"grad_norm": 21.236660119285595, |
|
"learning_rate": 1.1447311009737299e-07, |
|
"logits/chosen": 5334.93408203125, |
|
"logits/rejected": 4981.6474609375, |
|
"logps/chosen": -332.61944580078125, |
|
"logps/rejected": -423.7569274902344, |
|
"loss": 0.6253, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -124.39315032958984, |
|
"rewards/margins": 91.43525695800781, |
|
"rewards/rejected": -215.8284149169922, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 0.7221350078492935, |
|
"grad_norm": 14.521571204964287, |
|
"learning_rate": 1.0876578900107053e-07, |
|
"logits/chosen": 5887.0625, |
|
"logits/rejected": 4832.818359375, |
|
"logps/chosen": -360.1144714355469, |
|
"logps/rejected": -408.10321044921875, |
|
"loss": 0.6099, |
|
"rewards/accuracies": 0.824999988079071, |
|
"rewards/chosen": -127.6928939819336, |
|
"rewards/margins": 82.26800537109375, |
|
"rewards/rejected": -209.9608917236328, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 0.7299843014128728, |
|
"grad_norm": 12.269584700329853, |
|
"learning_rate": 1.0316459914033793e-07, |
|
"logits/chosen": 5838.90185546875, |
|
"logits/rejected": 4279.92529296875, |
|
"logps/chosen": -368.8111877441406, |
|
"logps/rejected": -407.0862121582031, |
|
"loss": 0.6068, |
|
"rewards/accuracies": 0.7833333015441895, |
|
"rewards/chosen": -132.82225036621094, |
|
"rewards/margins": 88.33971405029297, |
|
"rewards/rejected": -221.16195678710938, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 0.7378335949764521, |
|
"grad_norm": 16.95040924011293, |
|
"learning_rate": 9.767374956053584e-08, |
|
"logits/chosen": 5631.70751953125, |
|
"logits/rejected": 4867.74560546875, |
|
"logps/chosen": -354.5127868652344, |
|
"logps/rejected": -444.9837951660156, |
|
"loss": 0.6195, |
|
"rewards/accuracies": 0.7666667699813843, |
|
"rewards/chosen": -136.21914672851562, |
|
"rewards/margins": 103.7847671508789, |
|
"rewards/rejected": -240.00390625, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 0.7456828885400314, |
|
"grad_norm": 15.455216526066362, |
|
"learning_rate": 9.229736639120561e-08, |
|
"logits/chosen": 5815.58251953125, |
|
"logits/rejected": 5308.6826171875, |
|
"logps/chosen": -353.8769836425781, |
|
"logps/rejected": -416.19110107421875, |
|
"loss": 0.6285, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -136.0758056640625, |
|
"rewards/margins": 65.96086120605469, |
|
"rewards/rejected": -202.03665161132812, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 0.7535321821036107, |
|
"grad_norm": 21.025656579521215, |
|
"learning_rate": 8.70394897454659e-08, |
|
"logits/chosen": 5706.39208984375, |
|
"logits/rejected": 5004.17041015625, |
|
"logps/chosen": -332.70684814453125, |
|
"logps/rejected": -410.22308349609375, |
|
"loss": 0.6127, |
|
"rewards/accuracies": 0.8083332777023315, |
|
"rewards/chosen": -114.27842712402344, |
|
"rewards/margins": 93.46832275390625, |
|
"rewards/rejected": -207.7467498779297, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 0.7613814756671899, |
|
"grad_norm": 22.662244586868777, |
|
"learning_rate": 8.19040706840472e-08, |
|
"logits/chosen": 5792.12451171875, |
|
"logits/rejected": 4783.9365234375, |
|
"logps/chosen": -364.58013916015625, |
|
"logps/rejected": -429.33734130859375, |
|
"loss": 0.6128, |
|
"rewards/accuracies": 0.7500000596046448, |
|
"rewards/chosen": -124.36016845703125, |
|
"rewards/margins": 98.28662872314453, |
|
"rewards/rejected": -222.6468048095703, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 0.7692307692307693, |
|
"grad_norm": 16.294899633616406, |
|
"learning_rate": 7.689496824624525e-08, |
|
"logits/chosen": 5524.9208984375, |
|
"logits/rejected": 4383.34033203125, |
|
"logps/chosen": -343.7428283691406, |
|
"logps/rejected": -411.6790466308594, |
|
"loss": 0.6013, |
|
"rewards/accuracies": 0.7833333611488342, |
|
"rewards/chosen": -114.92622375488281, |
|
"rewards/margins": 106.25920104980469, |
|
"rewards/rejected": -221.18539428710938, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 0.7770800627943485, |
|
"grad_norm": 24.656802253568358, |
|
"learning_rate": 7.201594655002458e-08, |
|
"logits/chosen": 5839.876953125, |
|
"logits/rejected": 4818.5595703125, |
|
"logps/chosen": -359.73309326171875, |
|
"logps/rejected": -420.4481506347656, |
|
"loss": 0.6001, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -133.83816528320312, |
|
"rewards/margins": 92.54021453857422, |
|
"rewards/rejected": -226.3783721923828, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 0.7849293563579278, |
|
"grad_norm": 19.139612377869543, |
|
"learning_rate": 6.727067196345099e-08, |
|
"logits/chosen": 5519.62890625, |
|
"logits/rejected": 4601.50830078125, |
|
"logps/chosen": -347.6911315917969, |
|
"logps/rejected": -392.0986328125, |
|
"loss": 0.6166, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -133.935302734375, |
|
"rewards/margins": 77.81806945800781, |
|
"rewards/rejected": -211.7533416748047, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.792778649921507, |
|
"grad_norm": 18.4502462246621, |
|
"learning_rate": 6.26627103495786e-08, |
|
"logits/chosen": 5698.8525390625, |
|
"logits/rejected": 4684.4365234375, |
|
"logps/chosen": -344.99969482421875, |
|
"logps/rejected": -406.09136962890625, |
|
"loss": 0.626, |
|
"rewards/accuracies": 0.7166666984558105, |
|
"rewards/chosen": -130.29049682617188, |
|
"rewards/margins": 84.36732482910156, |
|
"rewards/rejected": -214.6577911376953, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 0.8006279434850864, |
|
"grad_norm": 13.030843283044183, |
|
"learning_rate": 5.8195524386862374e-08, |
|
"logits/chosen": 5804.3681640625, |
|
"logits/rejected": 5112.267578125, |
|
"logps/chosen": -367.52532958984375, |
|
"logps/rejected": -445.23822021484375, |
|
"loss": 0.6133, |
|
"rewards/accuracies": 0.8083332777023315, |
|
"rewards/chosen": -119.3467025756836, |
|
"rewards/margins": 99.81084442138672, |
|
"rewards/rejected": -219.15756225585938, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 0.8084772370486656, |
|
"grad_norm": 15.435517078983278, |
|
"learning_rate": 5.38724709671092e-08, |
|
"logits/chosen": 6208.55615234375, |
|
"logits/rejected": 5785.6513671875, |
|
"logps/chosen": -368.04693603515625, |
|
"logps/rejected": -461.67291259765625, |
|
"loss": 0.6042, |
|
"rewards/accuracies": 0.783333420753479, |
|
"rewards/chosen": -136.5785675048828, |
|
"rewards/margins": 91.97505950927734, |
|
"rewards/rejected": -228.5536346435547, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 0.8163265306122449, |
|
"grad_norm": 19.196729996375815, |
|
"learning_rate": 4.969679867292276e-08, |
|
"logits/chosen": 5469.60595703125, |
|
"logits/rejected": 4901.0498046875, |
|
"logps/chosen": -360.51763916015625, |
|
"logps/rejected": -447.8094177246094, |
|
"loss": 0.6096, |
|
"rewards/accuracies": 0.73333340883255, |
|
"rewards/chosen": -140.41592407226562, |
|
"rewards/margins": 98.20952606201172, |
|
"rewards/rejected": -238.62545776367188, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 0.8241758241758241, |
|
"grad_norm": 15.63885700925835, |
|
"learning_rate": 4.5671645336537416e-08, |
|
"logits/chosen": 5541.3017578125, |
|
"logits/rejected": 4979.87890625, |
|
"logps/chosen": -379.06878662109375, |
|
"logps/rejected": -450.43927001953125, |
|
"loss": 0.6099, |
|
"rewards/accuracies": 0.7833333611488342, |
|
"rewards/chosen": -141.58819580078125, |
|
"rewards/margins": 93.05669403076172, |
|
"rewards/rejected": -234.6448974609375, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 0.8320251177394035, |
|
"grad_norm": 71.6555736263913, |
|
"learning_rate": 4.180003568187776e-08, |
|
"logits/chosen": 6864.1591796875, |
|
"logits/rejected": 5329.9736328125, |
|
"logps/chosen": -406.0521240234375, |
|
"logps/rejected": -433.7730407714844, |
|
"loss": 0.6267, |
|
"rewards/accuracies": 0.7083333730697632, |
|
"rewards/chosen": -145.0647430419922, |
|
"rewards/margins": 68.66908264160156, |
|
"rewards/rejected": -213.7338409423828, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 0.8398744113029827, |
|
"grad_norm": 17.29969333705125, |
|
"learning_rate": 3.8084879051612144e-08, |
|
"logits/chosen": 5685.4697265625, |
|
"logits/rejected": 5125.021484375, |
|
"logps/chosen": -358.2772521972656, |
|
"logps/rejected": -418.6520080566406, |
|
"loss": 0.6134, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -133.0342559814453, |
|
"rewards/margins": 92.45338439941406, |
|
"rewards/rejected": -225.48764038085938, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 0.847723704866562, |
|
"grad_norm": 16.62350096619966, |
|
"learning_rate": 3.452896722091128e-08, |
|
"logits/chosen": 6247.75927734375, |
|
"logits/rejected": 4747.95556640625, |
|
"logps/chosen": -398.6324462890625, |
|
"logps/rejected": -429.2178649902344, |
|
"loss": 0.6032, |
|
"rewards/accuracies": 0.7833333611488342, |
|
"rewards/chosen": -132.06317138671875, |
|
"rewards/margins": 96.17476654052734, |
|
"rewards/rejected": -228.23794555664062, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 0.8555729984301413, |
|
"grad_norm": 20.56158228380985, |
|
"learning_rate": 3.11349722995527e-08, |
|
"logits/chosen": 6290.5673828125, |
|
"logits/rejected": 4655.23974609375, |
|
"logps/chosen": -363.40936279296875, |
|
"logps/rejected": -426.296630859375, |
|
"loss": 0.6032, |
|
"rewards/accuracies": 0.7333332896232605, |
|
"rewards/chosen": -135.9347381591797, |
|
"rewards/margins": 77.47732543945312, |
|
"rewards/rejected": -213.4120635986328, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 0.8634222919937206, |
|
"grad_norm": 17.638845220024372, |
|
"learning_rate": 2.7905444723949762e-08, |
|
"logits/chosen": 6115.4248046875, |
|
"logits/rejected": 4982.52685546875, |
|
"logps/chosen": -377.2098388671875, |
|
"logps/rejected": -429.16619873046875, |
|
"loss": 0.6167, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -138.67349243164062, |
|
"rewards/margins": 100.99529266357422, |
|
"rewards/rejected": -239.66879272460938, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.8712715855572999, |
|
"grad_norm": 18.82352427435552, |
|
"learning_rate": 2.484281134061142e-08, |
|
"logits/chosen": 6460.3466796875, |
|
"logits/rejected": 5133.16796875, |
|
"logps/chosen": -407.51690673828125, |
|
"logps/rejected": -459.20721435546875, |
|
"loss": 0.608, |
|
"rewards/accuracies": 0.7916666269302368, |
|
"rewards/chosen": -142.90335083007812, |
|
"rewards/margins": 94.03584289550781, |
|
"rewards/rejected": -236.939208984375, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 0.8791208791208791, |
|
"grad_norm": 38.731872878926325, |
|
"learning_rate": 2.194937358247506e-08, |
|
"logits/chosen": 6330.7939453125, |
|
"logits/rejected": 5062.67822265625, |
|
"logps/chosen": -387.6415100097656, |
|
"logps/rejected": -456.4393615722656, |
|
"loss": 0.611, |
|
"rewards/accuracies": 0.7500001192092896, |
|
"rewards/chosen": -142.5010223388672, |
|
"rewards/margins": 97.16749572753906, |
|
"rewards/rejected": -239.6685333251953, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 0.8869701726844584, |
|
"grad_norm": 21.93873798257166, |
|
"learning_rate": 1.9227305739481612e-08, |
|
"logits/chosen": 5757.74462890625, |
|
"logits/rejected": 4450.82373046875, |
|
"logps/chosen": -360.2721862792969, |
|
"logps/rejected": -406.9669189453125, |
|
"loss": 0.6027, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -125.49015045166016, |
|
"rewards/margins": 101.2511978149414, |
|
"rewards/rejected": -226.7413330078125, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 0.8948194662480377, |
|
"grad_norm": 21.229892802861798, |
|
"learning_rate": 1.6678653324693787e-08, |
|
"logits/chosen": 6381.58935546875, |
|
"logits/rejected": 5084.40478515625, |
|
"logps/chosen": -392.08349609375, |
|
"logps/rejected": -447.29827880859375, |
|
"loss": 0.5978, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -134.46275329589844, |
|
"rewards/margins": 92.02247619628906, |
|
"rewards/rejected": -226.48526000976562, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 0.902668759811617, |
|
"grad_norm": 11.998405903395946, |
|
"learning_rate": 1.4305331537183384e-08, |
|
"logits/chosen": 5625.76708984375, |
|
"logits/rejected": 5091.146484375, |
|
"logps/chosen": -361.97503662109375, |
|
"logps/rejected": -438.71697998046875, |
|
"loss": 0.6003, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -134.9916229248047, |
|
"rewards/margins": 84.23737335205078, |
|
"rewards/rejected": -219.22900390625, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 0.9105180533751962, |
|
"grad_norm": 20.001421395632878, |
|
"learning_rate": 1.2109123822844653e-08, |
|
"logits/chosen": 5778.00634765625, |
|
"logits/rejected": 4505.45849609375, |
|
"logps/chosen": -367.82598876953125, |
|
"logps/rejected": -416.082275390625, |
|
"loss": 0.6137, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -139.2692108154297, |
|
"rewards/margins": 80.87498474121094, |
|
"rewards/rejected": -220.14419555664062, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 0.9183673469387755, |
|
"grad_norm": 18.383575962086507, |
|
"learning_rate": 1.0091680534213387e-08, |
|
"logits/chosen": 6360.2177734375, |
|
"logits/rejected": 6036.2529296875, |
|
"logps/chosen": -372.2303771972656, |
|
"logps/rejected": -475.2777404785156, |
|
"loss": 0.6101, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -129.28201293945312, |
|
"rewards/margins": 99.24874877929688, |
|
"rewards/rejected": -228.53073120117188, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 0.9262166405023547, |
|
"grad_norm": 26.20378993221473, |
|
"learning_rate": 8.254517690300944e-09, |
|
"logits/chosen": 5551.3759765625, |
|
"logits/rejected": 4948.4052734375, |
|
"logps/chosen": -368.40777587890625, |
|
"logps/rejected": -442.9326171875, |
|
"loss": 0.6049, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -128.46182250976562, |
|
"rewards/margins": 97.64901733398438, |
|
"rewards/rejected": -226.11083984375, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 0.9340659340659341, |
|
"grad_norm": 12.632107820641844, |
|
"learning_rate": 6.599015837372907e-09, |
|
"logits/chosen": 6045.20654296875, |
|
"logits/rejected": 5189.9599609375, |
|
"logps/chosen": -404.6395568847656, |
|
"logps/rejected": -457.1123046875, |
|
"loss": 0.6143, |
|
"rewards/accuracies": 0.7333332896232605, |
|
"rewards/chosen": -155.75405883789062, |
|
"rewards/margins": 84.20575714111328, |
|
"rewards/rejected": -239.95980834960938, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 0.9419152276295133, |
|
"grad_norm": 32.82134972014968, |
|
"learning_rate": 5.126419011529992e-09, |
|
"logits/chosen": 6301.2685546875, |
|
"logits/rejected": 5249.4482421875, |
|
"logps/chosen": -383.8853454589844, |
|
"logps/rejected": -460.536865234375, |
|
"loss": 0.6054, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": -128.22366333007812, |
|
"rewards/margins": 113.64762878417969, |
|
"rewards/rejected": -241.8712921142578, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.9497645211930926, |
|
"grad_norm": 22.184372964812642, |
|
"learning_rate": 3.837833803870177e-09, |
|
"logits/chosen": 5859.86279296875, |
|
"logits/rejected": 5063.07080078125, |
|
"logps/chosen": -382.39556884765625, |
|
"logps/rejected": -460.39459228515625, |
|
"loss": 0.6185, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -142.2100830078125, |
|
"rewards/margins": 99.98322296142578, |
|
"rewards/rejected": -242.1933135986328, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 0.957613814756672, |
|
"grad_norm": 24.474760694660887, |
|
"learning_rate": 2.734228528934679e-09, |
|
"logits/chosen": 7345.8125, |
|
"logits/rejected": 5270.705078125, |
|
"logps/chosen": -443.3330993652344, |
|
"logps/rejected": -489.99658203125, |
|
"loss": 0.6063, |
|
"rewards/accuracies": 0.73333340883255, |
|
"rewards/chosen": -146.56466674804688, |
|
"rewards/margins": 98.23472595214844, |
|
"rewards/rejected": -244.7993621826172, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 0.9654631083202512, |
|
"grad_norm": 23.890847226085217, |
|
"learning_rate": 1.8164324970625645e-09, |
|
"logits/chosen": 6513.2109375, |
|
"logits/rejected": 5010.6240234375, |
|
"logps/chosen": -391.34930419921875, |
|
"logps/rejected": -439.3382263183594, |
|
"loss": 0.6264, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -130.76763916015625, |
|
"rewards/margins": 95.57390594482422, |
|
"rewards/rejected": -226.341552734375, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 0.9733124018838305, |
|
"grad_norm": 20.592256484362352, |
|
"learning_rate": 1.0851353912008642e-09, |
|
"logits/chosen": 5547.0556640625, |
|
"logits/rejected": 5059.66015625, |
|
"logps/chosen": -377.5496826171875, |
|
"logps/rejected": -469.046630859375, |
|
"loss": 0.6146, |
|
"rewards/accuracies": 0.7333332896232605, |
|
"rewards/chosen": -145.93728637695312, |
|
"rewards/margins": 87.95149230957031, |
|
"rewards/rejected": -233.88876342773438, |
|
"step": 1240 |
|
}, |
|
{ |
|
"epoch": 0.9811616954474097, |
|
"grad_norm": 15.042680729243902, |
|
"learning_rate": 5.408867486384471e-10, |
|
"logits/chosen": 5707.8701171875, |
|
"logits/rejected": 4727.94921875, |
|
"logps/chosen": -356.4093017578125, |
|
"logps/rejected": -407.896728515625, |
|
"loss": 0.6075, |
|
"rewards/accuracies": 0.7916666269302368, |
|
"rewards/chosen": -126.1011734008789, |
|
"rewards/margins": 92.36528778076172, |
|
"rewards/rejected": -218.46646118164062, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 0.989010989010989, |
|
"grad_norm": 17.914310013144693, |
|
"learning_rate": 1.840955480532924e-10, |
|
"logits/chosen": 5406.544921875, |
|
"logits/rejected": 5018.2861328125, |
|
"logps/chosen": -360.8941955566406, |
|
"logps/rejected": -435.3475646972656, |
|
"loss": 0.6055, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -128.73361206054688, |
|
"rewards/margins": 90.288818359375, |
|
"rewards/rejected": -219.02243041992188, |
|
"step": 1260 |
|
}, |
|
{ |
|
"epoch": 0.9968602825745683, |
|
"grad_norm": 19.6748421836041, |
|
"learning_rate": 1.502990218302247e-11, |
|
"logits/chosen": 5662.17626953125, |
|
"logits/rejected": 4531.2763671875, |
|
"logps/chosen": -363.6100158691406, |
|
"logps/rejected": -417.36590576171875, |
|
"loss": 0.6148, |
|
"rewards/accuracies": 0.7916666269302368, |
|
"rewards/chosen": -139.1360321044922, |
|
"rewards/margins": 91.9256362915039, |
|
"rewards/rejected": -231.06167602539062, |
|
"step": 1270 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"step": 1274, |
|
"total_flos": 0.0, |
|
"train_loss": 0.6281878652527718, |
|
"train_runtime": 14442.3019, |
|
"train_samples_per_second": 4.233, |
|
"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 |
|
} |
|
|