| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9984301412872841, | |
| "eval_steps": 1000, | |
| "global_step": 477, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.0416666666666666e-08, | |
| "logits/chosen": -2.7851884365081787, | |
| "logits/rejected": -2.649298906326294, | |
| "logps/chosen": -296.01092529296875, | |
| "logps/rejected": -290.09039306640625, | |
| "loss": 0.7406, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.083333333333333e-08, | |
| "logits/chosen": -2.681438446044922, | |
| "logits/rejected": -2.7385971546173096, | |
| "logps/chosen": -339.8030090332031, | |
| "logps/rejected": -247.9777374267578, | |
| "loss": 0.829, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.166666666666666e-08, | |
| "logits/chosen": -2.6828837394714355, | |
| "logits/rejected": -2.717893123626709, | |
| "logps/chosen": -246.95411682128906, | |
| "logps/rejected": -230.5216522216797, | |
| "loss": 0.7515, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -1.3528901035897434e-05, | |
| "rewards/margins": -0.00024310783192049712, | |
| "rewards/rejected": 0.00022957894543651491, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.25e-08, | |
| "logits/chosen": -2.6765449047088623, | |
| "logits/rejected": -2.7000393867492676, | |
| "logps/chosen": -300.23077392578125, | |
| "logps/rejected": -223.28292846679688, | |
| "loss": 0.8154, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -9.068308281712234e-05, | |
| "rewards/margins": -0.00012977616279385984, | |
| "rewards/rejected": 3.9093021769076586e-05, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.333333333333333e-08, | |
| "logits/chosen": -2.599257469177246, | |
| "logits/rejected": -2.6808290481567383, | |
| "logps/chosen": -276.9845886230469, | |
| "logps/rejected": -264.2096252441406, | |
| "loss": 0.7941, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.0009412524523213506, | |
| "rewards/margins": -0.00016639442765153944, | |
| "rewards/rejected": 0.0011076468508690596, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0416666666666667e-07, | |
| "logits/chosen": -2.6191229820251465, | |
| "logits/rejected": -2.5384933948516846, | |
| "logps/chosen": -259.14239501953125, | |
| "logps/rejected": -247.5746307373047, | |
| "loss": 0.8451, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.002248743548989296, | |
| "rewards/margins": 0.00199419679120183, | |
| "rewards/rejected": 0.0002545467286836356, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.25e-07, | |
| "logits/chosen": -2.6755356788635254, | |
| "logits/rejected": -2.6597650051116943, | |
| "logps/chosen": -240.25408935546875, | |
| "logps/rejected": -215.57801818847656, | |
| "loss": 0.7591, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": 0.003154001198709011, | |
| "rewards/margins": 0.0002609168004710227, | |
| "rewards/rejected": 0.002893084194511175, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.4583333333333335e-07, | |
| "logits/chosen": -2.7294678688049316, | |
| "logits/rejected": -2.6904141902923584, | |
| "logps/chosen": -315.6171875, | |
| "logps/rejected": -283.3052978515625, | |
| "loss": 0.7765, | |
| "rewards/accuracies": 0.546875, | |
| "rewards/chosen": 0.007693930994719267, | |
| "rewards/margins": 0.0004415067960508168, | |
| "rewards/rejected": 0.007252424024045467, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.6666666666666665e-07, | |
| "logits/chosen": -2.674433469772339, | |
| "logits/rejected": -2.635763168334961, | |
| "logps/chosen": -280.9161071777344, | |
| "logps/rejected": -286.4499816894531, | |
| "loss": 0.7959, | |
| "rewards/accuracies": 0.390625, | |
| "rewards/chosen": 0.010045092552900314, | |
| "rewards/margins": -0.0013794752303510904, | |
| "rewards/rejected": 0.011424567550420761, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.875e-07, | |
| "logits/chosen": -2.6775548458099365, | |
| "logits/rejected": -2.6488728523254395, | |
| "logps/chosen": -281.817138671875, | |
| "logps/rejected": -226.1727294921875, | |
| "loss": 0.7651, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": 0.021169912070035934, | |
| "rewards/margins": 0.003915595356374979, | |
| "rewards/rejected": 0.017254315316677094, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "logits/chosen": -2.6540913581848145, | |
| "logits/rejected": -2.652208089828491, | |
| "logps/chosen": -255.7197265625, | |
| "logps/rejected": -211.753662109375, | |
| "loss": 0.812, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": 0.031227679923176765, | |
| "rewards/margins": 0.004203509539365768, | |
| "rewards/rejected": 0.027024172246456146, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.2916666666666663e-07, | |
| "logits/chosen": -2.7285475730895996, | |
| "logits/rejected": -2.690129280090332, | |
| "logps/chosen": -283.28863525390625, | |
| "logps/rejected": -303.4035339355469, | |
| "loss": 0.8139, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.04402179270982742, | |
| "rewards/margins": 0.0007559025543741882, | |
| "rewards/rejected": 0.04326589033007622, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.5e-07, | |
| "logits/chosen": -2.625800609588623, | |
| "logits/rejected": -2.6034979820251465, | |
| "logps/chosen": -281.58795166015625, | |
| "logps/rejected": -234.80442810058594, | |
| "loss": 0.9627, | |
| "rewards/accuracies": 0.546875, | |
| "rewards/chosen": 0.063897043466568, | |
| "rewards/margins": 0.006137948948889971, | |
| "rewards/rejected": 0.05775909498333931, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.708333333333333e-07, | |
| "logits/chosen": -2.7099432945251465, | |
| "logits/rejected": -2.715182304382324, | |
| "logps/chosen": -313.54315185546875, | |
| "logps/rejected": -269.708984375, | |
| "loss": 0.7945, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": 0.09343694150447845, | |
| "rewards/margins": 0.011352106928825378, | |
| "rewards/rejected": 0.08208482712507248, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.916666666666667e-07, | |
| "logits/chosen": -2.683866024017334, | |
| "logits/rejected": -2.666459083557129, | |
| "logps/chosen": -289.9292297363281, | |
| "logps/rejected": -266.2529602050781, | |
| "loss": 0.963, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.11043840646743774, | |
| "rewards/margins": 0.009698072448372841, | |
| "rewards/rejected": 0.10074033588171005, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": -2.639746904373169, | |
| "logits/rejected": -2.635244369506836, | |
| "logps/chosen": -250.9815673828125, | |
| "logps/rejected": -256.66644287109375, | |
| "loss": 0.782, | |
| "rewards/accuracies": 0.609375, | |
| "rewards/chosen": 0.11398710310459137, | |
| "rewards/margins": 0.006070964969694614, | |
| "rewards/rejected": 0.10791613161563873, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.333333333333333e-07, | |
| "logits/chosen": -2.6974880695343018, | |
| "logits/rejected": -2.791822910308838, | |
| "logps/chosen": -311.99462890625, | |
| "logps/rejected": -263.93084716796875, | |
| "loss": 0.7708, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.13753816485404968, | |
| "rewards/margins": 0.02569321170449257, | |
| "rewards/rejected": 0.11184494942426682, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.541666666666667e-07, | |
| "logits/chosen": -2.683668613433838, | |
| "logits/rejected": -2.696622610092163, | |
| "logps/chosen": -290.19091796875, | |
| "logps/rejected": -280.1837158203125, | |
| "loss": 0.8665, | |
| "rewards/accuracies": 0.640625, | |
| "rewards/chosen": 0.13525506854057312, | |
| "rewards/margins": 0.01829436421394348, | |
| "rewards/rejected": 0.11696070432662964, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.75e-07, | |
| "logits/chosen": -2.703218698501587, | |
| "logits/rejected": -2.66160249710083, | |
| "logps/chosen": -331.1356201171875, | |
| "logps/rejected": -277.3096618652344, | |
| "loss": 0.8742, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": 0.14532874524593353, | |
| "rewards/margins": 0.02477584406733513, | |
| "rewards/rejected": 0.1205528974533081, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.958333333333333e-07, | |
| "logits/chosen": -2.605548620223999, | |
| "logits/rejected": -2.507476806640625, | |
| "logps/chosen": -275.4030456542969, | |
| "logps/rejected": -268.85137939453125, | |
| "loss": 0.784, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": 0.14142535626888275, | |
| "rewards/margins": 0.016934942454099655, | |
| "rewards/rejected": 0.1244904026389122, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "logits/chosen": -2.599735736846924, | |
| "logits/rejected": -2.6093332767486572, | |
| "logps/chosen": -259.8871154785156, | |
| "logps/rejected": -238.1406707763672, | |
| "loss": 0.7943, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": 0.1309652328491211, | |
| "rewards/margins": 0.03087472915649414, | |
| "rewards/rejected": 0.10009051114320755, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.375e-07, | |
| "logits/chosen": -2.678443670272827, | |
| "logits/rejected": -2.6648781299591064, | |
| "logps/chosen": -245.9468994140625, | |
| "logps/rejected": -246.77980041503906, | |
| "loss": 0.7385, | |
| "rewards/accuracies": 0.609375, | |
| "rewards/chosen": 0.12853887677192688, | |
| "rewards/margins": 0.01891843043267727, | |
| "rewards/rejected": 0.10962046682834625, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5833333333333327e-07, | |
| "logits/chosen": -2.603628158569336, | |
| "logits/rejected": -2.5344913005828857, | |
| "logps/chosen": -331.3908996582031, | |
| "logps/rejected": -259.43768310546875, | |
| "loss": 0.7334, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": 0.17403613030910492, | |
| "rewards/margins": 0.08045163750648499, | |
| "rewards/rejected": 0.09358450770378113, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.791666666666667e-07, | |
| "logits/chosen": -2.6753909587860107, | |
| "logits/rejected": -2.6965551376342773, | |
| "logps/chosen": -255.45181274414062, | |
| "logps/rejected": -280.82305908203125, | |
| "loss": 0.7035, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": 0.1322551965713501, | |
| "rewards/margins": 0.049186140298843384, | |
| "rewards/rejected": 0.0830690786242485, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5e-07, | |
| "logits/chosen": -2.5472424030303955, | |
| "logits/rejected": -2.5995569229125977, | |
| "logps/chosen": -269.2645568847656, | |
| "logps/rejected": -236.20904541015625, | |
| "loss": 0.7093, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.11034946143627167, | |
| "rewards/margins": 0.03396361321210861, | |
| "rewards/rejected": 0.07638585567474365, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.999731868769026e-07, | |
| "logits/chosen": -2.683974266052246, | |
| "logits/rejected": -2.6413497924804688, | |
| "logps/chosen": -319.9888916015625, | |
| "logps/rejected": -328.697265625, | |
| "loss": 0.938, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": 0.1664372831583023, | |
| "rewards/margins": 0.0846775621175766, | |
| "rewards/rejected": 0.08175972104072571, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.998927532591591e-07, | |
| "logits/chosen": -2.695955514907837, | |
| "logits/rejected": -2.660606861114502, | |
| "logps/chosen": -229.35089111328125, | |
| "logps/rejected": -203.45083618164062, | |
| "loss": 0.7139, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": 0.10423125326633453, | |
| "rewards/margins": 0.0388810932636261, | |
| "rewards/rejected": 0.06535016745328903, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.997587164001815e-07, | |
| "logits/chosen": -2.6461551189422607, | |
| "logits/rejected": -2.5729639530181885, | |
| "logps/chosen": -267.51885986328125, | |
| "logps/rejected": -215.0835723876953, | |
| "loss": 0.7405, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.12834982573986053, | |
| "rewards/margins": 0.09162335097789764, | |
| "rewards/rejected": 0.03672647103667259, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.99571105051544e-07, | |
| "logits/chosen": -2.5733587741851807, | |
| "logits/rejected": -2.5592525005340576, | |
| "logps/chosen": -275.8002014160156, | |
| "logps/rejected": -224.9879150390625, | |
| "loss": 0.7809, | |
| "rewards/accuracies": 0.640625, | |
| "rewards/chosen": 0.12791308760643005, | |
| "rewards/margins": 0.06309904903173447, | |
| "rewards/rejected": 0.0648140236735344, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.993299594568162e-07, | |
| "logits/chosen": -2.6415131092071533, | |
| "logits/rejected": -2.591508626937866, | |
| "logps/chosen": -280.14813232421875, | |
| "logps/rejected": -254.98541259765625, | |
| "loss": 1.0116, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": 0.12386615574359894, | |
| "rewards/margins": 0.06462900340557098, | |
| "rewards/rejected": 0.05923713743686676, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.990353313429303e-07, | |
| "logits/chosen": -2.635352611541748, | |
| "logits/rejected": -2.581376314163208, | |
| "logps/chosen": -300.7115173339844, | |
| "logps/rejected": -253.92111206054688, | |
| "loss": 0.7298, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": 0.12783905863761902, | |
| "rewards/margins": 0.11859284341335297, | |
| "rewards/rejected": 0.00924623105674982, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.986872839090852e-07, | |
| "logits/chosen": -2.7381272315979004, | |
| "logits/rejected": -2.6615114212036133, | |
| "logps/chosen": -310.81658935546875, | |
| "logps/rejected": -246.3937530517578, | |
| "loss": 0.7147, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": 0.08360395580530167, | |
| "rewards/margins": 0.0849369466304779, | |
| "rewards/rejected": -0.0013329838402569294, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.982858918131906e-07, | |
| "logits/chosen": -2.7732505798339844, | |
| "logits/rejected": -2.7643208503723145, | |
| "logps/chosen": -309.32196044921875, | |
| "logps/rejected": -286.745361328125, | |
| "loss": 0.7717, | |
| "rewards/accuracies": 0.640625, | |
| "rewards/chosen": 0.10828227549791336, | |
| "rewards/margins": 0.12604959309101105, | |
| "rewards/rejected": -0.01776731386780739, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.978312411558517e-07, | |
| "logits/chosen": -2.7538211345672607, | |
| "logits/rejected": -2.766498565673828, | |
| "logps/chosen": -260.09765625, | |
| "logps/rejected": -237.29440307617188, | |
| "loss": 0.7561, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": 0.09531789273023605, | |
| "rewards/margins": 0.13862968981266022, | |
| "rewards/rejected": -0.043311797082424164, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.97323429461901e-07, | |
| "logits/chosen": -2.736222505569458, | |
| "logits/rejected": -2.6993401050567627, | |
| "logps/chosen": -218.64276123046875, | |
| "logps/rejected": -247.08609008789062, | |
| "loss": 0.7268, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": 0.102709099650383, | |
| "rewards/margins": 0.15037541091442108, | |
| "rewards/rejected": -0.047666311264038086, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.967625656594781e-07, | |
| "logits/chosen": -2.6768088340759277, | |
| "logits/rejected": -2.720437526702881, | |
| "logps/chosen": -255.47183227539062, | |
| "logps/rejected": -234.1626739501953, | |
| "loss": 0.7297, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.06858564168214798, | |
| "rewards/margins": 0.10109753906726837, | |
| "rewards/rejected": -0.032511889934539795, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.961487700566646e-07, | |
| "logits/chosen": -2.685685157775879, | |
| "logits/rejected": -2.728847026824951, | |
| "logps/chosen": -277.7132873535156, | |
| "logps/rejected": -273.0968017578125, | |
| "loss": 0.843, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.08615937829017639, | |
| "rewards/margins": 0.09151909500360489, | |
| "rewards/rejected": -0.0053597185760736465, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.954821743156767e-07, | |
| "logits/chosen": -2.749175548553467, | |
| "logits/rejected": -2.7879812717437744, | |
| "logps/chosen": -277.5850524902344, | |
| "logps/rejected": -289.2405090332031, | |
| "loss": 0.7339, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": 0.037747230380773544, | |
| "rewards/margins": 0.07498619705438614, | |
| "rewards/rejected": -0.03723897039890289, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.947629214246236e-07, | |
| "logits/chosen": -2.7705671787261963, | |
| "logits/rejected": -2.763953447341919, | |
| "logps/chosen": -252.2857666015625, | |
| "logps/rejected": -292.7469177246094, | |
| "loss": 0.7657, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": 0.017970919609069824, | |
| "rewards/margins": 0.09476744383573532, | |
| "rewards/rejected": -0.0767965242266655, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.939911656668361e-07, | |
| "logits/chosen": -2.682727336883545, | |
| "logits/rejected": -2.6415486335754395, | |
| "logps/chosen": -290.85369873046875, | |
| "logps/rejected": -280.05572509765625, | |
| "loss": 0.7518, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": 0.0839323028922081, | |
| "rewards/margins": 0.10557042807340622, | |
| "rewards/rejected": -0.02163812704384327, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.93167072587771e-07, | |
| "logits/chosen": -2.7483415603637695, | |
| "logits/rejected": -2.7103254795074463, | |
| "logps/chosen": -307.75933837890625, | |
| "logps/rejected": -324.3096923828125, | |
| "loss": 0.7522, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": 0.03351260721683502, | |
| "rewards/margins": 0.10701370239257812, | |
| "rewards/rejected": -0.0735010951757431, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.922908189595017e-07, | |
| "logits/chosen": -2.8289828300476074, | |
| "logits/rejected": -2.8077309131622314, | |
| "logps/chosen": -252.59323120117188, | |
| "logps/rejected": -278.0732727050781, | |
| "loss": 0.7646, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": 0.07634499669075012, | |
| "rewards/margins": 0.1510338932275772, | |
| "rewards/rejected": -0.0746888816356659, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.913625927427995e-07, | |
| "logits/chosen": -2.673996686935425, | |
| "logits/rejected": -2.7177605628967285, | |
| "logps/chosen": -283.3115234375, | |
| "logps/rejected": -235.6822967529297, | |
| "loss": 0.7361, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": 0.027728475630283356, | |
| "rewards/margins": 0.19945986568927765, | |
| "rewards/rejected": -0.1717313826084137, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.903825930468148e-07, | |
| "logits/chosen": -2.790689468383789, | |
| "logits/rejected": -2.8265206813812256, | |
| "logps/chosen": -271.7046813964844, | |
| "logps/rejected": -289.9587097167969, | |
| "loss": 0.7311, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": 0.07139424234628677, | |
| "rewards/margins": 0.22358588874340057, | |
| "rewards/rejected": -0.152191624045372, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.893510300863676e-07, | |
| "logits/chosen": -2.6858608722686768, | |
| "logits/rejected": -2.6849381923675537, | |
| "logps/chosen": -234.48670959472656, | |
| "logps/rejected": -243.48057556152344, | |
| "loss": 0.6433, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": 0.03533104807138443, | |
| "rewards/margins": 0.19699391722679138, | |
| "rewards/rejected": -0.16166287660598755, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.882681251368548e-07, | |
| "logits/chosen": -2.6847596168518066, | |
| "logits/rejected": -2.627260684967041, | |
| "logps/chosen": -242.67559814453125, | |
| "logps/rejected": -233.02520751953125, | |
| "loss": 0.7216, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.0019026286900043488, | |
| "rewards/margins": 0.2595129609107971, | |
| "rewards/rejected": -0.26141557097435, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.871341104867864e-07, | |
| "logits/chosen": -2.861926794052124, | |
| "logits/rejected": -2.8458104133605957, | |
| "logps/chosen": -313.4859619140625, | |
| "logps/rejected": -299.7122497558594, | |
| "loss": 0.6911, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.011806448921561241, | |
| "rewards/margins": 0.1803116649389267, | |
| "rewards/rejected": -0.1921180933713913, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.859492293879573e-07, | |
| "logits/chosen": -2.814488410949707, | |
| "logits/rejected": -2.763936758041382, | |
| "logps/chosen": -251.5836181640625, | |
| "logps/rejected": -268.69342041015625, | |
| "loss": 0.6986, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": 0.0034225257113575935, | |
| "rewards/margins": 0.2998045086860657, | |
| "rewards/rejected": -0.29638200998306274, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.847137360032699e-07, | |
| "logits/chosen": -2.740324020385742, | |
| "logits/rejected": -2.6555376052856445, | |
| "logps/chosen": -267.32647705078125, | |
| "logps/rejected": -311.2066955566406, | |
| "loss": 0.6782, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": 0.04199559986591339, | |
| "rewards/margins": 0.22325468063354492, | |
| "rewards/rejected": -0.18125906586647034, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.834278953522137e-07, | |
| "logits/chosen": -2.788088798522949, | |
| "logits/rejected": -2.769721508026123, | |
| "logps/chosen": -259.531005859375, | |
| "logps/rejected": -282.8291931152344, | |
| "loss": 0.6506, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.01584019511938095, | |
| "rewards/margins": 0.3319249749183655, | |
| "rewards/rejected": -0.347765177488327, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.820919832540181e-07, | |
| "logits/chosen": -2.8391647338867188, | |
| "logits/rejected": -2.7747175693511963, | |
| "logps/chosen": -323.83184814453125, | |
| "logps/rejected": -294.0677795410156, | |
| "loss": 0.7505, | |
| "rewards/accuracies": 0.640625, | |
| "rewards/chosen": -0.0377802811563015, | |
| "rewards/margins": 0.21989765763282776, | |
| "rewards/rejected": -0.25767797231674194, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807062862684873e-07, | |
| "logits/chosen": -2.8058762550354004, | |
| "logits/rejected": -2.7266106605529785, | |
| "logps/chosen": -286.4516906738281, | |
| "logps/rejected": -269.9130554199219, | |
| "loss": 0.7046, | |
| "rewards/accuracies": 0.609375, | |
| "rewards/chosen": -0.2667284309864044, | |
| "rewards/margins": 0.10346898436546326, | |
| "rewards/rejected": -0.3701974153518677, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.792711016345321e-07, | |
| "logits/chosen": -2.8397815227508545, | |
| "logits/rejected": -2.845283031463623, | |
| "logps/chosen": -359.9697570800781, | |
| "logps/rejected": -326.89508056640625, | |
| "loss": 0.7683, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.029000351205468178, | |
| "rewards/margins": 0.4069358706474304, | |
| "rewards/rejected": -0.43593621253967285, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.777867372064105e-07, | |
| "logits/chosen": -2.812641143798828, | |
| "logits/rejected": -2.7523622512817383, | |
| "logps/chosen": -313.5459289550781, | |
| "logps/rejected": -299.8595275878906, | |
| "loss": 0.7817, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.13160283863544464, | |
| "rewards/margins": 0.31865066289901733, | |
| "rewards/rejected": -0.4502534866333008, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7625351138769166e-07, | |
| "logits/chosen": -2.7898976802825928, | |
| "logits/rejected": -2.7862889766693115, | |
| "logps/chosen": -327.0060119628906, | |
| "logps/rejected": -322.96783447265625, | |
| "loss": 0.7708, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.08684093505144119, | |
| "rewards/margins": 0.4070696532726288, | |
| "rewards/rejected": -0.49391061067581177, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7467175306295647e-07, | |
| "logits/chosen": -2.6663591861724854, | |
| "logits/rejected": -2.5961318016052246, | |
| "logps/chosen": -261.4271240234375, | |
| "logps/rejected": -309.5064697265625, | |
| "loss": 0.7365, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.0995909571647644, | |
| "rewards/margins": 0.18599346280097961, | |
| "rewards/rejected": -0.2855844497680664, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7304180152725024e-07, | |
| "logits/chosen": -2.7205381393432617, | |
| "logits/rejected": -2.632664680480957, | |
| "logps/chosen": -324.07098388671875, | |
| "logps/rejected": -303.51422119140625, | |
| "loss": 0.6995, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.09924812614917755, | |
| "rewards/margins": 0.3877854347229004, | |
| "rewards/rejected": -0.48703351616859436, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7136400641330245e-07, | |
| "logits/chosen": -2.7004384994506836, | |
| "logits/rejected": -2.6967031955718994, | |
| "logps/chosen": -273.4053039550781, | |
| "logps/rejected": -285.5749206542969, | |
| "loss": 0.6863, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.04174273461103439, | |
| "rewards/margins": 0.429397314786911, | |
| "rewards/rejected": -0.471140056848526, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.6963872761652834e-07, | |
| "logits/chosen": -2.7354209423065186, | |
| "logits/rejected": -2.6918838024139404, | |
| "logps/chosen": -313.76708984375, | |
| "logps/rejected": -291.7537536621094, | |
| "loss": 0.7426, | |
| "rewards/accuracies": 0.640625, | |
| "rewards/chosen": -0.0618835911154747, | |
| "rewards/margins": 0.23076358437538147, | |
| "rewards/rejected": -0.29264718294143677, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.6786633521783005e-07, | |
| "logits/chosen": -2.6053059101104736, | |
| "logits/rejected": -2.5342700481414795, | |
| "logps/chosen": -244.05995178222656, | |
| "logps/rejected": -259.55023193359375, | |
| "loss": 0.7067, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": 0.07344765961170197, | |
| "rewards/margins": 0.4161744713783264, | |
| "rewards/rejected": -0.34272682666778564, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.6604720940421207e-07, | |
| "logits/chosen": -2.674229145050049, | |
| "logits/rejected": -2.6460635662078857, | |
| "logps/chosen": -293.4000244140625, | |
| "logps/rejected": -237.8984375, | |
| "loss": 0.6553, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.08343453705310822, | |
| "rewards/margins": 0.23878370225429535, | |
| "rewards/rejected": -0.32221823930740356, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.6418174038722924e-07, | |
| "logits/chosen": -2.7431387901306152, | |
| "logits/rejected": -2.6589853763580322, | |
| "logps/chosen": -258.30517578125, | |
| "logps/rejected": -337.9000244140625, | |
| "loss": 0.7977, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.044009700417518616, | |
| "rewards/margins": 0.3760453462600708, | |
| "rewards/rejected": -0.42005497217178345, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.6227032831928483e-07, | |
| "logits/chosen": -2.644824981689453, | |
| "logits/rejected": -2.6608684062957764, | |
| "logps/chosen": -300.009033203125, | |
| "logps/rejected": -324.9970397949219, | |
| "loss": 0.6938, | |
| "rewards/accuracies": 0.640625, | |
| "rewards/chosen": -0.11674674600362778, | |
| "rewards/margins": 0.20092640817165375, | |
| "rewards/rejected": -0.31767311692237854, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.603133832077953e-07, | |
| "logits/chosen": -2.586251974105835, | |
| "logits/rejected": -2.579716920852661, | |
| "logps/chosen": -283.69183349609375, | |
| "logps/rejected": -308.4037780761719, | |
| "loss": 0.6959, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.10065947473049164, | |
| "rewards/margins": 0.31346243619918823, | |
| "rewards/rejected": -0.4141218662261963, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5831132482724193e-07, | |
| "logits/chosen": -2.683396577835083, | |
| "logits/rejected": -2.652515172958374, | |
| "logps/chosen": -294.0975646972656, | |
| "logps/rejected": -284.01007080078125, | |
| "loss": 0.6737, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -0.09628719091415405, | |
| "rewards/margins": 0.3271844983100891, | |
| "rewards/rejected": -0.4234716594219208, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5626458262912735e-07, | |
| "logits/chosen": -2.662360191345215, | |
| "logits/rejected": -2.656447410583496, | |
| "logps/chosen": -292.4847412109375, | |
| "logps/rejected": -313.1526184082031, | |
| "loss": 0.7135, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.13693922758102417, | |
| "rewards/margins": 0.4902416467666626, | |
| "rewards/rejected": -0.6271809339523315, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.541735956498554e-07, | |
| "logits/chosen": -2.7437210083007812, | |
| "logits/rejected": -2.72233247756958, | |
| "logps/chosen": -291.6712646484375, | |
| "logps/rejected": -298.3407287597656, | |
| "loss": 0.6636, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": 0.008928142488002777, | |
| "rewards/margins": 0.6095040440559387, | |
| "rewards/rejected": -0.6005759239196777, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.520388124165564e-07, | |
| "logits/chosen": -2.68888521194458, | |
| "logits/rejected": -2.673058271408081, | |
| "logps/chosen": -303.44732666015625, | |
| "logps/rejected": -304.8455810546875, | |
| "loss": 0.7579, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.08947059512138367, | |
| "rewards/margins": 0.37194037437438965, | |
| "rewards/rejected": -0.4614109992980957, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.498606908508753e-07, | |
| "logits/chosen": -2.6336781978607178, | |
| "logits/rejected": -2.641366958618164, | |
| "logps/chosen": -310.1975402832031, | |
| "logps/rejected": -321.4742736816406, | |
| "loss": 0.7511, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.1359916627407074, | |
| "rewards/margins": 0.3668138384819031, | |
| "rewards/rejected": -0.5028054714202881, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.476396981707453e-07, | |
| "logits/chosen": -2.741149425506592, | |
| "logits/rejected": -2.744832992553711, | |
| "logps/chosen": -283.7612609863281, | |
| "logps/rejected": -319.18011474609375, | |
| "loss": 0.7273, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.14727121591567993, | |
| "rewards/margins": 0.3416295051574707, | |
| "rewards/rejected": -0.48890072107315063, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.453763107901675e-07, | |
| "logits/chosen": -2.769141674041748, | |
| "logits/rejected": -2.6976804733276367, | |
| "logps/chosen": -343.0176086425781, | |
| "logps/rejected": -339.2708740234375, | |
| "loss": 0.684, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.13109339773654938, | |
| "rewards/margins": 0.3063846528530121, | |
| "rewards/rejected": -0.43747806549072266, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4307101421701755e-07, | |
| "logits/chosen": -2.628458023071289, | |
| "logits/rejected": -2.610029458999634, | |
| "logps/chosen": -299.93060302734375, | |
| "logps/rejected": -290.9799499511719, | |
| "loss": 0.6642, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.17024263739585876, | |
| "rewards/margins": 0.28561675548553467, | |
| "rewards/rejected": -0.45585939288139343, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4072430294890166e-07, | |
| "logits/chosen": -2.5924410820007324, | |
| "logits/rejected": -2.5785272121429443, | |
| "logps/chosen": -262.089599609375, | |
| "logps/rejected": -269.3330383300781, | |
| "loss": 0.6441, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": 0.034515030682086945, | |
| "rewards/margins": 0.3341864347457886, | |
| "rewards/rejected": -0.2996714115142822, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3833668036708483e-07, | |
| "logits/chosen": -2.625112771987915, | |
| "logits/rejected": -2.6584105491638184, | |
| "logps/chosen": -306.502685546875, | |
| "logps/rejected": -345.8301086425781, | |
| "loss": 0.6941, | |
| "rewards/accuracies": 0.640625, | |
| "rewards/chosen": -0.10175588726997375, | |
| "rewards/margins": 0.30893003940582275, | |
| "rewards/rejected": -0.4106859564781189, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3590865862851263e-07, | |
| "logits/chosen": -2.617330551147461, | |
| "logits/rejected": -2.5643858909606934, | |
| "logps/chosen": -298.26141357421875, | |
| "logps/rejected": -267.2621154785156, | |
| "loss": 0.6176, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.04226139560341835, | |
| "rewards/margins": 0.45013248920440674, | |
| "rewards/rejected": -0.49239394068717957, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3344075855595097e-07, | |
| "logits/chosen": -2.4896128177642822, | |
| "logits/rejected": -2.4251513481140137, | |
| "logps/chosen": -294.1082763671875, | |
| "logps/rejected": -305.47332763671875, | |
| "loss": 0.6897, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -0.11761671304702759, | |
| "rewards/margins": 0.43049168586730957, | |
| "rewards/rejected": -0.5481083989143372, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.309335095262675e-07, | |
| "logits/chosen": -2.3300864696502686, | |
| "logits/rejected": -2.2978248596191406, | |
| "logps/chosen": -284.1370849609375, | |
| "logps/rejected": -352.7837219238281, | |
| "loss": 0.695, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.26125186681747437, | |
| "rewards/margins": 0.5214322209358215, | |
| "rewards/rejected": -0.7826840877532959, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.2838744935687716e-07, | |
| "logits/chosen": -2.383730888366699, | |
| "logits/rejected": -2.2686848640441895, | |
| "logps/chosen": -285.3652038574219, | |
| "logps/rejected": -313.19207763671875, | |
| "loss": 0.6355, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.1003919169306755, | |
| "rewards/margins": 0.49685102701187134, | |
| "rewards/rejected": -0.5972429513931274, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.258031241903777e-07, | |
| "logits/chosen": -2.2691562175750732, | |
| "logits/rejected": -2.1100411415100098, | |
| "logps/chosen": -295.82708740234375, | |
| "logps/rejected": -299.9715881347656, | |
| "loss": 0.6957, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.07401493936777115, | |
| "rewards/margins": 0.35916173458099365, | |
| "rewards/rejected": -0.4331766366958618, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.2318108837739986e-07, | |
| "logits/chosen": -2.101970672607422, | |
| "logits/rejected": -2.052926540374756, | |
| "logps/chosen": -275.3291931152344, | |
| "logps/rejected": -346.45135498046875, | |
| "loss": 0.7169, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.17267772555351257, | |
| "rewards/margins": 0.42162394523620605, | |
| "rewards/rejected": -0.594301700592041, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.2052190435769554e-07, | |
| "logits/chosen": -2.161992073059082, | |
| "logits/rejected": -2.0577569007873535, | |
| "logps/chosen": -329.9812316894531, | |
| "logps/rejected": -323.7291259765625, | |
| "loss": 0.6115, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.0013326676562428474, | |
| "rewards/margins": 0.4901263415813446, | |
| "rewards/rejected": -0.4914590120315552, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.1782614253949255e-07, | |
| "logits/chosen": -1.8093454837799072, | |
| "logits/rejected": -1.6997994184494019, | |
| "logps/chosen": -252.4735565185547, | |
| "logps/rejected": -264.23297119140625, | |
| "loss": 0.7009, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.2441805750131607, | |
| "rewards/margins": 0.25481969118118286, | |
| "rewards/rejected": -0.49900031089782715, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.1509438117713863e-07, | |
| "logits/chosen": -1.5648189783096313, | |
| "logits/rejected": -1.543358325958252, | |
| "logps/chosen": -320.0599060058594, | |
| "logps/rejected": -306.7493896484375, | |
| "loss": 0.6848, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.039286356419324875, | |
| "rewards/margins": 0.518799901008606, | |
| "rewards/rejected": -0.47951358556747437, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.123272062470633e-07, | |
| "logits/chosen": -1.3095282316207886, | |
| "logits/rejected": -1.038232684135437, | |
| "logps/chosen": -291.6402282714844, | |
| "logps/rejected": -298.0380554199219, | |
| "loss": 0.6891, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.02619929611682892, | |
| "rewards/margins": 0.43957069516181946, | |
| "rewards/rejected": -0.4657699763774872, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0952521132208267e-07, | |
| "logits/chosen": -0.8358496427536011, | |
| "logits/rejected": -0.8311337828636169, | |
| "logps/chosen": -288.705078125, | |
| "logps/rejected": -253.3335723876953, | |
| "loss": 0.6821, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.1505282074213028, | |
| "rewards/margins": 0.26645365357398987, | |
| "rewards/rejected": -0.41698184609413147, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.0668899744407567e-07, | |
| "logits/chosen": -0.6292511820793152, | |
| "logits/rejected": 0.11955951899290085, | |
| "logps/chosen": -217.69808959960938, | |
| "logps/rejected": -266.32525634765625, | |
| "loss": 0.7194, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.19091403484344482, | |
| "rewards/margins": 0.39874905347824097, | |
| "rewards/rejected": -0.589663028717041, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.0381917299505686e-07, | |
| "logits/chosen": -0.4040292203426361, | |
| "logits/rejected": -0.05061528459191322, | |
| "logps/chosen": -281.4028015136719, | |
| "logps/rejected": -348.06854248046875, | |
| "loss": 0.6614, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.022851143032312393, | |
| "rewards/margins": 0.5195820331573486, | |
| "rewards/rejected": -0.5424332022666931, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.00916353566676e-07, | |
| "logits/chosen": 0.4304525554180145, | |
| "logits/rejected": 0.37021252512931824, | |
| "logps/chosen": -268.30401611328125, | |
| "logps/rejected": -340.36376953125, | |
| "loss": 0.6454, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.2065473347902298, | |
| "rewards/margins": 0.4695810079574585, | |
| "rewards/rejected": -0.6761283874511719, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.979811618281705e-07, | |
| "logits/chosen": 0.9382424354553223, | |
| "logits/rejected": -0.011063352227210999, | |
| "logps/chosen": -303.7531433105469, | |
| "logps/rejected": -319.4146423339844, | |
| "loss": 0.7602, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.31537631154060364, | |
| "rewards/margins": 0.3811299502849579, | |
| "rewards/rejected": -0.6965062618255615, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9501422739279953e-07, | |
| "logits/chosen": 0.4460795521736145, | |
| "logits/rejected": 0.570868968963623, | |
| "logps/chosen": -307.8954772949219, | |
| "logps/rejected": -339.8791198730469, | |
| "loss": 0.6882, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.10756506770849228, | |
| "rewards/margins": 0.5310941934585571, | |
| "rewards/rejected": -0.6386592388153076, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.920161866827889e-07, | |
| "logits/chosen": 0.07825072109699249, | |
| "logits/rejected": 0.6348622441291809, | |
| "logps/chosen": -309.1721496582031, | |
| "logps/rejected": -288.8408203125, | |
| "loss": 0.6567, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.1812726855278015, | |
| "rewards/margins": 0.5873675346374512, | |
| "rewards/rejected": -0.7686402797698975, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.889876827928156e-07, | |
| "logits/chosen": 0.37462693452835083, | |
| "logits/rejected": 0.3019832670688629, | |
| "logps/chosen": -253.56881713867188, | |
| "logps/rejected": -316.3982238769531, | |
| "loss": 0.5877, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.04077187180519104, | |
| "rewards/margins": 0.5590379238128662, | |
| "rewards/rejected": -0.5182660222053528, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.859293653520604e-07, | |
| "logits/chosen": 0.6516929268836975, | |
| "logits/rejected": 0.2800888419151306, | |
| "logps/chosen": -276.3404541015625, | |
| "logps/rejected": -317.9354248046875, | |
| "loss": 0.7184, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.1999325156211853, | |
| "rewards/margins": 0.5805739164352417, | |
| "rewards/rejected": -0.780506432056427, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.828418903848593e-07, | |
| "logits/chosen": -0.025208771228790283, | |
| "logits/rejected": 0.8900982737541199, | |
| "logps/chosen": -265.7961120605469, | |
| "logps/rejected": -337.1889343261719, | |
| "loss": 0.6888, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3380471467971802, | |
| "rewards/margins": 0.5418481826782227, | |
| "rewards/rejected": -0.8798953294754028, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.797259201699833e-07, | |
| "logits/chosen": 0.05765300244092941, | |
| "logits/rejected": 0.576537013053894, | |
| "logps/chosen": -331.6053466796875, | |
| "logps/rejected": -351.186767578125, | |
| "loss": 0.6485, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3514668941497803, | |
| "rewards/margins": 0.43950024247169495, | |
| "rewards/rejected": -0.7909672260284424, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.765821230985757e-07, | |
| "logits/chosen": 0.3566046953201294, | |
| "logits/rejected": 0.24242231249809265, | |
| "logps/chosen": -333.481689453125, | |
| "logps/rejected": -358.510986328125, | |
| "loss": 0.6358, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.3405124843120575, | |
| "rewards/margins": 0.591222882270813, | |
| "rewards/rejected": -0.9317353963851929, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.734111735307796e-07, | |
| "logits/chosen": 0.5186017155647278, | |
| "logits/rejected": 0.6986968517303467, | |
| "logps/chosen": -291.35626220703125, | |
| "logps/rejected": -325.29412841796875, | |
| "loss": 0.6586, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -0.3297085464000702, | |
| "rewards/margins": 0.641376256942749, | |
| "rewards/rejected": -0.9710848331451416, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.7021375165108377e-07, | |
| "logits/chosen": 0.40191513299942017, | |
| "logits/rejected": 0.24818360805511475, | |
| "logps/chosen": -289.5572509765625, | |
| "logps/rejected": -323.7060241699219, | |
| "loss": 0.6506, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.2648981511592865, | |
| "rewards/margins": 0.49969348311424255, | |
| "rewards/rejected": -0.7645915746688843, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.6699054332241985e-07, | |
| "logits/chosen": -0.4896765351295471, | |
| "logits/rejected": -0.1318652182817459, | |
| "logps/chosen": -350.1439208984375, | |
| "logps/rejected": -288.9870910644531, | |
| "loss": 0.7068, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.22379332780838013, | |
| "rewards/margins": 0.44880181550979614, | |
| "rewards/rejected": -0.6725951433181763, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.6374223993904124e-07, | |
| "logits/chosen": -0.05389963835477829, | |
| "logits/rejected": 0.3870914578437805, | |
| "logps/chosen": -330.797119140625, | |
| "logps/rejected": -357.25347900390625, | |
| "loss": 0.6192, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.20667432248592377, | |
| "rewards/margins": 0.6827113032341003, | |
| "rewards/rejected": -0.8893855810165405, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.604695382782159e-07, | |
| "logits/chosen": -0.1867658495903015, | |
| "logits/rejected": 0.04192525893449783, | |
| "logps/chosen": -306.168212890625, | |
| "logps/rejected": -340.8019104003906, | |
| "loss": 0.594, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.1515503227710724, | |
| "rewards/margins": 0.3953551650047302, | |
| "rewards/rejected": -0.546905517578125, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.571731403507635e-07, | |
| "logits/chosen": -0.3911605179309845, | |
| "logits/rejected": 0.24688732624053955, | |
| "logps/chosen": -315.493408203125, | |
| "logps/rejected": -362.08209228515625, | |
| "loss": 0.667, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.19705705344676971, | |
| "rewards/margins": 0.5905351638793945, | |
| "rewards/rejected": -0.7875921726226807, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.5385375325047163e-07, | |
| "logits/chosen": 0.018862135708332062, | |
| "logits/rejected": 0.042824022471904755, | |
| "logps/chosen": -346.4378662109375, | |
| "logps/rejected": -371.5933837890625, | |
| "loss": 0.7231, | |
| "rewards/accuracies": 0.640625, | |
| "rewards/chosen": -0.16377538442611694, | |
| "rewards/margins": 0.5193297863006592, | |
| "rewards/rejected": -0.6831051707267761, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.505120890024195e-07, | |
| "logits/chosen": -0.5048182010650635, | |
| "logits/rejected": 0.798430323600769, | |
| "logps/chosen": -311.4440002441406, | |
| "logps/rejected": -332.12078857421875, | |
| "loss": 0.5901, | |
| "rewards/accuracies": 0.609375, | |
| "rewards/chosen": -0.1354539841413498, | |
| "rewards/margins": 0.5432121753692627, | |
| "rewards/rejected": -0.6786662340164185, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4714886441024573e-07, | |
| "logits/chosen": 0.18471813201904297, | |
| "logits/rejected": 0.5488951206207275, | |
| "logps/chosen": -312.63348388671875, | |
| "logps/rejected": -304.498291015625, | |
| "loss": 0.6626, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.16734659671783447, | |
| "rewards/margins": 0.5686672925949097, | |
| "rewards/rejected": -0.7360140085220337, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4376480090239047e-07, | |
| "logits/chosen": 0.2744285464286804, | |
| "logits/rejected": 0.882749080657959, | |
| "logps/chosen": -260.8684997558594, | |
| "logps/rejected": -292.6208190917969, | |
| "loss": 0.7122, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.3560115098953247, | |
| "rewards/margins": 0.36396852135658264, | |
| "rewards/rejected": -0.719980001449585, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.403606243773448e-07, | |
| "logits/chosen": 0.10863735526800156, | |
| "logits/rejected": 0.9924963116645813, | |
| "logps/chosen": -277.6872863769531, | |
| "logps/rejected": -296.84375, | |
| "loss": 0.5924, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.19410894811153412, | |
| "rewards/margins": 0.49725741147994995, | |
| "rewards/rejected": -0.6913663744926453, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3693706504794243e-07, | |
| "logits/chosen": -0.0544244647026062, | |
| "logits/rejected": 0.6754695773124695, | |
| "logps/chosen": -276.8150634765625, | |
| "logps/rejected": -309.6656494140625, | |
| "loss": 0.6933, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.18632037937641144, | |
| "rewards/margins": 0.4902322292327881, | |
| "rewards/rejected": -0.6765525937080383, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.334948572847253e-07, | |
| "logits/chosen": 0.12526825070381165, | |
| "logits/rejected": 0.2164376825094223, | |
| "logps/chosen": -288.12420654296875, | |
| "logps/rejected": -301.5098876953125, | |
| "loss": 0.644, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.2477717101573944, | |
| "rewards/margins": 0.510245680809021, | |
| "rewards/rejected": -0.7580174207687378, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.300347394584172e-07, | |
| "logits/chosen": -0.517288863658905, | |
| "logits/rejected": -0.44046011567115784, | |
| "logps/chosen": -271.6811218261719, | |
| "logps/rejected": -325.8108215332031, | |
| "loss": 0.6557, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.2316058874130249, | |
| "rewards/margins": 0.4710268974304199, | |
| "rewards/rejected": -0.7026327848434448, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.265574537815398e-07, | |
| "logits/chosen": -0.6968811750411987, | |
| "logits/rejected": -0.789453387260437, | |
| "logps/chosen": -290.6319580078125, | |
| "logps/rejected": -339.7631530761719, | |
| "loss": 0.635, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.15982553362846375, | |
| "rewards/margins": 0.5405714511871338, | |
| "rewards/rejected": -0.7003970146179199, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.230637461492043e-07, | |
| "logits/chosen": -0.4320661425590515, | |
| "logits/rejected": -0.4025118052959442, | |
| "logps/chosen": -280.6953430175781, | |
| "logps/rejected": -280.9771728515625, | |
| "loss": 0.6636, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.15578195452690125, | |
| "rewards/margins": 0.5092009902000427, | |
| "rewards/rejected": -0.6649829745292664, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.1955436597911315e-07, | |
| "logits/chosen": -0.9900290369987488, | |
| "logits/rejected": -0.5774145126342773, | |
| "logps/chosen": -257.3419494628906, | |
| "logps/rejected": -318.7354431152344, | |
| "loss": 0.6765, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": 0.020992910489439964, | |
| "rewards/margins": 0.5579670071601868, | |
| "rewards/rejected": -0.5369741320610046, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.160300660508064e-07, | |
| "logits/chosen": -1.4757821559906006, | |
| "logits/rejected": -0.815751314163208, | |
| "logps/chosen": -307.86767578125, | |
| "logps/rejected": -318.9004821777344, | |
| "loss": 0.661, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -0.030079301446676254, | |
| "rewards/margins": 0.5298373103141785, | |
| "rewards/rejected": -0.5599166750907898, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.1249160234418644e-07, | |
| "logits/chosen": -0.9963082671165466, | |
| "logits/rejected": -0.20200899243354797, | |
| "logps/chosen": -275.8343200683594, | |
| "logps/rejected": -320.10943603515625, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.09299668669700623, | |
| "rewards/margins": 0.47309738397598267, | |
| "rewards/rejected": -0.5660940408706665, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.0893973387735683e-07, | |
| "logits/chosen": -0.29111677408218384, | |
| "logits/rejected": -0.016862310469150543, | |
| "logps/chosen": -243.76763916015625, | |
| "logps/rejected": -311.628662109375, | |
| "loss": 0.6769, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.1675092577934265, | |
| "rewards/margins": 0.49733197689056396, | |
| "rewards/rejected": -0.6648411750793457, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.05375222543809e-07, | |
| "logits/chosen": -0.8505387306213379, | |
| "logits/rejected": -0.23501801490783691, | |
| "logps/chosen": -294.7572021484375, | |
| "logps/rejected": -310.2570495605469, | |
| "loss": 0.7179, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.1568153351545334, | |
| "rewards/margins": 0.5180144906044006, | |
| "rewards/rejected": -0.6748298406600952, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.017988329489923e-07, | |
| "logits/chosen": -0.2259301245212555, | |
| "logits/rejected": 0.060025911778211594, | |
| "logps/chosen": -325.3514099121094, | |
| "logps/rejected": -357.01263427734375, | |
| "loss": 0.7001, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -0.1160479336977005, | |
| "rewards/margins": 0.5354103446006775, | |
| "rewards/rejected": -0.6514582633972168, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.9821133224630223e-07, | |
| "logits/chosen": -0.24249279499053955, | |
| "logits/rejected": 0.2804643511772156, | |
| "logps/chosen": -306.503662109375, | |
| "logps/rejected": -328.2669372558594, | |
| "loss": 0.6416, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.3325122892856598, | |
| "rewards/margins": 0.5438024401664734, | |
| "rewards/rejected": -0.8763147592544556, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.946134899725226e-07, | |
| "logits/chosen": -0.21587029099464417, | |
| "logits/rejected": 0.22806911170482635, | |
| "logps/chosen": -272.1225891113281, | |
| "logps/rejected": -317.8914489746094, | |
| "loss": 0.7015, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.2932060956954956, | |
| "rewards/margins": 0.544987142086029, | |
| "rewards/rejected": -0.8381932377815247, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.910060778827554e-07, | |
| "logits/chosen": 0.44503313302993774, | |
| "logits/rejected": 0.7824755311012268, | |
| "logps/chosen": -242.06944274902344, | |
| "logps/rejected": -281.43572998046875, | |
| "loss": 0.6554, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.23596498370170593, | |
| "rewards/margins": 0.4384329915046692, | |
| "rewards/rejected": -0.6743979454040527, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.873898697848762e-07, | |
| "logits/chosen": 0.17147797346115112, | |
| "logits/rejected": 0.5953517556190491, | |
| "logps/chosen": -300.029296875, | |
| "logps/rejected": -331.022216796875, | |
| "loss": 0.6662, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.23151984810829163, | |
| "rewards/margins": 0.441323846578598, | |
| "rewards/rejected": -0.6728437542915344, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.837656413735479e-07, | |
| "logits/chosen": 0.24648509919643402, | |
| "logits/rejected": 0.7520670294761658, | |
| "logps/chosen": -287.30377197265625, | |
| "logps/rejected": -344.3001708984375, | |
| "loss": 0.6139, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.20726029574871063, | |
| "rewards/margins": 0.7070189118385315, | |
| "rewards/rejected": -0.9142792820930481, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.801341700638307e-07, | |
| "logits/chosen": 0.4327443242073059, | |
| "logits/rejected": 0.28958457708358765, | |
| "logps/chosen": -314.9183349609375, | |
| "logps/rejected": -296.54376220703125, | |
| "loss": 0.6514, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.24840864539146423, | |
| "rewards/margins": 0.4947468638420105, | |
| "rewards/rejected": -0.7431554794311523, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7649623482442274e-07, | |
| "logits/chosen": 0.5047562718391418, | |
| "logits/rejected": 1.313357949256897, | |
| "logps/chosen": -284.20257568359375, | |
| "logps/rejected": -354.37872314453125, | |
| "loss": 0.6634, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.3625330328941345, | |
| "rewards/margins": 0.5466379523277283, | |
| "rewards/rejected": -0.9091709852218628, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7285261601056697e-07, | |
| "logits/chosen": 0.19208696484565735, | |
| "logits/rejected": 0.3003866970539093, | |
| "logps/chosen": -306.0182189941406, | |
| "logps/rejected": -343.263671875, | |
| "loss": 0.6481, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.1929151564836502, | |
| "rewards/margins": 0.7904263138771057, | |
| "rewards/rejected": -0.9833414554595947, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.692040951966617e-07, | |
| "logits/chosen": 0.5424776077270508, | |
| "logits/rejected": 0.7004806995391846, | |
| "logps/chosen": -312.521728515625, | |
| "logps/rejected": -328.9450378417969, | |
| "loss": 0.5763, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.0869397521018982, | |
| "rewards/margins": 0.742694616317749, | |
| "rewards/rejected": -0.8296343088150024, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.655514550086086e-07, | |
| "logits/chosen": 0.3900845944881439, | |
| "logits/rejected": 1.1437915563583374, | |
| "logps/chosen": -270.412353515625, | |
| "logps/rejected": -285.83740234375, | |
| "loss": 0.6635, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 0.039971936494112015, | |
| "rewards/margins": 0.6791943907737732, | |
| "rewards/rejected": -0.6392224431037903, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.618954789559356e-07, | |
| "logits/chosen": -0.6006118059158325, | |
| "logits/rejected": 0.0894058495759964, | |
| "logps/chosen": -329.71051025390625, | |
| "logps/rejected": -360.0182800292969, | |
| "loss": 0.7222, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.2027689814567566, | |
| "rewards/margins": 0.3933987617492676, | |
| "rewards/rejected": -0.5961677432060242, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.582369512637302e-07, | |
| "logits/chosen": 0.022289007902145386, | |
| "logits/rejected": 0.7651575207710266, | |
| "logps/chosen": -239.86917114257812, | |
| "logps/rejected": -276.7130126953125, | |
| "loss": 0.6653, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": 0.07738906890153885, | |
| "rewards/margins": 0.4737430810928345, | |
| "rewards/rejected": -0.3963540196418762, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.5457665670441937e-07, | |
| "logits/chosen": 0.08083158731460571, | |
| "logits/rejected": 0.5649360418319702, | |
| "logps/chosen": -276.38232421875, | |
| "logps/rejected": -330.9521789550781, | |
| "loss": 0.7062, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.10466089844703674, | |
| "rewards/margins": 0.6767792701721191, | |
| "rewards/rejected": -0.7814401984214783, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.509153804294318e-07, | |
| "logits/chosen": 0.3234708309173584, | |
| "logits/rejected": 0.7615450024604797, | |
| "logps/chosen": -269.09417724609375, | |
| "logps/rejected": -293.904052734375, | |
| "loss": 0.6256, | |
| "rewards/accuracies": 0.609375, | |
| "rewards/chosen": -0.07213997095823288, | |
| "rewards/margins": 0.43272051215171814, | |
| "rewards/rejected": -0.504860520362854, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4725390780077905e-07, | |
| "logits/chosen": 0.46226564049720764, | |
| "logits/rejected": 0.5075880885124207, | |
| "logps/chosen": -314.9215087890625, | |
| "logps/rejected": -332.1287536621094, | |
| "loss": 0.7012, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.053367726504802704, | |
| "rewards/margins": 0.6606994867324829, | |
| "rewards/rejected": -0.7140672206878662, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.435930242225919e-07, | |
| "logits/chosen": 0.2664536237716675, | |
| "logits/rejected": 1.190063714981079, | |
| "logps/chosen": -302.0753479003906, | |
| "logps/rejected": -365.6280822753906, | |
| "loss": 0.6695, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.07531341910362244, | |
| "rewards/margins": 0.6120296120643616, | |
| "rewards/rejected": -0.6873430609703064, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.399335149726463e-07, | |
| "logits/chosen": 0.32580170035362244, | |
| "logits/rejected": 0.5576289892196655, | |
| "logps/chosen": -325.13726806640625, | |
| "logps/rejected": -381.2005615234375, | |
| "loss": 0.6745, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.10618472844362259, | |
| "rewards/margins": 0.5702216625213623, | |
| "rewards/rejected": -0.6764063835144043, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3627616503391812e-07, | |
| "logits/chosen": 0.29107674956321716, | |
| "logits/rejected": 0.5633612871170044, | |
| "logps/chosen": -277.61590576171875, | |
| "logps/rejected": -348.748291015625, | |
| "loss": 0.6212, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.12373210489749908, | |
| "rewards/margins": 0.578803300857544, | |
| "rewards/rejected": -0.7025353908538818, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3262175892620062e-07, | |
| "logits/chosen": 0.6590377688407898, | |
| "logits/rejected": 0.6782903075218201, | |
| "logps/chosen": -301.8876647949219, | |
| "logps/rejected": -348.36968994140625, | |
| "loss": 0.732, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.06287754327058792, | |
| "rewards/margins": 0.7469431757926941, | |
| "rewards/rejected": -0.8098206520080566, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.2897108053782e-07, | |
| "logits/chosen": 0.5858335494995117, | |
| "logits/rejected": 1.1848359107971191, | |
| "logps/chosen": -278.7359619140625, | |
| "logps/rejected": -322.556884765625, | |
| "loss": 0.6291, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.1839427649974823, | |
| "rewards/margins": 0.5925223231315613, | |
| "rewards/rejected": -0.7764650583267212, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2532491295748865e-07, | |
| "logits/chosen": 0.5746503472328186, | |
| "logits/rejected": 1.5114972591400146, | |
| "logps/chosen": -256.6669616699219, | |
| "logps/rejected": -291.76373291015625, | |
| "loss": 0.677, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.17329725623130798, | |
| "rewards/margins": 0.6033150553703308, | |
| "rewards/rejected": -0.7766122817993164, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2168403830632769e-07, | |
| "logits/chosen": 1.2941956520080566, | |
| "logits/rejected": 1.145210862159729, | |
| "logps/chosen": -253.27915954589844, | |
| "logps/rejected": -313.1325378417969, | |
| "loss": 0.63, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.18560369312763214, | |
| "rewards/margins": 0.5705838203430176, | |
| "rewards/rejected": -0.7561875581741333, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1804923757009882e-07, | |
| "logits/chosen": 0.9151768684387207, | |
| "logits/rejected": 1.2065914869308472, | |
| "logps/chosen": -335.712890625, | |
| "logps/rejected": -334.68817138671875, | |
| "loss": 0.5678, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.24476921558380127, | |
| "rewards/margins": 0.7057139873504639, | |
| "rewards/rejected": -0.9504832029342651, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1442129043167873e-07, | |
| "logits/chosen": 0.8291666507720947, | |
| "logits/rejected": 1.7541477680206299, | |
| "logps/chosen": -330.841064453125, | |
| "logps/rejected": -347.6577453613281, | |
| "loss": 0.6978, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.29731371998786926, | |
| "rewards/margins": 0.6203598380088806, | |
| "rewards/rejected": -0.917673647403717, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1080097510381294e-07, | |
| "logits/chosen": 0.8476366400718689, | |
| "logits/rejected": 1.6469848155975342, | |
| "logps/chosen": -290.6492004394531, | |
| "logps/rejected": -338.7218017578125, | |
| "loss": 0.7014, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.2652449905872345, | |
| "rewards/margins": 0.51957768201828, | |
| "rewards/rejected": -0.7848227024078369, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0718906816218595e-07, | |
| "logits/chosen": 1.3053678274154663, | |
| "logits/rejected": 1.4397318363189697, | |
| "logps/chosen": -312.151123046875, | |
| "logps/rejected": -353.7760009765625, | |
| "loss": 0.5969, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.3702758550643921, | |
| "rewards/margins": 0.655584454536438, | |
| "rewards/rejected": -1.02586030960083, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.035863443788411e-07, | |
| "logits/chosen": 1.1411147117614746, | |
| "logits/rejected": 1.3497332334518433, | |
| "logps/chosen": -334.48223876953125, | |
| "logps/rejected": -358.02911376953125, | |
| "loss": 0.6894, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.24735981225967407, | |
| "rewards/margins": 0.7227092981338501, | |
| "rewards/rejected": -0.9700690507888794, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9999357655598891e-07, | |
| "logits/chosen": 1.477621078491211, | |
| "logits/rejected": 2.1223461627960205, | |
| "logps/chosen": -277.4604187011719, | |
| "logps/rejected": -316.42803955078125, | |
| "loss": 0.6614, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.5241543650627136, | |
| "rewards/margins": 0.5040414333343506, | |
| "rewards/rejected": -1.0281956195831299, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9641153536023642e-07, | |
| "logits/chosen": 0.708802342414856, | |
| "logits/rejected": 1.7905950546264648, | |
| "logps/chosen": -260.52166748046875, | |
| "logps/rejected": -309.6279602050781, | |
| "loss": 0.6517, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.22097410261631012, | |
| "rewards/margins": 0.6726060509681702, | |
| "rewards/rejected": -0.8935801982879639, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.928409891572757e-07, | |
| "logits/chosen": 0.6345776915550232, | |
| "logits/rejected": 1.2965819835662842, | |
| "logps/chosen": -296.1861572265625, | |
| "logps/rejected": -325.6953430175781, | |
| "loss": 0.6571, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.15726947784423828, | |
| "rewards/margins": 0.33030712604522705, | |
| "rewards/rejected": -0.48757660388946533, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8928270384706582e-07, | |
| "logits/chosen": 0.06032524257898331, | |
| "logits/rejected": 0.7254040241241455, | |
| "logps/chosen": -320.2947998046875, | |
| "logps/rejected": -417.827392578125, | |
| "loss": 0.823, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.05728043615818024, | |
| "rewards/margins": 0.4396788477897644, | |
| "rewards/rejected": -0.49695926904678345, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8573744269954297e-07, | |
| "logits/chosen": -0.6211626529693604, | |
| "logits/rejected": 0.5601530075073242, | |
| "logps/chosen": -387.068359375, | |
| "logps/rejected": -362.6905517578125, | |
| "loss": 0.7065, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": 0.007479578256607056, | |
| "rewards/margins": 0.5313453078269958, | |
| "rewards/rejected": -0.5238656997680664, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8220596619089573e-07, | |
| "logits/chosen": 0.012296931818127632, | |
| "logits/rejected": 0.47643163800239563, | |
| "logps/chosen": -278.79449462890625, | |
| "logps/rejected": -358.36175537109375, | |
| "loss": 0.6023, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -0.02508646249771118, | |
| "rewards/margins": 0.529964029788971, | |
| "rewards/rejected": -0.5550505518913269, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.7868903184043885e-07, | |
| "logits/chosen": 0.22885966300964355, | |
| "logits/rejected": 0.6849693655967712, | |
| "logps/chosen": -245.32431030273438, | |
| "logps/rejected": -260.42095947265625, | |
| "loss": 0.6808, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.2312414050102234, | |
| "rewards/margins": 0.5006843209266663, | |
| "rewards/rejected": -0.7319257855415344, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7518739404812155e-07, | |
| "logits/chosen": 0.10696160048246384, | |
| "logits/rejected": 0.900860071182251, | |
| "logps/chosen": -304.24365234375, | |
| "logps/rejected": -300.4913330078125, | |
| "loss": 0.6239, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": 0.06586351990699768, | |
| "rewards/margins": 0.6996278762817383, | |
| "rewards/rejected": -0.6337643265724182, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.717018039327053e-07, | |
| "logits/chosen": 0.37065258622169495, | |
| "logits/rejected": 1.0153319835662842, | |
| "logps/chosen": -274.7767639160156, | |
| "logps/rejected": -315.3498840332031, | |
| "loss": 0.6113, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.13434213399887085, | |
| "rewards/margins": 0.5673910975456238, | |
| "rewards/rejected": -0.7017332315444946, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.6823300917064458e-07, | |
| "logits/chosen": 0.0030569732189178467, | |
| "logits/rejected": 1.1464507579803467, | |
| "logps/chosen": -292.8100280761719, | |
| "logps/rejected": -329.35333251953125, | |
| "loss": 0.5978, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.3217441141605377, | |
| "rewards/margins": 0.6382726430892944, | |
| "rewards/rejected": -0.9600168466567993, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.647817538357072e-07, | |
| "logits/chosen": 0.3857354521751404, | |
| "logits/rejected": 1.0852301120758057, | |
| "logps/chosen": -329.72882080078125, | |
| "logps/rejected": -318.942138671875, | |
| "loss": 0.7355, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.1711738556623459, | |
| "rewards/margins": 0.6827853322029114, | |
| "rewards/rejected": -0.8539592027664185, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6134877823936607e-07, | |
| "logits/chosen": 0.8364083766937256, | |
| "logits/rejected": 1.8597906827926636, | |
| "logps/chosen": -264.943359375, | |
| "logps/rejected": -348.2406921386719, | |
| "loss": 0.5214, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.09745172411203384, | |
| "rewards/margins": 1.012548565864563, | |
| "rewards/rejected": -1.1100002527236938, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5793481877199943e-07, | |
| "logits/chosen": 0.5367469191551208, | |
| "logits/rejected": 1.0159227848052979, | |
| "logps/chosen": -320.8637390136719, | |
| "logps/rejected": -364.44586181640625, | |
| "loss": 0.6954, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.26059049367904663, | |
| "rewards/margins": 0.5368437170982361, | |
| "rewards/rejected": -0.7974342107772827, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5454060774493065e-07, | |
| "logits/chosen": 1.5845757722854614, | |
| "logits/rejected": 0.7877135872840881, | |
| "logps/chosen": -269.8088684082031, | |
| "logps/rejected": -316.855712890625, | |
| "loss": 0.6368, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.14322423934936523, | |
| "rewards/margins": 0.6599938869476318, | |
| "rewards/rejected": -0.8032180070877075, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5116687323334464e-07, | |
| "logits/chosen": 0.47069552540779114, | |
| "logits/rejected": 1.2589174509048462, | |
| "logps/chosen": -289.40130615234375, | |
| "logps/rejected": -347.61376953125, | |
| "loss": 0.6596, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.15590085089206696, | |
| "rewards/margins": 0.6152348518371582, | |
| "rewards/rejected": -0.771135687828064, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.478143389201113e-07, | |
| "logits/chosen": 0.5592608451843262, | |
| "logits/rejected": 1.0087881088256836, | |
| "logps/chosen": -281.7051696777344, | |
| "logps/rejected": -335.6897277832031, | |
| "loss": 0.5898, | |
| "rewards/accuracies": 0.859375, | |
| "rewards/chosen": 0.045632533729076385, | |
| "rewards/margins": 0.8765643835067749, | |
| "rewards/rejected": -0.8309319019317627, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4448372394055246e-07, | |
| "logits/chosen": -0.018945127725601196, | |
| "logits/rejected": 0.802770733833313, | |
| "logps/chosen": -328.23980712890625, | |
| "logps/rejected": -333.0008239746094, | |
| "loss": 0.6108, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.1781616359949112, | |
| "rewards/margins": 0.6047887206077576, | |
| "rewards/rejected": -0.7829503417015076, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.4117574272818386e-07, | |
| "logits/chosen": -0.5540781021118164, | |
| "logits/rejected": 0.035438332706689835, | |
| "logps/chosen": -317.64404296875, | |
| "logps/rejected": -343.2810974121094, | |
| "loss": 0.6491, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.0927032008767128, | |
| "rewards/margins": 0.4850231111049652, | |
| "rewards/rejected": -0.577726423740387, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.3789110486146468e-07, | |
| "logits/chosen": 0.01489407941699028, | |
| "logits/rejected": 0.62424635887146, | |
| "logps/chosen": -293.1672668457031, | |
| "logps/rejected": -329.47479248046875, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.16148582100868225, | |
| "rewards/margins": 0.5662094950675964, | |
| "rewards/rejected": -0.7276952862739563, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3463051491159093e-07, | |
| "logits/chosen": 0.06474697589874268, | |
| "logits/rejected": -0.06373190879821777, | |
| "logps/chosen": -314.8403015136719, | |
| "logps/rejected": -331.58941650390625, | |
| "loss": 0.6393, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.08766806125640869, | |
| "rewards/margins": 0.6701959371566772, | |
| "rewards/rejected": -0.7578639984130859, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3139467229135998e-07, | |
| "logits/chosen": -0.7029250860214233, | |
| "logits/rejected": -0.01946442574262619, | |
| "logps/chosen": -334.6876220703125, | |
| "logps/rejected": -347.177734375, | |
| "loss": 0.63, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.16892662644386292, | |
| "rewards/margins": 0.5833321809768677, | |
| "rewards/rejected": -0.752258837223053, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.281842711051438e-07, | |
| "logits/chosen": -0.23442219197750092, | |
| "logits/rejected": 0.7187111973762512, | |
| "logps/chosen": -295.4104309082031, | |
| "logps/rejected": -344.3884582519531, | |
| "loss": 0.5594, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.11131522059440613, | |
| "rewards/margins": 0.7158193588256836, | |
| "rewards/rejected": -0.8271346092224121, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2500000000000005e-07, | |
| "logits/chosen": -0.23394931852817535, | |
| "logits/rejected": 0.5850412845611572, | |
| "logps/chosen": -316.489990234375, | |
| "logps/rejected": -380.6622009277344, | |
| "loss": 0.695, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.21047508716583252, | |
| "rewards/margins": 0.6145602464675903, | |
| "rewards/rejected": -0.8250353336334229, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2184254201795363e-07, | |
| "logits/chosen": 0.016806162893772125, | |
| "logits/rejected": 0.5278490781784058, | |
| "logps/chosen": -292.6138916015625, | |
| "logps/rejected": -314.90130615234375, | |
| "loss": 0.6083, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.21455985307693481, | |
| "rewards/margins": 0.4614453911781311, | |
| "rewards/rejected": -0.6760051846504211, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1871257444948096e-07, | |
| "logits/chosen": 0.26967114210128784, | |
| "logits/rejected": 0.47079068422317505, | |
| "logps/chosen": -297.1998291015625, | |
| "logps/rejected": -293.2596435546875, | |
| "loss": 0.744, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.2197016477584839, | |
| "rewards/margins": 0.3860716223716736, | |
| "rewards/rejected": -0.6057732105255127, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1561076868822755e-07, | |
| "logits/chosen": 0.01600978896021843, | |
| "logits/rejected": 0.6568925380706787, | |
| "logps/chosen": -254.8614044189453, | |
| "logps/rejected": -302.18023681640625, | |
| "loss": 0.6576, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.19405773282051086, | |
| "rewards/margins": 0.5532448887825012, | |
| "rewards/rejected": -0.7473027110099792, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.125377900869913e-07, | |
| "logits/chosen": -0.40245479345321655, | |
| "logits/rejected": 0.1633588671684265, | |
| "logps/chosen": -301.2410888671875, | |
| "logps/rejected": -386.735107421875, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.09816699475049973, | |
| "rewards/margins": 0.6523302793502808, | |
| "rewards/rejected": -0.7504973411560059, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.09494297815e-07, | |
| "logits/chosen": -0.19187738001346588, | |
| "logits/rejected": 1.2669118642807007, | |
| "logps/chosen": -223.18153381347656, | |
| "logps/rejected": -265.8275146484375, | |
| "loss": 0.6018, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.05469350516796112, | |
| "rewards/margins": 0.8093068599700928, | |
| "rewards/rejected": -0.8640003800392151, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.0648094471651722e-07, | |
| "logits/chosen": 0.06107745319604874, | |
| "logits/rejected": 1.0187910795211792, | |
| "logps/chosen": -298.7318420410156, | |
| "logps/rejected": -298.97607421875, | |
| "loss": 0.615, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.29442766308784485, | |
| "rewards/margins": 0.5528799295425415, | |
| "rewards/rejected": -0.8473076820373535, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.0349837717080347e-07, | |
| "logits/chosen": 0.2968660891056061, | |
| "logits/rejected": 1.533956527709961, | |
| "logps/chosen": -308.34906005859375, | |
| "logps/rejected": -352.53582763671875, | |
| "loss": 0.7653, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.2456754446029663, | |
| "rewards/margins": 0.691182553768158, | |
| "rewards/rejected": -0.9368579983711243, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.0054723495346482e-07, | |
| "logits/chosen": 0.6172692775726318, | |
| "logits/rejected": 1.4260393381118774, | |
| "logps/chosen": -255.27175903320312, | |
| "logps/rejected": -352.9705505371094, | |
| "loss": 0.5945, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -0.2834804654121399, | |
| "rewards/margins": 0.7214651107788086, | |
| "rewards/rejected": -1.0049455165863037, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.76281510992176e-08, | |
| "logits/chosen": 0.1523429900407791, | |
| "logits/rejected": 1.0697251558303833, | |
| "logps/chosen": -341.0992736816406, | |
| "logps/rejected": -314.7637023925781, | |
| "loss": 0.5933, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.2510679364204407, | |
| "rewards/margins": 0.7420140504837036, | |
| "rewards/rejected": -0.9930820465087891, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.474175176609956e-08, | |
| "logits/chosen": -0.027563797309994698, | |
| "logits/rejected": 0.730286717414856, | |
| "logps/chosen": -306.8202209472656, | |
| "logps/rejected": -328.6500244140625, | |
| "loss": 0.6264, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.21240682899951935, | |
| "rewards/margins": 0.526252269744873, | |
| "rewards/rejected": -0.7386590838432312, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.18886561011557e-08, | |
| "logits/chosen": 0.010053783655166626, | |
| "logits/rejected": 0.9134060144424438, | |
| "logps/chosen": -343.5555419921875, | |
| "logps/rejected": -344.8569030761719, | |
| "loss": 0.5614, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.1430027186870575, | |
| "rewards/margins": 0.8304474949836731, | |
| "rewards/rejected": -0.9734501838684082, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.906947610762825e-08, | |
| "logits/chosen": 0.48395290970802307, | |
| "logits/rejected": 0.8562638759613037, | |
| "logps/chosen": -303.1999816894531, | |
| "logps/rejected": -394.3343505859375, | |
| "loss": 0.6914, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.22828936576843262, | |
| "rewards/margins": 0.8299635648727417, | |
| "rewards/rejected": -1.0582529306411743, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.628481651367875e-08, | |
| "logits/chosen": 0.9650541543960571, | |
| "logits/rejected": 1.530713438987732, | |
| "logps/chosen": -317.75604248046875, | |
| "logps/rejected": -357.28875732421875, | |
| "loss": 0.569, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.3810655474662781, | |
| "rewards/margins": 0.5778117775917053, | |
| "rewards/rejected": -0.9588773250579834, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.353527464267104e-08, | |
| "logits/chosen": 0.648531436920166, | |
| "logits/rejected": 0.5499491095542908, | |
| "logps/chosen": -314.2027282714844, | |
| "logps/rejected": -314.53021240234375, | |
| "loss": 0.6003, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2263791263103485, | |
| "rewards/margins": 0.5237205624580383, | |
| "rewards/rejected": -0.7500997185707092, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.082144028504231e-08, | |
| "logits/chosen": 0.57253497838974, | |
| "logits/rejected": 1.7951874732971191, | |
| "logps/chosen": -329.2058410644531, | |
| "logps/rejected": -354.29132080078125, | |
| "loss": 0.6177, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.18844135105609894, | |
| "rewards/margins": 0.7768537402153015, | |
| "rewards/rejected": -0.9652950763702393, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.814389557179016e-08, | |
| "logits/chosen": 0.21237725019454956, | |
| "logits/rejected": 1.0711424350738525, | |
| "logps/chosen": -355.3842468261719, | |
| "logps/rejected": -369.0294494628906, | |
| "loss": 0.6145, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.2611796259880066, | |
| "rewards/margins": 0.7826185822486877, | |
| "rewards/rejected": -1.0437982082366943, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.550321484960251e-08, | |
| "logits/chosen": 0.7930352687835693, | |
| "logits/rejected": 1.6684021949768066, | |
| "logps/chosen": -339.7545471191406, | |
| "logps/rejected": -346.7519226074219, | |
| "loss": 0.6387, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -0.27698060870170593, | |
| "rewards/margins": 0.8290520906448364, | |
| "rewards/rejected": -1.1060327291488647, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.289996455765748e-08, | |
| "logits/chosen": 0.9145855903625488, | |
| "logits/rejected": 1.413263201713562, | |
| "logps/chosen": -331.789794921875, | |
| "logps/rejected": -378.784912109375, | |
| "loss": 0.7024, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.29554906487464905, | |
| "rewards/margins": 0.6969149708747864, | |
| "rewards/rejected": -0.9924640655517578, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.033470310611945e-08, | |
| "logits/chosen": 0.3435116112232208, | |
| "logits/rejected": 0.8178513050079346, | |
| "logps/chosen": -300.52178955078125, | |
| "logps/rejected": -318.1594543457031, | |
| "loss": 0.5883, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.3312419652938843, | |
| "rewards/margins": 0.560550332069397, | |
| "rewards/rejected": -0.8917922973632812, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.780798075635675e-08, | |
| "logits/chosen": 0.1530955582857132, | |
| "logits/rejected": 1.2445425987243652, | |
| "logps/chosen": -338.9671936035156, | |
| "logps/rejected": -394.33905029296875, | |
| "loss": 0.6703, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.10936611145734787, | |
| "rewards/margins": 0.7992250919342041, | |
| "rewards/rejected": -0.9085911512374878, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.532033950290885e-08, | |
| "logits/chosen": 0.3828258514404297, | |
| "logits/rejected": -0.013105906546115875, | |
| "logps/chosen": -247.27374267578125, | |
| "logps/rejected": -330.0480041503906, | |
| "loss": 0.7178, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.16955241560935974, | |
| "rewards/margins": 0.5796515941619873, | |
| "rewards/rejected": -0.7492039799690247, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.28723129572247e-08, | |
| "logits/chosen": -0.008059903979301453, | |
| "logits/rejected": 0.4361591339111328, | |
| "logps/chosen": -293.2403259277344, | |
| "logps/rejected": -323.6603698730469, | |
| "loss": 0.6125, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.1709737777709961, | |
| "rewards/margins": 0.5192134380340576, | |
| "rewards/rejected": -0.6901871562004089, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.046442623320145e-08, | |
| "logits/chosen": 0.3876532316207886, | |
| "logits/rejected": 0.8840723037719727, | |
| "logps/chosen": -317.1882019042969, | |
| "logps/rejected": -368.5491943359375, | |
| "loss": 0.5847, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.15435317158699036, | |
| "rewards/margins": 0.7234021425247192, | |
| "rewards/rejected": -0.8777552843093872, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.809719583454414e-08, | |
| "logits/chosen": 0.1544257402420044, | |
| "logits/rejected": 0.5524401664733887, | |
| "logps/chosen": -320.5746154785156, | |
| "logps/rejected": -355.8383483886719, | |
| "loss": 0.6153, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.30984005331993103, | |
| "rewards/margins": 0.5343888998031616, | |
| "rewards/rejected": -0.844228982925415, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.57711295439732e-08, | |
| "logits/chosen": -0.23453757166862488, | |
| "logits/rejected": 0.108016237616539, | |
| "logps/chosen": -317.7830810546875, | |
| "logps/rejected": -392.8485107421875, | |
| "loss": 0.7964, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.26463451981544495, | |
| "rewards/margins": 0.3382280468940735, | |
| "rewards/rejected": -0.6028625965118408, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.3486726314303175e-08, | |
| "logits/chosen": -0.37305283546447754, | |
| "logits/rejected": 0.704928994178772, | |
| "logps/chosen": -298.06561279296875, | |
| "logps/rejected": -325.4785461425781, | |
| "loss": 0.6078, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.00762226153165102, | |
| "rewards/margins": 0.6488328576087952, | |
| "rewards/rejected": -0.6564551591873169, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.1244476161413806e-08, | |
| "logits/chosen": 0.41815757751464844, | |
| "logits/rejected": 1.4270755052566528, | |
| "logps/chosen": -233.82427978515625, | |
| "logps/rejected": -301.65826416015625, | |
| "loss": 0.6295, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.19570133090019226, | |
| "rewards/margins": 0.6876392960548401, | |
| "rewards/rejected": -0.8833405375480652, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.904486005914027e-08, | |
| "logits/chosen": -0.3861064612865448, | |
| "logits/rejected": 0.04176158830523491, | |
| "logps/chosen": -294.285400390625, | |
| "logps/rejected": -333.8177185058594, | |
| "loss": 0.6621, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.11257711052894592, | |
| "rewards/margins": 0.5992448329925537, | |
| "rewards/rejected": -0.7118219137191772, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.688834983610082e-08, | |
| "logits/chosen": 0.12746843695640564, | |
| "logits/rejected": -0.0012746751308441162, | |
| "logps/chosen": -273.5867919921875, | |
| "logps/rejected": -360.38873291015625, | |
| "loss": 0.6068, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.17792265117168427, | |
| "rewards/margins": 0.6180278658866882, | |
| "rewards/rejected": -0.7959505915641785, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.477540807448832e-08, | |
| "logits/chosen": -0.5536242127418518, | |
| "logits/rejected": 0.9800190925598145, | |
| "logps/chosen": -342.70782470703125, | |
| "logps/rejected": -373.4602355957031, | |
| "loss": 0.6496, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.16559717059135437, | |
| "rewards/margins": 0.689116358757019, | |
| "rewards/rejected": -0.8547135591506958, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.270648801084295e-08, | |
| "logits/chosen": 0.006685473024845123, | |
| "logits/rejected": 0.7440766096115112, | |
| "logps/chosen": -366.8838195800781, | |
| "logps/rejected": -356.4771728515625, | |
| "loss": 0.628, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.31455904245376587, | |
| "rewards/margins": 0.559075653553009, | |
| "rewards/rejected": -0.8736346364021301, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.0682033438831584e-08, | |
| "logits/chosen": -0.7373438477516174, | |
| "logits/rejected": 0.6611166000366211, | |
| "logps/chosen": -362.8245849609375, | |
| "logps/rejected": -376.4985046386719, | |
| "loss": 0.603, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.004348123446106911, | |
| "rewards/margins": 0.7439255714416504, | |
| "rewards/rejected": -0.7482736706733704, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.8702478614051345e-08, | |
| "logits/chosen": -0.5010265707969666, | |
| "logits/rejected": 0.7126686573028564, | |
| "logps/chosen": -333.6928405761719, | |
| "logps/rejected": -362.8348388671875, | |
| "loss": 0.5581, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.09581595659255981, | |
| "rewards/margins": 0.7684364914894104, | |
| "rewards/rejected": -0.8642523884773254, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.676824816087978e-08, | |
| "logits/chosen": 0.4850161671638489, | |
| "logits/rejected": 1.3615456819534302, | |
| "logps/chosen": -282.8792419433594, | |
| "logps/rejected": -272.2681579589844, | |
| "loss": 0.628, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2777165174484253, | |
| "rewards/margins": 0.5231444835662842, | |
| "rewards/rejected": -0.8008609414100647, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.487975698139084e-08, | |
| "logits/chosen": 0.2012457251548767, | |
| "logits/rejected": 0.999991774559021, | |
| "logps/chosen": -308.87738037109375, | |
| "logps/rejected": -351.0428466796875, | |
| "loss": 0.6757, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.27788233757019043, | |
| "rewards/margins": 0.4390586316585541, | |
| "rewards/rejected": -0.7169409394264221, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.303741016635614e-08, | |
| "logits/chosen": 0.05412872135639191, | |
| "logits/rejected": 0.9229797124862671, | |
| "logps/chosen": -312.1148681640625, | |
| "logps/rejected": -305.1954650878906, | |
| "loss": 0.7495, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.04841495305299759, | |
| "rewards/margins": 0.6764097809791565, | |
| "rewards/rejected": -0.7248247861862183, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.12416029083514e-08, | |
| "logits/chosen": 0.9064224362373352, | |
| "logits/rejected": 0.9631800651550293, | |
| "logps/chosen": -251.423583984375, | |
| "logps/rejected": -331.3157958984375, | |
| "loss": 0.6625, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.3316621482372284, | |
| "rewards/margins": 0.5902231335639954, | |
| "rewards/rejected": -0.9218851923942566, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.9492720416985e-08, | |
| "logits/chosen": 0.768115222454071, | |
| "logits/rejected": 1.219833493232727, | |
| "logps/chosen": -290.3539733886719, | |
| "logps/rejected": -339.0583801269531, | |
| "loss": 0.6143, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.2172953188419342, | |
| "rewards/margins": 0.7515923380851746, | |
| "rewards/rejected": -0.9688878059387207, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.7791137836269158e-08, | |
| "logits/chosen": -0.08727936446666718, | |
| "logits/rejected": 1.5329773426055908, | |
| "logps/chosen": -255.40463256835938, | |
| "logps/rejected": -344.9212951660156, | |
| "loss": 0.6984, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.2787282168865204, | |
| "rewards/margins": 0.8177152276039124, | |
| "rewards/rejected": -1.0964434146881104, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.613722016414943e-08, | |
| "logits/chosen": 0.10900969803333282, | |
| "logits/rejected": 1.1554454565048218, | |
| "logps/chosen": -302.16650390625, | |
| "logps/rejected": -325.09332275390625, | |
| "loss": 0.6676, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.15435893833637238, | |
| "rewards/margins": 0.6768224835395813, | |
| "rewards/rejected": -0.8311813473701477, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.4531322174210973e-08, | |
| "logits/chosen": 0.2667168378829956, | |
| "logits/rejected": 0.06283846497535706, | |
| "logps/chosen": -270.20196533203125, | |
| "logps/rejected": -340.9342041015625, | |
| "loss": 0.6368, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.2986586093902588, | |
| "rewards/margins": 0.41698285937309265, | |
| "rewards/rejected": -0.7156413793563843, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.297378833957761e-08, | |
| "logits/chosen": 0.11617850512266159, | |
| "logits/rejected": 1.1254775524139404, | |
| "logps/chosen": -359.0841064453125, | |
| "logps/rejected": -370.67474365234375, | |
| "loss": 0.7271, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.12873072922229767, | |
| "rewards/margins": 0.7082916498184204, | |
| "rewards/rejected": -0.8370224237442017, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.1464952759020856e-08, | |
| "logits/chosen": 0.8191357851028442, | |
| "logits/rejected": 0.4740172028541565, | |
| "logps/chosen": -314.5917053222656, | |
| "logps/rejected": -338.27239990234375, | |
| "loss": 0.6364, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.10390536487102509, | |
| "rewards/margins": 0.45885345339775085, | |
| "rewards/rejected": -0.5627588033676147, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.0005139085293942e-08, | |
| "logits/chosen": -0.050498366355895996, | |
| "logits/rejected": 1.2883732318878174, | |
| "logps/chosen": -322.22113037109375, | |
| "logps/rejected": -332.8543701171875, | |
| "loss": 0.6629, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.1322225034236908, | |
| "rewards/margins": 0.6378248333930969, | |
| "rewards/rejected": -0.7700474262237549, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.8594660455706763e-08, | |
| "logits/chosen": 0.42545390129089355, | |
| "logits/rejected": 0.9098062515258789, | |
| "logps/chosen": -289.212646484375, | |
| "logps/rejected": -291.0245666503906, | |
| "loss": 0.6363, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.17995241284370422, | |
| "rewards/margins": 0.5931330323219299, | |
| "rewards/rejected": -0.7730855345726013, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.7233819424956247e-08, | |
| "logits/chosen": 0.331084281206131, | |
| "logits/rejected": 0.931097686290741, | |
| "logps/chosen": -289.7149353027344, | |
| "logps/rejected": -335.7164306640625, | |
| "loss": 0.5931, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.14608880877494812, | |
| "rewards/margins": 0.6044370532035828, | |
| "rewards/rejected": -0.7505258321762085, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.5922907900227017e-08, | |
| "logits/chosen": -0.002596009522676468, | |
| "logits/rejected": 1.004823088645935, | |
| "logps/chosen": -328.34918212890625, | |
| "logps/rejected": -338.5020751953125, | |
| "loss": 0.6564, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.28392377495765686, | |
| "rewards/margins": 0.8925819993019104, | |
| "rewards/rejected": -1.1765056848526, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.4662207078575684e-08, | |
| "logits/chosen": 0.22775022685527802, | |
| "logits/rejected": 0.9491074681282043, | |
| "logps/chosen": -307.7417907714844, | |
| "logps/rejected": -377.443115234375, | |
| "loss": 0.6774, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -0.16252580285072327, | |
| "rewards/margins": 0.724989116191864, | |
| "rewards/rejected": -0.8875149488449097, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.345198738661285e-08, | |
| "logits/chosen": 0.4222959280014038, | |
| "logits/rejected": 1.0754032135009766, | |
| "logps/chosen": -275.5315856933594, | |
| "logps/rejected": -338.4194030761719, | |
| "loss": 0.607, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.19549164175987244, | |
| "rewards/margins": 0.6255277991294861, | |
| "rewards/rejected": -0.8210194110870361, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.2292508422495157e-08, | |
| "logits/chosen": 0.35431140661239624, | |
| "logits/rejected": 0.6342911124229431, | |
| "logps/chosen": -245.69967651367188, | |
| "logps/rejected": -307.9548034667969, | |
| "loss": 0.5931, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.08900564908981323, | |
| "rewards/margins": 0.7763758897781372, | |
| "rewards/rejected": -0.8653814792633057, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.118401890024001e-08, | |
| "logits/chosen": -0.6884260773658752, | |
| "logits/rejected": 0.5942091941833496, | |
| "logps/chosen": -309.41900634765625, | |
| "logps/rejected": -392.107421875, | |
| "loss": 0.8082, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.12636291980743408, | |
| "rewards/margins": 0.7045049667358398, | |
| "rewards/rejected": -0.8308678865432739, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0126756596375685e-08, | |
| "logits/chosen": 0.28693687915802, | |
| "logits/rejected": 0.19982758164405823, | |
| "logps/chosen": -284.8650207519531, | |
| "logps/rejected": -322.2666015625, | |
| "loss": 0.7014, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.29686933755874634, | |
| "rewards/margins": 0.5921163558959961, | |
| "rewards/rejected": -0.8889857530593872, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.12094829893642e-09, | |
| "logits/chosen": -0.22270016372203827, | |
| "logits/rejected": 1.3697696924209595, | |
| "logps/chosen": -282.0764465332031, | |
| "logps/rejected": -308.25970458984375, | |
| "loss": 0.7014, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.24073350429534912, | |
| "rewards/margins": 0.5527843236923218, | |
| "rewards/rejected": -0.7935178279876709, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.166809758815895e-09, | |
| "logits/chosen": -0.42256081104278564, | |
| "logits/rejected": 0.24574501812458038, | |
| "logps/chosen": -287.09100341796875, | |
| "logps/rejected": -344.0694580078125, | |
| "loss": 0.6457, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.2197842299938202, | |
| "rewards/margins": 0.44125646352767944, | |
| "rewards/rejected": -0.661040723323822, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.2645456434869965e-09, | |
| "logits/chosen": -0.25060874223709106, | |
| "logits/rejected": 0.9384374618530273, | |
| "logps/chosen": -278.2327880859375, | |
| "logps/rejected": -328.0905456542969, | |
| "loss": 0.6304, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.1905202716588974, | |
| "rewards/margins": 0.6406797766685486, | |
| "rewards/rejected": -0.8312000036239624, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.414349493100129e-09, | |
| "logits/chosen": 0.5399929881095886, | |
| "logits/rejected": 1.2095367908477783, | |
| "logps/chosen": -292.74786376953125, | |
| "logps/rejected": -348.2053527832031, | |
| "loss": 0.6079, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.0946485847234726, | |
| "rewards/margins": 0.6198921799659729, | |
| "rewards/rejected": -0.7145407795906067, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.616403678967624e-09, | |
| "logits/chosen": -0.07524357736110687, | |
| "logits/rejected": 1.0452533960342407, | |
| "logps/chosen": -356.8501281738281, | |
| "logps/rejected": -411.29931640625, | |
| "loss": 0.707, | |
| "rewards/accuracies": 0.609375, | |
| "rewards/chosen": -0.12035229802131653, | |
| "rewards/margins": 0.5983991622924805, | |
| "rewards/rejected": -0.7187514901161194, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.8708793644441086e-09, | |
| "logits/chosen": -0.13995477557182312, | |
| "logits/rejected": 0.6635825634002686, | |
| "logps/chosen": -278.5004577636719, | |
| "logps/rejected": -310.28857421875, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.18752513825893402, | |
| "rewards/margins": 0.5632314085960388, | |
| "rewards/rejected": -0.750756561756134, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.1779364682113794e-09, | |
| "logits/chosen": 0.5096856951713562, | |
| "logits/rejected": 1.3543939590454102, | |
| "logps/chosen": -241.704833984375, | |
| "logps/rejected": -314.39495849609375, | |
| "loss": 0.6309, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.21836420893669128, | |
| "rewards/margins": 0.6935679912567139, | |
| "rewards/rejected": -0.911932110786438, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.5377236299748147e-09, | |
| "logits/chosen": 0.5073993802070618, | |
| "logits/rejected": 0.9775093197822571, | |
| "logps/chosen": -270.1071472167969, | |
| "logps/rejected": -317.7763977050781, | |
| "loss": 0.6822, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -0.27697494626045227, | |
| "rewards/margins": 0.5462560653686523, | |
| "rewards/rejected": -0.8232309818267822, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.9503781785795713e-09, | |
| "logits/chosen": -0.3081734776496887, | |
| "logits/rejected": 0.7983647584915161, | |
| "logps/chosen": -324.9632568359375, | |
| "logps/rejected": -354.402099609375, | |
| "loss": 0.5723, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.0758206844329834, | |
| "rewards/margins": 0.8021882772445679, | |
| "rewards/rejected": -0.878009021282196, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.416026102552732e-09, | |
| "logits/chosen": 0.3920813202857971, | |
| "logits/rejected": 1.4421265125274658, | |
| "logps/chosen": -302.9536437988281, | |
| "logps/rejected": -352.2631530761719, | |
| "loss": 0.6248, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.22983548045158386, | |
| "rewards/margins": 0.8996194005012512, | |
| "rewards/rejected": -1.1294549703598022, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.9347820230782295e-09, | |
| "logits/chosen": 0.027735590934753418, | |
| "logits/rejected": 0.36365261673927307, | |
| "logps/chosen": -318.6025390625, | |
| "logps/rejected": -359.25799560546875, | |
| "loss": 0.6865, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.2018052041530609, | |
| "rewards/margins": 0.5466940402984619, | |
| "rewards/rejected": -0.7484991550445557, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.5067491694100153e-09, | |
| "logits/chosen": -0.1223829835653305, | |
| "logits/rejected": 0.2881450355052948, | |
| "logps/chosen": -289.8230285644531, | |
| "logps/rejected": -360.09906005859375, | |
| "loss": 0.6747, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.1440747082233429, | |
| "rewards/margins": 0.8571825623512268, | |
| "rewards/rejected": -1.0012574195861816, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.1320193567288527e-09, | |
| "logits/chosen": 0.2805168032646179, | |
| "logits/rejected": 0.7300337553024292, | |
| "logps/chosen": -256.3475341796875, | |
| "logps/rejected": -297.0839538574219, | |
| "loss": 0.564, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2623736560344696, | |
| "rewards/margins": 0.49820947647094727, | |
| "rewards/rejected": -0.7605831027030945, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.106729664475176e-10, | |
| "logits/chosen": -0.15276291966438293, | |
| "logits/rejected": 0.6538082957267761, | |
| "logps/chosen": -352.9091796875, | |
| "logps/rejected": -393.73797607421875, | |
| "loss": 0.6624, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.07146578282117844, | |
| "rewards/margins": 0.7627599239349365, | |
| "rewards/rejected": -0.8342257738113403, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.427789289685347e-10, | |
| "logits/chosen": 0.2313707172870636, | |
| "logits/rejected": 1.0570056438446045, | |
| "logps/chosen": -367.5287170410156, | |
| "logps/rejected": -360.8330078125, | |
| "loss": 0.6087, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.035427868366241455, | |
| "rewards/margins": 0.8285888433456421, | |
| "rewards/rejected": -0.8640167713165283, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.2839470889836627e-10, | |
| "logits/chosen": -0.49842333793640137, | |
| "logits/rejected": 0.5836061239242554, | |
| "logps/chosen": -309.240234375, | |
| "logps/rejected": -367.9534912109375, | |
| "loss": 0.5799, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -0.18272840976715088, | |
| "rewards/margins": 0.7705845236778259, | |
| "rewards/rejected": -0.9533129930496216, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.6756629272085544e-10, | |
| "logits/chosen": -0.1911655068397522, | |
| "logits/rejected": 0.9592544436454773, | |
| "logps/chosen": -297.9132385253906, | |
| "logps/rejected": -368.43060302734375, | |
| "loss": 0.6029, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.12253910303115845, | |
| "rewards/margins": 0.7276279926300049, | |
| "rewards/rejected": -0.8501670956611633, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.032817893297793e-11, | |
| "logits/chosen": -0.25915825366973877, | |
| "logits/rejected": 1.385116696357727, | |
| "logps/chosen": -289.7322998046875, | |
| "logps/rejected": -346.7291564941406, | |
| "loss": 0.685, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.21489854156970978, | |
| "rewards/margins": 0.8257070779800415, | |
| "rewards/rejected": -1.0406055450439453, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.7033706447061635e-12, | |
| "logits/chosen": 0.4867172837257385, | |
| "logits/rejected": 1.1571818590164185, | |
| "logps/chosen": -337.714599609375, | |
| "logps/rejected": -358.3683166503906, | |
| "loss": 0.6851, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.18006440997123718, | |
| "rewards/margins": 0.681747317314148, | |
| "rewards/rejected": -0.8618118166923523, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 477, | |
| "total_flos": 0.0, | |
| "train_loss": 0.6847315603082285, | |
| "train_runtime": 17806.8146, | |
| "train_samples_per_second": 3.433, | |
| "train_steps_per_second": 0.027 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 477, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |