| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 4.0, |
| "eval_steps": 500, |
| "global_step": 516, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.07782101167315175, |
| "grad_norm": 309.9635314941406, |
| "learning_rate": 3.461538461538461e-07, |
| "logits/chosen": 2.9046390056610107, |
| "logits/rejected": 2.998429775238037, |
| "logps/chosen": -250.68215942382812, |
| "logps/rejected": -271.45562744140625, |
| "loss": 0.8341, |
| "rewards/accuracies": 0.534375011920929, |
| "rewards/chosen": 0.07394347339868546, |
| "rewards/margins": 0.05555616691708565, |
| "rewards/rejected": 0.018387287855148315, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.1556420233463035, |
| "grad_norm": 180.70631408691406, |
| "learning_rate": 7.307692307692307e-07, |
| "logits/chosen": 2.893648386001587, |
| "logits/rejected": 2.9738540649414062, |
| "logps/chosen": -255.9049835205078, |
| "logps/rejected": -271.83197021484375, |
| "loss": 0.6594, |
| "rewards/accuracies": 0.6656249761581421, |
| "rewards/chosen": 0.6080068349838257, |
| "rewards/margins": 0.7042483687400818, |
| "rewards/rejected": -0.09624147415161133, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.23346303501945526, |
| "grad_norm": 152.26849365234375, |
| "learning_rate": 1.1153846153846154e-06, |
| "logits/chosen": 2.938114643096924, |
| "logits/rejected": 2.979564666748047, |
| "logps/chosen": -260.2184143066406, |
| "logps/rejected": -274.0084228515625, |
| "loss": 0.4358, |
| "rewards/accuracies": 0.78125, |
| "rewards/chosen": 2.1423935890197754, |
| "rewards/margins": 2.729647159576416, |
| "rewards/rejected": -0.5872532725334167, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.311284046692607, |
| "grad_norm": 61.16145324707031, |
| "learning_rate": 1.5e-06, |
| "logits/chosen": 2.962891101837158, |
| "logits/rejected": 3.0372633934020996, |
| "logps/chosen": -264.4953918457031, |
| "logps/rejected": -278.98541259765625, |
| "loss": 0.2424, |
| "rewards/accuracies": 0.8843749761581421, |
| "rewards/chosen": 3.3210906982421875, |
| "rewards/margins": 5.616915225982666, |
| "rewards/rejected": -2.2958242893218994, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.38910505836575876, |
| "grad_norm": 72.79554748535156, |
| "learning_rate": 1.8846153846153845e-06, |
| "logits/chosen": 2.916743278503418, |
| "logits/rejected": 2.9901554584503174, |
| "logps/chosen": -242.43307495117188, |
| "logps/rejected": -276.2446594238281, |
| "loss": 0.0823, |
| "rewards/accuracies": 0.9671875238418579, |
| "rewards/chosen": 4.046940803527832, |
| "rewards/margins": 9.120622634887695, |
| "rewards/rejected": -5.0736823081970215, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.4669260700389105, |
| "grad_norm": 51.30080795288086, |
| "learning_rate": 1.998877080632712e-06, |
| "logits/chosen": 2.911020517349243, |
| "logits/rejected": 2.9875214099884033, |
| "logps/chosen": -252.1035614013672, |
| "logps/rejected": -278.92047119140625, |
| "loss": 0.0745, |
| "rewards/accuracies": 0.973437488079071, |
| "rewards/chosen": 3.5658841133117676, |
| "rewards/margins": 11.655838966369629, |
| "rewards/rejected": -8.089953422546387, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.5447470817120622, |
| "grad_norm": 69.88111877441406, |
| "learning_rate": 1.9933831375786213e-06, |
| "logits/chosen": 2.9574735164642334, |
| "logits/rejected": 3.0135583877563477, |
| "logps/chosen": -250.85684204101562, |
| "logps/rejected": -278.03582763671875, |
| "loss": 0.0516, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": 2.9169883728027344, |
| "rewards/margins": 13.218663215637207, |
| "rewards/rejected": -10.301675796508789, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.622568093385214, |
| "grad_norm": 16.288284301757812, |
| "learning_rate": 1.9833370645493046e-06, |
| "logits/chosen": 2.9377493858337402, |
| "logits/rejected": 3.007570743560791, |
| "logps/chosen": -253.7940216064453, |
| "logps/rejected": -284.8679504394531, |
| "loss": 0.0387, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": 3.284533977508545, |
| "rewards/margins": 16.070585250854492, |
| "rewards/rejected": -12.786050796508789, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.7003891050583657, |
| "grad_norm": 8.968023300170898, |
| "learning_rate": 1.9687848971864386e-06, |
| "logits/chosen": 2.9306931495666504, |
| "logits/rejected": 2.964722156524658, |
| "logps/chosen": -252.9971160888672, |
| "logps/rejected": -275.87188720703125, |
| "loss": 0.0492, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": 3.5943214893341064, |
| "rewards/margins": 17.51387596130371, |
| "rewards/rejected": -13.9195556640625, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.7782101167315175, |
| "grad_norm": 2.590059757232666, |
| "learning_rate": 1.9497933200901047e-06, |
| "logits/chosen": 2.9647717475891113, |
| "logits/rejected": 3.0007081031799316, |
| "logps/chosen": -267.41375732421875, |
| "logps/rejected": -281.07513427734375, |
| "loss": 0.0337, |
| "rewards/accuracies": 0.9937499761581421, |
| "rewards/chosen": 4.1861348152160645, |
| "rewards/margins": 18.587297439575195, |
| "rewards/rejected": -14.401159286499023, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.8560311284046692, |
| "grad_norm": 77.77639770507812, |
| "learning_rate": 1.926449361239848e-06, |
| "logits/chosen": 2.8810315132141113, |
| "logits/rejected": 2.9851112365722656, |
| "logps/chosen": -253.496337890625, |
| "logps/rejected": -284.06610107421875, |
| "loss": 0.0533, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": 4.9546074867248535, |
| "rewards/margins": 20.860849380493164, |
| "rewards/rejected": -15.906242370605469, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.933852140077821, |
| "grad_norm": 127.29902648925781, |
| "learning_rate": 1.8988599931931866e-06, |
| "logits/chosen": 2.8698949813842773, |
| "logits/rejected": 2.943873405456543, |
| "logps/chosen": -246.2525177001953, |
| "logps/rejected": -279.6893310546875, |
| "loss": 0.0179, |
| "rewards/accuracies": 0.995312511920929, |
| "rewards/chosen": 3.602506160736084, |
| "rewards/margins": 21.305601119995117, |
| "rewards/rejected": -17.703094482421875, |
| "step": 120 |
| }, |
| { |
| "epoch": 1.0077821011673151, |
| "grad_norm": 0.10526127368211746, |
| "learning_rate": 1.8671516428890646e-06, |
| "logits/chosen": 2.960531711578369, |
| "logits/rejected": 3.0081727504730225, |
| "logps/chosen": -251.44546508789062, |
| "logps/rejected": -297.5830383300781, |
| "loss": 0.0043, |
| "rewards/accuracies": 0.9983552694320679, |
| "rewards/chosen": 5.055600166320801, |
| "rewards/margins": 22.011268615722656, |
| "rewards/rejected": -16.955669403076172, |
| "step": 130 |
| }, |
| { |
| "epoch": 1.0856031128404668, |
| "grad_norm": 3.8229660987854004, |
| "learning_rate": 1.8314696123025452e-06, |
| "logits/chosen": 2.9409918785095215, |
| "logits/rejected": 2.994568109512329, |
| "logps/chosen": -251.5284423828125, |
| "logps/rejected": -283.5034484863281, |
| "loss": 0.002, |
| "rewards/accuracies": 0.9984375238418579, |
| "rewards/chosen": 5.045574188232422, |
| "rewards/margins": 24.76313591003418, |
| "rewards/rejected": -19.717559814453125, |
| "step": 140 |
| }, |
| { |
| "epoch": 1.1634241245136188, |
| "grad_norm": 0.1067415103316307, |
| "learning_rate": 1.7919774126055671e-06, |
| "logits/chosen": 2.9014945030212402, |
| "logits/rejected": 3.0171058177948, |
| "logps/chosen": -244.58154296875, |
| "logps/rejected": -300.16827392578125, |
| "loss": 0.0005, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 5.922303199768066, |
| "rewards/margins": 27.960514068603516, |
| "rewards/rejected": -22.038206100463867, |
| "step": 150 |
| }, |
| { |
| "epoch": 1.2412451361867705, |
| "grad_norm": 0.11893969029188156, |
| "learning_rate": 1.7488560148849425e-06, |
| "logits/chosen": 2.930441379547119, |
| "logits/rejected": 2.9720261096954346, |
| "logps/chosen": -257.78741455078125, |
| "logps/rejected": -283.66986083984375, |
| "loss": 0.0044, |
| "rewards/accuracies": 0.9984375238418579, |
| "rewards/chosen": 4.894805908203125, |
| "rewards/margins": 27.222713470458984, |
| "rewards/rejected": -22.32790756225586, |
| "step": 160 |
| }, |
| { |
| "epoch": 1.3190661478599222, |
| "grad_norm": 0.019848737865686417, |
| "learning_rate": 1.7023030208511418e-06, |
| "logits/chosen": 2.9215731620788574, |
| "logits/rejected": 2.998058795928955, |
| "logps/chosen": -257.8537902832031, |
| "logps/rejected": -300.69781494140625, |
| "loss": 0.0005, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.8431878089904785, |
| "rewards/margins": 27.765335083007812, |
| "rewards/rejected": -22.92214584350586, |
| "step": 170 |
| }, |
| { |
| "epoch": 1.3968871595330739, |
| "grad_norm": 0.013338716700673103, |
| "learning_rate": 1.6525317573380523e-06, |
| "logits/chosen": 2.905754566192627, |
| "logits/rejected": 2.9750256538391113, |
| "logps/chosen": -256.52801513671875, |
| "logps/rejected": -292.92559814453125, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 5.072903633117676, |
| "rewards/margins": 28.44639015197754, |
| "rewards/rejected": -23.373485565185547, |
| "step": 180 |
| }, |
| { |
| "epoch": 1.4747081712062258, |
| "grad_norm": 1.3543317317962646, |
| "learning_rate": 1.5997702987431278e-06, |
| "logits/chosen": 2.8545498847961426, |
| "logits/rejected": 2.93253755569458, |
| "logps/chosen": -249.04165649414062, |
| "logps/rejected": -290.2785949707031, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.263331413269043, |
| "rewards/margins": 29.8680362701416, |
| "rewards/rejected": -25.604705810546875, |
| "step": 190 |
| }, |
| { |
| "epoch": 1.5525291828793775, |
| "grad_norm": 0.0023172160144895315, |
| "learning_rate": 1.5442604218875369e-06, |
| "logits/chosen": 2.8726515769958496, |
| "logits/rejected": 2.96691632270813, |
| "logps/chosen": -235.4767608642578, |
| "logps/rejected": -299.6068420410156, |
| "loss": 0.0174, |
| "rewards/accuracies": 0.9984375238418579, |
| "rewards/chosen": 4.076581001281738, |
| "rewards/margins": 29.738256454467773, |
| "rewards/rejected": -25.661672592163086, |
| "step": 200 |
| }, |
| { |
| "epoch": 1.6303501945525292, |
| "grad_norm": 0.12606221437454224, |
| "learning_rate": 1.4862564980856257e-06, |
| "logits/chosen": 2.866564989089966, |
| "logits/rejected": 2.9289443492889404, |
| "logps/chosen": -251.0595245361328, |
| "logps/rejected": -291.28009033203125, |
| "loss": 0.0002, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 2.828310012817383, |
| "rewards/margins": 29.65151023864746, |
| "rewards/rejected": -26.823205947875977, |
| "step": 210 |
| }, |
| { |
| "epoch": 1.708171206225681, |
| "grad_norm": 1.1696350574493408, |
| "learning_rate": 1.4260243275007265e-06, |
| "logits/chosen": 2.876115083694458, |
| "logits/rejected": 2.9629712104797363, |
| "logps/chosen": -252.755126953125, |
| "logps/rejected": -293.724609375, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 2.9194681644439697, |
| "rewards/margins": 30.779850006103516, |
| "rewards/rejected": -27.86037826538086, |
| "step": 220 |
| }, |
| { |
| "epoch": 1.7859922178988326, |
| "grad_norm": 0.7943425178527832, |
| "learning_rate": 1.3638399211288185e-06, |
| "logits/chosen": 2.853001356124878, |
| "logits/rejected": 2.91581392288208, |
| "logps/chosen": -248.50790405273438, |
| "logps/rejected": -303.24749755859375, |
| "loss": 0.0009, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 3.9409661293029785, |
| "rewards/margins": 31.501480102539062, |
| "rewards/rejected": -27.560516357421875, |
| "step": 230 |
| }, |
| { |
| "epoch": 1.8638132295719845, |
| "grad_norm": 22.544157028198242, |
| "learning_rate": 1.2999882359915478e-06, |
| "logits/chosen": 2.838768482208252, |
| "logits/rejected": 2.8869760036468506, |
| "logps/chosen": -257.80755615234375, |
| "logps/rejected": -301.2661437988281, |
| "loss": 0.0009, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.118237018585205, |
| "rewards/margins": 31.00071144104004, |
| "rewards/rejected": -26.882476806640625, |
| "step": 240 |
| }, |
| { |
| "epoch": 1.9416342412451362, |
| "grad_norm": 0.08505785465240479, |
| "learning_rate": 1.2347618693345148e-06, |
| "logits/chosen": 2.8469579219818115, |
| "logits/rejected": 2.91058349609375, |
| "logps/chosen": -274.753173828125, |
| "logps/rejected": -310.84820556640625, |
| "loss": 0.0097, |
| "rewards/accuracies": 0.996874988079071, |
| "rewards/chosen": 3.213207960128784, |
| "rewards/margins": 32.35531234741211, |
| "rewards/rejected": -29.142101287841797, |
| "step": 250 |
| }, |
| { |
| "epoch": 2.0155642023346303, |
| "grad_norm": 0.15910857915878296, |
| "learning_rate": 1.1684597178146115e-06, |
| "logits/chosen": 2.820605516433716, |
| "logits/rejected": 2.9221818447113037, |
| "logps/chosen": -240.77276611328125, |
| "logps/rejected": -284.52947998046875, |
| "loss": 0.0095, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.895946979522705, |
| "rewards/margins": 31.758358001708984, |
| "rewards/rejected": -26.862411499023438, |
| "step": 260 |
| }, |
| { |
| "epoch": 2.093385214007782, |
| "grad_norm": 0.036736179143190384, |
| "learning_rate": 1.1013856078206078e-06, |
| "logits/chosen": 2.7885546684265137, |
| "logits/rejected": 2.9099419116973877, |
| "logps/chosen": -255.7447509765625, |
| "logps/rejected": -300.6316223144531, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.1806111335754395, |
| "rewards/margins": 32.345706939697266, |
| "rewards/rejected": -28.16509437561035, |
| "step": 270 |
| }, |
| { |
| "epoch": 2.1712062256809337, |
| "grad_norm": 0.010115005075931549, |
| "learning_rate": 1.0338469032034844e-06, |
| "logits/chosen": 2.880408763885498, |
| "logits/rejected": 2.913715124130249, |
| "logps/chosen": -244.33837890625, |
| "logps/rejected": -284.74456787109375, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 5.522616386413574, |
| "rewards/margins": 32.65098571777344, |
| "rewards/rejected": -27.128368377685547, |
| "step": 280 |
| }, |
| { |
| "epoch": 2.2490272373540856, |
| "grad_norm": 0.0006411768263205886, |
| "learning_rate": 9.661530967965155e-07, |
| "logits/chosen": 2.8637478351593018, |
| "logits/rejected": 2.8776774406433105, |
| "logps/chosen": -251.5993194580078, |
| "logps/rejected": -285.7321472167969, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 5.7556328773498535, |
| "rewards/margins": 32.75992202758789, |
| "rewards/rejected": -27.004287719726562, |
| "step": 290 |
| }, |
| { |
| "epoch": 2.3268482490272375, |
| "grad_norm": 0.010583557188510895, |
| "learning_rate": 8.986143921793921e-07, |
| "logits/chosen": 2.8255324363708496, |
| "logits/rejected": 2.884343385696411, |
| "logps/chosen": -256.0443115234375, |
| "logps/rejected": -298.6083068847656, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.289700984954834, |
| "rewards/margins": 32.55815887451172, |
| "rewards/rejected": -28.26845359802246, |
| "step": 300 |
| }, |
| { |
| "epoch": 2.404669260700389, |
| "grad_norm": 0.05514160171151161, |
| "learning_rate": 8.315402821853885e-07, |
| "logits/chosen": 2.8168251514434814, |
| "logits/rejected": 2.911100387573242, |
| "logps/chosen": -242.12930297851562, |
| "logps/rejected": -311.6902770996094, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.304950714111328, |
| "rewards/margins": 33.0163459777832, |
| "rewards/rejected": -28.71139144897461, |
| "step": 310 |
| }, |
| { |
| "epoch": 2.482490272373541, |
| "grad_norm": 0.0014977785758674145, |
| "learning_rate": 7.65238130665485e-07, |
| "logits/chosen": 2.8739066123962402, |
| "logits/rejected": 2.9437661170959473, |
| "logps/chosen": -254.6911163330078, |
| "logps/rejected": -298.3764953613281, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.761944770812988, |
| "rewards/margins": 32.24135208129883, |
| "rewards/rejected": -27.479406356811523, |
| "step": 320 |
| }, |
| { |
| "epoch": 2.5603112840466924, |
| "grad_norm": 0.003176012309268117, |
| "learning_rate": 7.000117640084525e-07, |
| "logits/chosen": 2.849517822265625, |
| "logits/rejected": 2.9144535064697266, |
| "logps/chosen": -253.174560546875, |
| "logps/rejected": -302.97296142578125, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 5.072056770324707, |
| "rewards/margins": 33.25294876098633, |
| "rewards/rejected": -28.180889129638672, |
| "step": 330 |
| }, |
| { |
| "epoch": 2.6381322957198443, |
| "grad_norm": 0.007333577610552311, |
| "learning_rate": 6.361600788711815e-07, |
| "logits/chosen": 2.801302671432495, |
| "logits/rejected": 2.9121482372283936, |
| "logps/chosen": -259.02398681640625, |
| "logps/rejected": -303.2242431640625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.999932289123535, |
| "rewards/margins": 32.66350555419922, |
| "rewards/rejected": -27.66357421875, |
| "step": 340 |
| }, |
| { |
| "epoch": 2.7159533073929962, |
| "grad_norm": 0.0016177126672118902, |
| "learning_rate": 5.739756724992736e-07, |
| "logits/chosen": 2.85249662399292, |
| "logits/rejected": 2.9361824989318848, |
| "logps/chosen": -244.018310546875, |
| "logps/rejected": -286.9032897949219, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.892648220062256, |
| "rewards/margins": 32.217613220214844, |
| "rewards/rejected": -27.324966430664062, |
| "step": 350 |
| }, |
| { |
| "epoch": 2.7937743190661477, |
| "grad_norm": 0.3721546530723572, |
| "learning_rate": 5.137435019143744e-07, |
| "logits/chosen": 2.801509141921997, |
| "logits/rejected": 2.8811545372009277, |
| "logps/chosen": -257.4078063964844, |
| "logps/rejected": -306.99542236328125, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.106388092041016, |
| "rewards/margins": 33.20668411254883, |
| "rewards/rejected": -29.100292205810547, |
| "step": 360 |
| }, |
| { |
| "epoch": 2.8715953307392996, |
| "grad_norm": 0.004263872280716896, |
| "learning_rate": 4.557395781124631e-07, |
| "logits/chosen": 2.853585958480835, |
| "logits/rejected": 2.9362740516662598, |
| "logps/chosen": -265.7266845703125, |
| "logps/rejected": -301.75091552734375, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.696002960205078, |
| "rewards/margins": 32.666351318359375, |
| "rewards/rejected": -27.9703426361084, |
| "step": 370 |
| }, |
| { |
| "epoch": 2.9494163424124515, |
| "grad_norm": 6.243278503417969, |
| "learning_rate": 4.0022970125687217e-07, |
| "logits/chosen": 2.8308067321777344, |
| "logits/rejected": 2.924168586730957, |
| "logps/chosen": -253.6175537109375, |
| "logps/rejected": -304.56488037109375, |
| "loss": 0.0001, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.487698554992676, |
| "rewards/margins": 33.75385284423828, |
| "rewards/rejected": -29.26615333557129, |
| "step": 380 |
| }, |
| { |
| "epoch": 3.0233463035019454, |
| "grad_norm": 0.010610430501401424, |
| "learning_rate": 3.474682426619474e-07, |
| "logits/chosen": 2.86039662361145, |
| "logits/rejected": 2.955193042755127, |
| "logps/chosen": -244.86038208007812, |
| "logps/rejected": -298.5942077636719, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 5.364467144012451, |
| "rewards/margins": 33.471588134765625, |
| "rewards/rejected": -28.107120513916016, |
| "step": 390 |
| }, |
| { |
| "epoch": 3.1011673151750974, |
| "grad_norm": 0.0002077793178614229, |
| "learning_rate": 2.976969791488586e-07, |
| "logits/chosen": 2.794825792312622, |
| "logits/rejected": 2.913062572479248, |
| "logps/chosen": -249.8114471435547, |
| "logps/rejected": -304.4586181640625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.951796054840088, |
| "rewards/margins": 33.71596908569336, |
| "rewards/rejected": -28.764175415039062, |
| "step": 400 |
| }, |
| { |
| "epoch": 3.178988326848249, |
| "grad_norm": 0.0018190924311056733, |
| "learning_rate": 2.511439851150573e-07, |
| "logits/chosen": 2.846440076828003, |
| "logits/rejected": 2.9116950035095215, |
| "logps/chosen": -245.32241821289062, |
| "logps/rejected": -299.29278564453125, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.315257549285889, |
| "rewards/margins": 33.424171447753906, |
| "rewards/rejected": -29.10890769958496, |
| "step": 410 |
| }, |
| { |
| "epoch": 3.2568093385214008, |
| "grad_norm": 0.0022656808141618967, |
| "learning_rate": 2.080225873944328e-07, |
| "logits/chosen": 2.8200035095214844, |
| "logits/rejected": 2.9111595153808594, |
| "logps/chosen": -258.51519775390625, |
| "logps/rejected": -303.6025390625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.961642265319824, |
| "rewards/margins": 33.01478958129883, |
| "rewards/rejected": -28.053142547607422, |
| "step": 420 |
| }, |
| { |
| "epoch": 3.3346303501945527, |
| "grad_norm": 0.006798978429287672, |
| "learning_rate": 1.6853038769745465e-07, |
| "logits/chosen": 2.8389997482299805, |
| "logits/rejected": 2.8933396339416504, |
| "logps/chosen": -245.7606658935547, |
| "logps/rejected": -295.68115234375, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.181079387664795, |
| "rewards/margins": 32.27184295654297, |
| "rewards/rejected": -28.090763092041016, |
| "step": 430 |
| }, |
| { |
| "epoch": 3.412451361867704, |
| "grad_norm": 0.00100744201336056, |
| "learning_rate": 1.3284835711093535e-07, |
| "logits/chosen": 2.8530194759368896, |
| "logits/rejected": 2.9206631183624268, |
| "logps/chosen": -258.67431640625, |
| "logps/rejected": -298.80755615234375, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.105883598327637, |
| "rewards/margins": 32.130775451660156, |
| "rewards/rejected": -28.024892807006836, |
| "step": 440 |
| }, |
| { |
| "epoch": 3.490272373540856, |
| "grad_norm": 0.0001550566521473229, |
| "learning_rate": 1.0114000680681356e-07, |
| "logits/chosen": 2.81189227104187, |
| "logits/rejected": 2.925079107284546, |
| "logps/chosen": -253.63217163085938, |
| "logps/rejected": -299.74273681640625, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 6.281905174255371, |
| "rewards/margins": 33.87714385986328, |
| "rewards/rejected": -27.595239639282227, |
| "step": 450 |
| }, |
| { |
| "epoch": 3.5680933852140075, |
| "grad_norm": 0.01041385531425476, |
| "learning_rate": 7.355063876015199e-08, |
| "logits/chosen": 2.877830982208252, |
| "logits/rejected": 2.9270596504211426, |
| "logps/chosen": -277.89544677734375, |
| "logps/rejected": -303.5425109863281, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 5.2588934898376465, |
| "rewards/margins": 32.801124572753906, |
| "rewards/rejected": -27.5422306060791, |
| "step": 460 |
| }, |
| { |
| "epoch": 3.6459143968871595, |
| "grad_norm": 0.01067538931965828, |
| "learning_rate": 5.020667990989525e-08, |
| "logits/chosen": 2.8863813877105713, |
| "logits/rejected": 2.9612936973571777, |
| "logps/chosen": -259.4737854003906, |
| "logps/rejected": -296.72637939453125, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.224932670593262, |
| "rewards/margins": 32.182594299316406, |
| "rewards/rejected": -27.957666397094727, |
| "step": 470 |
| }, |
| { |
| "epoch": 3.7237354085603114, |
| "grad_norm": 0.004288141615688801, |
| "learning_rate": 3.121510281356143e-08, |
| "logits/chosen": 2.824970245361328, |
| "logits/rejected": 2.8980326652526855, |
| "logps/chosen": -251.7251434326172, |
| "logps/rejected": -293.5643615722656, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 5.086753845214844, |
| "rewards/margins": 32.74205780029297, |
| "rewards/rejected": -27.65530776977539, |
| "step": 480 |
| }, |
| { |
| "epoch": 3.801556420233463, |
| "grad_norm": 9.263599349651486e-05, |
| "learning_rate": 1.6662935450695416e-08, |
| "logits/chosen": 2.7916643619537354, |
| "logits/rejected": 2.8788905143737793, |
| "logps/chosen": -239.5885467529297, |
| "logps/rejected": -293.6639099121094, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 5.8184919357299805, |
| "rewards/margins": 32.92034912109375, |
| "rewards/rejected": -27.101858139038086, |
| "step": 490 |
| }, |
| { |
| "epoch": 3.8793774319066148, |
| "grad_norm": 0.003024149453267455, |
| "learning_rate": 6.6168624213784885e-09, |
| "logits/chosen": 2.828496217727661, |
| "logits/rejected": 2.8872532844543457, |
| "logps/chosen": -242.9530792236328, |
| "logps/rejected": -291.443359375, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.576481342315674, |
| "rewards/margins": 31.855998992919922, |
| "rewards/rejected": -27.279516220092773, |
| "step": 500 |
| }, |
| { |
| "epoch": 3.9571984435797667, |
| "grad_norm": 0.003526742337271571, |
| "learning_rate": 1.1229193672881444e-09, |
| "logits/chosen": 2.8279409408569336, |
| "logits/rejected": 2.9110004901885986, |
| "logps/chosen": -243.04946899414062, |
| "logps/rejected": -305.1340026855469, |
| "loss": 0.0, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": 4.348401069641113, |
| "rewards/margins": 32.896034240722656, |
| "rewards/rejected": -28.547632217407227, |
| "step": 510 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 516, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 4, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 2, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
