phi3-medium-spin-phi3-data / trainer_state.json
AmberYifan's picture
Model save
b6f8dd8 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 3126,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.597444089456869e-09,
"logits/generated": 6.076260089874268,
"logits/real": 4.217202663421631,
"logps/generated": -793.698486328125,
"logps/real": -221.5892333984375,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/generated": 0.0,
"rewards/margins": 0.0,
"rewards/real": 0.0,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 1.597444089456869e-08,
"logits/generated": 5.937064170837402,
"logits/real": 4.38163423538208,
"logps/generated": -943.8297729492188,
"logps/real": -251.95458984375,
"loss": 0.7034,
"rewards/accuracies": 0.1111111119389534,
"rewards/generated": 0.008679242804646492,
"rewards/margins": -0.012087766081094742,
"rewards/real": -0.0034085246734321117,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 3.194888178913738e-08,
"logits/generated": 6.016368389129639,
"logits/real": 4.054781913757324,
"logps/generated": -866.6522216796875,
"logps/real": -244.7873992919922,
"loss": 0.6959,
"rewards/accuracies": 0.5249999761581421,
"rewards/generated": -0.04229893162846565,
"rewards/margins": 0.02972055971622467,
"rewards/real": -0.01257836353033781,
"step": 20
},
{
"epoch": 0.02,
"learning_rate": 4.7923322683706064e-08,
"logits/generated": 6.003698348999023,
"logits/real": 4.410984992980957,
"logps/generated": -772.19970703125,
"logps/real": -265.46051025390625,
"loss": 0.6868,
"rewards/accuracies": 0.5249999761581421,
"rewards/generated": -0.002037657890468836,
"rewards/margins": 0.010990817099809647,
"rewards/real": 0.008953156881034374,
"step": 30
},
{
"epoch": 0.03,
"learning_rate": 6.389776357827476e-08,
"logits/generated": 6.05269718170166,
"logits/real": 4.136534214019775,
"logps/generated": -916.8504028320312,
"logps/real": -240.4838409423828,
"loss": 0.7385,
"rewards/accuracies": 0.550000011920929,
"rewards/generated": 0.05573784187436104,
"rewards/margins": -0.06023125723004341,
"rewards/real": -0.004493414890021086,
"step": 40
},
{
"epoch": 0.03,
"learning_rate": 7.987220447284344e-08,
"logits/generated": 6.0863237380981445,
"logits/real": 4.26947546005249,
"logps/generated": -910.7779541015625,
"logps/real": -224.7820281982422,
"loss": 0.7306,
"rewards/accuracies": 0.4375,
"rewards/generated": 0.05095089226961136,
"rewards/margins": -0.05160406976938248,
"rewards/real": -0.0006531793624162674,
"step": 50
},
{
"epoch": 0.04,
"learning_rate": 9.584664536741213e-08,
"logits/generated": 5.816224098205566,
"logits/real": 4.575616359710693,
"logps/generated": -828.2184448242188,
"logps/real": -243.4587860107422,
"loss": 0.7114,
"rewards/accuracies": 0.4625000059604645,
"rewards/generated": 0.006193811539560556,
"rewards/margins": -0.02020972967147827,
"rewards/real": -0.014015915803611279,
"step": 60
},
{
"epoch": 0.04,
"learning_rate": 1.1182108626198082e-07,
"logits/generated": 5.910815238952637,
"logits/real": 4.322142601013184,
"logps/generated": -864.3304443359375,
"logps/real": -267.9166564941406,
"loss": 0.7368,
"rewards/accuracies": 0.42500001192092896,
"rewards/generated": 0.10107575356960297,
"rewards/margins": -0.11164508759975433,
"rewards/real": -0.010569351725280285,
"step": 70
},
{
"epoch": 0.05,
"learning_rate": 1.2779552715654952e-07,
"logits/generated": 5.944818496704102,
"logits/real": 4.048387050628662,
"logps/generated": -961.0286865234375,
"logps/real": -237.78018188476562,
"loss": 0.678,
"rewards/accuracies": 0.637499988079071,
"rewards/generated": -0.04871482402086258,
"rewards/margins": 0.04645959287881851,
"rewards/real": -0.002255239523947239,
"step": 80
},
{
"epoch": 0.06,
"learning_rate": 1.437699680511182e-07,
"logits/generated": 6.106154441833496,
"logits/real": 4.130688667297363,
"logps/generated": -899.5615234375,
"logps/real": -257.23492431640625,
"loss": 0.663,
"rewards/accuracies": 0.6000000238418579,
"rewards/generated": -0.09308083355426788,
"rewards/margins": 0.07399795949459076,
"rewards/real": -0.01908286102116108,
"step": 90
},
{
"epoch": 0.06,
"learning_rate": 1.5974440894568688e-07,
"logits/generated": 5.771765232086182,
"logits/real": 4.03811502456665,
"logps/generated": -820.4385986328125,
"logps/real": -240.3477325439453,
"loss": 0.721,
"rewards/accuracies": 0.42500001192092896,
"rewards/generated": 0.0457366518676281,
"rewards/margins": -0.05348087102174759,
"rewards/real": -0.007744210306555033,
"step": 100
},
{
"epoch": 0.07,
"learning_rate": 1.757188498402556e-07,
"logits/generated": 5.8602399826049805,
"logits/real": 4.352300643920898,
"logps/generated": -949.75537109375,
"logps/real": -256.1276550292969,
"loss": 0.6804,
"rewards/accuracies": 0.6625000238418579,
"rewards/generated": -0.08218346536159515,
"rewards/margins": 0.0958067774772644,
"rewards/real": 0.013623319566249847,
"step": 110
},
{
"epoch": 0.08,
"learning_rate": 1.9169329073482426e-07,
"logits/generated": 5.985620021820068,
"logits/real": 4.446410179138184,
"logps/generated": -827.1920776367188,
"logps/real": -240.00082397460938,
"loss": 0.6858,
"rewards/accuracies": 0.5249999761581421,
"rewards/generated": -0.024061355739831924,
"rewards/margins": 0.02317228354513645,
"rewards/real": -0.0008890745230019093,
"step": 120
},
{
"epoch": 0.08,
"learning_rate": 2.0766773162939297e-07,
"logits/generated": 5.976418972015381,
"logits/real": 4.385097980499268,
"logps/generated": -903.66357421875,
"logps/real": -253.0922393798828,
"loss": 0.6812,
"rewards/accuracies": 0.574999988079071,
"rewards/generated": -0.03790752962231636,
"rewards/margins": 0.04170641303062439,
"rewards/real": 0.0037988885305821896,
"step": 130
},
{
"epoch": 0.09,
"learning_rate": 2.2364217252396164e-07,
"logits/generated": 5.8814568519592285,
"logits/real": 4.022303581237793,
"logps/generated": -872.1270751953125,
"logps/real": -239.0474090576172,
"loss": 0.6462,
"rewards/accuracies": 0.6000000238418579,
"rewards/generated": -0.08364593237638474,
"rewards/margins": 0.07660557329654694,
"rewards/real": -0.007040367461740971,
"step": 140
},
{
"epoch": 0.1,
"learning_rate": 2.3961661341853033e-07,
"logits/generated": 5.902901649475098,
"logits/real": 4.500279426574707,
"logps/generated": -909.7518310546875,
"logps/real": -252.1283416748047,
"loss": 0.6449,
"rewards/accuracies": 0.550000011920929,
"rewards/generated": -0.1788884699344635,
"rewards/margins": 0.1675419807434082,
"rewards/real": -0.011346508748829365,
"step": 150
},
{
"epoch": 0.1,
"learning_rate": 2.5559105431309904e-07,
"logits/generated": 6.0863142013549805,
"logits/real": 4.278581142425537,
"logps/generated": -862.6253662109375,
"logps/real": -249.1087188720703,
"loss": 0.6591,
"rewards/accuracies": 0.5874999761581421,
"rewards/generated": -0.06084873527288437,
"rewards/margins": 0.06349506974220276,
"rewards/real": 0.0026463475078344345,
"step": 160
},
{
"epoch": 0.11,
"learning_rate": 2.715654952076677e-07,
"logits/generated": 5.8610758781433105,
"logits/real": 4.366555213928223,
"logps/generated": -893.8683471679688,
"logps/real": -238.64443969726562,
"loss": 0.6079,
"rewards/accuracies": 0.762499988079071,
"rewards/generated": -0.22141461074352264,
"rewards/margins": 0.23539376258850098,
"rewards/real": 0.01397914718836546,
"step": 170
},
{
"epoch": 0.12,
"learning_rate": 2.875399361022364e-07,
"logits/generated": 5.8870625495910645,
"logits/real": 4.547116279602051,
"logps/generated": -906.3629150390625,
"logps/real": -263.76654052734375,
"loss": 0.5942,
"rewards/accuracies": 0.6625000238418579,
"rewards/generated": -0.2859794497489929,
"rewards/margins": 0.2683030664920807,
"rewards/real": -0.017676372081041336,
"step": 180
},
{
"epoch": 0.12,
"learning_rate": 3.035143769968051e-07,
"logits/generated": 6.080388069152832,
"logits/real": 4.27817440032959,
"logps/generated": -907.4010009765625,
"logps/real": -263.9908447265625,
"loss": 0.5832,
"rewards/accuracies": 0.675000011920929,
"rewards/generated": -0.28283512592315674,
"rewards/margins": 0.27541953325271606,
"rewards/real": -0.007415570318698883,
"step": 190
},
{
"epoch": 0.13,
"learning_rate": 3.1948881789137375e-07,
"logits/generated": 6.112253665924072,
"logits/real": 4.260523796081543,
"logps/generated": -938.8836059570312,
"logps/real": -248.11587524414062,
"loss": 0.5842,
"rewards/accuracies": 0.6625000238418579,
"rewards/generated": -0.30967745184898376,
"rewards/margins": 0.29746749997138977,
"rewards/real": -0.012209964916110039,
"step": 200
},
{
"epoch": 0.13,
"learning_rate": 3.354632587859425e-07,
"logits/generated": 5.790924549102783,
"logits/real": 4.103597640991211,
"logps/generated": -907.6522216796875,
"logps/real": -227.3259735107422,
"loss": 0.6164,
"rewards/accuracies": 0.6499999761581421,
"rewards/generated": -0.22061340510845184,
"rewards/margins": 0.21634364128112793,
"rewards/real": -0.004269786179065704,
"step": 210
},
{
"epoch": 0.14,
"learning_rate": 3.514376996805112e-07,
"logits/generated": 5.998345375061035,
"logits/real": 4.214221000671387,
"logps/generated": -942.66552734375,
"logps/real": -222.0115509033203,
"loss": 0.5932,
"rewards/accuracies": 0.75,
"rewards/generated": -0.32513970136642456,
"rewards/margins": 0.32551324367523193,
"rewards/real": 0.0003735637292265892,
"step": 220
},
{
"epoch": 0.15,
"learning_rate": 3.6741214057507985e-07,
"logits/generated": 6.107602596282959,
"logits/real": 4.030685901641846,
"logps/generated": -899.4200439453125,
"logps/real": -238.90316772460938,
"loss": 0.5673,
"rewards/accuracies": 0.7250000238418579,
"rewards/generated": -0.3097040355205536,
"rewards/margins": 0.31323733925819397,
"rewards/real": 0.003533291397616267,
"step": 230
},
{
"epoch": 0.15,
"learning_rate": 3.833865814696485e-07,
"logits/generated": 6.146908283233643,
"logits/real": 4.572686672210693,
"logps/generated": -849.7062377929688,
"logps/real": -287.7304382324219,
"loss": 0.6028,
"rewards/accuracies": 0.612500011920929,
"rewards/generated": -0.2624419331550598,
"rewards/margins": 0.24513304233551025,
"rewards/real": -0.017308901995420456,
"step": 240
},
{
"epoch": 0.16,
"learning_rate": 3.993610223642173e-07,
"logits/generated": 6.0791335105896,
"logits/real": 4.294358730316162,
"logps/generated": -928.7376708984375,
"logps/real": -228.5765380859375,
"loss": 0.5843,
"rewards/accuracies": 0.7749999761581421,
"rewards/generated": -0.3841578960418701,
"rewards/margins": 0.3720285892486572,
"rewards/real": -0.012129291892051697,
"step": 250
},
{
"epoch": 0.17,
"learning_rate": 4.1533546325878595e-07,
"logits/generated": 6.011288642883301,
"logits/real": 3.944628953933716,
"logps/generated": -920.6962890625,
"logps/real": -224.2725372314453,
"loss": 0.5626,
"rewards/accuracies": 0.7250000238418579,
"rewards/generated": -0.4267396926879883,
"rewards/margins": 0.42054229974746704,
"rewards/real": -0.006197371985763311,
"step": 260
},
{
"epoch": 0.17,
"learning_rate": 4.313099041533546e-07,
"logits/generated": 5.999421119689941,
"logits/real": 4.240687370300293,
"logps/generated": -839.8876953125,
"logps/real": -239.6907196044922,
"loss": 0.5742,
"rewards/accuracies": 0.7124999761581421,
"rewards/generated": -0.29800790548324585,
"rewards/margins": 0.30169859528541565,
"rewards/real": 0.003690724028274417,
"step": 270
},
{
"epoch": 0.18,
"learning_rate": 4.472843450479233e-07,
"logits/generated": 6.157493591308594,
"logits/real": 4.233771324157715,
"logps/generated": -919.6163330078125,
"logps/real": -256.02667236328125,
"loss": 0.5615,
"rewards/accuracies": 0.7749999761581421,
"rewards/generated": -0.436979204416275,
"rewards/margins": 0.4427516460418701,
"rewards/real": 0.00577241787686944,
"step": 280
},
{
"epoch": 0.19,
"learning_rate": 4.63258785942492e-07,
"logits/generated": 5.8831915855407715,
"logits/real": 4.111518859863281,
"logps/generated": -915.33642578125,
"logps/real": -250.9635772705078,
"loss": 0.5469,
"rewards/accuracies": 0.762499988079071,
"rewards/generated": -0.44483208656311035,
"rewards/margins": 0.45819053053855896,
"rewards/real": 0.013358525931835175,
"step": 290
},
{
"epoch": 0.19,
"learning_rate": 4.792332268370607e-07,
"logits/generated": 6.069736480712891,
"logits/real": 4.062201499938965,
"logps/generated": -1062.180419921875,
"logps/real": -239.5394744873047,
"loss": 0.5112,
"rewards/accuracies": 0.8500000238418579,
"rewards/generated": -0.7640186548233032,
"rewards/margins": 0.7979092597961426,
"rewards/real": 0.03389066457748413,
"step": 300
},
{
"epoch": 0.2,
"learning_rate": 4.952076677316294e-07,
"logits/generated": 6.100470542907715,
"logits/real": 4.4795241355896,
"logps/generated": -883.9137573242188,
"logps/real": -260.9756164550781,
"loss": 0.5219,
"rewards/accuracies": 0.6625000238418579,
"rewards/generated": -0.466789186000824,
"rewards/margins": 0.47098153829574585,
"rewards/real": 0.0041923513635993,
"step": 310
},
{
"epoch": 0.2,
"learning_rate": 4.987557767507998e-07,
"logits/generated": 5.904444694519043,
"logits/real": 4.497801303863525,
"logps/generated": -920.7421875,
"logps/real": -258.6832580566406,
"loss": 0.5076,
"rewards/accuracies": 0.875,
"rewards/generated": -0.6512015461921692,
"rewards/margins": 0.6806803941726685,
"rewards/real": 0.02947883866727352,
"step": 320
},
{
"epoch": 0.21,
"learning_rate": 4.969783149662282e-07,
"logits/generated": 5.837555885314941,
"logits/real": 4.39646053314209,
"logps/generated": -887.0213623046875,
"logps/real": -259.81976318359375,
"loss": 0.4883,
"rewards/accuracies": 0.8500000238418579,
"rewards/generated": -0.6606628894805908,
"rewards/margins": 0.6972694396972656,
"rewards/real": 0.036606594920158386,
"step": 330
},
{
"epoch": 0.22,
"learning_rate": 4.952008531816565e-07,
"logits/generated": 6.078923225402832,
"logits/real": 4.288995265960693,
"logps/generated": -755.9134521484375,
"logps/real": -254.28958129882812,
"loss": 0.5167,
"rewards/accuracies": 0.7875000238418579,
"rewards/generated": -0.4952046275138855,
"rewards/margins": 0.5218986868858337,
"rewards/real": 0.026694035157561302,
"step": 340
},
{
"epoch": 0.22,
"learning_rate": 4.93423391397085e-07,
"logits/generated": 6.069632530212402,
"logits/real": 4.2718353271484375,
"logps/generated": -885.2103271484375,
"logps/real": -246.41549682617188,
"loss": 0.494,
"rewards/accuracies": 0.75,
"rewards/generated": -0.8642350435256958,
"rewards/margins": 0.8856006860733032,
"rewards/real": 0.021365612745285034,
"step": 350
},
{
"epoch": 0.23,
"learning_rate": 4.916459296125133e-07,
"logits/generated": 6.0406084060668945,
"logits/real": 4.213619232177734,
"logps/generated": -862.3121337890625,
"logps/real": -237.30038452148438,
"loss": 0.4344,
"rewards/accuracies": 0.875,
"rewards/generated": -0.819617748260498,
"rewards/margins": 0.8550539016723633,
"rewards/real": 0.03543621301651001,
"step": 360
},
{
"epoch": 0.24,
"learning_rate": 4.898684678279417e-07,
"logits/generated": 5.9395928382873535,
"logits/real": 4.514933109283447,
"logps/generated": -784.0012817382812,
"logps/real": -274.2304992675781,
"loss": 0.482,
"rewards/accuracies": 0.875,
"rewards/generated": -0.666130006313324,
"rewards/margins": 0.7213379144668579,
"rewards/real": 0.055207859724760056,
"step": 370
},
{
"epoch": 0.24,
"learning_rate": 4.8809100604337e-07,
"logits/generated": 5.9382734298706055,
"logits/real": 4.106156826019287,
"logps/generated": -868.2293090820312,
"logps/real": -257.92828369140625,
"loss": 0.4572,
"rewards/accuracies": 0.949999988079071,
"rewards/generated": -0.924160361289978,
"rewards/margins": 0.9683877229690552,
"rewards/real": 0.044227343052625656,
"step": 380
},
{
"epoch": 0.25,
"learning_rate": 4.863135442587984e-07,
"logits/generated": 6.171679496765137,
"logits/real": 4.2907819747924805,
"logps/generated": -979.2257080078125,
"logps/real": -244.9657745361328,
"loss": 0.4354,
"rewards/accuracies": 0.8999999761581421,
"rewards/generated": -1.2005796432495117,
"rewards/margins": 1.255791187286377,
"rewards/real": 0.05521152541041374,
"step": 390
},
{
"epoch": 0.26,
"learning_rate": 4.845360824742267e-07,
"logits/generated": 6.069159984588623,
"logits/real": 4.032870292663574,
"logps/generated": -804.0493774414062,
"logps/real": -250.74868774414062,
"loss": 0.447,
"rewards/accuracies": 0.8999999761581421,
"rewards/generated": -0.7962594032287598,
"rewards/margins": 0.8555679321289062,
"rewards/real": 0.05930844694375992,
"step": 400
},
{
"epoch": 0.26,
"learning_rate": 4.827586206896552e-07,
"logits/generated": 6.105320453643799,
"logits/real": 4.221116542816162,
"logps/generated": -797.972900390625,
"logps/real": -244.3699188232422,
"loss": 0.471,
"rewards/accuracies": 0.9375,
"rewards/generated": -0.7431430816650391,
"rewards/margins": 0.8322073817253113,
"rewards/real": 0.08906435966491699,
"step": 410
},
{
"epoch": 0.27,
"learning_rate": 4.809811589050835e-07,
"logits/generated": 5.868694305419922,
"logits/real": 4.118344306945801,
"logps/generated": -944.1519775390625,
"logps/real": -250.259765625,
"loss": 0.4199,
"rewards/accuracies": 0.9375,
"rewards/generated": -1.201935887336731,
"rewards/margins": 1.2849478721618652,
"rewards/real": 0.08301188051700592,
"step": 420
},
{
"epoch": 0.28,
"learning_rate": 4.792036971205119e-07,
"logits/generated": 6.04774284362793,
"logits/real": 4.184063911437988,
"logps/generated": -943.5716552734375,
"logps/real": -233.9786834716797,
"loss": 0.3929,
"rewards/accuracies": 0.9750000238418579,
"rewards/generated": -1.1643246412277222,
"rewards/margins": 1.2494871616363525,
"rewards/real": 0.08516237884759903,
"step": 430
},
{
"epoch": 0.28,
"learning_rate": 4.774262353359402e-07,
"logits/generated": 6.015780925750732,
"logits/real": 4.106911659240723,
"logps/generated": -873.0906982421875,
"logps/real": -245.89151000976562,
"loss": 0.4092,
"rewards/accuracies": 0.949999988079071,
"rewards/generated": -1.0168969631195068,
"rewards/margins": 1.096966028213501,
"rewards/real": 0.08006921410560608,
"step": 440
},
{
"epoch": 0.29,
"learning_rate": 4.7564877355136863e-07,
"logits/generated": 6.1327619552612305,
"logits/real": 4.286798000335693,
"logps/generated": -913.6229248046875,
"logps/real": -235.7820281982422,
"loss": 0.3918,
"rewards/accuracies": 0.9624999761581421,
"rewards/generated": -1.1738783121109009,
"rewards/margins": 1.2836366891860962,
"rewards/real": 0.1097583994269371,
"step": 450
},
{
"epoch": 0.29,
"learning_rate": 4.73871311766797e-07,
"logits/generated": 5.843601226806641,
"logits/real": 4.894453048706055,
"logps/generated": -836.0406494140625,
"logps/real": -262.8157653808594,
"loss": 0.3896,
"rewards/accuracies": 0.9624999761581421,
"rewards/generated": -1.122596263885498,
"rewards/margins": 1.2202900648117065,
"rewards/real": 0.09769367426633835,
"step": 460
},
{
"epoch": 0.3,
"learning_rate": 4.7209384998222536e-07,
"logits/generated": 5.926724433898926,
"logits/real": 4.339844226837158,
"logps/generated": -835.6307373046875,
"logps/real": -243.3745880126953,
"loss": 0.3792,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -1.0902034044265747,
"rewards/margins": 1.2509477138519287,
"rewards/real": 0.16074436902999878,
"step": 470
},
{
"epoch": 0.31,
"learning_rate": 4.7031638819765373e-07,
"logits/generated": 6.176774501800537,
"logits/real": 4.369190692901611,
"logps/generated": -1035.437255859375,
"logps/real": -241.65518188476562,
"loss": 0.3517,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -1.6769460439682007,
"rewards/margins": 1.8335769176483154,
"rewards/real": 0.15663087368011475,
"step": 480
},
{
"epoch": 0.31,
"learning_rate": 4.6853892641308215e-07,
"logits/generated": 6.054518699645996,
"logits/real": 4.261479377746582,
"logps/generated": -991.4581909179688,
"logps/real": -227.12417602539062,
"loss": 0.3452,
"rewards/accuracies": 0.9750000238418579,
"rewards/generated": -1.612874984741211,
"rewards/margins": 1.7803363800048828,
"rewards/real": 0.1674613654613495,
"step": 490
},
{
"epoch": 0.32,
"learning_rate": 4.6676146462851046e-07,
"logits/generated": 5.990649700164795,
"logits/real": 4.494472026824951,
"logps/generated": -877.47412109375,
"logps/real": -245.2666015625,
"loss": 0.3664,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -1.3127868175506592,
"rewards/margins": 1.4855382442474365,
"rewards/real": 0.1727515310049057,
"step": 500
},
{
"epoch": 0.33,
"learning_rate": 4.649840028439388e-07,
"logits/generated": 6.058517932891846,
"logits/real": 4.018253803253174,
"logps/generated": -849.6085815429688,
"logps/real": -239.42333984375,
"loss": 0.3424,
"rewards/accuracies": 1.0,
"rewards/generated": -1.2417773008346558,
"rewards/margins": 1.4562361240386963,
"rewards/real": 0.21445894241333008,
"step": 510
},
{
"epoch": 0.33,
"learning_rate": 4.632065410593672e-07,
"logits/generated": 6.090991020202637,
"logits/real": 4.206871509552002,
"logps/generated": -858.64306640625,
"logps/real": -231.8360595703125,
"loss": 0.3043,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -1.3968095779418945,
"rewards/margins": 1.6249290704727173,
"rewards/real": 0.22811949253082275,
"step": 520
},
{
"epoch": 0.34,
"learning_rate": 4.6142907927479556e-07,
"logits/generated": 5.986910820007324,
"logits/real": 4.2822465896606445,
"logps/generated": -1048.8765869140625,
"logps/real": -231.92068481445312,
"loss": 0.2983,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -2.237921714782715,
"rewards/margins": 2.462768793106079,
"rewards/real": 0.22484686970710754,
"step": 530
},
{
"epoch": 0.35,
"learning_rate": 4.59651617490224e-07,
"logits/generated": 5.905124187469482,
"logits/real": 4.177928447723389,
"logps/generated": -821.7687377929688,
"logps/real": -246.280517578125,
"loss": 0.3157,
"rewards/accuracies": 1.0,
"rewards/generated": -1.3176755905151367,
"rewards/margins": 1.5714536905288696,
"rewards/real": 0.2537779211997986,
"step": 540
},
{
"epoch": 0.35,
"learning_rate": 4.578741557056523e-07,
"logits/generated": 5.976016044616699,
"logits/real": 4.194876194000244,
"logps/generated": -845.0601806640625,
"logps/real": -246.12539672851562,
"loss": 0.2847,
"rewards/accuracies": 1.0,
"rewards/generated": -1.4623727798461914,
"rewards/margins": 1.7181476354599,
"rewards/real": 0.25577467679977417,
"step": 550
},
{
"epoch": 0.36,
"learning_rate": 4.560966939210807e-07,
"logits/generated": 6.045588493347168,
"logits/real": 4.3586626052856445,
"logps/generated": -811.4501953125,
"logps/real": -235.3633270263672,
"loss": 0.2894,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -1.3929394483566284,
"rewards/margins": 1.660951018333435,
"rewards/real": 0.2680116295814514,
"step": 560
},
{
"epoch": 0.36,
"learning_rate": 4.54319232136509e-07,
"logits/generated": 5.941956520080566,
"logits/real": 3.9948837757110596,
"logps/generated": -939.8425903320312,
"logps/real": -252.4247283935547,
"loss": 0.2729,
"rewards/accuracies": 1.0,
"rewards/generated": -1.8964020013809204,
"rewards/margins": 2.198000907897949,
"rewards/real": 0.3015987277030945,
"step": 570
},
{
"epoch": 0.37,
"learning_rate": 4.525417703519374e-07,
"logits/generated": 5.852833271026611,
"logits/real": 4.229121208190918,
"logps/generated": -947.67333984375,
"logps/real": -247.6307830810547,
"loss": 0.2615,
"rewards/accuracies": 1.0,
"rewards/generated": -1.9862486124038696,
"rewards/margins": 2.275512933731079,
"rewards/real": 0.28926438093185425,
"step": 580
},
{
"epoch": 0.38,
"learning_rate": 4.507643085673658e-07,
"logits/generated": 5.97081995010376,
"logits/real": 4.184626579284668,
"logps/generated": -939.2525634765625,
"logps/real": -249.10293579101562,
"loss": 0.2501,
"rewards/accuracies": 1.0,
"rewards/generated": -2.023348331451416,
"rewards/margins": 2.3503973484039307,
"rewards/real": 0.3270490765571594,
"step": 590
},
{
"epoch": 0.38,
"learning_rate": 4.489868467827941e-07,
"logits/generated": 5.946187496185303,
"logits/real": 3.98576021194458,
"logps/generated": -868.517578125,
"logps/real": -234.7804412841797,
"loss": 0.2434,
"rewards/accuracies": 1.0,
"rewards/generated": -1.8050874471664429,
"rewards/margins": 2.1618123054504395,
"rewards/real": 0.35672444105148315,
"step": 600
},
{
"epoch": 0.39,
"learning_rate": 4.4720938499822254e-07,
"logits/generated": 6.001513481140137,
"logits/real": 4.200368881225586,
"logps/generated": -938.6095581054688,
"logps/real": -236.7099609375,
"loss": 0.2342,
"rewards/accuracies": 1.0,
"rewards/generated": -2.136950731277466,
"rewards/margins": 2.4946799278259277,
"rewards/real": 0.3577292263507843,
"step": 610
},
{
"epoch": 0.4,
"learning_rate": 4.4543192321365085e-07,
"logits/generated": 6.019902229309082,
"logits/real": 4.102996349334717,
"logps/generated": -877.0018310546875,
"logps/real": -231.89395141601562,
"loss": 0.2346,
"rewards/accuracies": 1.0,
"rewards/generated": -1.801348328590393,
"rewards/margins": 2.227733850479126,
"rewards/real": 0.42638540267944336,
"step": 620
},
{
"epoch": 0.4,
"learning_rate": 4.4365446142907927e-07,
"logits/generated": 5.965851783752441,
"logits/real": 4.574382781982422,
"logps/generated": -923.7843627929688,
"logps/real": -251.86672973632812,
"loss": 0.2313,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -1.9577804803848267,
"rewards/margins": 2.378849744796753,
"rewards/real": 0.4210694432258606,
"step": 630
},
{
"epoch": 0.41,
"learning_rate": 4.4187699964450764e-07,
"logits/generated": 6.019913196563721,
"logits/real": 4.158315181732178,
"logps/generated": -858.7822265625,
"logps/real": -240.2414093017578,
"loss": 0.1937,
"rewards/accuracies": 1.0,
"rewards/generated": -1.8835636377334595,
"rewards/margins": 2.340017795562744,
"rewards/real": 0.45645445585250854,
"step": 640
},
{
"epoch": 0.42,
"learning_rate": 4.4009953785993595e-07,
"logits/generated": 6.02773380279541,
"logits/real": 4.274472713470459,
"logps/generated": -930.7633056640625,
"logps/real": -252.86410522460938,
"loss": 0.2073,
"rewards/accuracies": 1.0,
"rewards/generated": -2.2265586853027344,
"rewards/margins": 2.6945266723632812,
"rewards/real": 0.4679679274559021,
"step": 650
},
{
"epoch": 0.42,
"learning_rate": 4.3832207607536437e-07,
"logits/generated": 6.097564220428467,
"logits/real": 4.119298458099365,
"logps/generated": -823.2561645507812,
"logps/real": -236.7152557373047,
"loss": 0.2025,
"rewards/accuracies": 1.0,
"rewards/generated": -1.7512767314910889,
"rewards/margins": 2.2316291332244873,
"rewards/real": 0.48035264015197754,
"step": 660
},
{
"epoch": 0.43,
"learning_rate": 4.365446142907927e-07,
"logits/generated": 6.0582170486450195,
"logits/real": 4.083959579467773,
"logps/generated": -952.2449340820312,
"logps/real": -228.7444305419922,
"loss": 0.2019,
"rewards/accuracies": 1.0,
"rewards/generated": -2.267892599105835,
"rewards/margins": 2.736295461654663,
"rewards/real": 0.4684027135372162,
"step": 670
},
{
"epoch": 0.44,
"learning_rate": 4.347671525062211e-07,
"logits/generated": 5.989724159240723,
"logits/real": 4.3230791091918945,
"logps/generated": -932.2943115234375,
"logps/real": -237.28170776367188,
"loss": 0.1903,
"rewards/accuracies": 1.0,
"rewards/generated": -2.2489707469940186,
"rewards/margins": 2.7711246013641357,
"rewards/real": 0.5221537947654724,
"step": 680
},
{
"epoch": 0.44,
"learning_rate": 4.3298969072164947e-07,
"logits/generated": 5.9734578132629395,
"logits/real": 4.305315971374512,
"logps/generated": -898.1632080078125,
"logps/real": -249.6607666015625,
"loss": 0.1736,
"rewards/accuracies": 1.0,
"rewards/generated": -2.172926664352417,
"rewards/margins": 2.7069246768951416,
"rewards/real": 0.5339978337287903,
"step": 690
},
{
"epoch": 0.45,
"learning_rate": 4.3121222893707783e-07,
"logits/generated": 5.79467248916626,
"logits/real": 4.179832458496094,
"logps/generated": -853.1522216796875,
"logps/real": -225.7552490234375,
"loss": 0.1769,
"rewards/accuracies": 1.0,
"rewards/generated": -2.0059478282928467,
"rewards/margins": 2.5637969970703125,
"rewards/real": 0.5578492879867554,
"step": 700
},
{
"epoch": 0.45,
"learning_rate": 4.294347671525062e-07,
"logits/generated": 6.031177997589111,
"logits/real": 4.149500846862793,
"logps/generated": -914.326171875,
"logps/real": -237.4337158203125,
"loss": 0.1682,
"rewards/accuracies": 1.0,
"rewards/generated": -2.2514748573303223,
"rewards/margins": 2.817643642425537,
"rewards/real": 0.5661691427230835,
"step": 710
},
{
"epoch": 0.46,
"learning_rate": 4.276573053679346e-07,
"logits/generated": 6.02950382232666,
"logits/real": 4.436091423034668,
"logps/generated": -924.0319213867188,
"logps/real": -244.99362182617188,
"loss": 0.1649,
"rewards/accuracies": 1.0,
"rewards/generated": -2.3590235710144043,
"rewards/margins": 2.9966607093811035,
"rewards/real": 0.6376368403434753,
"step": 720
},
{
"epoch": 0.47,
"learning_rate": 4.2587984358336293e-07,
"logits/generated": 6.004773139953613,
"logits/real": 4.440590858459473,
"logps/generated": -880.44970703125,
"logps/real": -252.4867401123047,
"loss": 0.148,
"rewards/accuracies": 1.0,
"rewards/generated": -2.2073614597320557,
"rewards/margins": 2.841465473175049,
"rewards/real": 0.6341038942337036,
"step": 730
},
{
"epoch": 0.47,
"learning_rate": 4.241023817987913e-07,
"logits/generated": 6.170002460479736,
"logits/real": 4.281262397766113,
"logps/generated": -804.2672119140625,
"logps/real": -246.38973999023438,
"loss": 0.1503,
"rewards/accuracies": 1.0,
"rewards/generated": -2.0122382640838623,
"rewards/margins": 2.680762529373169,
"rewards/real": 0.668523907661438,
"step": 740
},
{
"epoch": 0.48,
"learning_rate": 4.2232492001421966e-07,
"logits/generated": 5.92216157913208,
"logits/real": 4.2551093101501465,
"logps/generated": -862.2703857421875,
"logps/real": -258.5106201171875,
"loss": 0.1396,
"rewards/accuracies": 1.0,
"rewards/generated": -2.400254249572754,
"rewards/margins": 3.064797878265381,
"rewards/real": 0.6645434498786926,
"step": 750
},
{
"epoch": 0.49,
"learning_rate": 4.2054745822964803e-07,
"logits/generated": 5.7719502449035645,
"logits/real": 4.651850700378418,
"logps/generated": -891.9352416992188,
"logps/real": -271.16015625,
"loss": 0.1323,
"rewards/accuracies": 1.0,
"rewards/generated": -2.4314839839935303,
"rewards/margins": 3.151881694793701,
"rewards/real": 0.7203975319862366,
"step": 760
},
{
"epoch": 0.49,
"learning_rate": 4.1876999644507645e-07,
"logits/generated": 6.078845024108887,
"logits/real": 4.190167427062988,
"logps/generated": -986.5220947265625,
"logps/real": -244.7814483642578,
"loss": 0.1271,
"rewards/accuracies": 1.0,
"rewards/generated": -2.8437342643737793,
"rewards/margins": 3.5870704650878906,
"rewards/real": 0.7433363795280457,
"step": 770
},
{
"epoch": 0.5,
"learning_rate": 4.1699253466050476e-07,
"logits/generated": 5.826613903045654,
"logits/real": 4.3144378662109375,
"logps/generated": -897.0218505859375,
"logps/real": -217.27749633789062,
"loss": 0.1252,
"rewards/accuracies": 1.0,
"rewards/generated": -2.579515218734741,
"rewards/margins": 3.3170909881591797,
"rewards/real": 0.737575888633728,
"step": 780
},
{
"epoch": 0.51,
"learning_rate": 4.152150728759332e-07,
"logits/generated": 5.925543785095215,
"logits/real": 4.433269500732422,
"logps/generated": -918.3221435546875,
"logps/real": -254.0979766845703,
"loss": 0.1275,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -2.530599594116211,
"rewards/margins": 3.2494373321533203,
"rewards/real": 0.7188378572463989,
"step": 790
},
{
"epoch": 0.51,
"learning_rate": 4.134376110913615e-07,
"logits/generated": 5.992081642150879,
"logits/real": 4.210297107696533,
"logps/generated": -960.2722778320312,
"logps/real": -226.6741943359375,
"loss": 0.1126,
"rewards/accuracies": 1.0,
"rewards/generated": -2.807277202606201,
"rewards/margins": 3.636183261871338,
"rewards/real": 0.8289061784744263,
"step": 800
},
{
"epoch": 0.52,
"learning_rate": 4.1166014930678986e-07,
"logits/generated": 6.153465747833252,
"logits/real": 4.260440349578857,
"logps/generated": -966.8284912109375,
"logps/real": -226.76773071289062,
"loss": 0.1122,
"rewards/accuracies": 1.0,
"rewards/generated": -2.9602110385894775,
"rewards/margins": 3.7220828533172607,
"rewards/real": 0.7618720531463623,
"step": 810
},
{
"epoch": 0.52,
"learning_rate": 4.098826875222183e-07,
"logits/generated": 5.99381160736084,
"logits/real": 4.221859931945801,
"logps/generated": -903.6896362304688,
"logps/real": -233.30349731445312,
"loss": 0.1167,
"rewards/accuracies": 1.0,
"rewards/generated": -2.786527633666992,
"rewards/margins": 3.6095592975616455,
"rewards/real": 0.823030948638916,
"step": 820
},
{
"epoch": 0.53,
"learning_rate": 4.081052257376466e-07,
"logits/generated": 5.980920314788818,
"logits/real": 4.316037178039551,
"logps/generated": -979.0846557617188,
"logps/real": -240.3975830078125,
"loss": 0.1015,
"rewards/accuracies": 1.0,
"rewards/generated": -3.2098584175109863,
"rewards/margins": 4.0826005935668945,
"rewards/real": 0.8727418780326843,
"step": 830
},
{
"epoch": 0.54,
"learning_rate": 4.06327763953075e-07,
"logits/generated": 6.098940849304199,
"logits/real": 4.4456987380981445,
"logps/generated": -979.9505004882812,
"logps/real": -263.83966064453125,
"loss": 0.0951,
"rewards/accuracies": 1.0,
"rewards/generated": -3.254727840423584,
"rewards/margins": 4.050951957702637,
"rewards/real": 0.7962234616279602,
"step": 840
},
{
"epoch": 0.54,
"learning_rate": 4.045503021685033e-07,
"logits/generated": 6.151425361633301,
"logits/real": 4.164916038513184,
"logps/generated": -920.0028076171875,
"logps/real": -246.395263671875,
"loss": 0.0947,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -2.9581751823425293,
"rewards/margins": 3.8200302124023438,
"rewards/real": 0.861855149269104,
"step": 850
},
{
"epoch": 0.55,
"learning_rate": 4.0277284038393174e-07,
"logits/generated": 6.034658908843994,
"logits/real": 4.340263366699219,
"logps/generated": -912.1318359375,
"logps/real": -238.3675537109375,
"loss": 0.0904,
"rewards/accuracies": 1.0,
"rewards/generated": -2.746849536895752,
"rewards/margins": 3.6980957984924316,
"rewards/real": 0.9512465596199036,
"step": 860
},
{
"epoch": 0.56,
"learning_rate": 4.009953785993601e-07,
"logits/generated": 5.890770435333252,
"logits/real": 4.256333351135254,
"logps/generated": -841.7523193359375,
"logps/real": -251.98605346679688,
"loss": 0.0886,
"rewards/accuracies": 1.0,
"rewards/generated": -2.5989060401916504,
"rewards/margins": 3.512964963912964,
"rewards/real": 0.9140589833259583,
"step": 870
},
{
"epoch": 0.56,
"learning_rate": 3.992179168147884e-07,
"logits/generated": 5.99869441986084,
"logits/real": 4.36562442779541,
"logps/generated": -1008.8406372070312,
"logps/real": -242.4652862548828,
"loss": 0.0893,
"rewards/accuracies": 1.0,
"rewards/generated": -3.4944260120391846,
"rewards/margins": 4.502266883850098,
"rewards/real": 1.0078411102294922,
"step": 880
},
{
"epoch": 0.57,
"learning_rate": 3.9744045503021684e-07,
"logits/generated": 5.980597496032715,
"logits/real": 4.342673301696777,
"logps/generated": -875.7488403320312,
"logps/real": -243.07528686523438,
"loss": 0.0872,
"rewards/accuracies": 1.0,
"rewards/generated": -2.8209424018859863,
"rewards/margins": 3.841762065887451,
"rewards/real": 1.0208194255828857,
"step": 890
},
{
"epoch": 0.58,
"learning_rate": 3.956629932456452e-07,
"logits/generated": 5.762509346008301,
"logits/real": 4.160869598388672,
"logps/generated": -963.9265747070312,
"logps/real": -229.8173828125,
"loss": 0.072,
"rewards/accuracies": 1.0,
"rewards/generated": -3.1484429836273193,
"rewards/margins": 4.151796340942383,
"rewards/real": 1.0033533573150635,
"step": 900
},
{
"epoch": 0.58,
"learning_rate": 3.938855314610736e-07,
"logits/generated": 6.13181209564209,
"logits/real": 4.697200775146484,
"logps/generated": -839.2921752929688,
"logps/real": -260.0076904296875,
"loss": 0.0733,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -2.6962039470672607,
"rewards/margins": 3.7993292808532715,
"rewards/real": 1.103124976158142,
"step": 910
},
{
"epoch": 0.59,
"learning_rate": 3.9210806967650194e-07,
"logits/generated": 5.9377360343933105,
"logits/real": 4.226620674133301,
"logps/generated": -908.3089599609375,
"logps/real": -254.44149780273438,
"loss": 0.0785,
"rewards/accuracies": 1.0,
"rewards/generated": -2.9502556324005127,
"rewards/margins": 3.9924514293670654,
"rewards/real": 1.0421960353851318,
"step": 920
},
{
"epoch": 0.6,
"learning_rate": 3.903306078919303e-07,
"logits/generated": 6.1085710525512695,
"logits/real": 4.342537879943848,
"logps/generated": -1006.3308715820312,
"logps/real": -224.99795532226562,
"loss": 0.0711,
"rewards/accuracies": 1.0,
"rewards/generated": -3.459972381591797,
"rewards/margins": 4.4290771484375,
"rewards/real": 0.9691041707992554,
"step": 930
},
{
"epoch": 0.6,
"learning_rate": 3.8855314610735867e-07,
"logits/generated": 6.049094200134277,
"logits/real": 4.437540531158447,
"logps/generated": -905.2625732421875,
"logps/real": -250.00546264648438,
"loss": 0.0654,
"rewards/accuracies": 1.0,
"rewards/generated": -3.0280721187591553,
"rewards/margins": 4.146524429321289,
"rewards/real": 1.1184518337249756,
"step": 940
},
{
"epoch": 0.61,
"learning_rate": 3.867756843227871e-07,
"logits/generated": 6.0964884757995605,
"logits/real": 4.308868408203125,
"logps/generated": -988.3762817382812,
"logps/real": -213.0933074951172,
"loss": 0.0675,
"rewards/accuracies": 1.0,
"rewards/generated": -3.6333088874816895,
"rewards/margins": 4.766198635101318,
"rewards/real": 1.132889986038208,
"step": 950
},
{
"epoch": 0.61,
"learning_rate": 3.849982225382154e-07,
"logits/generated": 6.094457626342773,
"logits/real": 4.595139980316162,
"logps/generated": -874.7532348632812,
"logps/real": -239.2107391357422,
"loss": 0.071,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -2.91304612159729,
"rewards/margins": 4.163543701171875,
"rewards/real": 1.2504980564117432,
"step": 960
},
{
"epoch": 0.62,
"learning_rate": 3.8322076075364377e-07,
"logits/generated": 5.932214260101318,
"logits/real": 4.410408020019531,
"logps/generated": -892.4168090820312,
"logps/real": -231.3613739013672,
"loss": 0.0559,
"rewards/accuracies": 1.0,
"rewards/generated": -3.2558155059814453,
"rewards/margins": 4.5149030685424805,
"rewards/real": 1.2590879201889038,
"step": 970
},
{
"epoch": 0.63,
"learning_rate": 3.8144329896907214e-07,
"logits/generated": 5.978908538818359,
"logits/real": 3.7718472480773926,
"logps/generated": -904.0095825195312,
"logps/real": -221.69680786132812,
"loss": 0.0523,
"rewards/accuracies": 1.0,
"rewards/generated": -3.370866298675537,
"rewards/margins": 4.739681243896484,
"rewards/real": 1.3688147068023682,
"step": 980
},
{
"epoch": 0.63,
"learning_rate": 3.796658371845005e-07,
"logits/generated": 5.914143085479736,
"logits/real": 4.5271077156066895,
"logps/generated": -804.7938232421875,
"logps/real": -251.84097290039062,
"loss": 0.0613,
"rewards/accuracies": 1.0,
"rewards/generated": -2.7886099815368652,
"rewards/margins": 4.072835445404053,
"rewards/real": 1.2842260599136353,
"step": 990
},
{
"epoch": 0.64,
"learning_rate": 3.778883753999289e-07,
"logits/generated": 6.046614646911621,
"logits/real": 4.076377868652344,
"logps/generated": -828.2825317382812,
"logps/real": -222.2332000732422,
"loss": 0.058,
"rewards/accuracies": 1.0,
"rewards/generated": -3.041111469268799,
"rewards/margins": 4.31611967086792,
"rewards/real": 1.2750083208084106,
"step": 1000
},
{
"epoch": 0.65,
"learning_rate": 3.7611091361535723e-07,
"logits/generated": 5.892106056213379,
"logits/real": 4.29874324798584,
"logps/generated": -991.6868286132812,
"logps/real": -222.35653686523438,
"loss": 0.055,
"rewards/accuracies": 1.0,
"rewards/generated": -3.95190167427063,
"rewards/margins": 5.255926132202148,
"rewards/real": 1.3040244579315186,
"step": 1010
},
{
"epoch": 0.65,
"learning_rate": 3.7433345183078565e-07,
"logits/generated": 5.954197883605957,
"logits/real": 3.857102632522583,
"logps/generated": -923.1248168945312,
"logps/real": -203.76199340820312,
"loss": 0.0545,
"rewards/accuracies": 1.0,
"rewards/generated": -3.505117893218994,
"rewards/margins": 4.928933143615723,
"rewards/real": 1.4238157272338867,
"step": 1020
},
{
"epoch": 0.66,
"learning_rate": 3.7255599004621397e-07,
"logits/generated": 6.068178653717041,
"logits/real": 4.234717845916748,
"logps/generated": -960.6267700195312,
"logps/real": -242.6820831298828,
"loss": 0.0501,
"rewards/accuracies": 1.0,
"rewards/generated": -3.7443442344665527,
"rewards/margins": 5.048287868499756,
"rewards/real": 1.3039430379867554,
"step": 1030
},
{
"epoch": 0.67,
"learning_rate": 3.7077852826164233e-07,
"logits/generated": 6.109949588775635,
"logits/real": 4.280592441558838,
"logps/generated": -789.5339965820312,
"logps/real": -248.1779022216797,
"loss": 0.0447,
"rewards/accuracies": 1.0,
"rewards/generated": -3.201239824295044,
"rewards/margins": 4.559616565704346,
"rewards/real": 1.358377456665039,
"step": 1040
},
{
"epoch": 0.67,
"learning_rate": 3.6900106647707075e-07,
"logits/generated": 6.069329261779785,
"logits/real": 4.076823711395264,
"logps/generated": -924.1849365234375,
"logps/real": -228.2224578857422,
"loss": 0.0377,
"rewards/accuracies": 1.0,
"rewards/generated": -3.9169020652770996,
"rewards/margins": 5.305215358734131,
"rewards/real": 1.3883137702941895,
"step": 1050
},
{
"epoch": 0.68,
"learning_rate": 3.6722360469249906e-07,
"logits/generated": 6.091455936431885,
"logits/real": 4.33205509185791,
"logps/generated": -1016.7650146484375,
"logps/real": -233.74496459960938,
"loss": 0.0414,
"rewards/accuracies": 1.0,
"rewards/generated": -4.456690311431885,
"rewards/margins": 5.847256183624268,
"rewards/real": 1.3905656337738037,
"step": 1060
},
{
"epoch": 0.68,
"learning_rate": 3.654461429079275e-07,
"logits/generated": 6.010404109954834,
"logits/real": 4.235081672668457,
"logps/generated": -920.1266479492188,
"logps/real": -240.25985717773438,
"loss": 0.0433,
"rewards/accuracies": 1.0,
"rewards/generated": -3.6452674865722656,
"rewards/margins": 5.162905693054199,
"rewards/real": 1.5176377296447754,
"step": 1070
},
{
"epoch": 0.69,
"learning_rate": 3.6366868112335585e-07,
"logits/generated": 5.845242500305176,
"logits/real": 4.197909355163574,
"logps/generated": -866.0949096679688,
"logps/real": -226.61862182617188,
"loss": 0.037,
"rewards/accuracies": 1.0,
"rewards/generated": -3.452317476272583,
"rewards/margins": 5.058979511260986,
"rewards/real": 1.6066612005233765,
"step": 1080
},
{
"epoch": 0.7,
"learning_rate": 3.618912193387842e-07,
"logits/generated": 6.059525012969971,
"logits/real": 4.2445173263549805,
"logps/generated": -1042.3199462890625,
"logps/real": -222.6482696533203,
"loss": 0.0398,
"rewards/accuracies": 1.0,
"rewards/generated": -4.495182514190674,
"rewards/margins": 6.020852088928223,
"rewards/real": 1.5256696939468384,
"step": 1090
},
{
"epoch": 0.7,
"learning_rate": 3.601137575542126e-07,
"logits/generated": 6.201509952545166,
"logits/real": 4.5010085105896,
"logps/generated": -1009.8512573242188,
"logps/real": -234.58670043945312,
"loss": 0.0445,
"rewards/accuracies": 1.0,
"rewards/generated": -4.315596580505371,
"rewards/margins": 5.6330060958862305,
"rewards/real": 1.317409634590149,
"step": 1100
},
{
"epoch": 0.71,
"learning_rate": 3.583362957696409e-07,
"logits/generated": 6.0568976402282715,
"logits/real": 4.108635425567627,
"logps/generated": -997.2374267578125,
"logps/real": -220.4910430908203,
"loss": 0.0422,
"rewards/accuracies": 1.0,
"rewards/generated": -4.327475070953369,
"rewards/margins": 5.861368179321289,
"rewards/real": 1.5338925123214722,
"step": 1110
},
{
"epoch": 0.72,
"learning_rate": 3.565588339850693e-07,
"logits/generated": 6.054575443267822,
"logits/real": 4.0250701904296875,
"logps/generated": -973.4529418945312,
"logps/real": -234.23666381835938,
"loss": 0.0386,
"rewards/accuracies": 1.0,
"rewards/generated": -4.307805061340332,
"rewards/margins": 5.763493537902832,
"rewards/real": 1.455688714981079,
"step": 1120
},
{
"epoch": 0.72,
"learning_rate": 3.547813722004977e-07,
"logits/generated": 5.811452865600586,
"logits/real": 4.26353120803833,
"logps/generated": -964.6253051757812,
"logps/real": -233.96548461914062,
"loss": 0.0324,
"rewards/accuracies": 1.0,
"rewards/generated": -4.142745018005371,
"rewards/margins": 5.800713539123535,
"rewards/real": 1.6579687595367432,
"step": 1130
},
{
"epoch": 0.73,
"learning_rate": 3.5300391041592605e-07,
"logits/generated": 5.870312213897705,
"logits/real": 4.17107629776001,
"logps/generated": -863.7185668945312,
"logps/real": -227.08627319335938,
"loss": 0.032,
"rewards/accuracies": 1.0,
"rewards/generated": -3.778026580810547,
"rewards/margins": 5.291924476623535,
"rewards/real": 1.5138972997665405,
"step": 1140
},
{
"epoch": 0.74,
"learning_rate": 3.512264486313544e-07,
"logits/generated": 5.952390670776367,
"logits/real": 4.353332996368408,
"logps/generated": -848.7848510742188,
"logps/real": -239.64132690429688,
"loss": 0.0287,
"rewards/accuracies": 1.0,
"rewards/generated": -3.684706449508667,
"rewards/margins": 5.332028388977051,
"rewards/real": 1.6473219394683838,
"step": 1150
},
{
"epoch": 0.74,
"learning_rate": 3.494489868467828e-07,
"logits/generated": 5.984662055969238,
"logits/real": 4.688880443572998,
"logps/generated": -926.8069458007812,
"logps/real": -222.42788696289062,
"loss": 0.0319,
"rewards/accuracies": 1.0,
"rewards/generated": -4.1294732093811035,
"rewards/margins": 5.777259349822998,
"rewards/real": 1.6477859020233154,
"step": 1160
},
{
"epoch": 0.75,
"learning_rate": 3.4767152506221114e-07,
"logits/generated": 6.093937873840332,
"logits/real": 4.4692511558532715,
"logps/generated": -893.0154418945312,
"logps/real": -249.5071563720703,
"loss": 0.035,
"rewards/accuracies": 1.0,
"rewards/generated": -3.984126567840576,
"rewards/margins": 5.740622520446777,
"rewards/real": 1.7564961910247803,
"step": 1170
},
{
"epoch": 0.75,
"learning_rate": 3.458940632776395e-07,
"logits/generated": 6.014351844787598,
"logits/real": 4.509620666503906,
"logps/generated": -945.0499877929688,
"logps/real": -208.66506958007812,
"loss": 0.0301,
"rewards/accuracies": 1.0,
"rewards/generated": -4.119633674621582,
"rewards/margins": 5.723677635192871,
"rewards/real": 1.6040436029434204,
"step": 1180
},
{
"epoch": 0.76,
"learning_rate": 3.441166014930679e-07,
"logits/generated": 5.890795707702637,
"logits/real": 4.306885719299316,
"logps/generated": -920.9514770507812,
"logps/real": -219.43441772460938,
"loss": 0.03,
"rewards/accuracies": 1.0,
"rewards/generated": -4.295720100402832,
"rewards/margins": 5.946078300476074,
"rewards/real": 1.6503584384918213,
"step": 1190
},
{
"epoch": 0.77,
"learning_rate": 3.4233913970849624e-07,
"logits/generated": 6.092907905578613,
"logits/real": 4.382439613342285,
"logps/generated": -869.8155517578125,
"logps/real": -213.0802459716797,
"loss": 0.0317,
"rewards/accuracies": 1.0,
"rewards/generated": -3.7826313972473145,
"rewards/margins": 5.640969276428223,
"rewards/real": 1.858338713645935,
"step": 1200
},
{
"epoch": 0.77,
"learning_rate": 3.405616779239246e-07,
"logits/generated": 5.900615692138672,
"logits/real": 4.525036334991455,
"logps/generated": -850.4383544921875,
"logps/real": -246.38864135742188,
"loss": 0.0283,
"rewards/accuracies": 1.0,
"rewards/generated": -3.64056396484375,
"rewards/margins": 5.370596885681152,
"rewards/real": 1.7300331592559814,
"step": 1210
},
{
"epoch": 0.78,
"learning_rate": 3.38784216139353e-07,
"logits/generated": 6.163350582122803,
"logits/real": 4.153713226318359,
"logps/generated": -1029.7822265625,
"logps/real": -221.838134765625,
"loss": 0.0266,
"rewards/accuracies": 1.0,
"rewards/generated": -4.886613368988037,
"rewards/margins": 6.8068389892578125,
"rewards/real": 1.9202255010604858,
"step": 1220
},
{
"epoch": 0.79,
"learning_rate": 3.370067543547814e-07,
"logits/generated": 5.858575344085693,
"logits/real": 4.069919109344482,
"logps/generated": -794.5404663085938,
"logps/real": -215.64620971679688,
"loss": 0.026,
"rewards/accuracies": 1.0,
"rewards/generated": -3.5341129302978516,
"rewards/margins": 5.426337242126465,
"rewards/real": 1.8922239542007446,
"step": 1230
},
{
"epoch": 0.79,
"learning_rate": 3.352292925702097e-07,
"logits/generated": 6.137094497680664,
"logits/real": 4.2704057693481445,
"logps/generated": -988.0245361328125,
"logps/real": -224.5236358642578,
"loss": 0.0247,
"rewards/accuracies": 1.0,
"rewards/generated": -4.90227746963501,
"rewards/margins": 6.6614089012146,
"rewards/real": 1.7591317892074585,
"step": 1240
},
{
"epoch": 0.8,
"learning_rate": 3.334518307856381e-07,
"logits/generated": 5.977653503417969,
"logits/real": 4.102234363555908,
"logps/generated": -871.8009643554688,
"logps/real": -239.428955078125,
"loss": 0.0261,
"rewards/accuracies": 1.0,
"rewards/generated": -3.9591381549835205,
"rewards/margins": 5.843039512634277,
"rewards/real": 1.883901834487915,
"step": 1250
},
{
"epoch": 0.81,
"learning_rate": 3.316743690010665e-07,
"logits/generated": 5.898265361785889,
"logits/real": 3.970461368560791,
"logps/generated": -968.3258056640625,
"logps/real": -222.0578155517578,
"loss": 0.0248,
"rewards/accuracies": 1.0,
"rewards/generated": -4.621912479400635,
"rewards/margins": 6.498594760894775,
"rewards/real": 1.8766825199127197,
"step": 1260
},
{
"epoch": 0.81,
"learning_rate": 3.298969072164948e-07,
"logits/generated": 5.933172225952148,
"logits/real": 4.601564407348633,
"logps/generated": -935.9641723632812,
"logps/real": -240.28604125976562,
"loss": 0.0219,
"rewards/accuracies": 1.0,
"rewards/generated": -4.39015007019043,
"rewards/margins": 6.343320846557617,
"rewards/real": 1.9531705379486084,
"step": 1270
},
{
"epoch": 0.82,
"learning_rate": 3.281194454319232e-07,
"logits/generated": 6.0884270668029785,
"logits/real": 4.594856262207031,
"logps/generated": -970.0340576171875,
"logps/real": -238.2774658203125,
"loss": 0.0198,
"rewards/accuracies": 1.0,
"rewards/generated": -4.799683094024658,
"rewards/margins": 6.7805304527282715,
"rewards/real": 1.980847716331482,
"step": 1280
},
{
"epoch": 0.83,
"learning_rate": 3.2634198364735154e-07,
"logits/generated": 6.068641185760498,
"logits/real": 4.550690650939941,
"logps/generated": -885.751953125,
"logps/real": -224.6710205078125,
"loss": 0.0204,
"rewards/accuracies": 1.0,
"rewards/generated": -4.135249137878418,
"rewards/margins": 5.933065891265869,
"rewards/real": 1.797816514968872,
"step": 1290
},
{
"epoch": 0.83,
"learning_rate": 3.2456452186277996e-07,
"logits/generated": 5.979355812072754,
"logits/real": 4.242101192474365,
"logps/generated": -818.2512817382812,
"logps/real": -227.32369995117188,
"loss": 0.0215,
"rewards/accuracies": 1.0,
"rewards/generated": -3.9465458393096924,
"rewards/margins": 6.159433364868164,
"rewards/real": 2.2128875255584717,
"step": 1300
},
{
"epoch": 0.84,
"learning_rate": 3.227870600782083e-07,
"logits/generated": 6.055127143859863,
"logits/real": 4.004499912261963,
"logps/generated": -912.0535278320312,
"logps/real": -221.82894897460938,
"loss": 0.0182,
"rewards/accuracies": 1.0,
"rewards/generated": -4.248215675354004,
"rewards/margins": 6.312341690063477,
"rewards/real": 2.0641255378723145,
"step": 1310
},
{
"epoch": 0.84,
"learning_rate": 3.210095982936367e-07,
"logits/generated": 5.85116720199585,
"logits/real": 4.39048433303833,
"logps/generated": -948.9578857421875,
"logps/real": -230.64669799804688,
"loss": 0.0183,
"rewards/accuracies": 1.0,
"rewards/generated": -4.669097900390625,
"rewards/margins": 6.7218804359436035,
"rewards/real": 2.052781581878662,
"step": 1320
},
{
"epoch": 0.85,
"learning_rate": 3.1923213650906505e-07,
"logits/generated": 5.8959479331970215,
"logits/real": 4.247786521911621,
"logps/generated": -939.8728637695312,
"logps/real": -230.90365600585938,
"loss": 0.0206,
"rewards/accuracies": 1.0,
"rewards/generated": -4.3719964027404785,
"rewards/margins": 6.431820869445801,
"rewards/real": 2.0598244667053223,
"step": 1330
},
{
"epoch": 0.86,
"learning_rate": 3.1745467472449337e-07,
"logits/generated": 5.981402397155762,
"logits/real": 4.520476341247559,
"logps/generated": -1016.5609130859375,
"logps/real": -234.46566772460938,
"loss": 0.0206,
"rewards/accuracies": 1.0,
"rewards/generated": -5.179906368255615,
"rewards/margins": 7.442817687988281,
"rewards/real": 2.262911319732666,
"step": 1340
},
{
"epoch": 0.86,
"learning_rate": 3.156772129399218e-07,
"logits/generated": 5.772179126739502,
"logits/real": 4.208775520324707,
"logps/generated": -790.708984375,
"logps/real": -210.6012420654297,
"loss": 0.0221,
"rewards/accuracies": 1.0,
"rewards/generated": -3.570807933807373,
"rewards/margins": 5.612199306488037,
"rewards/real": 2.041391134262085,
"step": 1350
},
{
"epoch": 0.87,
"learning_rate": 3.1389975115535015e-07,
"logits/generated": 6.091591835021973,
"logits/real": 4.308236122131348,
"logps/generated": -937.48095703125,
"logps/real": -217.81820678710938,
"loss": 0.017,
"rewards/accuracies": 1.0,
"rewards/generated": -4.688237190246582,
"rewards/margins": 6.935911655426025,
"rewards/real": 2.2476744651794434,
"step": 1360
},
{
"epoch": 0.88,
"learning_rate": 3.121222893707785e-07,
"logits/generated": 5.910862922668457,
"logits/real": 4.276075839996338,
"logps/generated": -890.7863159179688,
"logps/real": -241.32388305664062,
"loss": 0.0155,
"rewards/accuracies": 1.0,
"rewards/generated": -4.329002380371094,
"rewards/margins": 6.619669437408447,
"rewards/real": 2.290666341781616,
"step": 1370
},
{
"epoch": 0.88,
"learning_rate": 3.103448275862069e-07,
"logits/generated": 6.195101261138916,
"logits/real": 4.239165782928467,
"logps/generated": -918.6710815429688,
"logps/real": -233.1322479248047,
"loss": 0.0161,
"rewards/accuracies": 1.0,
"rewards/generated": -4.716291904449463,
"rewards/margins": 6.934659481048584,
"rewards/real": 2.2183680534362793,
"step": 1380
},
{
"epoch": 0.89,
"learning_rate": 3.0856736580163525e-07,
"logits/generated": 6.146795749664307,
"logits/real": 4.251899719238281,
"logps/generated": -896.61669921875,
"logps/real": -236.81289672851562,
"loss": 0.0172,
"rewards/accuracies": 1.0,
"rewards/generated": -4.417421817779541,
"rewards/margins": 6.579673767089844,
"rewards/real": 2.162252426147461,
"step": 1390
},
{
"epoch": 0.9,
"learning_rate": 3.067899040170636e-07,
"logits/generated": 5.921133995056152,
"logits/real": 4.311642646789551,
"logps/generated": -1017.5269775390625,
"logps/real": -231.27236938476562,
"loss": 0.0163,
"rewards/accuracies": 1.0,
"rewards/generated": -5.362940311431885,
"rewards/margins": 7.530875205993652,
"rewards/real": 2.1679351329803467,
"step": 1400
},
{
"epoch": 0.9,
"learning_rate": 3.05012442232492e-07,
"logits/generated": 6.124849319458008,
"logits/real": 4.260175704956055,
"logps/generated": -1089.728759765625,
"logps/real": -230.3025665283203,
"loss": 0.0129,
"rewards/accuracies": 1.0,
"rewards/generated": -6.250988960266113,
"rewards/margins": 8.456911087036133,
"rewards/real": 2.2059216499328613,
"step": 1410
},
{
"epoch": 0.91,
"learning_rate": 3.0323498044792035e-07,
"logits/generated": 5.7611541748046875,
"logits/real": 3.985320568084717,
"logps/generated": -856.4290771484375,
"logps/real": -219.84048461914062,
"loss": 0.019,
"rewards/accuracies": 1.0,
"rewards/generated": -4.234499931335449,
"rewards/margins": 6.346347808837891,
"rewards/real": 2.1118481159210205,
"step": 1420
},
{
"epoch": 0.91,
"learning_rate": 3.014575186633487e-07,
"logits/generated": 6.099104881286621,
"logits/real": 4.165801525115967,
"logps/generated": -1022.8517456054688,
"logps/real": -225.13583374023438,
"loss": 0.0097,
"rewards/accuracies": 1.0,
"rewards/generated": -5.710265636444092,
"rewards/margins": 8.138273239135742,
"rewards/real": 2.428006649017334,
"step": 1430
},
{
"epoch": 0.92,
"learning_rate": 2.996800568787771e-07,
"logits/generated": 5.9258551597595215,
"logits/real": 4.278181076049805,
"logps/generated": -991.6013793945312,
"logps/real": -224.7017822265625,
"loss": 0.0131,
"rewards/accuracies": 1.0,
"rewards/generated": -5.471372604370117,
"rewards/margins": 7.898287773132324,
"rewards/real": 2.426915168762207,
"step": 1440
},
{
"epoch": 0.93,
"learning_rate": 2.9790259509420545e-07,
"logits/generated": 6.0136613845825195,
"logits/real": 4.205864906311035,
"logps/generated": -879.1495971679688,
"logps/real": -233.78097534179688,
"loss": 0.0158,
"rewards/accuracies": 1.0,
"rewards/generated": -4.704037189483643,
"rewards/margins": 6.998660087585449,
"rewards/real": 2.2946221828460693,
"step": 1450
},
{
"epoch": 0.93,
"learning_rate": 2.9612513330963387e-07,
"logits/generated": 5.957737922668457,
"logits/real": 4.087791919708252,
"logps/generated": -961.3818359375,
"logps/real": -189.39364624023438,
"loss": 0.0164,
"rewards/accuracies": 1.0,
"rewards/generated": -5.026106834411621,
"rewards/margins": 7.373734951019287,
"rewards/real": 2.347627639770508,
"step": 1460
},
{
"epoch": 0.94,
"learning_rate": 2.943476715250622e-07,
"logits/generated": 5.931456565856934,
"logits/real": 4.309812545776367,
"logps/generated": -880.28955078125,
"logps/real": -227.7857666015625,
"loss": 0.0142,
"rewards/accuracies": 1.0,
"rewards/generated": -4.744623184204102,
"rewards/margins": 6.99019718170166,
"rewards/real": 2.2455737590789795,
"step": 1470
},
{
"epoch": 0.95,
"learning_rate": 2.9257020974049054e-07,
"logits/generated": 6.020743370056152,
"logits/real": 4.34235143661499,
"logps/generated": -918.6298828125,
"logps/real": -233.72021484375,
"loss": 0.014,
"rewards/accuracies": 1.0,
"rewards/generated": -4.72176456451416,
"rewards/margins": 7.291744232177734,
"rewards/real": 2.5699801445007324,
"step": 1480
},
{
"epoch": 0.95,
"learning_rate": 2.9079274795591896e-07,
"logits/generated": 6.1958794593811035,
"logits/real": 4.313004970550537,
"logps/generated": -913.8150634765625,
"logps/real": -237.1126251220703,
"loss": 0.0125,
"rewards/accuracies": 1.0,
"rewards/generated": -5.31519889831543,
"rewards/margins": 7.788308143615723,
"rewards/real": 2.473109483718872,
"step": 1490
},
{
"epoch": 0.96,
"learning_rate": 2.890152861713473e-07,
"logits/generated": 5.900928497314453,
"logits/real": 4.231713771820068,
"logps/generated": -1001.9522705078125,
"logps/real": -226.7261199951172,
"loss": 0.011,
"rewards/accuracies": 1.0,
"rewards/generated": -5.419561862945557,
"rewards/margins": 7.794165134429932,
"rewards/real": 2.3746023178100586,
"step": 1500
},
{
"epoch": 0.97,
"learning_rate": 2.872378243867757e-07,
"logits/generated": 5.979315280914307,
"logits/real": 4.155876159667969,
"logps/generated": -896.203125,
"logps/real": -224.9453582763672,
"loss": 0.0136,
"rewards/accuracies": 1.0,
"rewards/generated": -4.840609550476074,
"rewards/margins": 7.281617164611816,
"rewards/real": 2.441007137298584,
"step": 1510
},
{
"epoch": 0.97,
"learning_rate": 2.85460362602204e-07,
"logits/generated": 5.99771785736084,
"logits/real": 4.335788249969482,
"logps/generated": -985.0953369140625,
"logps/real": -216.6324920654297,
"loss": 0.0112,
"rewards/accuracies": 1.0,
"rewards/generated": -5.507411479949951,
"rewards/margins": 8.085838317871094,
"rewards/real": 2.5784270763397217,
"step": 1520
},
{
"epoch": 0.98,
"learning_rate": 2.8368290081763243e-07,
"logits/generated": 5.952843189239502,
"logits/real": 4.346559524536133,
"logps/generated": -945.6209106445312,
"logps/real": -227.56997680664062,
"loss": 0.0113,
"rewards/accuracies": 1.0,
"rewards/generated": -5.429326057434082,
"rewards/margins": 8.042230606079102,
"rewards/real": 2.612903118133545,
"step": 1530
},
{
"epoch": 0.99,
"learning_rate": 2.819054390330608e-07,
"logits/generated": 6.0241618156433105,
"logits/real": 4.244063377380371,
"logps/generated": -932.4935302734375,
"logps/real": -231.8907012939453,
"loss": 0.0101,
"rewards/accuracies": 1.0,
"rewards/generated": -5.1778459548950195,
"rewards/margins": 7.901692867279053,
"rewards/real": 2.723846435546875,
"step": 1540
},
{
"epoch": 0.99,
"learning_rate": 2.801279772484891e-07,
"logits/generated": 5.867781639099121,
"logits/real": 4.314043045043945,
"logps/generated": -851.2517700195312,
"logps/real": -235.279296875,
"loss": 0.0126,
"rewards/accuracies": 1.0,
"rewards/generated": -4.752983570098877,
"rewards/margins": 7.189466953277588,
"rewards/real": 2.436483383178711,
"step": 1550
},
{
"epoch": 1.0,
"learning_rate": 2.783505154639175e-07,
"logits/generated": 5.901434898376465,
"logits/real": 4.186619281768799,
"logps/generated": -1009.0958862304688,
"logps/real": -212.00369262695312,
"loss": 0.0079,
"rewards/accuracies": 1.0,
"rewards/generated": -5.709371089935303,
"rewards/margins": 8.269929885864258,
"rewards/real": 2.560558795928955,
"step": 1560
},
{
"epoch": 1.0,
"learning_rate": 2.7657305367934584e-07,
"logits/generated": 5.8600664138793945,
"logits/real": 4.312530040740967,
"logps/generated": -903.2091674804688,
"logps/real": -217.84249877929688,
"loss": 0.0091,
"rewards/accuracies": 1.0,
"rewards/generated": -5.082972049713135,
"rewards/margins": 7.53509521484375,
"rewards/real": 2.4521236419677734,
"step": 1570
},
{
"epoch": 1.01,
"learning_rate": 2.7479559189477426e-07,
"logits/generated": 6.000660419464111,
"logits/real": 4.141573905944824,
"logps/generated": -989.6737060546875,
"logps/real": -216.5168914794922,
"loss": 0.0089,
"rewards/accuracies": 1.0,
"rewards/generated": -5.764129161834717,
"rewards/margins": 8.395139694213867,
"rewards/real": 2.6310102939605713,
"step": 1580
},
{
"epoch": 1.02,
"learning_rate": 2.730181301102026e-07,
"logits/generated": 5.965339660644531,
"logits/real": 4.219322204589844,
"logps/generated": -854.77783203125,
"logps/real": -235.1974639892578,
"loss": 0.0086,
"rewards/accuracies": 1.0,
"rewards/generated": -4.901017665863037,
"rewards/margins": 7.7865166664123535,
"rewards/real": 2.8854994773864746,
"step": 1590
},
{
"epoch": 1.02,
"learning_rate": 2.71240668325631e-07,
"logits/generated": 6.012763023376465,
"logits/real": 4.389718055725098,
"logps/generated": -984.3873291015625,
"logps/real": -239.3701629638672,
"loss": 0.0062,
"rewards/accuracies": 1.0,
"rewards/generated": -5.781050682067871,
"rewards/margins": 8.516217231750488,
"rewards/real": 2.73516583442688,
"step": 1600
},
{
"epoch": 1.03,
"learning_rate": 2.6946320654105936e-07,
"logits/generated": 5.979636192321777,
"logits/real": 4.670422554016113,
"logps/generated": -967.5466918945312,
"logps/real": -236.21444702148438,
"loss": 0.0078,
"rewards/accuracies": 1.0,
"rewards/generated": -5.631641387939453,
"rewards/margins": 8.182405471801758,
"rewards/real": 2.5507636070251465,
"step": 1610
},
{
"epoch": 1.04,
"learning_rate": 2.676857447564877e-07,
"logits/generated": 6.084554195404053,
"logits/real": 4.161642551422119,
"logps/generated": -987.7828979492188,
"logps/real": -225.3826446533203,
"loss": 0.0083,
"rewards/accuracies": 1.0,
"rewards/generated": -5.720307350158691,
"rewards/margins": 8.545812606811523,
"rewards/real": 2.8255059719085693,
"step": 1620
},
{
"epoch": 1.04,
"learning_rate": 2.659082829719161e-07,
"logits/generated": 5.93519926071167,
"logits/real": 4.27223539352417,
"logps/generated": -930.1174926757812,
"logps/real": -235.65243530273438,
"loss": 0.0111,
"rewards/accuracies": 1.0,
"rewards/generated": -5.5342912673950195,
"rewards/margins": 8.177888870239258,
"rewards/real": 2.6435976028442383,
"step": 1630
},
{
"epoch": 1.05,
"learning_rate": 2.6413082118734445e-07,
"logits/generated": 6.053906440734863,
"logits/real": 4.279298305511475,
"logps/generated": -873.9427490234375,
"logps/real": -234.75802612304688,
"loss": 0.0103,
"rewards/accuracies": 1.0,
"rewards/generated": -5.114261627197266,
"rewards/margins": 7.79233455657959,
"rewards/real": 2.678072929382324,
"step": 1640
},
{
"epoch": 1.06,
"learning_rate": 2.623533594027728e-07,
"logits/generated": 5.621105194091797,
"logits/real": 4.41392707824707,
"logps/generated": -888.2545776367188,
"logps/real": -219.32546997070312,
"loss": 0.0103,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -5.186315536499023,
"rewards/margins": 7.618834495544434,
"rewards/real": 2.432518720626831,
"step": 1650
},
{
"epoch": 1.06,
"learning_rate": 2.605758976182012e-07,
"logits/generated": 6.062514781951904,
"logits/real": 4.3643798828125,
"logps/generated": -1034.1868896484375,
"logps/real": -212.97476196289062,
"loss": 0.0094,
"rewards/accuracies": 1.0,
"rewards/generated": -6.618842124938965,
"rewards/margins": 9.310908317565918,
"rewards/real": 2.6920673847198486,
"step": 1660
},
{
"epoch": 1.07,
"learning_rate": 2.587984358336296e-07,
"logits/generated": 5.926046848297119,
"logits/real": 4.445965766906738,
"logps/generated": -821.0653076171875,
"logps/real": -250.19674682617188,
"loss": 0.0098,
"rewards/accuracies": 1.0,
"rewards/generated": -4.612605094909668,
"rewards/margins": 7.3182783126831055,
"rewards/real": 2.7056736946105957,
"step": 1670
},
{
"epoch": 1.07,
"learning_rate": 2.570209740490579e-07,
"logits/generated": 5.849274635314941,
"logits/real": 4.148129463195801,
"logps/generated": -878.97021484375,
"logps/real": -214.24234008789062,
"loss": 0.0079,
"rewards/accuracies": 1.0,
"rewards/generated": -5.2036943435668945,
"rewards/margins": 7.895529747009277,
"rewards/real": 2.691835403442383,
"step": 1680
},
{
"epoch": 1.08,
"learning_rate": 2.5524351226448634e-07,
"logits/generated": 6.104997634887695,
"logits/real": 4.673565864562988,
"logps/generated": -864.9569091796875,
"logps/real": -246.60794067382812,
"loss": 0.0088,
"rewards/accuracies": 1.0,
"rewards/generated": -5.325832366943359,
"rewards/margins": 8.104931831359863,
"rewards/real": 2.7790980339050293,
"step": 1690
},
{
"epoch": 1.09,
"learning_rate": 2.5346605047991465e-07,
"logits/generated": 6.075316429138184,
"logits/real": 4.084663391113281,
"logps/generated": -872.2156372070312,
"logps/real": -223.63442993164062,
"loss": 0.0056,
"rewards/accuracies": 1.0,
"rewards/generated": -5.434564113616943,
"rewards/margins": 8.21035385131836,
"rewards/real": 2.7757906913757324,
"step": 1700
},
{
"epoch": 1.09,
"learning_rate": 2.51688588695343e-07,
"logits/generated": 6.089809894561768,
"logits/real": 4.060254096984863,
"logps/generated": -923.2381591796875,
"logps/real": -191.72105407714844,
"loss": 0.009,
"rewards/accuracies": 1.0,
"rewards/generated": -5.731915473937988,
"rewards/margins": 8.29061222076416,
"rewards/real": 2.5586960315704346,
"step": 1710
},
{
"epoch": 1.1,
"learning_rate": 2.499111269107714e-07,
"logits/generated": 6.061944007873535,
"logits/real": 4.62351131439209,
"logps/generated": -841.89013671875,
"logps/real": -263.7066955566406,
"loss": 0.0087,
"rewards/accuracies": 1.0,
"rewards/generated": -5.006676197052002,
"rewards/margins": 7.827229976654053,
"rewards/real": 2.820553779602051,
"step": 1720
},
{
"epoch": 1.11,
"learning_rate": 2.4813366512619975e-07,
"logits/generated": 5.924272537231445,
"logits/real": 4.0540995597839355,
"logps/generated": -903.25,
"logps/real": -217.80874633789062,
"loss": 0.007,
"rewards/accuracies": 1.0,
"rewards/generated": -5.3574018478393555,
"rewards/margins": 8.290170669555664,
"rewards/real": 2.9327681064605713,
"step": 1730
},
{
"epoch": 1.11,
"learning_rate": 2.4635620334162817e-07,
"logits/generated": 6.056190013885498,
"logits/real": 4.321590900421143,
"logps/generated": -932.0973510742188,
"logps/real": -225.2566375732422,
"loss": 0.0064,
"rewards/accuracies": 1.0,
"rewards/generated": -5.739737510681152,
"rewards/margins": 8.39258861541748,
"rewards/real": 2.652851104736328,
"step": 1740
},
{
"epoch": 1.12,
"learning_rate": 2.4457874155705653e-07,
"logits/generated": 6.022032737731934,
"logits/real": 4.070498466491699,
"logps/generated": -932.9993286132812,
"logps/real": -211.3143768310547,
"loss": 0.0093,
"rewards/accuracies": 1.0,
"rewards/generated": -5.863167762756348,
"rewards/margins": 8.546854972839355,
"rewards/real": 2.683687448501587,
"step": 1750
},
{
"epoch": 1.13,
"learning_rate": 2.428012797724849e-07,
"logits/generated": 6.195624351501465,
"logits/real": 4.141390800476074,
"logps/generated": -957.7415161132812,
"logps/real": -210.617431640625,
"loss": 0.0069,
"rewards/accuracies": 1.0,
"rewards/generated": -5.7170329093933105,
"rewards/margins": 8.636775016784668,
"rewards/real": 2.919740915298462,
"step": 1760
},
{
"epoch": 1.13,
"learning_rate": 2.4102381798791327e-07,
"logits/generated": 6.033434867858887,
"logits/real": 4.116759777069092,
"logps/generated": -842.8616333007812,
"logps/real": -230.01284790039062,
"loss": 0.0072,
"rewards/accuracies": 1.0,
"rewards/generated": -5.187349319458008,
"rewards/margins": 8.29853343963623,
"rewards/real": 3.1111843585968018,
"step": 1770
},
{
"epoch": 1.14,
"learning_rate": 2.392463562033416e-07,
"logits/generated": 6.119296550750732,
"logits/real": 4.1949968338012695,
"logps/generated": -1000.3076171875,
"logps/real": -228.13479614257812,
"loss": 0.0053,
"rewards/accuracies": 1.0,
"rewards/generated": -6.538392066955566,
"rewards/margins": 9.60668659210205,
"rewards/real": 3.0682942867279053,
"step": 1780
},
{
"epoch": 1.15,
"learning_rate": 2.3746889441877e-07,
"logits/generated": 5.990731716156006,
"logits/real": 4.539961814880371,
"logps/generated": -782.185791015625,
"logps/real": -251.5903778076172,
"loss": 0.0054,
"rewards/accuracies": 1.0,
"rewards/generated": -4.510619163513184,
"rewards/margins": 7.38437557220459,
"rewards/real": 2.873756170272827,
"step": 1790
},
{
"epoch": 1.15,
"learning_rate": 2.3569143263419836e-07,
"logits/generated": 6.0371599197387695,
"logits/real": 4.193976402282715,
"logps/generated": -992.1876220703125,
"logps/real": -220.98495483398438,
"loss": 0.0047,
"rewards/accuracies": 1.0,
"rewards/generated": -6.436119079589844,
"rewards/margins": 9.288522720336914,
"rewards/real": 2.852402687072754,
"step": 1800
},
{
"epoch": 1.16,
"learning_rate": 2.3391397084962673e-07,
"logits/generated": 6.059049129486084,
"logits/real": 4.382033348083496,
"logps/generated": -784.0813598632812,
"logps/real": -219.2273712158203,
"loss": 0.0097,
"rewards/accuracies": 1.0,
"rewards/generated": -4.49301815032959,
"rewards/margins": 7.499096870422363,
"rewards/real": 3.0060791969299316,
"step": 1810
},
{
"epoch": 1.16,
"learning_rate": 2.3213650906505507e-07,
"logits/generated": 5.9765849113464355,
"logits/real": 4.078164577484131,
"logps/generated": -911.6340942382812,
"logps/real": -189.57408142089844,
"loss": 0.0055,
"rewards/accuracies": 1.0,
"rewards/generated": -5.761612415313721,
"rewards/margins": 8.555708885192871,
"rewards/real": 2.794097900390625,
"step": 1820
},
{
"epoch": 1.17,
"learning_rate": 2.3035904728048346e-07,
"logits/generated": 5.922765254974365,
"logits/real": 4.084896564483643,
"logps/generated": -967.2000122070312,
"logps/real": -221.4872589111328,
"loss": 0.0057,
"rewards/accuracies": 1.0,
"rewards/generated": -6.081551551818848,
"rewards/margins": 9.249353408813477,
"rewards/real": 3.167802572250366,
"step": 1830
},
{
"epoch": 1.18,
"learning_rate": 2.2858158549591183e-07,
"logits/generated": 6.103907585144043,
"logits/real": 4.329716682434082,
"logps/generated": -922.7809448242188,
"logps/real": -229.83584594726562,
"loss": 0.0055,
"rewards/accuracies": 1.0,
"rewards/generated": -5.810357570648193,
"rewards/margins": 8.797757148742676,
"rewards/real": 2.9873995780944824,
"step": 1840
},
{
"epoch": 1.18,
"learning_rate": 2.268041237113402e-07,
"logits/generated": 5.952910423278809,
"logits/real": 4.254477024078369,
"logps/generated": -829.0789184570312,
"logps/real": -191.4152374267578,
"loss": 0.0097,
"rewards/accuracies": 1.0,
"rewards/generated": -4.8961920738220215,
"rewards/margins": 7.95013952255249,
"rewards/real": 3.053947925567627,
"step": 1850
},
{
"epoch": 1.19,
"learning_rate": 2.2502666192676856e-07,
"logits/generated": 6.104610919952393,
"logits/real": 4.344254016876221,
"logps/generated": -927.4021606445312,
"logps/real": -209.98464965820312,
"loss": 0.008,
"rewards/accuracies": 1.0,
"rewards/generated": -6.0140275955200195,
"rewards/margins": 8.953369140625,
"rewards/real": 2.939342737197876,
"step": 1860
},
{
"epoch": 1.2,
"learning_rate": 2.2324920014219693e-07,
"logits/generated": 5.963999271392822,
"logits/real": 4.088013648986816,
"logps/generated": -981.2156982421875,
"logps/real": -223.4355010986328,
"loss": 0.0062,
"rewards/accuracies": 1.0,
"rewards/generated": -6.4049482345581055,
"rewards/margins": 9.413580894470215,
"rewards/real": 3.0086326599121094,
"step": 1870
},
{
"epoch": 1.2,
"learning_rate": 2.2147173835762532e-07,
"logits/generated": 6.034571647644043,
"logits/real": 3.9964118003845215,
"logps/generated": -919.56103515625,
"logps/real": -212.9126739501953,
"loss": 0.0082,
"rewards/accuracies": 1.0,
"rewards/generated": -5.920588970184326,
"rewards/margins": 8.81024169921875,
"rewards/real": 2.889652967453003,
"step": 1880
},
{
"epoch": 1.21,
"learning_rate": 2.1969427657305366e-07,
"logits/generated": 5.908297061920166,
"logits/real": 4.290045738220215,
"logps/generated": -903.4417724609375,
"logps/real": -216.17636108398438,
"loss": 0.0051,
"rewards/accuracies": 1.0,
"rewards/generated": -6.024001598358154,
"rewards/margins": 8.847755432128906,
"rewards/real": 2.8237533569335938,
"step": 1890
},
{
"epoch": 1.22,
"learning_rate": 2.1791681478848203e-07,
"logits/generated": 5.78277587890625,
"logits/real": 4.178603172302246,
"logps/generated": -897.1768798828125,
"logps/real": -224.93814086914062,
"loss": 0.0048,
"rewards/accuracies": 1.0,
"rewards/generated": -5.952237129211426,
"rewards/margins": 9.044183731079102,
"rewards/real": 3.091947555541992,
"step": 1900
},
{
"epoch": 1.22,
"learning_rate": 2.161393530039104e-07,
"logits/generated": 5.6169939041137695,
"logits/real": 4.14485502243042,
"logps/generated": -836.0411376953125,
"logps/real": -214.97607421875,
"loss": 0.0063,
"rewards/accuracies": 1.0,
"rewards/generated": -5.351627826690674,
"rewards/margins": 8.435457229614258,
"rewards/real": 3.0838301181793213,
"step": 1910
},
{
"epoch": 1.23,
"learning_rate": 2.1436189121933878e-07,
"logits/generated": 5.942612171173096,
"logits/real": 4.3120036125183105,
"logps/generated": -1033.1988525390625,
"logps/real": -231.5688018798828,
"loss": 0.0057,
"rewards/accuracies": 1.0,
"rewards/generated": -6.671760559082031,
"rewards/margins": 9.734092712402344,
"rewards/real": 3.0623319149017334,
"step": 1920
},
{
"epoch": 1.23,
"learning_rate": 2.1258442943476715e-07,
"logits/generated": 6.000979423522949,
"logits/real": 3.862889051437378,
"logps/generated": -996.46875,
"logps/real": -188.82571411132812,
"loss": 0.008,
"rewards/accuracies": 1.0,
"rewards/generated": -6.690791130065918,
"rewards/margins": 9.859224319458008,
"rewards/real": 3.168431520462036,
"step": 1930
},
{
"epoch": 1.24,
"learning_rate": 2.1080696765019552e-07,
"logits/generated": 6.081242084503174,
"logits/real": 4.578803062438965,
"logps/generated": -1046.6558837890625,
"logps/real": -228.24392700195312,
"loss": 0.005,
"rewards/accuracies": 1.0,
"rewards/generated": -7.087597846984863,
"rewards/margins": 10.14047622680664,
"rewards/real": 3.0528788566589355,
"step": 1940
},
{
"epoch": 1.25,
"learning_rate": 2.0902950586562388e-07,
"logits/generated": 5.984295845031738,
"logits/real": 4.308724880218506,
"logps/generated": -935.5621948242188,
"logps/real": -218.9115753173828,
"loss": 0.0079,
"rewards/accuracies": 1.0,
"rewards/generated": -6.082803726196289,
"rewards/margins": 9.108025550842285,
"rewards/real": 3.025221347808838,
"step": 1950
},
{
"epoch": 1.25,
"learning_rate": 2.0725204408105225e-07,
"logits/generated": 5.89461612701416,
"logits/real": 4.198281288146973,
"logps/generated": -961.24560546875,
"logps/real": -220.1450653076172,
"loss": 0.0045,
"rewards/accuracies": 1.0,
"rewards/generated": -6.416398048400879,
"rewards/margins": 9.537870407104492,
"rewards/real": 3.1214730739593506,
"step": 1960
},
{
"epoch": 1.26,
"learning_rate": 2.0547458229648061e-07,
"logits/generated": 6.187463283538818,
"logits/real": 4.732372283935547,
"logps/generated": -980.1790161132812,
"logps/real": -251.7775421142578,
"loss": 0.0048,
"rewards/accuracies": 1.0,
"rewards/generated": -6.859214782714844,
"rewards/margins": 9.723730087280273,
"rewards/real": 2.8645150661468506,
"step": 1970
},
{
"epoch": 1.27,
"learning_rate": 2.0369712051190898e-07,
"logits/generated": 5.920609951019287,
"logits/real": 4.184802055358887,
"logps/generated": -969.5177612304688,
"logps/real": -220.58566284179688,
"loss": 0.0037,
"rewards/accuracies": 1.0,
"rewards/generated": -6.475825309753418,
"rewards/margins": 9.5746431350708,
"rewards/real": 3.098818063735962,
"step": 1980
},
{
"epoch": 1.27,
"learning_rate": 2.0191965872733735e-07,
"logits/generated": 6.136545181274414,
"logits/real": 4.201231002807617,
"logps/generated": -993.6239013671875,
"logps/real": -216.8855438232422,
"loss": 0.0054,
"rewards/accuracies": 1.0,
"rewards/generated": -6.71942138671875,
"rewards/margins": 9.831668853759766,
"rewards/real": 3.1122474670410156,
"step": 1990
},
{
"epoch": 1.28,
"learning_rate": 2.001421969427657e-07,
"logits/generated": 5.892589569091797,
"logits/real": 4.182487487792969,
"logps/generated": -942.9978637695312,
"logps/real": -218.97500610351562,
"loss": 0.0049,
"rewards/accuracies": 1.0,
"rewards/generated": -6.580080986022949,
"rewards/margins": 9.862131118774414,
"rewards/real": 3.2820498943328857,
"step": 2000
},
{
"epoch": 1.29,
"learning_rate": 1.983647351581941e-07,
"logits/generated": 5.94655179977417,
"logits/real": 4.0822672843933105,
"logps/generated": -957.2570190429688,
"logps/real": -196.55645751953125,
"loss": 0.0029,
"rewards/accuracies": 1.0,
"rewards/generated": -6.333019256591797,
"rewards/margins": 9.670191764831543,
"rewards/real": 3.3371729850769043,
"step": 2010
},
{
"epoch": 1.29,
"learning_rate": 1.9658727337362247e-07,
"logits/generated": 5.719095230102539,
"logits/real": 4.3129563331604,
"logps/generated": -879.9132080078125,
"logps/real": -218.69076538085938,
"loss": 0.0042,
"rewards/accuracies": 1.0,
"rewards/generated": -6.011284828186035,
"rewards/margins": 9.261569023132324,
"rewards/real": 3.2502846717834473,
"step": 2020
},
{
"epoch": 1.3,
"learning_rate": 1.9480981158905084e-07,
"logits/generated": 5.970053195953369,
"logits/real": 4.292898654937744,
"logps/generated": -888.4221801757812,
"logps/real": -221.7141876220703,
"loss": 0.0073,
"rewards/accuracies": 1.0,
"rewards/generated": -5.928704738616943,
"rewards/margins": 9.284444808959961,
"rewards/real": 3.3557403087615967,
"step": 2030
},
{
"epoch": 1.31,
"learning_rate": 1.9303234980447918e-07,
"logits/generated": 5.851454257965088,
"logits/real": 3.963120222091675,
"logps/generated": -1002.041015625,
"logps/real": -187.06405639648438,
"loss": 0.0043,
"rewards/accuracies": 1.0,
"rewards/generated": -6.853819370269775,
"rewards/margins": 10.24903678894043,
"rewards/real": 3.3952178955078125,
"step": 2040
},
{
"epoch": 1.31,
"learning_rate": 1.9125488801990754e-07,
"logits/generated": 5.893982887268066,
"logits/real": 4.2695136070251465,
"logps/generated": -952.0750732421875,
"logps/real": -224.6346435546875,
"loss": 0.0039,
"rewards/accuracies": 1.0,
"rewards/generated": -6.340090751647949,
"rewards/margins": 9.606979370117188,
"rewards/real": 3.26688814163208,
"step": 2050
},
{
"epoch": 1.32,
"learning_rate": 1.8947742623533593e-07,
"logits/generated": 6.079073905944824,
"logits/real": 4.082903861999512,
"logps/generated": -1034.026123046875,
"logps/real": -209.4253692626953,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/generated": -7.287435054779053,
"rewards/margins": 10.478463172912598,
"rewards/real": 3.191028356552124,
"step": 2060
},
{
"epoch": 1.32,
"learning_rate": 1.876999644507643e-07,
"logits/generated": 6.078752517700195,
"logits/real": 4.00618314743042,
"logps/generated": -836.09326171875,
"logps/real": -218.91085815429688,
"loss": 0.0052,
"rewards/accuracies": 1.0,
"rewards/generated": -5.683650970458984,
"rewards/margins": 8.769237518310547,
"rewards/real": 3.0855870246887207,
"step": 2070
},
{
"epoch": 1.33,
"learning_rate": 1.8592250266619267e-07,
"logits/generated": 6.1372151374816895,
"logits/real": 4.399873733520508,
"logps/generated": -925.8645629882812,
"logps/real": -231.36962890625,
"loss": 0.0032,
"rewards/accuracies": 1.0,
"rewards/generated": -6.495853424072266,
"rewards/margins": 9.849853515625,
"rewards/real": 3.353998899459839,
"step": 2080
},
{
"epoch": 1.34,
"learning_rate": 1.8414504088162103e-07,
"logits/generated": 6.150097370147705,
"logits/real": 4.1030778884887695,
"logps/generated": -969.1099853515625,
"logps/real": -210.23385620117188,
"loss": 0.0035,
"rewards/accuracies": 1.0,
"rewards/generated": -6.695557594299316,
"rewards/margins": 9.90971851348877,
"rewards/real": 3.214160203933716,
"step": 2090
},
{
"epoch": 1.34,
"learning_rate": 1.8236757909704943e-07,
"logits/generated": 5.9813079833984375,
"logits/real": 4.158692359924316,
"logps/generated": -843.7550659179688,
"logps/real": -198.7861328125,
"loss": 0.0043,
"rewards/accuracies": 1.0,
"rewards/generated": -5.507570743560791,
"rewards/margins": 8.458051681518555,
"rewards/real": 2.950481414794922,
"step": 2100
},
{
"epoch": 1.35,
"learning_rate": 1.805901173124778e-07,
"logits/generated": 6.074183940887451,
"logits/real": 4.167660236358643,
"logps/generated": -904.7354736328125,
"logps/real": -216.07406616210938,
"loss": 0.0057,
"rewards/accuracies": 1.0,
"rewards/generated": -5.98261833190918,
"rewards/margins": 9.305191993713379,
"rewards/real": 3.3225739002227783,
"step": 2110
},
{
"epoch": 1.36,
"learning_rate": 1.7881265552790613e-07,
"logits/generated": 6.014121055603027,
"logits/real": 4.315189838409424,
"logps/generated": -998.3967895507812,
"logps/real": -213.5117950439453,
"loss": 0.0037,
"rewards/accuracies": 1.0,
"rewards/generated": -6.877680778503418,
"rewards/margins": 10.27568244934082,
"rewards/real": 3.398001194000244,
"step": 2120
},
{
"epoch": 1.36,
"learning_rate": 1.770351937433345e-07,
"logits/generated": 6.0427069664001465,
"logits/real": 4.096532344818115,
"logps/generated": -962.1192626953125,
"logps/real": -211.24386596679688,
"loss": 0.0039,
"rewards/accuracies": 1.0,
"rewards/generated": -6.622589111328125,
"rewards/margins": 9.86971664428711,
"rewards/real": 3.24712872505188,
"step": 2130
},
{
"epoch": 1.37,
"learning_rate": 1.7525773195876286e-07,
"logits/generated": 5.948111534118652,
"logits/real": 4.327864170074463,
"logps/generated": -862.5738525390625,
"logps/real": -236.5964813232422,
"loss": 0.0034,
"rewards/accuracies": 1.0,
"rewards/generated": -5.988071441650391,
"rewards/margins": 9.170713424682617,
"rewards/real": 3.1826424598693848,
"step": 2140
},
{
"epoch": 1.38,
"learning_rate": 1.7348027017419126e-07,
"logits/generated": 6.061889171600342,
"logits/real": 4.250190258026123,
"logps/generated": -962.5604248046875,
"logps/real": -212.2547149658203,
"loss": 0.0056,
"rewards/accuracies": 1.0,
"rewards/generated": -6.656978607177734,
"rewards/margins": 10.204740524291992,
"rewards/real": 3.547761917114258,
"step": 2150
},
{
"epoch": 1.38,
"learning_rate": 1.7170280838961962e-07,
"logits/generated": 6.133549690246582,
"logits/real": 4.258768081665039,
"logps/generated": -847.89306640625,
"logps/real": -213.8179931640625,
"loss": 0.0024,
"rewards/accuracies": 1.0,
"rewards/generated": -5.887804985046387,
"rewards/margins": 9.027533531188965,
"rewards/real": 3.1397290229797363,
"step": 2160
},
{
"epoch": 1.39,
"learning_rate": 1.69925346605048e-07,
"logits/generated": 6.092435359954834,
"logits/real": 3.95710825920105,
"logps/generated": -1076.5191650390625,
"logps/real": -204.7867889404297,
"loss": 0.0046,
"rewards/accuracies": 1.0,
"rewards/generated": -7.864550590515137,
"rewards/margins": 11.004741668701172,
"rewards/real": 3.1401913166046143,
"step": 2170
},
{
"epoch": 1.39,
"learning_rate": 1.6814788482047635e-07,
"logits/generated": 6.069060325622559,
"logits/real": 4.248088836669922,
"logps/generated": -1034.2674560546875,
"logps/real": -239.43045043945312,
"loss": 0.003,
"rewards/accuracies": 1.0,
"rewards/generated": -7.654637813568115,
"rewards/margins": 11.039973258972168,
"rewards/real": 3.385335922241211,
"step": 2180
},
{
"epoch": 1.4,
"learning_rate": 1.6637042303590475e-07,
"logits/generated": 6.127971649169922,
"logits/real": 4.42900276184082,
"logps/generated": -1067.667724609375,
"logps/real": -212.1188507080078,
"loss": 0.0028,
"rewards/accuracies": 1.0,
"rewards/generated": -7.7270331382751465,
"rewards/margins": 11.030082702636719,
"rewards/real": 3.3030498027801514,
"step": 2190
},
{
"epoch": 1.41,
"learning_rate": 1.6459296125133309e-07,
"logits/generated": 6.039584159851074,
"logits/real": 4.363356113433838,
"logps/generated": -967.5791015625,
"logps/real": -234.0696258544922,
"loss": 0.002,
"rewards/accuracies": 1.0,
"rewards/generated": -6.919652462005615,
"rewards/margins": 10.246850967407227,
"rewards/real": 3.3271987438201904,
"step": 2200
},
{
"epoch": 1.41,
"learning_rate": 1.6281549946676145e-07,
"logits/generated": 5.848420143127441,
"logits/real": 4.181252479553223,
"logps/generated": -879.5675659179688,
"logps/real": -204.01536560058594,
"loss": 0.0029,
"rewards/accuracies": 1.0,
"rewards/generated": -6.058112144470215,
"rewards/margins": 9.286849975585938,
"rewards/real": 3.2287373542785645,
"step": 2210
},
{
"epoch": 1.42,
"learning_rate": 1.6103803768218982e-07,
"logits/generated": 6.0381035804748535,
"logits/real": 4.147576332092285,
"logps/generated": -888.4138793945312,
"logps/real": -218.30050659179688,
"loss": 0.0046,
"rewards/accuracies": 1.0,
"rewards/generated": -6.2195258140563965,
"rewards/margins": 9.568934440612793,
"rewards/real": 3.3494086265563965,
"step": 2220
},
{
"epoch": 1.43,
"learning_rate": 1.5926057589761818e-07,
"logits/generated": 6.021380424499512,
"logits/real": 4.303619384765625,
"logps/generated": -751.9827880859375,
"logps/real": -225.69869995117188,
"loss": 0.0028,
"rewards/accuracies": 1.0,
"rewards/generated": -5.188918113708496,
"rewards/margins": 8.795283317565918,
"rewards/real": 3.606365203857422,
"step": 2230
},
{
"epoch": 1.43,
"learning_rate": 1.5748311411304658e-07,
"logits/generated": 5.8918375968933105,
"logits/real": 4.548980712890625,
"logps/generated": -925.9884033203125,
"logps/real": -212.60781860351562,
"loss": 0.0044,
"rewards/accuracies": 1.0,
"rewards/generated": -6.482697486877441,
"rewards/margins": 9.820296287536621,
"rewards/real": 3.337599992752075,
"step": 2240
},
{
"epoch": 1.44,
"learning_rate": 1.5570565232847494e-07,
"logits/generated": 5.850405216217041,
"logits/real": 4.177556037902832,
"logps/generated": -941.3117065429688,
"logps/real": -207.508544921875,
"loss": 0.0035,
"rewards/accuracies": 1.0,
"rewards/generated": -6.867137908935547,
"rewards/margins": 10.460260391235352,
"rewards/real": 3.5931217670440674,
"step": 2250
},
{
"epoch": 1.45,
"learning_rate": 1.539281905439033e-07,
"logits/generated": 6.011968612670898,
"logits/real": 4.186308860778809,
"logps/generated": -850.3424072265625,
"logps/real": -213.4459991455078,
"loss": 0.004,
"rewards/accuracies": 1.0,
"rewards/generated": -6.205097198486328,
"rewards/margins": 9.331758499145508,
"rewards/real": 3.1266608238220215,
"step": 2260
},
{
"epoch": 1.45,
"learning_rate": 1.5215072875933165e-07,
"logits/generated": 5.9976019859313965,
"logits/real": 4.6466498374938965,
"logps/generated": -959.1027221679688,
"logps/real": -230.4766845703125,
"loss": 0.0033,
"rewards/accuracies": 1.0,
"rewards/generated": -6.896917819976807,
"rewards/margins": 10.29023551940918,
"rewards/real": 3.3933181762695312,
"step": 2270
},
{
"epoch": 1.46,
"learning_rate": 1.5037326697476004e-07,
"logits/generated": 5.940140247344971,
"logits/real": 3.9636178016662598,
"logps/generated": -811.2144775390625,
"logps/real": -191.14474487304688,
"loss": 0.0044,
"rewards/accuracies": 1.0,
"rewards/generated": -5.637009620666504,
"rewards/margins": 9.112122535705566,
"rewards/real": 3.4751129150390625,
"step": 2280
},
{
"epoch": 1.47,
"learning_rate": 1.485958051901884e-07,
"logits/generated": 6.092099189758301,
"logits/real": 4.396188259124756,
"logps/generated": -1047.5501708984375,
"logps/real": -235.4965362548828,
"loss": 0.004,
"rewards/accuracies": 1.0,
"rewards/generated": -7.702287197113037,
"rewards/margins": 11.126907348632812,
"rewards/real": 3.42461895942688,
"step": 2290
},
{
"epoch": 1.47,
"learning_rate": 1.4681834340561677e-07,
"logits/generated": 6.009620666503906,
"logits/real": 4.459410667419434,
"logps/generated": -1028.2261962890625,
"logps/real": -224.2885284423828,
"loss": 0.0025,
"rewards/accuracies": 1.0,
"rewards/generated": -7.595302581787109,
"rewards/margins": 11.32234001159668,
"rewards/real": 3.7270379066467285,
"step": 2300
},
{
"epoch": 1.48,
"learning_rate": 1.4504088162104514e-07,
"logits/generated": 6.041359901428223,
"logits/real": 4.407500267028809,
"logps/generated": -1056.5250244140625,
"logps/real": -230.5962677001953,
"loss": 0.0051,
"rewards/accuracies": 1.0,
"rewards/generated": -8.136027336120605,
"rewards/margins": 11.669519424438477,
"rewards/real": 3.533491611480713,
"step": 2310
},
{
"epoch": 1.48,
"learning_rate": 1.432634198364735e-07,
"logits/generated": 6.096525192260742,
"logits/real": 4.020089149475098,
"logps/generated": -940.7005615234375,
"logps/real": -217.18240356445312,
"loss": 0.002,
"rewards/accuracies": 1.0,
"rewards/generated": -6.886039733886719,
"rewards/margins": 10.523492813110352,
"rewards/real": 3.637453079223633,
"step": 2320
},
{
"epoch": 1.49,
"learning_rate": 1.414859580519019e-07,
"logits/generated": 5.8838653564453125,
"logits/real": 4.09322452545166,
"logps/generated": -871.5797729492188,
"logps/real": -189.5806427001953,
"loss": 0.0043,
"rewards/accuracies": 1.0,
"rewards/generated": -5.9241437911987305,
"rewards/margins": 9.4077730178833,
"rewards/real": 3.483628511428833,
"step": 2330
},
{
"epoch": 1.5,
"learning_rate": 1.3970849626733024e-07,
"logits/generated": 6.006686687469482,
"logits/real": 4.329590797424316,
"logps/generated": -1036.4449462890625,
"logps/real": -228.02566528320312,
"loss": 0.0033,
"rewards/accuracies": 1.0,
"rewards/generated": -7.656490325927734,
"rewards/margins": 10.992205619812012,
"rewards/real": 3.3357155323028564,
"step": 2340
},
{
"epoch": 1.5,
"learning_rate": 1.379310344827586e-07,
"logits/generated": 6.243993282318115,
"logits/real": 4.345526695251465,
"logps/generated": -972.0113525390625,
"logps/real": -213.86898803710938,
"loss": 0.0086,
"rewards/accuracies": 1.0,
"rewards/generated": -7.251333713531494,
"rewards/margins": 10.62246322631836,
"rewards/real": 3.371129274368286,
"step": 2350
},
{
"epoch": 1.51,
"learning_rate": 1.3615357269818697e-07,
"logits/generated": 6.056420803070068,
"logits/real": 4.507519245147705,
"logps/generated": -959.6647338867188,
"logps/real": -228.41311645507812,
"loss": 0.005,
"rewards/accuracies": 1.0,
"rewards/generated": -7.193532466888428,
"rewards/margins": 10.772246360778809,
"rewards/real": 3.578713893890381,
"step": 2360
},
{
"epoch": 1.52,
"learning_rate": 1.3437611091361536e-07,
"logits/generated": 5.848100662231445,
"logits/real": 4.477749824523926,
"logps/generated": -885.1476440429688,
"logps/real": -217.71826171875,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/generated": -6.187966823577881,
"rewards/margins": 9.537440299987793,
"rewards/real": 3.349473476409912,
"step": 2370
},
{
"epoch": 1.52,
"learning_rate": 1.3259864912904373e-07,
"logits/generated": 6.177857875823975,
"logits/real": 3.999253511428833,
"logps/generated": -849.5914916992188,
"logps/real": -186.1688690185547,
"loss": 0.0021,
"rewards/accuracies": 1.0,
"rewards/generated": -5.99931526184082,
"rewards/margins": 9.616048812866211,
"rewards/real": 3.6167335510253906,
"step": 2380
},
{
"epoch": 1.53,
"learning_rate": 1.308211873444721e-07,
"logits/generated": 5.982209205627441,
"logits/real": 4.022818565368652,
"logps/generated": -943.8885498046875,
"logps/real": -191.94265747070312,
"loss": 0.0032,
"rewards/accuracies": 1.0,
"rewards/generated": -7.117369174957275,
"rewards/margins": 10.699490547180176,
"rewards/real": 3.5821213722229004,
"step": 2390
},
{
"epoch": 1.54,
"learning_rate": 1.2904372555990046e-07,
"logits/generated": 6.1165690422058105,
"logits/real": 4.515404224395752,
"logps/generated": -1075.3916015625,
"logps/real": -223.2095489501953,
"loss": 0.0074,
"rewards/accuracies": 1.0,
"rewards/generated": -7.828172206878662,
"rewards/margins": 11.10367202758789,
"rewards/real": 3.2754998207092285,
"step": 2400
},
{
"epoch": 1.54,
"learning_rate": 1.2726626377532883e-07,
"logits/generated": 5.962019920349121,
"logits/real": 4.166319847106934,
"logps/generated": -1018.0177001953125,
"logps/real": -193.1107177734375,
"loss": 0.005,
"rewards/accuracies": 1.0,
"rewards/generated": -7.824549198150635,
"rewards/margins": 11.278244972229004,
"rewards/real": 3.4536960124969482,
"step": 2410
},
{
"epoch": 1.55,
"learning_rate": 1.254888019907572e-07,
"logits/generated": 5.941873073577881,
"logits/real": 4.457114219665527,
"logps/generated": -839.1423950195312,
"logps/real": -225.7877960205078,
"loss": 0.0021,
"rewards/accuracies": 1.0,
"rewards/generated": -6.322813987731934,
"rewards/margins": 9.847611427307129,
"rewards/real": 3.5247981548309326,
"step": 2420
},
{
"epoch": 1.55,
"learning_rate": 1.2371134020618556e-07,
"logits/generated": 6.1170220375061035,
"logits/real": 4.145179748535156,
"logps/generated": -971.7708740234375,
"logps/real": -213.4941864013672,
"loss": 0.0021,
"rewards/accuracies": 1.0,
"rewards/generated": -7.316300392150879,
"rewards/margins": 10.892778396606445,
"rewards/real": 3.5764777660369873,
"step": 2430
},
{
"epoch": 1.56,
"learning_rate": 1.2193387842161392e-07,
"logits/generated": 5.986542701721191,
"logits/real": 4.025667190551758,
"logps/generated": -830.68359375,
"logps/real": -206.99441528320312,
"loss": 0.0063,
"rewards/accuracies": 1.0,
"rewards/generated": -5.811788082122803,
"rewards/margins": 9.347892761230469,
"rewards/real": 3.5361053943634033,
"step": 2440
},
{
"epoch": 1.57,
"learning_rate": 1.2015641663704232e-07,
"logits/generated": 5.972803592681885,
"logits/real": 4.296057224273682,
"logps/generated": -1000.8699340820312,
"logps/real": -219.41787719726562,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/generated": -7.70371150970459,
"rewards/margins": 11.355711936950684,
"rewards/real": 3.6520004272460938,
"step": 2450
},
{
"epoch": 1.57,
"learning_rate": 1.1837895485247067e-07,
"logits/generated": 6.02608585357666,
"logits/real": 4.351847171783447,
"logps/generated": -974.85693359375,
"logps/real": -215.2225341796875,
"loss": 0.0028,
"rewards/accuracies": 1.0,
"rewards/generated": -7.276649475097656,
"rewards/margins": 10.775941848754883,
"rewards/real": 3.4992916584014893,
"step": 2460
},
{
"epoch": 1.58,
"learning_rate": 1.1660149306789902e-07,
"logits/generated": 5.748597145080566,
"logits/real": 4.363656997680664,
"logps/generated": -966.9947509765625,
"logps/real": -231.53848266601562,
"loss": 0.0033,
"rewards/accuracies": 1.0,
"rewards/generated": -7.087010860443115,
"rewards/margins": 10.6896390914917,
"rewards/real": 3.6026291847229004,
"step": 2470
},
{
"epoch": 1.59,
"learning_rate": 1.148240312833274e-07,
"logits/generated": 6.186932563781738,
"logits/real": 4.215136528015137,
"logps/generated": -960.4109497070312,
"logps/real": -200.14752197265625,
"loss": 0.0026,
"rewards/accuracies": 1.0,
"rewards/generated": -7.6329665184021,
"rewards/margins": 11.175156593322754,
"rewards/real": 3.5421881675720215,
"step": 2480
},
{
"epoch": 1.59,
"learning_rate": 1.1304656949875577e-07,
"logits/generated": 5.933046340942383,
"logits/real": 4.051640510559082,
"logps/generated": -947.9317626953125,
"logps/real": -187.852783203125,
"loss": 0.0033,
"rewards/accuracies": 1.0,
"rewards/generated": -6.863749027252197,
"rewards/margins": 10.43601131439209,
"rewards/real": 3.57226300239563,
"step": 2490
},
{
"epoch": 1.6,
"learning_rate": 1.1126910771418415e-07,
"logits/generated": 6.067012786865234,
"logits/real": 4.100882530212402,
"logps/generated": -922.0003662109375,
"logps/real": -218.1196746826172,
"loss": 0.0038,
"rewards/accuracies": 1.0,
"rewards/generated": -6.713512420654297,
"rewards/margins": 10.470623016357422,
"rewards/real": 3.7571117877960205,
"step": 2500
},
{
"epoch": 1.61,
"learning_rate": 1.094916459296125e-07,
"logits/generated": 5.888331413269043,
"logits/real": 4.533158302307129,
"logps/generated": -941.0359497070312,
"logps/real": -235.983154296875,
"loss": 0.0019,
"rewards/accuracies": 1.0,
"rewards/generated": -7.0601677894592285,
"rewards/margins": 10.506025314331055,
"rewards/real": 3.445856809616089,
"step": 2510
},
{
"epoch": 1.61,
"learning_rate": 1.0771418414504088e-07,
"logits/generated": 6.122256278991699,
"logits/real": 4.691792964935303,
"logps/generated": -1032.4281005859375,
"logps/real": -242.85391235351562,
"loss": 0.002,
"rewards/accuracies": 1.0,
"rewards/generated": -7.817018032073975,
"rewards/margins": 11.140421867370605,
"rewards/real": 3.3234035968780518,
"step": 2520
},
{
"epoch": 1.62,
"learning_rate": 1.0593672236046925e-07,
"logits/generated": 6.091986179351807,
"logits/real": 4.298600196838379,
"logps/generated": -979.9375,
"logps/real": -207.18222045898438,
"loss": 0.0019,
"rewards/accuracies": 1.0,
"rewards/generated": -7.318638801574707,
"rewards/margins": 10.859162330627441,
"rewards/real": 3.540522813796997,
"step": 2530
},
{
"epoch": 1.63,
"learning_rate": 1.0415926057589762e-07,
"logits/generated": 5.999621868133545,
"logits/real": 4.219559192657471,
"logps/generated": -949.9793090820312,
"logps/real": -210.6415557861328,
"loss": 0.003,
"rewards/accuracies": 1.0,
"rewards/generated": -7.1169939041137695,
"rewards/margins": 10.84516716003418,
"rewards/real": 3.7281742095947266,
"step": 2540
},
{
"epoch": 1.63,
"learning_rate": 1.0238179879132598e-07,
"logits/generated": 5.945744514465332,
"logits/real": 4.196887969970703,
"logps/generated": -999.5768432617188,
"logps/real": -197.3428192138672,
"loss": 0.0015,
"rewards/accuracies": 1.0,
"rewards/generated": -7.847522735595703,
"rewards/margins": 11.279948234558105,
"rewards/real": 3.4324257373809814,
"step": 2550
},
{
"epoch": 1.64,
"learning_rate": 1.0060433700675434e-07,
"logits/generated": 6.025457859039307,
"logits/real": 4.405503273010254,
"logps/generated": -911.369140625,
"logps/real": -226.285400390625,
"loss": 0.0031,
"rewards/accuracies": 1.0,
"rewards/generated": -6.9073333740234375,
"rewards/margins": 10.433539390563965,
"rewards/real": 3.526205539703369,
"step": 2560
},
{
"epoch": 1.64,
"learning_rate": 9.882687522218272e-08,
"logits/generated": 5.8123369216918945,
"logits/real": 4.084736347198486,
"logps/generated": -892.3932495117188,
"logps/real": -207.2658233642578,
"loss": 0.0027,
"rewards/accuracies": 1.0,
"rewards/generated": -6.392351150512695,
"rewards/margins": 10.307291030883789,
"rewards/real": 3.914940595626831,
"step": 2570
},
{
"epoch": 1.65,
"learning_rate": 9.704941343761109e-08,
"logits/generated": 6.000952243804932,
"logits/real": 4.174363136291504,
"logps/generated": -1008.0392456054688,
"logps/real": -216.5176239013672,
"loss": 0.0026,
"rewards/accuracies": 1.0,
"rewards/generated": -7.5495429039001465,
"rewards/margins": 11.444334030151367,
"rewards/real": 3.8947906494140625,
"step": 2580
},
{
"epoch": 1.66,
"learning_rate": 9.527195165303946e-08,
"logits/generated": 5.935231685638428,
"logits/real": 4.574646949768066,
"logps/generated": -1026.3360595703125,
"logps/real": -224.52346801757812,
"loss": 0.0025,
"rewards/accuracies": 1.0,
"rewards/generated": -7.995066165924072,
"rewards/margins": 11.642257690429688,
"rewards/real": 3.6471920013427734,
"step": 2590
},
{
"epoch": 1.66,
"learning_rate": 9.349448986846782e-08,
"logits/generated": 6.165256500244141,
"logits/real": 4.483449459075928,
"logps/generated": -1073.19873046875,
"logps/real": -215.8119354248047,
"loss": 0.0031,
"rewards/accuracies": 1.0,
"rewards/generated": -8.13926887512207,
"rewards/margins": 11.734591484069824,
"rewards/real": 3.595323085784912,
"step": 2600
},
{
"epoch": 1.67,
"learning_rate": 9.17170280838962e-08,
"logits/generated": 6.178045272827148,
"logits/real": 4.426362991333008,
"logps/generated": -981.1510009765625,
"logps/real": -221.8535614013672,
"loss": 0.0024,
"rewards/accuracies": 1.0,
"rewards/generated": -7.534024238586426,
"rewards/margins": 11.1114501953125,
"rewards/real": 3.5774269104003906,
"step": 2610
},
{
"epoch": 1.68,
"learning_rate": 8.993956629932455e-08,
"logits/generated": 6.002736568450928,
"logits/real": 4.408158302307129,
"logps/generated": -1037.64697265625,
"logps/real": -206.6656036376953,
"loss": 0.0019,
"rewards/accuracies": 1.0,
"rewards/generated": -7.90814733505249,
"rewards/margins": 11.406362533569336,
"rewards/real": 3.498215913772583,
"step": 2620
},
{
"epoch": 1.68,
"learning_rate": 8.816210451475293e-08,
"logits/generated": 5.922328948974609,
"logits/real": 4.30651330947876,
"logps/generated": -883.3487548828125,
"logps/real": -226.1007843017578,
"loss": 0.0042,
"rewards/accuracies": 1.0,
"rewards/generated": -6.406916618347168,
"rewards/margins": 10.177984237670898,
"rewards/real": 3.7710673809051514,
"step": 2630
},
{
"epoch": 1.69,
"learning_rate": 8.63846427301813e-08,
"logits/generated": 5.988922595977783,
"logits/real": 4.155168533325195,
"logps/generated": -918.8670654296875,
"logps/real": -203.3168487548828,
"loss": 0.0023,
"rewards/accuracies": 1.0,
"rewards/generated": -7.007664680480957,
"rewards/margins": 10.59145450592041,
"rewards/real": 3.5837910175323486,
"step": 2640
},
{
"epoch": 1.7,
"learning_rate": 8.460718094560966e-08,
"logits/generated": 5.974033832550049,
"logits/real": 4.323070526123047,
"logps/generated": -953.7561645507812,
"logps/real": -215.21804809570312,
"loss": 0.003,
"rewards/accuracies": 1.0,
"rewards/generated": -7.087655544281006,
"rewards/margins": 10.743782043457031,
"rewards/real": 3.656125545501709,
"step": 2650
},
{
"epoch": 1.7,
"learning_rate": 8.282971916103803e-08,
"logits/generated": 5.881442546844482,
"logits/real": 4.603425025939941,
"logps/generated": -941.333984375,
"logps/real": -229.0399932861328,
"loss": 0.0031,
"rewards/accuracies": 1.0,
"rewards/generated": -6.910244941711426,
"rewards/margins": 10.281949996948242,
"rewards/real": 3.3717052936553955,
"step": 2660
},
{
"epoch": 1.71,
"learning_rate": 8.10522573764664e-08,
"logits/generated": 6.160346031188965,
"logits/real": 4.405267238616943,
"logps/generated": -1009.9052734375,
"logps/real": -224.79244995117188,
"loss": 0.0035,
"rewards/accuracies": 1.0,
"rewards/generated": -7.713784694671631,
"rewards/margins": 11.180540084838867,
"rewards/real": 3.4667556285858154,
"step": 2670
},
{
"epoch": 1.71,
"learning_rate": 7.927479559189478e-08,
"logits/generated": 6.0979323387146,
"logits/real": 4.142810821533203,
"logps/generated": -919.0299072265625,
"logps/real": -200.34954833984375,
"loss": 0.0033,
"rewards/accuracies": 1.0,
"rewards/generated": -6.830052375793457,
"rewards/margins": 10.462355613708496,
"rewards/real": 3.6323037147521973,
"step": 2680
},
{
"epoch": 1.72,
"learning_rate": 7.749733380732314e-08,
"logits/generated": 5.893312454223633,
"logits/real": 4.082333564758301,
"logps/generated": -922.8045043945312,
"logps/real": -211.110107421875,
"loss": 0.0026,
"rewards/accuracies": 1.0,
"rewards/generated": -7.06411600112915,
"rewards/margins": 11.093643188476562,
"rewards/real": 4.029527187347412,
"step": 2690
},
{
"epoch": 1.73,
"learning_rate": 7.571987202275151e-08,
"logits/generated": 6.046978950500488,
"logits/real": 4.000611305236816,
"logps/generated": -851.3162231445312,
"logps/real": -213.77294921875,
"loss": 0.0044,
"rewards/accuracies": 1.0,
"rewards/generated": -6.470201015472412,
"rewards/margins": 10.399231910705566,
"rewards/real": 3.929030656814575,
"step": 2700
},
{
"epoch": 1.73,
"learning_rate": 7.394241023817987e-08,
"logits/generated": 6.083460807800293,
"logits/real": 4.147582054138184,
"logps/generated": -869.0367431640625,
"logps/real": -222.88449096679688,
"loss": 0.0023,
"rewards/accuracies": 1.0,
"rewards/generated": -6.679482936859131,
"rewards/margins": 10.497906684875488,
"rewards/real": 3.8184237480163574,
"step": 2710
},
{
"epoch": 1.74,
"learning_rate": 7.216494845360824e-08,
"logits/generated": 5.894058704376221,
"logits/real": 4.494288921356201,
"logps/generated": -1004.9658203125,
"logps/real": -224.24234008789062,
"loss": 0.0045,
"rewards/accuracies": 1.0,
"rewards/generated": -7.787278175354004,
"rewards/margins": 11.781339645385742,
"rewards/real": 3.994061231613159,
"step": 2720
},
{
"epoch": 1.75,
"learning_rate": 7.038748666903662e-08,
"logits/generated": 6.083498954772949,
"logits/real": 4.627425670623779,
"logps/generated": -1075.3470458984375,
"logps/real": -226.70223999023438,
"loss": 0.0016,
"rewards/accuracies": 1.0,
"rewards/generated": -8.463549613952637,
"rewards/margins": 12.042739868164062,
"rewards/real": 3.579190731048584,
"step": 2730
},
{
"epoch": 1.75,
"learning_rate": 6.861002488446497e-08,
"logits/generated": 5.915851593017578,
"logits/real": 3.9263007640838623,
"logps/generated": -1111.206298828125,
"logps/real": -196.06072998046875,
"loss": 0.0027,
"rewards/accuracies": 1.0,
"rewards/generated": -8.594639778137207,
"rewards/margins": 12.281318664550781,
"rewards/real": 3.6866791248321533,
"step": 2740
},
{
"epoch": 1.76,
"learning_rate": 6.683256309989335e-08,
"logits/generated": 5.970302581787109,
"logits/real": 4.437585830688477,
"logps/generated": -840.1096801757812,
"logps/real": -226.9420928955078,
"loss": 0.0017,
"rewards/accuracies": 1.0,
"rewards/generated": -6.3191237449646,
"rewards/margins": 10.023898124694824,
"rewards/real": 3.704774856567383,
"step": 2750
},
{
"epoch": 1.77,
"learning_rate": 6.505510131532172e-08,
"logits/generated": 6.018714904785156,
"logits/real": 4.127536296844482,
"logps/generated": -929.6011962890625,
"logps/real": -204.32046508789062,
"loss": 0.0024,
"rewards/accuracies": 1.0,
"rewards/generated": -7.2565131187438965,
"rewards/margins": 10.836641311645508,
"rewards/real": 3.5801289081573486,
"step": 2760
},
{
"epoch": 1.77,
"learning_rate": 6.32776395307501e-08,
"logits/generated": 5.825669765472412,
"logits/real": 4.458681106567383,
"logps/generated": -974.4544067382812,
"logps/real": -212.163818359375,
"loss": 0.0024,
"rewards/accuracies": 1.0,
"rewards/generated": -7.371243953704834,
"rewards/margins": 10.979463577270508,
"rewards/real": 3.6082186698913574,
"step": 2770
},
{
"epoch": 1.78,
"learning_rate": 6.150017774617845e-08,
"logits/generated": 5.914022922515869,
"logits/real": 4.451127529144287,
"logps/generated": -971.3380126953125,
"logps/real": -248.98635864257812,
"loss": 0.0053,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -7.341151237487793,
"rewards/margins": 11.175028800964355,
"rewards/real": 3.833878755569458,
"step": 2780
},
{
"epoch": 1.79,
"learning_rate": 5.972271596160682e-08,
"logits/generated": 6.177198886871338,
"logits/real": 4.612268924713135,
"logps/generated": -955.9030151367188,
"logps/real": -221.28634643554688,
"loss": 0.0022,
"rewards/accuracies": 1.0,
"rewards/generated": -7.392415523529053,
"rewards/margins": 10.885235786437988,
"rewards/real": 3.4928202629089355,
"step": 2790
},
{
"epoch": 1.79,
"learning_rate": 5.794525417703519e-08,
"logits/generated": 5.88433837890625,
"logits/real": 4.339784145355225,
"logps/generated": -990.3663330078125,
"logps/real": -223.5160369873047,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/generated": -7.730770111083984,
"rewards/margins": 11.330351829528809,
"rewards/real": 3.599581480026245,
"step": 2800
},
{
"epoch": 1.8,
"learning_rate": 5.616779239246356e-08,
"logits/generated": 6.089413642883301,
"logits/real": 4.758141994476318,
"logps/generated": -946.3424072265625,
"logps/real": -223.90957641601562,
"loss": 0.0016,
"rewards/accuracies": 1.0,
"rewards/generated": -7.317401885986328,
"rewards/margins": 11.000505447387695,
"rewards/real": 3.6831040382385254,
"step": 2810
},
{
"epoch": 1.8,
"learning_rate": 5.439033060789193e-08,
"logits/generated": 5.951352119445801,
"logits/real": 4.1157331466674805,
"logps/generated": -876.4993896484375,
"logps/real": -212.88046264648438,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/generated": -6.624871730804443,
"rewards/margins": 10.744012832641602,
"rewards/real": 4.119140625,
"step": 2820
},
{
"epoch": 1.81,
"learning_rate": 5.26128688233203e-08,
"logits/generated": 5.940617561340332,
"logits/real": 3.8113338947296143,
"logps/generated": -888.90087890625,
"logps/real": -203.17901611328125,
"loss": 0.0017,
"rewards/accuracies": 1.0,
"rewards/generated": -6.581952095031738,
"rewards/margins": 10.407301902770996,
"rewards/real": 3.825350522994995,
"step": 2830
},
{
"epoch": 1.82,
"learning_rate": 5.0835407038748666e-08,
"logits/generated": 5.964734077453613,
"logits/real": 4.341010093688965,
"logps/generated": -881.1970825195312,
"logps/real": -232.49813842773438,
"loss": 0.003,
"rewards/accuracies": 1.0,
"rewards/generated": -6.593221187591553,
"rewards/margins": 10.22175407409668,
"rewards/real": 3.628532886505127,
"step": 2840
},
{
"epoch": 1.82,
"learning_rate": 4.905794525417703e-08,
"logits/generated": 5.8404388427734375,
"logits/real": 4.435536861419678,
"logps/generated": -969.4814453125,
"logps/real": -215.01858520507812,
"loss": 0.0017,
"rewards/accuracies": 1.0,
"rewards/generated": -7.376192569732666,
"rewards/margins": 11.145563125610352,
"rewards/real": 3.769369602203369,
"step": 2850
},
{
"epoch": 1.83,
"learning_rate": 4.72804834696054e-08,
"logits/generated": 5.9956512451171875,
"logits/real": 4.474661827087402,
"logps/generated": -909.1708984375,
"logps/real": -216.86392211914062,
"loss": 0.0016,
"rewards/accuracies": 1.0,
"rewards/generated": -7.260798454284668,
"rewards/margins": 10.6810302734375,
"rewards/real": 3.4202327728271484,
"step": 2860
},
{
"epoch": 1.84,
"learning_rate": 4.550302168503377e-08,
"logits/generated": 6.177473545074463,
"logits/real": 4.074435234069824,
"logps/generated": -931.0374145507812,
"logps/real": -199.68115234375,
"loss": 0.0081,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -7.1744065284729,
"rewards/margins": 11.10449504852295,
"rewards/real": 3.9300880432128906,
"step": 2870
},
{
"epoch": 1.84,
"learning_rate": 4.372555990046214e-08,
"logits/generated": 6.158198356628418,
"logits/real": 4.45182466506958,
"logps/generated": -1053.023681640625,
"logps/real": -227.9947967529297,
"loss": 0.004,
"rewards/accuracies": 1.0,
"rewards/generated": -8.230512619018555,
"rewards/margins": 11.803583145141602,
"rewards/real": 3.5730698108673096,
"step": 2880
},
{
"epoch": 1.85,
"learning_rate": 4.194809811589051e-08,
"logits/generated": 5.763731956481934,
"logits/real": 4.512047290802002,
"logps/generated": -877.0777587890625,
"logps/real": -239.983642578125,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/generated": -6.7064409255981445,
"rewards/margins": 10.49316120147705,
"rewards/real": 3.7867202758789062,
"step": 2890
},
{
"epoch": 1.86,
"learning_rate": 4.0170636331318876e-08,
"logits/generated": 6.0582170486450195,
"logits/real": 4.427424430847168,
"logps/generated": -1025.2308349609375,
"logps/real": -215.9585418701172,
"loss": 0.0027,
"rewards/accuracies": 1.0,
"rewards/generated": -8.088347434997559,
"rewards/margins": 11.211756706237793,
"rewards/real": 3.1234092712402344,
"step": 2900
},
{
"epoch": 1.86,
"learning_rate": 3.839317454674725e-08,
"logits/generated": 6.159816741943359,
"logits/real": 4.5531134605407715,
"logps/generated": -914.5631713867188,
"logps/real": -223.43667602539062,
"loss": 0.0034,
"rewards/accuracies": 1.0,
"rewards/generated": -7.19748067855835,
"rewards/margins": 11.130581855773926,
"rewards/real": 3.933100938796997,
"step": 2910
},
{
"epoch": 1.87,
"learning_rate": 3.6615712762175614e-08,
"logits/generated": 6.062338829040527,
"logits/real": 3.9617767333984375,
"logps/generated": -1065.44189453125,
"logps/real": -200.53231811523438,
"loss": 0.0022,
"rewards/accuracies": 1.0,
"rewards/generated": -8.692277908325195,
"rewards/margins": 12.533926010131836,
"rewards/real": 3.841648578643799,
"step": 2920
},
{
"epoch": 1.87,
"learning_rate": 3.4838250977603974e-08,
"logits/generated": 5.966034889221191,
"logits/real": 4.169475555419922,
"logps/generated": -900.4786376953125,
"logps/real": -219.13211059570312,
"loss": 0.0023,
"rewards/accuracies": 1.0,
"rewards/generated": -6.983071804046631,
"rewards/margins": 10.687957763671875,
"rewards/real": 3.7048866748809814,
"step": 2930
},
{
"epoch": 1.88,
"learning_rate": 3.3060789193032346e-08,
"logits/generated": 6.001954555511475,
"logits/real": 4.039769649505615,
"logps/generated": -948.9764404296875,
"logps/real": -213.39022827148438,
"loss": 0.0019,
"rewards/accuracies": 1.0,
"rewards/generated": -7.106842041015625,
"rewards/margins": 10.968811988830566,
"rewards/real": 3.8619697093963623,
"step": 2940
},
{
"epoch": 1.89,
"learning_rate": 3.128332740846071e-08,
"logits/generated": 6.0790205001831055,
"logits/real": 4.268954753875732,
"logps/generated": -969.9235229492188,
"logps/real": -210.45849609375,
"loss": 0.0016,
"rewards/accuracies": 1.0,
"rewards/generated": -7.430506706237793,
"rewards/margins": 11.069308280944824,
"rewards/real": 3.638800859451294,
"step": 2950
},
{
"epoch": 1.89,
"learning_rate": 2.9505865623889085e-08,
"logits/generated": 6.002804756164551,
"logits/real": 4.1793718338012695,
"logps/generated": -1044.44287109375,
"logps/real": -208.38851928710938,
"loss": 0.0026,
"rewards/accuracies": 1.0,
"rewards/generated": -8.205821990966797,
"rewards/margins": 11.79900074005127,
"rewards/real": 3.593177318572998,
"step": 2960
},
{
"epoch": 1.9,
"learning_rate": 2.7728403839317454e-08,
"logits/generated": 6.175272464752197,
"logits/real": 4.347874641418457,
"logps/generated": -908.1828002929688,
"logps/real": -213.10421752929688,
"loss": 0.0025,
"rewards/accuracies": 1.0,
"rewards/generated": -7.079012870788574,
"rewards/margins": 10.570651054382324,
"rewards/real": 3.4916369915008545,
"step": 2970
},
{
"epoch": 1.91,
"learning_rate": 2.5950942054745824e-08,
"logits/generated": 5.989743232727051,
"logits/real": 4.590392112731934,
"logps/generated": -881.7191162109375,
"logps/real": -221.69577026367188,
"loss": 0.0017,
"rewards/accuracies": 1.0,
"rewards/generated": -6.744132995605469,
"rewards/margins": 10.367905616760254,
"rewards/real": 3.6237728595733643,
"step": 2980
},
{
"epoch": 1.91,
"learning_rate": 2.4173480270174193e-08,
"logits/generated": 5.84287166595459,
"logits/real": 4.579287052154541,
"logps/generated": -880.3978271484375,
"logps/real": -232.52279663085938,
"loss": 0.0031,
"rewards/accuracies": 1.0,
"rewards/generated": -6.6468071937561035,
"rewards/margins": 10.200614929199219,
"rewards/real": 3.553807020187378,
"step": 2990
},
{
"epoch": 1.92,
"learning_rate": 2.2396018485602556e-08,
"logits/generated": 6.122179985046387,
"logits/real": 4.397927761077881,
"logps/generated": -998.4055786132812,
"logps/real": -210.6758575439453,
"loss": 0.0024,
"rewards/accuracies": 1.0,
"rewards/generated": -7.7901740074157715,
"rewards/margins": 11.620798110961914,
"rewards/real": 3.830622434616089,
"step": 3000
},
{
"epoch": 1.93,
"learning_rate": 2.0618556701030925e-08,
"logits/generated": 5.882719993591309,
"logits/real": 4.193206787109375,
"logps/generated": -996.2776489257812,
"logps/real": -182.93798828125,
"loss": 0.0019,
"rewards/accuracies": 1.0,
"rewards/generated": -8.111333847045898,
"rewards/margins": 11.724261283874512,
"rewards/real": 3.612928867340088,
"step": 3010
},
{
"epoch": 1.93,
"learning_rate": 1.8841094916459295e-08,
"logits/generated": 6.127943992614746,
"logits/real": 4.209385395050049,
"logps/generated": -1060.7899169921875,
"logps/real": -205.1070098876953,
"loss": 0.002,
"rewards/accuracies": 1.0,
"rewards/generated": -8.274694442749023,
"rewards/margins": 11.977832794189453,
"rewards/real": 3.703138828277588,
"step": 3020
},
{
"epoch": 1.94,
"learning_rate": 1.7063633131887664e-08,
"logits/generated": 5.8941450119018555,
"logits/real": 4.013119697570801,
"logps/generated": -973.5771484375,
"logps/real": -188.08303833007812,
"loss": 0.0019,
"rewards/accuracies": 1.0,
"rewards/generated": -7.611605644226074,
"rewards/margins": 11.475560188293457,
"rewards/real": 3.8639538288116455,
"step": 3030
},
{
"epoch": 1.94,
"learning_rate": 1.5286171347316033e-08,
"logits/generated": 6.093392372131348,
"logits/real": 4.220031261444092,
"logps/generated": -948.8116455078125,
"logps/real": -216.5891876220703,
"loss": 0.0037,
"rewards/accuracies": 1.0,
"rewards/generated": -7.310966491699219,
"rewards/margins": 10.79468059539795,
"rewards/real": 3.4837143421173096,
"step": 3040
},
{
"epoch": 1.95,
"learning_rate": 1.35087095627444e-08,
"logits/generated": 6.005775451660156,
"logits/real": 4.406448841094971,
"logps/generated": -944.8621826171875,
"logps/real": -192.6377716064453,
"loss": 0.0028,
"rewards/accuracies": 1.0,
"rewards/generated": -7.452921390533447,
"rewards/margins": 11.0402193069458,
"rewards/real": 3.587296962738037,
"step": 3050
},
{
"epoch": 1.96,
"learning_rate": 1.1731247778172769e-08,
"logits/generated": 6.163327693939209,
"logits/real": 4.344714641571045,
"logps/generated": -917.35498046875,
"logps/real": -198.57122802734375,
"loss": 0.0016,
"rewards/accuracies": 1.0,
"rewards/generated": -7.304339408874512,
"rewards/margins": 10.936470985412598,
"rewards/real": 3.632131576538086,
"step": 3060
},
{
"epoch": 1.96,
"learning_rate": 9.953785993601137e-09,
"logits/generated": 6.108910083770752,
"logits/real": 4.15443754196167,
"logps/generated": -962.0145263671875,
"logps/real": -203.24107360839844,
"loss": 0.0027,
"rewards/accuracies": 1.0,
"rewards/generated": -7.498394012451172,
"rewards/margins": 10.694067001342773,
"rewards/real": 3.1956734657287598,
"step": 3070
},
{
"epoch": 1.97,
"learning_rate": 8.176324209029506e-09,
"logits/generated": 6.031611919403076,
"logits/real": 3.9185752868652344,
"logps/generated": -981.349609375,
"logps/real": -206.8091278076172,
"loss": 0.0025,
"rewards/accuracies": 1.0,
"rewards/generated": -7.681130409240723,
"rewards/margins": 11.426141738891602,
"rewards/real": 3.745011806488037,
"step": 3080
},
{
"epoch": 1.98,
"learning_rate": 6.398862424457874e-09,
"logits/generated": 5.970823764801025,
"logits/real": 4.175595283508301,
"logps/generated": -943.4645385742188,
"logps/real": -184.49136352539062,
"loss": 0.0026,
"rewards/accuracies": 1.0,
"rewards/generated": -6.951342582702637,
"rewards/margins": 10.521745681762695,
"rewards/real": 3.5704009532928467,
"step": 3090
},
{
"epoch": 1.98,
"learning_rate": 4.621400639886242e-09,
"logits/generated": 6.0879998207092285,
"logits/real": 4.444035530090332,
"logps/generated": -1086.6719970703125,
"logps/real": -244.66543579101562,
"loss": 0.0026,
"rewards/accuracies": 1.0,
"rewards/generated": -8.595160484313965,
"rewards/margins": 12.099470138549805,
"rewards/real": 3.5043106079101562,
"step": 3100
},
{
"epoch": 1.99,
"learning_rate": 2.8439388553146107e-09,
"logits/generated": 6.041680335998535,
"logits/real": 4.341583728790283,
"logps/generated": -994.8681640625,
"logps/real": -205.96835327148438,
"loss": 0.0038,
"rewards/accuracies": 1.0,
"rewards/generated": -7.797635078430176,
"rewards/margins": 11.64828872680664,
"rewards/real": 3.850653886795044,
"step": 3110
},
{
"epoch": 2.0,
"learning_rate": 1.066477070742979e-09,
"logits/generated": 6.018385887145996,
"logits/real": 3.9701621532440186,
"logps/generated": -901.3045654296875,
"logps/real": -204.19090270996094,
"loss": 0.0027,
"rewards/accuracies": 1.0,
"rewards/generated": -6.618832588195801,
"rewards/margins": 10.127131462097168,
"rewards/real": 3.5082993507385254,
"step": 3120
},
{
"epoch": 2.0,
"step": 3126,
"total_flos": 0.0,
"train_loss": 0.12108022763431954,
"train_runtime": 40273.1203,
"train_samples_per_second": 2.483,
"train_steps_per_second": 0.078
}
],
"logging_steps": 10,
"max_steps": 3126,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 500,
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}