MMR-DR_GRPO / trainer_state.json
kangdawei's picture
Model save
459d6a3 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.5714285714285714,
"eval_steps": 500,
"global_step": 500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"completion_length": 2571.2083587646484,
"epoch": 0.001142857142857143,
"grad_norm": 0.12744835019111633,
"kl": 0.0,
"learning_rate": 2e-08,
"loss": 0.0681,
"reward": 0.1723687592893839,
"reward_std": 0.7976016625761986,
"rewards/cosine_scaled_reward": -0.015534311532974243,
"rewards/format_reward": 0.5208333488553762,
"step": 1
},
{
"completion_length": 2804.395881652832,
"epoch": 0.002285714285714286,
"grad_norm": 0.061661649495363235,
"kl": 0.0,
"learning_rate": 4e-08,
"loss": 0.0245,
"reward": -0.018269629566930234,
"reward_std": 0.44402940198779106,
"rewards/cosine_scaled_reward": -0.04980122856795788,
"rewards/format_reward": 0.37500000558793545,
"step": 2
},
{
"completion_length": 3337.375030517578,
"epoch": 0.0034285714285714284,
"grad_norm": 0.10997606813907623,
"kl": 4.9442052841186523e-05,
"learning_rate": 6e-08,
"loss": -0.0096,
"reward": -0.3936590664088726,
"reward_std": 0.5575782060623169,
"rewards/cosine_scaled_reward": -0.19862568378448486,
"rewards/format_reward": 0.1458333395421505,
"step": 3
},
{
"completion_length": 2260.895881652832,
"epoch": 0.004571428571428572,
"grad_norm": 0.15913932025432587,
"kl": 2.993643283843994e-05,
"learning_rate": 8e-08,
"loss": 0.0445,
"reward": 0.13440400827676058,
"reward_std": 0.8942861538380384,
"rewards/cosine_scaled_reward": -0.10464579728432,
"rewards/format_reward": 0.6250000018626451,
"step": 4
},
{
"completion_length": 3328.2916870117188,
"epoch": 0.005714285714285714,
"grad_norm": 0.12190493941307068,
"kl": 4.523247480392456e-05,
"learning_rate": 1e-07,
"loss": 0.0609,
"reward": -0.27870709635317326,
"reward_std": 0.7129274010658264,
"rewards/cosine_scaled_reward": -0.2122665431816131,
"rewards/format_reward": 0.31250000931322575,
"step": 5
},
{
"completion_length": 3129.8333740234375,
"epoch": 0.006857142857142857,
"grad_norm": 0.14417913556098938,
"kl": 4.3526291847229004e-05,
"learning_rate": 1.2e-07,
"loss": -0.0215,
"reward": -0.07921706140041351,
"reward_std": 0.6614448297768831,
"rewards/cosine_scaled_reward": -0.06220451416447759,
"rewards/format_reward": 0.29166666977107525,
"step": 6
},
{
"completion_length": 3113.729217529297,
"epoch": 0.008,
"grad_norm": 0.1703435778617859,
"kl": 1.8851831555366516e-05,
"learning_rate": 1.4e-07,
"loss": 0.0934,
"reward": -0.0697940494865179,
"reward_std": 0.8283112272620201,
"rewards/cosine_scaled_reward": -0.14727119728922844,
"rewards/format_reward": 0.4583333395421505,
"step": 7
},
{
"completion_length": 2710.6041870117188,
"epoch": 0.009142857142857144,
"grad_norm": 0.1097821518778801,
"kl": 2.4922192096710205e-05,
"learning_rate": 1.6e-07,
"loss": 0.033,
"reward": 0.20415285229682922,
"reward_std": 0.7091612815856934,
"rewards/cosine_scaled_reward": 0.06203982699662447,
"rewards/format_reward": 0.41666667349636555,
"step": 8
},
{
"completion_length": 3129.687530517578,
"epoch": 0.010285714285714285,
"grad_norm": 0.1483897864818573,
"kl": 4.6446919441223145e-05,
"learning_rate": 1.8e-07,
"loss": 0.0837,
"reward": -0.2891757604666054,
"reward_std": 0.7400874830782413,
"rewards/cosine_scaled_reward": -0.21521726623177528,
"rewards/format_reward": 0.2916666753590107,
"step": 9
},
{
"completion_length": 2703.3750076293945,
"epoch": 0.011428571428571429,
"grad_norm": 0.11335786432027817,
"kl": 3.4183263778686523e-05,
"learning_rate": 2e-07,
"loss": 0.0303,
"reward": -0.11180419684387743,
"reward_std": 0.7680593952536583,
"rewards/cosine_scaled_reward": -0.1549013671465218,
"rewards/format_reward": 0.4166666679084301,
"step": 10
},
{
"completion_length": 3326.854217529297,
"epoch": 0.012571428571428572,
"grad_norm": 0.13309723138809204,
"kl": 3.510713577270508e-05,
"learning_rate": 2.1999999999999998e-07,
"loss": 0.0705,
"reward": -0.4452022071927786,
"reward_std": 0.6313146576285362,
"rewards/cosine_scaled_reward": -0.23017787747085094,
"rewards/format_reward": 0.12500000186264515,
"step": 11
},
{
"completion_length": 2544.666717529297,
"epoch": 0.013714285714285714,
"grad_norm": 0.14041368663311005,
"kl": 3.600120544433594e-05,
"learning_rate": 2.4e-07,
"loss": 0.0277,
"reward": 0.006800652190577239,
"reward_std": 0.7336893752217293,
"rewards/cosine_scaled_reward": -0.17795369494706392,
"rewards/format_reward": 0.6250000204890966,
"step": 12
},
{
"completion_length": 3018.8333587646484,
"epoch": 0.014857142857142857,
"grad_norm": 0.13021793961524963,
"kl": 3.663450479507446e-05,
"learning_rate": 2.6e-07,
"loss": 0.0269,
"reward": -0.03413328202441335,
"reward_std": 0.6794986762106419,
"rewards/cosine_scaled_reward": -0.06286561000160873,
"rewards/format_reward": 0.3541666828095913,
"step": 13
},
{
"completion_length": 2947.5208740234375,
"epoch": 0.016,
"grad_norm": 0.12336233258247375,
"kl": 2.9928982257843018e-05,
"learning_rate": 2.8e-07,
"loss": 0.0508,
"reward": 0.07709243893623352,
"reward_std": 0.8241597190499306,
"rewards/cosine_scaled_reward": -0.015953163150697947,
"rewards/format_reward": 0.39583334140479565,
"step": 14
},
{
"completion_length": 2676.5625228881836,
"epoch": 0.017142857142857144,
"grad_norm": 0.047696515917778015,
"kl": 3.113597631454468e-05,
"learning_rate": 3e-07,
"loss": 0.0069,
"reward": 0.036701809614896774,
"reward_std": 0.43677932769060135,
"rewards/cosine_scaled_reward": -0.021881014108657837,
"rewards/format_reward": 0.39583333395421505,
"step": 15
},
{
"completion_length": 3442.0000610351562,
"epoch": 0.018285714285714287,
"grad_norm": 0.12319929152727127,
"kl": 3.999471664428711e-05,
"learning_rate": 3.2e-07,
"loss": 0.0369,
"reward": -0.43698735162615776,
"reward_std": 0.556250561028719,
"rewards/cosine_scaled_reward": -0.20507843233644962,
"rewards/format_reward": 0.1041666679084301,
"step": 16
},
{
"completion_length": 2437.750011444092,
"epoch": 0.019428571428571427,
"grad_norm": 0.14334183931350708,
"kl": 3.676116466522217e-05,
"learning_rate": 3.4000000000000003e-07,
"loss": 0.0128,
"reward": 0.32777530141174793,
"reward_std": 0.8251273017376661,
"rewards/cosine_scaled_reward": 0.026506672613322735,
"rewards/format_reward": 0.6458333432674408,
"step": 17
},
{
"completion_length": 2874.2292098999023,
"epoch": 0.02057142857142857,
"grad_norm": 0.1185201108455658,
"kl": 1.5683472156524658e-05,
"learning_rate": 3.6e-07,
"loss": 0.0031,
"reward": -0.024667851626873016,
"reward_std": 0.8119602091610432,
"rewards/cosine_scaled_reward": -0.12354222661815584,
"rewards/format_reward": 0.4583333395421505,
"step": 18
},
{
"completion_length": 3046.6458587646484,
"epoch": 0.021714285714285714,
"grad_norm": 0.19736716151237488,
"kl": 2.751254942268133e-05,
"learning_rate": 3.7999999999999996e-07,
"loss": 0.0779,
"reward": 0.09869576059281826,
"reward_std": 0.8057069275528193,
"rewards/cosine_scaled_reward": 0.01759765949100256,
"rewards/format_reward": 0.3541666753590107,
"step": 19
},
{
"completion_length": 2335.291732788086,
"epoch": 0.022857142857142857,
"grad_norm": 0.21245542168617249,
"kl": 1.4175660908222198e-05,
"learning_rate": 4e-07,
"loss": 0.1185,
"reward": 0.38306641951203346,
"reward_std": 0.8724737018346786,
"rewards/cosine_scaled_reward": 0.04800319205969572,
"rewards/format_reward": 0.666666679084301,
"step": 20
},
{
"completion_length": 2711.3958435058594,
"epoch": 0.024,
"grad_norm": 0.15821056067943573,
"kl": 2.4184584617614746e-05,
"learning_rate": 4.1999999999999995e-07,
"loss": 0.0511,
"reward": 0.1797822918742895,
"reward_std": 0.9325026646256447,
"rewards/cosine_scaled_reward": -0.0036781951785087585,
"rewards/format_reward": 0.47916666977107525,
"step": 21
},
{
"completion_length": 1910.2083892822266,
"epoch": 0.025142857142857144,
"grad_norm": 0.14541339874267578,
"kl": 2.391217276453972e-05,
"learning_rate": 4.3999999999999997e-07,
"loss": 0.0295,
"reward": 0.347878472879529,
"reward_std": 0.6697478331625462,
"rewards/cosine_scaled_reward": -0.004899490624666214,
"rewards/format_reward": 0.7500000074505806,
"step": 22
},
{
"completion_length": 2642.2083740234375,
"epoch": 0.026285714285714287,
"grad_norm": 0.13504153490066528,
"kl": 2.520345151424408e-05,
"learning_rate": 4.6e-07,
"loss": 0.0768,
"reward": 0.019413750036619604,
"reward_std": 0.9948069639503956,
"rewards/cosine_scaled_reward": -0.10891764007828897,
"rewards/format_reward": 0.45833334513008595,
"step": 23
},
{
"completion_length": 2697.000015258789,
"epoch": 0.027428571428571427,
"grad_norm": 0.13816049695014954,
"kl": 2.109631896018982e-05,
"learning_rate": 4.8e-07,
"loss": 0.0697,
"reward": 0.3483977415598929,
"reward_std": 0.8868874367326498,
"rewards/cosine_scaled_reward": 0.06968503817915916,
"rewards/format_reward": 0.5833333432674408,
"step": 24
},
{
"completion_length": 2679.3958587646484,
"epoch": 0.02857142857142857,
"grad_norm": 0.10494664311408997,
"kl": 3.129430115222931e-05,
"learning_rate": 5e-07,
"loss": 0.001,
"reward": 0.1587733030319214,
"reward_std": 0.7055602557957172,
"rewards/cosine_scaled_reward": 0.02119587583001703,
"rewards/format_reward": 0.4375000074505806,
"step": 25
},
{
"completion_length": 3099.979217529297,
"epoch": 0.029714285714285714,
"grad_norm": 0.075159452855587,
"kl": 2.6673078536987305e-05,
"learning_rate": 5.2e-07,
"loss": 0.003,
"reward": -0.006685070693492889,
"reward_std": 0.5016829147934914,
"rewards/cosine_scaled_reward": -0.08887793682515621,
"rewards/format_reward": 0.45833333395421505,
"step": 26
},
{
"completion_length": 2958.8958740234375,
"epoch": 0.030857142857142857,
"grad_norm": 0.13963520526885986,
"kl": 1.3343989849090576e-05,
"learning_rate": 5.4e-07,
"loss": 0.0628,
"reward": 0.024143089074641466,
"reward_std": 0.7820315174758434,
"rewards/cosine_scaled_reward": -0.0528924111276865,
"rewards/format_reward": 0.3958333358168602,
"step": 27
},
{
"completion_length": 2831.958366394043,
"epoch": 0.032,
"grad_norm": 0.0931120440363884,
"kl": 2.434663474559784e-05,
"learning_rate": 5.6e-07,
"loss": 0.045,
"reward": -0.018430547177558765,
"reward_std": 0.626394847407937,
"rewards/cosine_scaled_reward": -0.051781938411295414,
"rewards/format_reward": 0.3541666753590107,
"step": 28
},
{
"completion_length": 3351.750030517578,
"epoch": 0.03314285714285714,
"grad_norm": 0.10650024563074112,
"kl": 2.886354923248291e-05,
"learning_rate": 5.8e-07,
"loss": 0.0438,
"reward": -0.466870941221714,
"reward_std": 0.4314545188099146,
"rewards/cosine_scaled_reward": -0.2414477914571762,
"rewards/format_reward": 0.1458333358168602,
"step": 29
},
{
"completion_length": 2859.812545776367,
"epoch": 0.03428571428571429,
"grad_norm": 0.18510645627975464,
"kl": 1.806020736694336e-05,
"learning_rate": 6e-07,
"loss": 0.1004,
"reward": 0.23486553132534027,
"reward_std": 1.0620297119021416,
"rewards/cosine_scaled_reward": 0.04186862939968705,
"rewards/format_reward": 0.4583333469927311,
"step": 30
},
{
"completion_length": 3150.250030517578,
"epoch": 0.03542857142857143,
"grad_norm": 0.11023243516683578,
"kl": 1.753866672515869e-05,
"learning_rate": 6.2e-07,
"loss": 0.0007,
"reward": -0.2325914899702184,
"reward_std": 0.6160467248409986,
"rewards/cosine_scaled_reward": -0.16382855689153075,
"rewards/format_reward": 0.29166667722165585,
"step": 31
},
{
"completion_length": 3241.166717529297,
"epoch": 0.036571428571428574,
"grad_norm": 0.16251012682914734,
"kl": 3.0582770705223083e-05,
"learning_rate": 6.4e-07,
"loss": 0.078,
"reward": -0.17190912459045649,
"reward_std": 0.7624128982424736,
"rewards/cosine_scaled_reward": -0.11033781431615353,
"rewards/format_reward": 0.25000000931322575,
"step": 32
},
{
"completion_length": 3253.625030517578,
"epoch": 0.037714285714285714,
"grad_norm": 0.1084270030260086,
"kl": 4.7340989112854004e-05,
"learning_rate": 6.6e-07,
"loss": 0.0213,
"reward": -0.2144899107515812,
"reward_std": 0.7134921476244926,
"rewards/cosine_scaled_reward": -0.15857082698494196,
"rewards/format_reward": 0.29166667349636555,
"step": 33
},
{
"completion_length": 2398.7708587646484,
"epoch": 0.038857142857142854,
"grad_norm": 0.14240942895412445,
"kl": 6.242096424102783e-05,
"learning_rate": 6.800000000000001e-07,
"loss": 0.0497,
"reward": 0.31558607518672943,
"reward_std": 0.8723539188504219,
"rewards/cosine_scaled_reward": 0.06751577369868755,
"rewards/format_reward": 0.541666679084301,
"step": 34
},
{
"completion_length": 3059.1667098999023,
"epoch": 0.04,
"grad_norm": 0.13201534748077393,
"kl": 4.059448838233948e-05,
"learning_rate": 7e-07,
"loss": 0.0401,
"reward": -0.0117220189422369,
"reward_std": 0.9031526632606983,
"rewards/cosine_scaled_reward": -0.04519825894385576,
"rewards/format_reward": 0.3125000037252903,
"step": 35
},
{
"completion_length": 3364.4791870117188,
"epoch": 0.04114285714285714,
"grad_norm": 0.09137182682752609,
"kl": 6.282329559326172e-05,
"learning_rate": 7.2e-07,
"loss": 0.0275,
"reward": -0.37299076607450843,
"reward_std": 0.5668806284666061,
"rewards/cosine_scaled_reward": -0.19870344595983624,
"rewards/format_reward": 0.16666667349636555,
"step": 36
},
{
"completion_length": 3180.062530517578,
"epoch": 0.04228571428571429,
"grad_norm": 0.06048983708024025,
"kl": 6.483495235443115e-05,
"learning_rate": 7.4e-07,
"loss": 0.0165,
"reward": -0.4461427731439471,
"reward_std": 0.4245890639722347,
"rewards/cosine_scaled_reward": -0.26771107502281666,
"rewards/format_reward": 0.22916666977107525,
"step": 37
},
{
"completion_length": 3264.937530517578,
"epoch": 0.04342857142857143,
"grad_norm": 0.09965316206216812,
"kl": 6.621703505516052e-05,
"learning_rate": 7.599999999999999e-07,
"loss": 0.0253,
"reward": -0.2411605268716812,
"reward_std": 0.5170722529292107,
"rewards/cosine_scaled_reward": -0.10409475707274396,
"rewards/format_reward": 0.1666666679084301,
"step": 38
},
{
"completion_length": 2869.1875534057617,
"epoch": 0.044571428571428574,
"grad_norm": 0.05097671225667,
"kl": 7.027853280305862e-05,
"learning_rate": 7.799999999999999e-07,
"loss": 0.0051,
"reward": 0.1772767100483179,
"reward_std": 0.425695575773716,
"rewards/cosine_scaled_reward": 0.008680417202413082,
"rewards/format_reward": 0.5208333358168602,
"step": 39
},
{
"completion_length": 2741.87508392334,
"epoch": 0.045714285714285714,
"grad_norm": 0.1285083144903183,
"kl": 0.00026188790798187256,
"learning_rate": 8e-07,
"loss": 0.0503,
"reward": -0.07532945368438959,
"reward_std": 0.5845174305140972,
"rewards/cosine_scaled_reward": -0.10031583718955517,
"rewards/format_reward": 0.3750000037252903,
"step": 40
},
{
"completion_length": 3102.6041870117188,
"epoch": 0.046857142857142854,
"grad_norm": 0.11375687271356583,
"kl": 9.709596633911133e-05,
"learning_rate": 8.199999999999999e-07,
"loss": 0.0088,
"reward": -0.23022597841918468,
"reward_std": 0.6951021775603294,
"rewards/cosine_scaled_reward": -0.16865359526127577,
"rewards/format_reward": 0.2916666716337204,
"step": 41
},
{
"completion_length": 2861.2708435058594,
"epoch": 0.048,
"grad_norm": 0.04051314666867256,
"kl": 8.571147918701172e-05,
"learning_rate": 8.399999999999999e-07,
"loss": 0.0008,
"reward": -0.4261672543361783,
"reward_std": 0.2807927541434765,
"rewards/cosine_scaled_reward": -0.277394006960094,
"rewards/format_reward": 0.2916666679084301,
"step": 42
},
{
"completion_length": 3074.937530517578,
"epoch": 0.04914285714285714,
"grad_norm": 0.099971242249012,
"kl": 7.382780313491821e-05,
"learning_rate": 8.599999999999999e-07,
"loss": 0.0538,
"reward": -0.23065751791000366,
"reward_std": 0.6040490940213203,
"rewards/cosine_scaled_reward": -0.13301679654978216,
"rewards/format_reward": 0.2291666679084301,
"step": 43
},
{
"completion_length": 2720.145881652832,
"epoch": 0.05028571428571429,
"grad_norm": 0.10246309638023376,
"kl": 0.0003544166684150696,
"learning_rate": 8.799999999999999e-07,
"loss": -0.011,
"reward": 0.011595025658607483,
"reward_std": 0.5640863105654716,
"rewards/cosine_scaled_reward": -0.05820102244615555,
"rewards/format_reward": 0.4166666679084301,
"step": 44
},
{
"completion_length": 3374.291717529297,
"epoch": 0.05142857142857143,
"grad_norm": 0.11695674061775208,
"kl": 0.00015322864055633545,
"learning_rate": 9e-07,
"loss": 0.0382,
"reward": -0.16595029830932617,
"reward_std": 0.7633817754685879,
"rewards/cosine_scaled_reward": -0.08942665439099073,
"rewards/format_reward": 0.20833333767950535,
"step": 45
},
{
"completion_length": 3254.958335876465,
"epoch": 0.052571428571428575,
"grad_norm": 0.06262751668691635,
"kl": 0.0002702465280890465,
"learning_rate": 9.2e-07,
"loss": 0.0021,
"reward": -0.45498750917613506,
"reward_std": 0.3716874625533819,
"rewards/cosine_scaled_reward": -0.24009231384843588,
"rewards/format_reward": 0.1666666679084301,
"step": 46
},
{
"completion_length": 2913.1250610351562,
"epoch": 0.053714285714285714,
"grad_norm": 0.13851620256900787,
"kl": 0.0001294887624680996,
"learning_rate": 9.399999999999999e-07,
"loss": 0.1169,
"reward": 0.13789150305092335,
"reward_std": 0.9271005466580391,
"rewards/cosine_scaled_reward": 0.006347283720970154,
"rewards/format_reward": 0.416666679084301,
"step": 47
},
{
"completion_length": 2858.250045776367,
"epoch": 0.054857142857142854,
"grad_norm": 0.12427303940057755,
"kl": 0.0005601570010185242,
"learning_rate": 9.6e-07,
"loss": 0.0265,
"reward": -0.051424789475277066,
"reward_std": 0.813489394262433,
"rewards/cosine_scaled_reward": -0.07831720494141337,
"rewards/format_reward": 0.3333333358168602,
"step": 48
},
{
"completion_length": 2364.562530517578,
"epoch": 0.056,
"grad_norm": 0.09100354462862015,
"kl": 0.0003460892476141453,
"learning_rate": 9.8e-07,
"loss": 0.0101,
"reward": 0.31277667777612805,
"reward_std": 0.7601135969161987,
"rewards/cosine_scaled_reward": 0.04609484411776066,
"rewards/format_reward": 0.583333333954215,
"step": 49
},
{
"completion_length": 2957.937530517578,
"epoch": 0.05714285714285714,
"grad_norm": 0.11310350149869919,
"kl": 0.0003506038337945938,
"learning_rate": 1e-06,
"loss": 0.0231,
"reward": 0.07819816470146179,
"reward_std": 0.5129979718476534,
"rewards/cosine_scaled_reward": 0.04247652553021908,
"rewards/format_reward": 0.3125,
"step": 50
},
{
"completion_length": 2296.833335876465,
"epoch": 0.05828571428571429,
"grad_norm": 0.09167315810918808,
"kl": 0.0011107921600341797,
"learning_rate": 9.999890338174275e-07,
"loss": 0.0018,
"reward": 0.04252960532903671,
"reward_std": 0.607517946511507,
"rewards/cosine_scaled_reward": -0.12472434528172016,
"rewards/format_reward": 0.5833333358168602,
"step": 51
},
{
"completion_length": 2910.0417289733887,
"epoch": 0.05942857142857143,
"grad_norm": 0.1661226898431778,
"kl": 0.0011932700872421265,
"learning_rate": 9.999561358041868e-07,
"loss": 0.0441,
"reward": 0.25355312041938305,
"reward_std": 1.1632588431239128,
"rewards/cosine_scaled_reward": 0.05952752288430929,
"rewards/format_reward": 0.4375000037252903,
"step": 52
},
{
"completion_length": 2785.2500610351562,
"epoch": 0.060571428571428575,
"grad_norm": 0.1787281036376953,
"kl": 0.0006063804030418396,
"learning_rate": 9.999013075636804e-07,
"loss": 0.0319,
"reward": 0.1302148699760437,
"reward_std": 0.8732462916523218,
"rewards/cosine_scaled_reward": -0.028057849034667015,
"rewards/format_reward": 0.4791666716337204,
"step": 53
},
{
"completion_length": 2941.104248046875,
"epoch": 0.061714285714285715,
"grad_norm": 0.21909195184707642,
"kl": 0.00023331446573138237,
"learning_rate": 9.998245517681593e-07,
"loss": 0.1682,
"reward": 0.3165966849774122,
"reward_std": 1.0916050113737583,
"rewards/cosine_scaled_reward": 0.09867796488106251,
"rewards/format_reward": 0.45833334885537624,
"step": 54
},
{
"completion_length": 3090.6458587646484,
"epoch": 0.06285714285714286,
"grad_norm": 0.1303940713405609,
"kl": 0.0007647275924682617,
"learning_rate": 9.997258721585931e-07,
"loss": 0.0367,
"reward": 0.04699781024828553,
"reward_std": 0.8400920890271664,
"rewards/cosine_scaled_reward": -0.015484219416975975,
"rewards/format_reward": 0.33333333767950535,
"step": 55
},
{
"completion_length": 2963.4584045410156,
"epoch": 0.064,
"grad_norm": 0.09380260854959488,
"kl": 0.0018551349639892578,
"learning_rate": 9.996052735444862e-07,
"loss": 0.0177,
"reward": -0.06758141331374645,
"reward_std": 0.6088844947516918,
"rewards/cosine_scaled_reward": -0.0913094412535429,
"rewards/format_reward": 0.3750000074505806,
"step": 56
},
{
"completion_length": 3346.7708435058594,
"epoch": 0.06514285714285714,
"grad_norm": 0.13119955360889435,
"kl": 0.00034799426794052124,
"learning_rate": 9.994627618036452e-07,
"loss": 0.0109,
"reward": -0.14502036944031715,
"reward_std": 0.866327153518796,
"rewards/cosine_scaled_reward": -0.12297584302723408,
"rewards/format_reward": 0.29166666977107525,
"step": 57
},
{
"completion_length": 2402.6458740234375,
"epoch": 0.06628571428571428,
"grad_norm": 0.11010728776454926,
"kl": 0.00420612096786499,
"learning_rate": 9.992983438818915e-07,
"loss": 0.0515,
"reward": 0.10594136267900467,
"reward_std": 0.6864449828863144,
"rewards/cosine_scaled_reward": -0.09731632098555565,
"rewards/format_reward": 0.604166679084301,
"step": 58
},
{
"completion_length": 3156.6458740234375,
"epoch": 0.06742857142857143,
"grad_norm": 0.14981059730052948,
"kl": 0.0015277080237865448,
"learning_rate": 9.991120277927223e-07,
"loss": 0.0349,
"reward": -0.15045391581952572,
"reward_std": 0.6176320239901543,
"rewards/cosine_scaled_reward": -0.09630896709859371,
"rewards/format_reward": 0.2708333432674408,
"step": 59
},
{
"completion_length": 3069.6041870117188,
"epoch": 0.06857142857142857,
"grad_norm": 0.08052746206521988,
"kl": 0.0005385726690292358,
"learning_rate": 9.989038226169207e-07,
"loss": 0.0124,
"reward": -0.26248200982809067,
"reward_std": 0.5521546974778175,
"rewards/cosine_scaled_reward": -0.19996057264506817,
"rewards/format_reward": 0.3333333395421505,
"step": 60
},
{
"completion_length": 3061.250030517578,
"epoch": 0.06971428571428571,
"grad_norm": 0.1755966693162918,
"kl": 0.0010443329811096191,
"learning_rate": 9.98673738502114e-07,
"loss": 0.106,
"reward": 0.027474643662571907,
"reward_std": 0.7631269134581089,
"rewards/cosine_scaled_reward": -0.0676457080990076,
"rewards/format_reward": 0.43750000931322575,
"step": 61
},
{
"completion_length": 2801.791778564453,
"epoch": 0.07085714285714285,
"grad_norm": 0.5502915382385254,
"kl": 0.06185805797576904,
"learning_rate": 9.98421786662277e-07,
"loss": 0.0777,
"reward": 0.16424999572336674,
"reward_std": 1.018867939710617,
"rewards/cosine_scaled_reward": -0.02840923797339201,
"rewards/format_reward": 0.5000000186264515,
"step": 62
},
{
"completion_length": 2495.2708740234375,
"epoch": 0.072,
"grad_norm": 0.19492964446544647,
"kl": 0.0023099184036254883,
"learning_rate": 9.981479793771866e-07,
"loss": 0.1135,
"reward": 0.40039824694395065,
"reward_std": 1.032276712357998,
"rewards/cosine_scaled_reward": 0.06732268328778446,
"rewards/format_reward": 0.6458333507180214,
"step": 63
},
{
"completion_length": 3155.354217529297,
"epoch": 0.07314285714285715,
"grad_norm": 0.13117599487304688,
"kl": 0.0020775794982910156,
"learning_rate": 9.97852329991824e-07,
"loss": 0.0771,
"reward": -0.09214674681425095,
"reward_std": 0.7898851484060287,
"rewards/cosine_scaled_reward": -0.06703834654763341,
"rewards/format_reward": 0.27083333767950535,
"step": 64
},
{
"completion_length": 2716.3542251586914,
"epoch": 0.07428571428571429,
"grad_norm": 0.11173497885465622,
"kl": 0.00382232666015625,
"learning_rate": 9.975348529157229e-07,
"loss": 0.0204,
"reward": -0.11022741347551346,
"reward_std": 0.672077115625143,
"rewards/cosine_scaled_reward": -0.14180615730583668,
"rewards/format_reward": 0.3958333358168602,
"step": 65
},
{
"completion_length": 2310.1041831970215,
"epoch": 0.07542857142857143,
"grad_norm": 0.06708691269159317,
"kl": 0.0025424957275390625,
"learning_rate": 9.971955636222684e-07,
"loss": 0.0339,
"reward": 0.20656662248075008,
"reward_std": 0.589762119576335,
"rewards/cosine_scaled_reward": 0.03981146775186062,
"rewards/format_reward": 0.4791666716337204,
"step": 66
},
{
"completion_length": 3528.8541870117188,
"epoch": 0.07657142857142857,
"grad_norm": 0.07865083962678909,
"kl": 0.0022742748260498047,
"learning_rate": 9.968344786479415e-07,
"loss": 0.0142,
"reward": -0.46846646815538406,
"reward_std": 0.5137450993061066,
"rewards/cosine_scaled_reward": -0.22528725676238537,
"rewards/format_reward": 0.1041666679084301,
"step": 67
},
{
"completion_length": 2388.875030517578,
"epoch": 0.07771428571428571,
"grad_norm": 0.1428786665201187,
"kl": 0.009924888610839844,
"learning_rate": 9.964516155915151e-07,
"loss": 0.0722,
"reward": 0.0902576670050621,
"reward_std": 0.7360437363386154,
"rewards/cosine_scaled_reward": -0.06687035039067268,
"rewards/format_reward": 0.5208333414047956,
"step": 68
},
{
"completion_length": 2723.7708740234375,
"epoch": 0.07885714285714286,
"grad_norm": 0.08696511387825012,
"kl": 0.004919290542602539,
"learning_rate": 9.960469931131936e-07,
"loss": 0.0544,
"reward": -0.24375398270785809,
"reward_std": 0.5107810720801353,
"rewards/cosine_scaled_reward": -0.21045764535665512,
"rewards/format_reward": 0.37500000186264515,
"step": 69
},
{
"completion_length": 3078.8125762939453,
"epoch": 0.08,
"grad_norm": 0.09773669391870499,
"kl": 0.0020369887351989746,
"learning_rate": 9.956206309337066e-07,
"loss": 0.0139,
"reward": -0.0818065321072936,
"reward_std": 0.6509491205215454,
"rewards/cosine_scaled_reward": -0.11709374003112316,
"rewards/format_reward": 0.39583334140479565,
"step": 70
},
{
"completion_length": 2727.979202270508,
"epoch": 0.08114285714285714,
"grad_norm": 0.07420142740011215,
"kl": 0.004831179976463318,
"learning_rate": 9.951725498333448e-07,
"loss": 0.002,
"reward": -0.03378719836473465,
"reward_std": 0.6056636273860931,
"rewards/cosine_scaled_reward": -0.07006076944526285,
"rewards/format_reward": 0.37500000558793545,
"step": 71
},
{
"completion_length": 3157.750030517578,
"epoch": 0.08228571428571428,
"grad_norm": 0.12406568974256516,
"kl": 0.006371498107910156,
"learning_rate": 9.947027716509488e-07,
"loss": 0.0799,
"reward": -0.2358899898827076,
"reward_std": 0.626086350530386,
"rewards/cosine_scaled_reward": -0.14525874331593513,
"rewards/format_reward": 0.25000000186264515,
"step": 72
},
{
"completion_length": 3562.5625,
"epoch": 0.08342857142857144,
"grad_norm": 0.07233923673629761,
"kl": 0.0012898445129394531,
"learning_rate": 9.942113192828444e-07,
"loss": 0.0053,
"reward": -0.47873567789793015,
"reward_std": 0.4625825770199299,
"rewards/cosine_scaled_reward": -0.23070307821035385,
"rewards/format_reward": 0.1041666679084301,
"step": 73
},
{
"completion_length": 3244.3334045410156,
"epoch": 0.08457142857142858,
"grad_norm": 0.1488964557647705,
"kl": 0.003963470458984375,
"learning_rate": 9.93698216681727e-07,
"loss": 0.0581,
"reward": -0.0020135529339313507,
"reward_std": 0.7993322685360909,
"rewards/cosine_scaled_reward": -0.022909073159098625,
"rewards/format_reward": 0.29166666977107525,
"step": 74
},
{
"completion_length": 3028.687545776367,
"epoch": 0.08571428571428572,
"grad_norm": 0.08981792628765106,
"kl": 0.005274057388305664,
"learning_rate": 9.931634888554935e-07,
"loss": 0.0046,
"reward": 0.03718305751681328,
"reward_std": 0.6846093349158764,
"rewards/cosine_scaled_reward": -0.010177075862884521,
"rewards/format_reward": 0.3333333395421505,
"step": 75
},
{
"completion_length": 2771.2083740234375,
"epoch": 0.08685714285714285,
"grad_norm": 0.10527168214321136,
"kl": 0.0019083023071289062,
"learning_rate": 9.926071618660237e-07,
"loss": 0.0698,
"reward": -0.13475606916472316,
"reward_std": 0.6331770308315754,
"rewards/cosine_scaled_reward": -0.19242265075445175,
"rewards/format_reward": 0.4791666753590107,
"step": 76
},
{
"completion_length": 3160.062545776367,
"epoch": 0.088,
"grad_norm": 0.053626783192157745,
"kl": 0.0024300217628479004,
"learning_rate": 9.9202926282791e-07,
"loss": 0.0089,
"reward": -0.07912399154156446,
"reward_std": 0.4063580594956875,
"rewards/cosine_scaled_reward": -0.07036877004429698,
"rewards/format_reward": 0.3333333358168602,
"step": 77
},
{
"completion_length": 3157.354217529297,
"epoch": 0.08914285714285715,
"grad_norm": 0.12693333625793457,
"kl": 0.0019735097885131836,
"learning_rate": 9.91429819907136e-07,
"loss": 0.043,
"reward": 0.10285145416855812,
"reward_std": 0.8885178752243519,
"rewards/cosine_scaled_reward": 0.01143704541027546,
"rewards/format_reward": 0.35416666977107525,
"step": 78
},
{
"completion_length": 2627.145866394043,
"epoch": 0.09028571428571429,
"grad_norm": 0.0993918776512146,
"kl": 0.005728721618652344,
"learning_rate": 9.908088623197048e-07,
"loss": 0.0208,
"reward": 0.005513674899702892,
"reward_std": 0.696781549602747,
"rewards/cosine_scaled_reward": -0.11277296394109726,
"rewards/format_reward": 0.5000000111758709,
"step": 79
},
{
"completion_length": 3378.6666870117188,
"epoch": 0.09142857142857143,
"grad_norm": 0.11993207037448883,
"kl": 0.0034465789794921875,
"learning_rate": 9.901664203302124e-07,
"loss": 0.0148,
"reward": -0.25228459760546684,
"reward_std": 0.6576736122369766,
"rewards/cosine_scaled_reward": -0.15029108710587025,
"rewards/format_reward": 0.2291666753590107,
"step": 80
},
{
"completion_length": 3124.9792098999023,
"epoch": 0.09257142857142857,
"grad_norm": 0.07844506949186325,
"kl": 0.0075626373291015625,
"learning_rate": 9.895025252503755e-07,
"loss": 0.013,
"reward": -0.3477923655882478,
"reward_std": 0.4749002642929554,
"rewards/cosine_scaled_reward": -0.2027296293526888,
"rewards/format_reward": 0.22916666977107525,
"step": 81
},
{
"completion_length": 2722.375045776367,
"epoch": 0.09371428571428571,
"grad_norm": 0.13039354979991913,
"kl": 0.006313920021057129,
"learning_rate": 9.888172094375033e-07,
"loss": -0.0104,
"reward": 0.08498383313417435,
"reward_std": 0.7605466395616531,
"rewards/cosine_scaled_reward": -0.03437880612909794,
"rewards/format_reward": 0.4375000074505806,
"step": 82
},
{
"completion_length": 2761.1250228881836,
"epoch": 0.09485714285714286,
"grad_norm": 0.11148078739643097,
"kl": 0.006354331970214844,
"learning_rate": 9.881105062929221e-07,
"loss": 0.0171,
"reward": -0.1586984060704708,
"reward_std": 0.6908692009747028,
"rewards/cosine_scaled_reward": -0.14450077898800373,
"rewards/format_reward": 0.3333333358168602,
"step": 83
},
{
"completion_length": 3147.3333587646484,
"epoch": 0.096,
"grad_norm": 0.15051236748695374,
"kl": 0.0017566680908203125,
"learning_rate": 9.873824502603459e-07,
"loss": 0.0437,
"reward": 0.20425852667540312,
"reward_std": 0.8553840257227421,
"rewards/cosine_scaled_reward": 0.06748985398371588,
"rewards/format_reward": 0.3958333395421505,
"step": 84
},
{
"completion_length": 3178.7500762939453,
"epoch": 0.09714285714285714,
"grad_norm": 0.12316413968801498,
"kl": 0.002705097198486328,
"learning_rate": 9.866330768241983e-07,
"loss": 0.0373,
"reward": -0.051826220005750656,
"reward_std": 0.8340367153286934,
"rewards/cosine_scaled_reward": -0.0864723757840693,
"rewards/format_reward": 0.35416666977107525,
"step": 85
},
{
"completion_length": 3106.2083740234375,
"epoch": 0.09828571428571428,
"grad_norm": 0.16358214616775513,
"kl": 0.005809783935546875,
"learning_rate": 9.85862422507884e-07,
"loss": 0.0502,
"reward": -0.21086052944883704,
"reward_std": 0.6448673270642757,
"rewards/cosine_scaled_reward": -0.1603346224874258,
"rewards/format_reward": 0.3125000074505806,
"step": 86
},
{
"completion_length": 2988.3334045410156,
"epoch": 0.09942857142857142,
"grad_norm": 0.18081220984458923,
"kl": 0.006814241409301758,
"learning_rate": 9.850705248720068e-07,
"loss": 0.0524,
"reward": 0.13324597105383873,
"reward_std": 0.8470934070646763,
"rewards/cosine_scaled_reward": -0.020617252215743065,
"rewards/format_reward": 0.47916668094694614,
"step": 87
},
{
"completion_length": 2812.625030517578,
"epoch": 0.10057142857142858,
"grad_norm": 0.1538206785917282,
"kl": 0.012554168701171875,
"learning_rate": 9.8425742251254e-07,
"loss": 0.0806,
"reward": 0.1440363209694624,
"reward_std": 0.7827430870383978,
"rewards/cosine_scaled_reward": -0.009355325251817703,
"rewards/format_reward": 0.4791666753590107,
"step": 88
},
{
"completion_length": 3243.562515258789,
"epoch": 0.10171428571428572,
"grad_norm": 0.10299122333526611,
"kl": 0.004688262939453125,
"learning_rate": 9.83423155058946e-07,
"loss": 0.0151,
"reward": -0.2134521808475256,
"reward_std": 0.6248220186680555,
"rewards/cosine_scaled_reward": -0.12200713902711868,
"rewards/format_reward": 0.22916667349636555,
"step": 89
},
{
"completion_length": 2706.8958587646484,
"epoch": 0.10285714285714286,
"grad_norm": 0.08340943604707718,
"kl": 0.009927749633789062,
"learning_rate": 9.825677631722435e-07,
"loss": 0.0087,
"reward": -0.25001570768654346,
"reward_std": 0.5442444123327732,
"rewards/cosine_scaled_reward": -0.24867095332592726,
"rewards/format_reward": 0.43750000558793545,
"step": 90
},
{
"completion_length": 2932.2500762939453,
"epoch": 0.104,
"grad_norm": 0.19951650500297546,
"kl": 0.004284858703613281,
"learning_rate": 9.816912885430258e-07,
"loss": 0.0723,
"reward": 0.07665663212537766,
"reward_std": 0.8412054367363453,
"rewards/cosine_scaled_reward": -0.04515018220990896,
"rewards/format_reward": 0.4375000037252903,
"step": 91
},
{
"completion_length": 2869.812545776367,
"epoch": 0.10514285714285715,
"grad_norm": 0.12959204614162445,
"kl": 0.010074138641357422,
"learning_rate": 9.807937738894303e-07,
"loss": 0.0231,
"reward": 0.038971804082393646,
"reward_std": 0.8026900477707386,
"rewards/cosine_scaled_reward": -0.07580470014363527,
"rewards/format_reward": 0.45833334140479565,
"step": 92
},
{
"completion_length": 3530.5625,
"epoch": 0.10628571428571429,
"grad_norm": 0.09351453185081482,
"kl": 0.005786895751953125,
"learning_rate": 9.798752629550546e-07,
"loss": 0.0137,
"reward": -0.5460330247879028,
"reward_std": 0.4978806171566248,
"rewards/cosine_scaled_reward": -0.25744735077023506,
"rewards/format_reward": 0.06250000186264515,
"step": 93
},
{
"completion_length": 3159.645835876465,
"epoch": 0.10742857142857143,
"grad_norm": 0.09856709837913513,
"kl": 0.009381294250488281,
"learning_rate": 9.78935800506826e-07,
"loss": 0.0331,
"reward": -0.2638606168329716,
"reward_std": 0.6384533829987049,
"rewards/cosine_scaled_reward": -0.1464436650276184,
"rewards/format_reward": 0.2083333358168602,
"step": 94
},
{
"completion_length": 3420.812530517578,
"epoch": 0.10857142857142857,
"grad_norm": 0.0972990095615387,
"kl": 0.0027017593383789062,
"learning_rate": 9.779754323328192e-07,
"loss": 0.0175,
"reward": -0.29529019072651863,
"reward_std": 0.6119959745556116,
"rewards/cosine_scaled_reward": -0.18471521139144897,
"rewards/format_reward": 0.2500000074505806,
"step": 95
},
{
"completion_length": 3003.437515258789,
"epoch": 0.10971428571428571,
"grad_norm": 0.10754227638244629,
"kl": 0.008491039276123047,
"learning_rate": 9.769942052400235e-07,
"loss": 0.0294,
"reward": -0.0011827312409877777,
"reward_std": 0.7199659757316113,
"rewards/cosine_scaled_reward": -0.056293437257409096,
"rewards/format_reward": 0.3750000037252903,
"step": 96
},
{
"completion_length": 3227.7708740234375,
"epoch": 0.11085714285714286,
"grad_norm": 0.055975738912820816,
"kl": 0.005445957183837891,
"learning_rate": 9.759921670520634e-07,
"loss": 0.0114,
"reward": -0.04315417259931564,
"reward_std": 0.39706853218376637,
"rewards/cosine_scaled_reward": -0.03440806642174721,
"rewards/format_reward": 0.31250000186264515,
"step": 97
},
{
"completion_length": 3059.354217529297,
"epoch": 0.112,
"grad_norm": 0.10543321073055267,
"kl": 0.005114555358886719,
"learning_rate": 9.749693666068663e-07,
"loss": 0.0567,
"reward": -0.14790542237460613,
"reward_std": 0.6289958246052265,
"rewards/cosine_scaled_reward": -0.12040873523801565,
"rewards/format_reward": 0.3125000111758709,
"step": 98
},
{
"completion_length": 2944.104217529297,
"epoch": 0.11314285714285714,
"grad_norm": 0.14774614572525024,
"kl": 0.007845878601074219,
"learning_rate": 9.739258537542835e-07,
"loss": 0.0208,
"reward": 0.024691712111234665,
"reward_std": 0.8161002658307552,
"rewards/cosine_scaled_reward": -0.026264889165759087,
"rewards/format_reward": 0.33333334140479565,
"step": 99
},
{
"completion_length": 3003.562515258789,
"epoch": 0.11428571428571428,
"grad_norm": 0.14759190380573273,
"kl": 0.00844573974609375,
"learning_rate": 9.728616793536587e-07,
"loss": 0.0498,
"reward": -0.03464020788669586,
"reward_std": 0.7968008350580931,
"rewards/cosine_scaled_reward": -0.07347729802131653,
"rewards/format_reward": 0.35416667349636555,
"step": 100
},
{
"completion_length": 2660.7500228881836,
"epoch": 0.11542857142857142,
"grad_norm": 0.05885668471455574,
"kl": 0.004193544387817383,
"learning_rate": 9.717768952713511e-07,
"loss": 0.0099,
"reward": -0.024280589073896408,
"reward_std": 0.478687334805727,
"rewards/cosine_scaled_reward": -0.06905538472346961,
"rewards/format_reward": 0.39583333395421505,
"step": 101
},
{
"completion_length": 3017.6459045410156,
"epoch": 0.11657142857142858,
"grad_norm": 0.19236327707767487,
"kl": 0.01230621337890625,
"learning_rate": 9.706715543782064e-07,
"loss": 0.1031,
"reward": 0.0067907206248492,
"reward_std": 0.7886966746300459,
"rewards/cosine_scaled_reward": -0.07641204819083214,
"rewards/format_reward": 0.41666667722165585,
"step": 102
},
{
"completion_length": 3141.437545776367,
"epoch": 0.11771428571428572,
"grad_norm": 0.12005823105573654,
"kl": 0.010993003845214844,
"learning_rate": 9.695457105469804e-07,
"loss": 0.0418,
"reward": 0.11525936797261238,
"reward_std": 0.7872378453612328,
"rewards/cosine_scaled_reward": -0.020046040415763855,
"rewards/format_reward": 0.4583333507180214,
"step": 103
},
{
"completion_length": 2757.520851135254,
"epoch": 0.11885714285714286,
"grad_norm": 0.16546285152435303,
"kl": 0.007067680358886719,
"learning_rate": 9.683994186497132e-07,
"loss": 0.029,
"reward": -0.012922056019306183,
"reward_std": 0.7183502614498138,
"rewards/cosine_scaled_reward": -0.08642422500997782,
"rewards/format_reward": 0.41666667349636555,
"step": 104
},
{
"completion_length": 2874.68758392334,
"epoch": 0.12,
"grad_norm": 0.15343305468559265,
"kl": 0.0062236785888671875,
"learning_rate": 9.672327345550543e-07,
"loss": -0.0011,
"reward": 0.3746343031525612,
"reward_std": 0.9715399444103241,
"rewards/cosine_scaled_reward": 0.13150912104174495,
"rewards/format_reward": 0.47916668094694614,
"step": 105
},
{
"completion_length": 2406.2292404174805,
"epoch": 0.12114285714285715,
"grad_norm": 0.11538145691156387,
"kl": 0.014267921447753906,
"learning_rate": 9.66045715125541e-07,
"loss": 0.0512,
"reward": 0.7112023187801242,
"reward_std": 0.7698421813547611,
"rewards/cosine_scaled_reward": 0.2695285137742758,
"rewards/format_reward": 0.6875000111758709,
"step": 106
},
{
"completion_length": 2906.083366394043,
"epoch": 0.12228571428571429,
"grad_norm": 0.1356947422027588,
"kl": 0.0071544647216796875,
"learning_rate": 9.648384182148252e-07,
"loss": 0.0559,
"reward": 0.04975247010588646,
"reward_std": 0.6387073248624802,
"rewards/cosine_scaled_reward": -0.04459162801504135,
"rewards/format_reward": 0.43750001303851604,
"step": 107
},
{
"completion_length": 2924.479202270508,
"epoch": 0.12342857142857143,
"grad_norm": 1.227287769317627,
"kl": 0.09261322021484375,
"learning_rate": 9.636109026648554e-07,
"loss": 0.0411,
"reward": -0.06936776265501976,
"reward_std": 0.6040569245815277,
"rewards/cosine_scaled_reward": -0.09396206960082054,
"rewards/format_reward": 0.3750000037252903,
"step": 108
},
{
"completion_length": 3159.4166870117188,
"epoch": 0.12457142857142857,
"grad_norm": 0.06464666873216629,
"kl": 0.0052165985107421875,
"learning_rate": 9.623632283030077e-07,
"loss": -0.0001,
"reward": -0.19677976984530687,
"reward_std": 0.3907594494521618,
"rewards/cosine_scaled_reward": -0.13640070147812366,
"rewards/format_reward": 0.3125,
"step": 109
},
{
"completion_length": 2941.979217529297,
"epoch": 0.12571428571428572,
"grad_norm": 0.11352120339870453,
"kl": 0.0075016021728515625,
"learning_rate": 9.610954559391704e-07,
"loss": 0.0559,
"reward": -0.02889834251254797,
"reward_std": 0.6912853047251701,
"rewards/cosine_scaled_reward": -0.09470331901684403,
"rewards/format_reward": 0.41666667349636555,
"step": 110
},
{
"completion_length": 3445.0625610351562,
"epoch": 0.12685714285714286,
"grad_norm": 0.1718732863664627,
"kl": 0.0095977783203125,
"learning_rate": 9.598076473627796e-07,
"loss": 0.0525,
"reward": -0.1273586554452777,
"reward_std": 0.7887241318821907,
"rewards/cosine_scaled_reward": -0.042100945487618446,
"rewards/format_reward": 0.1666666716337204,
"step": 111
},
{
"completion_length": 3372.2083740234375,
"epoch": 0.128,
"grad_norm": 0.13645856082439423,
"kl": 0.00563812255859375,
"learning_rate": 9.58499865339809e-07,
"loss": 0.0184,
"reward": -0.056365881115198135,
"reward_std": 0.7170692086219788,
"rewards/cosine_scaled_reward": -0.048214955255389214,
"rewards/format_reward": 0.2916666753590107,
"step": 112
},
{
"completion_length": 2884.791717529297,
"epoch": 0.12914285714285714,
"grad_norm": 0.20523111522197723,
"kl": 0.010105133056640625,
"learning_rate": 9.571721736097088e-07,
"loss": -0.0033,
"reward": 0.13258023280650377,
"reward_std": 0.8612403385341167,
"rewards/cosine_scaled_reward": -0.058843023143708706,
"rewards/format_reward": 0.5416666809469461,
"step": 113
},
{
"completion_length": 2625.458396911621,
"epoch": 0.13028571428571428,
"grad_norm": 0.08083964139223099,
"kl": 0.0072994232177734375,
"learning_rate": 9.55824636882301e-07,
"loss": 0.0276,
"reward": -0.11776229925453663,
"reward_std": 0.5373759977519512,
"rewards/cosine_scaled_reward": -0.2312552430666983,
"rewards/format_reward": 0.5833333414047956,
"step": 114
},
{
"completion_length": 3025.1041717529297,
"epoch": 0.13142857142857142,
"grad_norm": 0.06019025295972824,
"kl": 0.0069713592529296875,
"learning_rate": 9.54457320834625e-07,
"loss": 0.0315,
"reward": -0.2688702419400215,
"reward_std": 0.44118795450776815,
"rewards/cosine_scaled_reward": -0.14728060085326433,
"rewards/format_reward": 0.2291666716337204,
"step": 115
},
{
"completion_length": 3401.4166870117188,
"epoch": 0.13257142857142856,
"grad_norm": 0.1261359304189682,
"kl": 0.0067615509033203125,
"learning_rate": 9.530702921077358e-07,
"loss": 0.0071,
"reward": -0.318508867174387,
"reward_std": 0.602899955585599,
"rewards/cosine_scaled_reward": -0.16228950582444668,
"rewards/format_reward": 0.1666666716337204,
"step": 116
},
{
"completion_length": 3060.7916717529297,
"epoch": 0.1337142857142857,
"grad_norm": 0.0866546705365181,
"kl": 0.008612632751464844,
"learning_rate": 9.516636183034564e-07,
"loss": 0.0005,
"reward": -0.3146707344567403,
"reward_std": 0.5091048590838909,
"rewards/cosine_scaled_reward": -0.2149065202102065,
"rewards/format_reward": 0.29166666977107525,
"step": 117
},
{
"completion_length": 3120.0833435058594,
"epoch": 0.13485714285714287,
"grad_norm": 0.18219080567359924,
"kl": 0.00536346435546875,
"learning_rate": 9.502373679810839e-07,
"loss": 0.0522,
"reward": 0.4105420224368572,
"reward_std": 1.0126653835177422,
"rewards/cosine_scaled_reward": 0.17556388210505247,
"rewards/format_reward": 0.43750000931322575,
"step": 118
},
{
"completion_length": 2533.9167251586914,
"epoch": 0.136,
"grad_norm": 0.1974111646413803,
"kl": 0.054787635803222656,
"learning_rate": 9.487916106540465e-07,
"loss": 0.0693,
"reward": 0.0521804504096508,
"reward_std": 0.7259144820272923,
"rewards/cosine_scaled_reward": -0.07517153583467007,
"rewards/format_reward": 0.4791666679084301,
"step": 119
},
{
"completion_length": 2534.4167137145996,
"epoch": 0.13714285714285715,
"grad_norm": 0.11178319901227951,
"kl": 0.009916305541992188,
"learning_rate": 9.473264167865171e-07,
"loss": 0.0475,
"reward": 0.0007900348864495754,
"reward_std": 0.6627710647881031,
"rewards/cosine_scaled_reward": -0.09777611820027232,
"rewards/format_reward": 0.45833333767950535,
"step": 120
},
{
"completion_length": 2313.229248046875,
"epoch": 0.1382857142857143,
"grad_norm": 0.16405229270458221,
"kl": 0.013139724731445312,
"learning_rate": 9.458418577899774e-07,
"loss": 0.0402,
"reward": 0.3239856115542352,
"reward_std": 0.7424066159874201,
"rewards/cosine_scaled_reward": -0.005178395658731461,
"rewards/format_reward": 0.7083333469927311,
"step": 121
},
{
"completion_length": 2761.8958587646484,
"epoch": 0.13942857142857143,
"grad_norm": 0.12051185965538025,
"kl": 0.0068817138671875,
"learning_rate": 9.443380060197385e-07,
"loss": 0.026,
"reward": 0.3165215402841568,
"reward_std": 0.7121441401541233,
"rewards/cosine_scaled_reward": 0.07533489167690277,
"rewards/format_reward": 0.5416666697710752,
"step": 122
},
{
"completion_length": 3158.104217529297,
"epoch": 0.14057142857142857,
"grad_norm": 0.1131911501288414,
"kl": 0.007569313049316406,
"learning_rate": 9.428149347714143e-07,
"loss": 0.0551,
"reward": -0.16298508271574974,
"reward_std": 0.7209546975791454,
"rewards/cosine_scaled_reward": -0.13522059097886086,
"rewards/format_reward": 0.31250000558793545,
"step": 123
},
{
"completion_length": 2517.312530517578,
"epoch": 0.1417142857142857,
"grad_norm": 0.12443465739488602,
"kl": 0.009019851684570312,
"learning_rate": 9.412727182773486e-07,
"loss": -0.0164,
"reward": 0.19143125228583813,
"reward_std": 0.8426807429641485,
"rewards/cosine_scaled_reward": -0.0331303218845278,
"rewards/format_reward": 0.5625000055879354,
"step": 124
},
{
"completion_length": 2815.5833740234375,
"epoch": 0.14285714285714285,
"grad_norm": 0.06811921298503876,
"kl": 0.0053272247314453125,
"learning_rate": 9.397114317029974e-07,
"loss": 0.0266,
"reward": -0.04552587552461773,
"reward_std": 0.594413885846734,
"rewards/cosine_scaled_reward": -0.050011674873530865,
"rewards/format_reward": 0.31250000186264515,
"step": 125
},
{
"completion_length": 2926.750045776367,
"epoch": 0.144,
"grad_norm": 0.11453459411859512,
"kl": 0.005611419677734375,
"learning_rate": 9.381311511432658e-07,
"loss": 0.0231,
"reward": -0.027173910290002823,
"reward_std": 0.6515960693359375,
"rewards/cosine_scaled_reward": -0.11056716740131378,
"rewards/format_reward": 0.45833333395421505,
"step": 126
},
{
"completion_length": 3180.0416870117188,
"epoch": 0.14514285714285713,
"grad_norm": 0.13321727514266968,
"kl": 0.008800506591796875,
"learning_rate": 9.36531953618799e-07,
"loss": 0.0247,
"reward": -0.2067241296172142,
"reward_std": 0.7376469634473324,
"rewards/cosine_scaled_reward": -0.1875669350847602,
"rewards/format_reward": 0.35416667722165585,
"step": 127
},
{
"completion_length": 2854.145866394043,
"epoch": 0.1462857142857143,
"grad_norm": 0.13030071556568146,
"kl": 0.008055686950683594,
"learning_rate": 9.34913917072228e-07,
"loss": 0.0297,
"reward": 0.28113045543432236,
"reward_std": 0.7527024820446968,
"rewards/cosine_scaled_reward": 0.08081426518037915,
"rewards/format_reward": 0.47916666977107525,
"step": 128
},
{
"completion_length": 3468.4583435058594,
"epoch": 0.14742857142857144,
"grad_norm": 0.09160865843296051,
"kl": 0.009267807006835938,
"learning_rate": 9.332771203643714e-07,
"loss": 0.0182,
"reward": -0.3858581744134426,
"reward_std": 0.4741293340921402,
"rewards/cosine_scaled_reward": -0.17545855604112148,
"rewards/format_reward": 0.12500000186264515,
"step": 129
},
{
"completion_length": 3145.2083435058594,
"epoch": 0.14857142857142858,
"grad_norm": 0.11552103608846664,
"kl": 0.008371353149414062,
"learning_rate": 9.316216432703916e-07,
"loss": 0.0555,
"reward": -0.1405576877295971,
"reward_std": 0.6180392988026142,
"rewards/cosine_scaled_reward": -0.08423710052738898,
"rewards/format_reward": 0.2500000074505806,
"step": 130
},
{
"completion_length": 2990.750011444092,
"epoch": 0.14971428571428572,
"grad_norm": 0.1343425065279007,
"kl": 0.010997772216796875,
"learning_rate": 9.299475664759068e-07,
"loss": 0.0573,
"reward": -0.0012444127351045609,
"reward_std": 0.7613321915268898,
"rewards/cosine_scaled_reward": -0.0183742493391037,
"rewards/format_reward": 0.29166666977107525,
"step": 131
},
{
"completion_length": 2849.229217529297,
"epoch": 0.15085714285714286,
"grad_norm": 0.1754404902458191,
"kl": 0.007671356201171875,
"learning_rate": 9.282549715730579e-07,
"loss": 0.0039,
"reward": 0.15461664367467165,
"reward_std": 1.0978601425886154,
"rewards/cosine_scaled_reward": -0.016693929443135858,
"rewards/format_reward": 0.4583333395421505,
"step": 132
},
{
"completion_length": 3204.5208740234375,
"epoch": 0.152,
"grad_norm": 0.09437424689531326,
"kl": 0.010120391845703125,
"learning_rate": 9.265439410565328e-07,
"loss": 0.0371,
"reward": -0.3896838743239641,
"reward_std": 0.46856095641851425,
"rewards/cosine_scaled_reward": -0.23207763396203518,
"rewards/format_reward": 0.22916666977107525,
"step": 133
},
{
"completion_length": 2440.8125534057617,
"epoch": 0.15314285714285714,
"grad_norm": 0.11227884888648987,
"kl": 0.010713577270507812,
"learning_rate": 9.248145583195447e-07,
"loss": 0.0059,
"reward": 0.03150389529764652,
"reward_std": 0.6699380800127983,
"rewards/cosine_scaled_reward": -0.10194769129157066,
"rewards/format_reward": 0.5208333358168602,
"step": 134
},
{
"completion_length": 2048.7083740234375,
"epoch": 0.15428571428571428,
"grad_norm": 0.11956455558538437,
"kl": 0.009142875671386719,
"learning_rate": 9.230669076497687e-07,
"loss": 0.0417,
"reward": 0.611090637743473,
"reward_std": 0.8544140718877316,
"rewards/cosine_scaled_reward": 0.19521427806466818,
"rewards/format_reward": 0.6875,
"step": 135
},
{
"completion_length": 3031.7708587646484,
"epoch": 0.15542857142857142,
"grad_norm": 0.16711600124835968,
"kl": 0.012142181396484375,
"learning_rate": 9.213010742252327e-07,
"loss": 0.0697,
"reward": 0.3044010065495968,
"reward_std": 0.8896326720714569,
"rewards/cosine_scaled_reward": 0.08019791916012764,
"rewards/format_reward": 0.5000000093132257,
"step": 136
},
{
"completion_length": 3144.291717529297,
"epoch": 0.15657142857142858,
"grad_norm": 0.1869216412305832,
"kl": 0.01092529296875,
"learning_rate": 9.195171441101668e-07,
"loss": 0.0486,
"reward": -0.18480181868653744,
"reward_std": 0.7662193942815065,
"rewards/cosine_scaled_reward": -0.14503308382700197,
"rewards/format_reward": 0.2916666753590107,
"step": 137
},
{
"completion_length": 2823.6667251586914,
"epoch": 0.15771428571428572,
"grad_norm": 0.1232113316655159,
"kl": 0.010227203369140625,
"learning_rate": 9.177152042508077e-07,
"loss": 0.0399,
"reward": -0.14219614176545292,
"reward_std": 0.6205537635833025,
"rewards/cosine_scaled_reward": -0.1589539386332035,
"rewards/format_reward": 0.3958333432674408,
"step": 138
},
{
"completion_length": 3272.187530517578,
"epoch": 0.15885714285714286,
"grad_norm": 0.15304112434387207,
"kl": 0.0121917724609375,
"learning_rate": 9.158953424711624e-07,
"loss": 0.0143,
"reward": -0.028373660519719124,
"reward_std": 0.7605781219899654,
"rewards/cosine_scaled_reward": -0.07536444254219532,
"rewards/format_reward": 0.37500000558793545,
"step": 139
},
{
"completion_length": 3336.541717529297,
"epoch": 0.16,
"grad_norm": 0.1725020408630371,
"kl": 0.018383026123046875,
"learning_rate": 9.140576474687263e-07,
"loss": 0.0185,
"reward": -0.09492362570017576,
"reward_std": 0.739283999428153,
"rewards/cosine_scaled_reward": -0.052022709511220455,
"rewards/format_reward": 0.2291666753590107,
"step": 140
},
{
"completion_length": 2818.7500915527344,
"epoch": 0.16114285714285714,
"grad_norm": 0.1428345888853073,
"kl": 0.014141082763671875,
"learning_rate": 9.122022088101613e-07,
"loss": 0.0428,
"reward": 0.17010034061968327,
"reward_std": 0.8265232257544994,
"rewards/cosine_scaled_reward": -0.03258664230816066,
"rewards/format_reward": 0.5416666734963655,
"step": 141
},
{
"completion_length": 2865.041717529297,
"epoch": 0.16228571428571428,
"grad_norm": 0.132287859916687,
"kl": 0.012800216674804688,
"learning_rate": 9.103291169269299e-07,
"loss": -0.0266,
"reward": 0.14131132513284683,
"reward_std": 0.6835053861141205,
"rewards/cosine_scaled_reward": -0.061369335278868675,
"rewards/format_reward": 0.5833333414047956,
"step": 142
},
{
"completion_length": 2789.979217529297,
"epoch": 0.16342857142857142,
"grad_norm": 0.2371223270893097,
"kl": 0.01451873779296875,
"learning_rate": 9.084384631108882e-07,
"loss": 0.0726,
"reward": -0.21053062099963427,
"reward_std": 0.6688540205359459,
"rewards/cosine_scaled_reward": -0.22842608066275716,
"rewards/format_reward": 0.4375000111758709,
"step": 143
},
{
"completion_length": 3133.833366394043,
"epoch": 0.16457142857142856,
"grad_norm": 0.14418281614780426,
"kl": 0.0147552490234375,
"learning_rate": 9.065303395098358e-07,
"loss": 0.0211,
"reward": -0.19076000433415174,
"reward_std": 0.8512778505682945,
"rewards/cosine_scaled_reward": -0.1536552112083882,
"rewards/format_reward": 0.29166666977107525,
"step": 144
},
{
"completion_length": 2111.166679382324,
"epoch": 0.1657142857142857,
"grad_norm": 0.08913204073905945,
"kl": 0.011333465576171875,
"learning_rate": 9.046048391230247e-07,
"loss": 0.0067,
"reward": 0.37262871488928795,
"reward_std": 0.6370398961007595,
"rewards/cosine_scaled_reward": 0.05427891947329044,
"rewards/format_reward": 0.6666666679084301,
"step": 145
},
{
"completion_length": 2267.708351135254,
"epoch": 0.16685714285714287,
"grad_norm": 0.09961648285388947,
"kl": 0.008184432983398438,
"learning_rate": 9.026620557966279e-07,
"loss": 0.0193,
"reward": 0.021780904848128557,
"reward_std": 0.629846852272749,
"rewards/cosine_scaled_reward": -0.18135410267859697,
"rewards/format_reward": 0.666666679084301,
"step": 146
},
{
"completion_length": 2814.354202270508,
"epoch": 0.168,
"grad_norm": 0.1152271032333374,
"kl": 0.015193939208984375,
"learning_rate": 9.007020842191634e-07,
"loss": -0.0113,
"reward": 0.11335810273885727,
"reward_std": 0.75190694257617,
"rewards/cosine_scaled_reward": -0.01303301053121686,
"rewards/format_reward": 0.4375000149011612,
"step": 147
},
{
"completion_length": 2681.6458892822266,
"epoch": 0.16914285714285715,
"grad_norm": 0.10002895444631577,
"kl": 0.0156402587890625,
"learning_rate": 8.987250199168808e-07,
"loss": 0.0397,
"reward": 0.015451362356543541,
"reward_std": 0.6257824674248695,
"rewards/cosine_scaled_reward": -0.08100172178819776,
"rewards/format_reward": 0.4583333395421505,
"step": 148
},
{
"completion_length": 2944.9375610351562,
"epoch": 0.1702857142857143,
"grad_norm": 0.11802194267511368,
"kl": 0.012241363525390625,
"learning_rate": 8.967309592491052e-07,
"loss": 0.0548,
"reward": -0.029212953057140112,
"reward_std": 0.5978215262293816,
"rewards/cosine_scaled_reward": -0.11697033792734146,
"rewards/format_reward": 0.47916668094694614,
"step": 149
},
{
"completion_length": 3025.500045776367,
"epoch": 0.17142857142857143,
"grad_norm": 0.158779576420784,
"kl": 0.016357421875,
"learning_rate": 8.9471999940354e-07,
"loss": 0.075,
"reward": 0.17186034470796585,
"reward_std": 0.9030736871063709,
"rewards/cosine_scaled_reward": 0.03854179289191961,
"rewards/format_reward": 0.39583333395421505,
"step": 150
},
{
"completion_length": 2807.2709045410156,
"epoch": 0.17257142857142857,
"grad_norm": 0.2797929644584656,
"kl": 0.016078948974609375,
"learning_rate": 8.926922383915315e-07,
"loss": 0.1148,
"reward": 0.35514865489676595,
"reward_std": 1.013380728662014,
"rewards/cosine_scaled_reward": 0.09170010522939265,
"rewards/format_reward": 0.520833345130086,
"step": 151
},
{
"completion_length": 2813.312511444092,
"epoch": 0.1737142857142857,
"grad_norm": 0.15241533517837524,
"kl": 0.016437530517578125,
"learning_rate": 8.906477750432903e-07,
"loss": 0.0504,
"reward": -0.21058875182643533,
"reward_std": 0.6285759322345257,
"rewards/cosine_scaled_reward": -0.17488694563508034,
"rewards/format_reward": 0.33333333395421505,
"step": 152
},
{
"completion_length": 2995.937515258789,
"epoch": 0.17485714285714285,
"grad_norm": 0.09293405711650848,
"kl": 0.022808074951171875,
"learning_rate": 8.88586709003076e-07,
"loss": 0.0405,
"reward": -0.2887796126306057,
"reward_std": 0.4931412860751152,
"rewards/cosine_scaled_reward": -0.21798867918550968,
"rewards/format_reward": 0.3333333395421505,
"step": 153
},
{
"completion_length": 3390.1458435058594,
"epoch": 0.176,
"grad_norm": 0.17047946155071259,
"kl": 0.01389312744140625,
"learning_rate": 8.865091407243394e-07,
"loss": 0.0042,
"reward": 0.22278353199362755,
"reward_std": 0.9290903359651566,
"rewards/cosine_scaled_reward": 0.07489011948928237,
"rewards/format_reward": 0.3958333432674408,
"step": 154
},
{
"completion_length": 2628.625030517578,
"epoch": 0.17714285714285713,
"grad_norm": 0.16334204375743866,
"kl": 0.01788330078125,
"learning_rate": 8.844151714648274e-07,
"loss": 0.0719,
"reward": 0.07151136547327042,
"reward_std": 0.850743044167757,
"rewards/cosine_scaled_reward": -0.05793104809708893,
"rewards/format_reward": 0.4583333395421505,
"step": 155
},
{
"completion_length": 3115.0208740234375,
"epoch": 0.1782857142857143,
"grad_norm": 0.1153322234749794,
"kl": 0.015087127685546875,
"learning_rate": 8.823049032816478e-07,
"loss": 0.0422,
"reward": -0.11803282611072063,
"reward_std": 0.6224103905260563,
"rewards/cosine_scaled_reward": -0.07339339889585972,
"rewards/format_reward": 0.2708333432674408,
"step": 156
},
{
"completion_length": 3112.3333740234375,
"epoch": 0.17942857142857144,
"grad_norm": 0.11238392442464828,
"kl": 0.018207550048828125,
"learning_rate": 8.801784390262943e-07,
"loss": 0.0267,
"reward": -0.12678863108158112,
"reward_std": 0.6108014769852161,
"rewards/cosine_scaled_reward": -0.13557185977697372,
"rewards/format_reward": 0.3750000149011612,
"step": 157
},
{
"completion_length": 3134.9375610351562,
"epoch": 0.18057142857142858,
"grad_norm": 0.14315029978752136,
"kl": 0.016357421875,
"learning_rate": 8.780358823396352e-07,
"loss": 0.0291,
"reward": 0.31565938144922256,
"reward_std": 0.7341890074312687,
"rewards/cosine_scaled_reward": 0.12421234138309956,
"rewards/format_reward": 0.43750000931322575,
"step": 158
},
{
"completion_length": 3197.2916870117188,
"epoch": 0.18171428571428572,
"grad_norm": 0.08521895110607147,
"kl": 0.02240753173828125,
"learning_rate": 8.758773376468604e-07,
"loss": -0.0065,
"reward": -0.2289612852036953,
"reward_std": 0.45951012521982193,
"rewards/cosine_scaled_reward": -0.15268473327159882,
"rewards/format_reward": 0.29166666977107525,
"step": 159
},
{
"completion_length": 2835.1250076293945,
"epoch": 0.18285714285714286,
"grad_norm": 0.1391443908214569,
"kl": 0.0212249755859375,
"learning_rate": 8.737029101523929e-07,
"loss": 0.0389,
"reward": -0.09338995814323425,
"reward_std": 0.6676197461783886,
"rewards/cosine_scaled_reward": -0.08398129511624575,
"rewards/format_reward": 0.3125000074505806,
"step": 160
},
{
"completion_length": 2894.2500762939453,
"epoch": 0.184,
"grad_norm": 0.15109732747077942,
"kl": 0.022735595703125,
"learning_rate": 8.715127058347614e-07,
"loss": 0.0618,
"reward": 0.18475250899791718,
"reward_std": 0.7132500857114792,
"rewards/cosine_scaled_reward": 0.038985077291727066,
"rewards/format_reward": 0.4375000074505806,
"step": 161
},
{
"completion_length": 3240.104217529297,
"epoch": 0.18514285714285714,
"grad_norm": 0.25355350971221924,
"kl": 0.02893829345703125,
"learning_rate": 8.693068314414344e-07,
"loss": 0.03,
"reward": -0.08004653453826904,
"reward_std": 0.7777004204690456,
"rewards/cosine_scaled_reward": -0.050195490941405296,
"rewards/format_reward": 0.2500000074505806,
"step": 162
},
{
"completion_length": 2491.4166870117188,
"epoch": 0.18628571428571428,
"grad_norm": 0.10477182269096375,
"kl": 0.019779205322265625,
"learning_rate": 8.670853944836176e-07,
"loss": 0.0184,
"reward": 0.30021179956384003,
"reward_std": 0.5858964845538139,
"rewards/cosine_scaled_reward": 0.05349706672132015,
"rewards/format_reward": 0.5833333432674408,
"step": 163
},
{
"completion_length": 2634.3334045410156,
"epoch": 0.18742857142857142,
"grad_norm": 0.10012540221214294,
"kl": 0.02051544189453125,
"learning_rate": 8.648485032310144e-07,
"loss": 0.0372,
"reward": 0.18599897995591164,
"reward_std": 0.5746248178184032,
"rewards/cosine_scaled_reward": 0.004582956433296204,
"rewards/format_reward": 0.520833333954215,
"step": 164
},
{
"completion_length": 3389.854217529297,
"epoch": 0.18857142857142858,
"grad_norm": 0.17076270282268524,
"kl": 0.0308380126953125,
"learning_rate": 8.625962667065487e-07,
"loss": 0.0336,
"reward": -0.19457256980240345,
"reward_std": 0.8124973215162754,
"rewards/cosine_scaled_reward": -0.14226967841386795,
"rewards/format_reward": 0.2708333395421505,
"step": 165
},
{
"completion_length": 3060.5416870117188,
"epoch": 0.18971428571428572,
"grad_norm": 0.1075100377202034,
"kl": 0.01813507080078125,
"learning_rate": 8.603287946810513e-07,
"loss": 0.0452,
"reward": -0.002260749228298664,
"reward_std": 0.5852988548576832,
"rewards/cosine_scaled_reward": -0.04963597096502781,
"rewards/format_reward": 0.37500000931322575,
"step": 166
},
{
"completion_length": 2668.5625610351562,
"epoch": 0.19085714285714286,
"grad_norm": 0.17578865587711334,
"kl": 0.0214080810546875,
"learning_rate": 8.580461976679099e-07,
"loss": 0.0434,
"reward": -0.03433691617101431,
"reward_std": 0.7491964735090733,
"rewards/cosine_scaled_reward": -0.18043148797005415,
"rewards/format_reward": 0.5625000055879354,
"step": 167
},
{
"completion_length": 3086.2083892822266,
"epoch": 0.192,
"grad_norm": 0.1612425446510315,
"kl": 0.02228546142578125,
"learning_rate": 8.557485869176825e-07,
"loss": 0.0382,
"reward": 0.3381266240030527,
"reward_std": 0.7707905732095242,
"rewards/cosine_scaled_reward": 0.10119387321174145,
"rewards/format_reward": 0.5208333469927311,
"step": 168
},
{
"completion_length": 2517.1041946411133,
"epoch": 0.19314285714285714,
"grad_norm": 0.16143831610679626,
"kl": 0.025909423828125,
"learning_rate": 8.534360744126753e-07,
"loss": 0.0434,
"reward": 0.7395636513829231,
"reward_std": 0.8513101674616337,
"rewards/cosine_scaled_reward": 0.3049415610730648,
"rewards/format_reward": 0.6458333358168602,
"step": 169
},
{
"completion_length": 2514.916702270508,
"epoch": 0.19428571428571428,
"grad_norm": 0.10102265328168869,
"kl": 0.0198974609375,
"learning_rate": 8.511087728614862e-07,
"loss": 0.0574,
"reward": 0.13654466345906258,
"reward_std": 0.6086900364607573,
"rewards/cosine_scaled_reward": 0.003594242036342621,
"rewards/format_reward": 0.45833334140479565,
"step": 170
},
{
"completion_length": 2985.0208740234375,
"epoch": 0.19542857142857142,
"grad_norm": 0.13056401908397675,
"kl": 0.0213165283203125,
"learning_rate": 8.487667956935087e-07,
"loss": 0.0198,
"reward": 0.062067726626992226,
"reward_std": 0.7003943808376789,
"rewards/cosine_scaled_reward": -0.022239719983190298,
"rewards/format_reward": 0.3958333432674408,
"step": 171
},
{
"completion_length": 3023.25,
"epoch": 0.19657142857142856,
"grad_norm": 0.13006411492824554,
"kl": 0.027801513671875,
"learning_rate": 8.464102570534061e-07,
"loss": 0.0166,
"reward": -0.10689089074730873,
"reward_std": 0.7168557345867157,
"rewards/cosine_scaled_reward": -0.0742192417383194,
"rewards/format_reward": 0.2708333395421505,
"step": 172
},
{
"completion_length": 2487.312572479248,
"epoch": 0.1977142857142857,
"grad_norm": 0.15723471343517303,
"kl": 0.029430389404296875,
"learning_rate": 8.440392717955475e-07,
"loss": 0.0449,
"reward": 0.07556262612342834,
"reward_std": 0.7289031557738781,
"rewards/cosine_scaled_reward": -0.11148698627948761,
"rewards/format_reward": 0.5833333395421505,
"step": 173
},
{
"completion_length": 2791.6459045410156,
"epoch": 0.19885714285714284,
"grad_norm": 0.12100586295127869,
"kl": 0.03168487548828125,
"learning_rate": 8.416539554784089e-07,
"loss": 0.0243,
"reward": 0.15438630525022745,
"reward_std": 0.6935141123831272,
"rewards/cosine_scaled_reward": -0.043977076187729836,
"rewards/format_reward": 0.5625000111758709,
"step": 174
},
{
"completion_length": 2961.2083892822266,
"epoch": 0.2,
"grad_norm": 0.10910208523273468,
"kl": 0.027740478515625,
"learning_rate": 8.392544243589427e-07,
"loss": 0.0207,
"reward": 0.0747014251537621,
"reward_std": 0.6700296718627214,
"rewards/cosine_scaled_reward": -0.031223440542817116,
"rewards/format_reward": 0.43750001303851604,
"step": 175
},
{
"completion_length": 2736.104217529297,
"epoch": 0.20114285714285715,
"grad_norm": 0.22644685208797455,
"kl": 0.02967071533203125,
"learning_rate": 8.368407953869103e-07,
"loss": 0.0671,
"reward": 0.0621003326959908,
"reward_std": 0.8631531074643135,
"rewards/cosine_scaled_reward": -0.06792039083666168,
"rewards/format_reward": 0.45833333767950535,
"step": 176
},
{
"completion_length": 3160.0000610351562,
"epoch": 0.2022857142857143,
"grad_norm": 0.23566588759422302,
"kl": 0.03285980224609375,
"learning_rate": 8.344131861991828e-07,
"loss": 0.0637,
"reward": 0.023158524534665048,
"reward_std": 0.7583435364067554,
"rewards/cosine_scaled_reward": -0.10049578547477722,
"rewards/format_reward": 0.5000000186264515,
"step": 177
},
{
"completion_length": 2873.854232788086,
"epoch": 0.20342857142857143,
"grad_norm": 0.19544903934001923,
"kl": 0.0418548583984375,
"learning_rate": 8.319717151140072e-07,
"loss": 0.0085,
"reward": 0.14865556359291077,
"reward_std": 0.8756407424807549,
"rewards/cosine_scaled_reward": 0.016698965802788734,
"rewards/format_reward": 0.416666679084301,
"step": 178
},
{
"completion_length": 3137.1666870117188,
"epoch": 0.20457142857142857,
"grad_norm": 0.07909320294857025,
"kl": 0.03249359130859375,
"learning_rate": 8.295165011252396e-07,
"loss": 0.0049,
"reward": -0.36861060559749603,
"reward_std": 0.3870035018771887,
"rewards/cosine_scaled_reward": -0.24457938224077225,
"rewards/format_reward": 0.2916666679084301,
"step": 179
},
{
"completion_length": 2372.270866394043,
"epoch": 0.2057142857142857,
"grad_norm": 0.14237825572490692,
"kl": 0.03276824951171875,
"learning_rate": 8.270476638965461e-07,
"loss": 0.0399,
"reward": 0.31979808397591114,
"reward_std": 0.7863202355802059,
"rewards/cosine_scaled_reward": 0.06971612432971597,
"rewards/format_reward": 0.5416666679084301,
"step": 180
},
{
"completion_length": 3232.3958892822266,
"epoch": 0.20685714285714285,
"grad_norm": 0.11890272796154022,
"kl": 0.0372467041015625,
"learning_rate": 8.245653237555705e-07,
"loss": 0.0285,
"reward": -0.08456094935536385,
"reward_std": 0.6224741712212563,
"rewards/cosine_scaled_reward": -0.07428957521915436,
"rewards/format_reward": 0.3125000074505806,
"step": 181
},
{
"completion_length": 2675.750045776367,
"epoch": 0.208,
"grad_norm": 0.1544029265642166,
"kl": 0.026336669921875,
"learning_rate": 8.220696016880687e-07,
"loss": 0.0012,
"reward": 0.28332219598814845,
"reward_std": 0.8586058430373669,
"rewards/cosine_scaled_reward": 0.04892569035291672,
"rewards/format_reward": 0.541666679084301,
"step": 182
},
{
"completion_length": 2722.3334045410156,
"epoch": 0.20914285714285713,
"grad_norm": 0.2383163869380951,
"kl": 0.0434417724609375,
"learning_rate": 8.195606193320136e-07,
"loss": 0.0422,
"reward": 0.37364979088306427,
"reward_std": 0.7791223339736462,
"rewards/cosine_scaled_reward": 0.08836854621767998,
"rewards/format_reward": 0.583333345130086,
"step": 183
},
{
"completion_length": 2838.3125534057617,
"epoch": 0.2102857142857143,
"grad_norm": 0.08796233683824539,
"kl": 0.037139892578125,
"learning_rate": 8.170384989716657e-07,
"loss": 0.0153,
"reward": -0.2598969964310527,
"reward_std": 0.48821557871997356,
"rewards/cosine_scaled_reward": -0.2099758218973875,
"rewards/format_reward": 0.3541666679084301,
"step": 184
},
{
"completion_length": 2718.395851135254,
"epoch": 0.21142857142857144,
"grad_norm": 0.14351139962673187,
"kl": 0.0391998291015625,
"learning_rate": 8.145033635316128e-07,
"loss": 0.0168,
"reward": -0.022745168476831168,
"reward_std": 0.6752897650003433,
"rewards/cosine_scaled_reward": -0.14890163764357567,
"rewards/format_reward": 0.5416666828095913,
"step": 185
},
{
"completion_length": 3067.541717529297,
"epoch": 0.21257142857142858,
"grad_norm": 0.10349977016448975,
"kl": 0.038116455078125,
"learning_rate": 8.119553365707802e-07,
"loss": 0.0083,
"reward": 0.06632867828011513,
"reward_std": 0.6240403186529875,
"rewards/cosine_scaled_reward": 0.0068851374089717865,
"rewards/format_reward": 0.35416666977107525,
"step": 186
},
{
"completion_length": 2420.291702270508,
"epoch": 0.21371428571428572,
"grad_norm": 0.15342004597187042,
"kl": 0.046356201171875,
"learning_rate": 8.093945422764069e-07,
"loss": 0.0298,
"reward": 0.08685800805687904,
"reward_std": 0.6128952912986279,
"rewards/cosine_scaled_reward": -0.11734730005264282,
"rewards/format_reward": 0.6250000111758709,
"step": 187
},
{
"completion_length": 3567.3958435058594,
"epoch": 0.21485714285714286,
"grad_norm": 0.15315963327884674,
"kl": 0.05072021484375,
"learning_rate": 8.068211054579943e-07,
"loss": 0.0074,
"reward": -0.398057883605361,
"reward_std": 0.6201205141842365,
"rewards/cosine_scaled_reward": -0.18498908844776452,
"rewards/format_reward": 0.10416666977107525,
"step": 188
},
{
"completion_length": 3156.937545776367,
"epoch": 0.216,
"grad_norm": 0.22657519578933716,
"kl": 0.05267333984375,
"learning_rate": 8.04235151541222e-07,
"loss": 0.0596,
"reward": 0.08714086917461827,
"reward_std": 0.7904996033757925,
"rewards/cosine_scaled_reward": -0.008101928047835827,
"rewards/format_reward": 0.3958333469927311,
"step": 189
},
{
"completion_length": 2789.041702270508,
"epoch": 0.21714285714285714,
"grad_norm": 0.13739848136901855,
"kl": 0.051605224609375,
"learning_rate": 8.01636806561836e-07,
"loss": 0.0128,
"reward": 0.15377038344740868,
"reward_std": 0.7078918963670731,
"rewards/cosine_scaled_reward": -0.0018926504999399185,
"rewards/format_reward": 0.47916668094694614,
"step": 190
},
{
"completion_length": 2624.083366394043,
"epoch": 0.21828571428571428,
"grad_norm": 0.17836546897888184,
"kl": 0.05645751953125,
"learning_rate": 7.990261971595048e-07,
"loss": 0.0403,
"reward": 0.17610792024061084,
"reward_std": 0.7685041464865208,
"rewards/cosine_scaled_reward": 0.013859146274626255,
"rewards/format_reward": 0.4583333395421505,
"step": 191
},
{
"completion_length": 3288.916717529297,
"epoch": 0.21942857142857142,
"grad_norm": 0.20157144963741302,
"kl": 0.053802490234375,
"learning_rate": 7.964034505716476e-07,
"loss": 0.0343,
"reward": -0.18958128988742828,
"reward_std": 0.6780943684279919,
"rewards/cosine_scaled_reward": -0.14980729576200247,
"rewards/format_reward": 0.31250000931322575,
"step": 192
},
{
"completion_length": 3114.1458740234375,
"epoch": 0.22057142857142858,
"grad_norm": 0.18799538910388947,
"kl": 0.0482025146484375,
"learning_rate": 7.93768694627233e-07,
"loss": 0.0323,
"reward": -0.24548013135790825,
"reward_std": 0.6712213661521673,
"rewards/cosine_scaled_reward": -0.1577296955510974,
"rewards/format_reward": 0.25000000186264515,
"step": 193
},
{
"completion_length": 3140.9375610351562,
"epoch": 0.22171428571428572,
"grad_norm": 0.27719372510910034,
"kl": 0.0517578125,
"learning_rate": 7.911220577405484e-07,
"loss": 0.0627,
"reward": 0.3136282116174698,
"reward_std": 1.0058045387268066,
"rewards/cosine_scaled_reward": 0.1217548530548811,
"rewards/format_reward": 0.4166666716337204,
"step": 194
},
{
"completion_length": 3052.187545776367,
"epoch": 0.22285714285714286,
"grad_norm": 0.2891777753829956,
"kl": 0.058441162109375,
"learning_rate": 7.884636689049422e-07,
"loss": 0.0488,
"reward": -0.09272150322794914,
"reward_std": 0.6414213795214891,
"rewards/cosine_scaled_reward": -0.07150299660861492,
"rewards/format_reward": 0.2916666716337204,
"step": 195
},
{
"completion_length": 3178.437530517578,
"epoch": 0.224,
"grad_norm": 0.1800169199705124,
"kl": 0.06561279296875,
"learning_rate": 7.857936576865356e-07,
"loss": 0.0051,
"reward": -0.12664931640028954,
"reward_std": 0.584855318069458,
"rewards/cosine_scaled_reward": -0.08012674562633038,
"rewards/format_reward": 0.27083333395421505,
"step": 196
},
{
"completion_length": 2523.6667404174805,
"epoch": 0.22514285714285714,
"grad_norm": 0.29900649189949036,
"kl": 0.08282470703125,
"learning_rate": 7.831121542179086e-07,
"loss": 0.0404,
"reward": 0.32191772386431694,
"reward_std": 1.1363152042031288,
"rewards/cosine_scaled_reward": 0.052287210419308394,
"rewards/format_reward": 0.5416666809469461,
"step": 197
},
{
"completion_length": 2786.625045776367,
"epoch": 0.22628571428571428,
"grad_norm": 0.34645184874534607,
"kl": 0.10394287109375,
"learning_rate": 7.804192891917571e-07,
"loss": 0.0284,
"reward": -0.09496624395251274,
"reward_std": 0.6138490363955498,
"rewards/cosine_scaled_reward": -0.11393817700445652,
"rewards/format_reward": 0.37500000558793545,
"step": 198
},
{
"completion_length": 2917.729217529297,
"epoch": 0.22742857142857142,
"grad_norm": 0.27866116166114807,
"kl": 0.08343505859375,
"learning_rate": 7.777151938545235e-07,
"loss": 0.0601,
"reward": -0.2887334353290498,
"reward_std": 0.6878443285822868,
"rewards/cosine_scaled_reward": -0.18037102394737303,
"rewards/format_reward": 0.22916667349636555,
"step": 199
},
{
"completion_length": 2416.0833587646484,
"epoch": 0.22857142857142856,
"grad_norm": 0.29420676827430725,
"kl": 0.07171630859375,
"learning_rate": 7.75e-07,
"loss": 0.0608,
"reward": 0.38031474966555834,
"reward_std": 0.8896522857248783,
"rewards/cosine_scaled_reward": 0.08112852554768324,
"rewards/format_reward": 0.604166679084301,
"step": 200
},
{
"completion_length": 2426.1667098999023,
"epoch": 0.2297142857142857,
"grad_norm": 0.3698631525039673,
"kl": 0.069580078125,
"learning_rate": 7.72273839962904e-07,
"loss": 0.0396,
"reward": 0.5210681445896626,
"reward_std": 0.8631033673882484,
"rewards/cosine_scaled_reward": 0.18454162776470184,
"rewards/format_reward": 0.5833333395421505,
"step": 201
},
{
"completion_length": 2384.5416946411133,
"epoch": 0.23085714285714284,
"grad_norm": 0.21861740946769714,
"kl": 0.09124755859375,
"learning_rate": 7.695368466124296e-07,
"loss": 0.0316,
"reward": 0.4668930694460869,
"reward_std": 0.7846425659954548,
"rewards/cosine_scaled_reward": 0.17546686669811606,
"rewards/format_reward": 0.541666679084301,
"step": 202
},
{
"completion_length": 3118.041717529297,
"epoch": 0.232,
"grad_norm": 0.32266831398010254,
"kl": 0.115875244140625,
"learning_rate": 7.667891533457718e-07,
"loss": 0.058,
"reward": 0.050868467427790165,
"reward_std": 0.8579942099750042,
"rewards/cosine_scaled_reward": 0.037848809035494924,
"rewards/format_reward": 0.22916666977107525,
"step": 203
},
{
"completion_length": 2304.9583587646484,
"epoch": 0.23314285714285715,
"grad_norm": 0.18206676840782166,
"kl": 0.109619140625,
"learning_rate": 7.640308940816239e-07,
"loss": 0.0048,
"reward": 0.30574227310717106,
"reward_std": 0.6170836389064789,
"rewards/cosine_scaled_reward": -0.008607452735304832,
"rewards/format_reward": 0.708333358168602,
"step": 204
},
{
"completion_length": 2729.062545776367,
"epoch": 0.2342857142857143,
"grad_norm": 0.3828980624675751,
"kl": 0.11627197265625,
"learning_rate": 7.612622032536507e-07,
"loss": 0.0304,
"reward": 0.5650580562651157,
"reward_std": 1.018170591443777,
"rewards/cosine_scaled_reward": 0.22851963574066758,
"rewards/format_reward": 0.541666679084301,
"step": 205
},
{
"completion_length": 3118.2500534057617,
"epoch": 0.23542857142857143,
"grad_norm": 0.38659653067588806,
"kl": 0.12823486328125,
"learning_rate": 7.584832158039378e-07,
"loss": 0.0409,
"reward": -0.19641774892807007,
"reward_std": 0.7758117392659187,
"rewards/cosine_scaled_reward": -0.16738836327567697,
"rewards/format_reward": 0.3125000037252903,
"step": 206
},
{
"completion_length": 3061.0000610351562,
"epoch": 0.23657142857142857,
"grad_norm": 0.5246978998184204,
"kl": 0.1700439453125,
"learning_rate": 7.556940671764124e-07,
"loss": 0.0753,
"reward": -0.06954369135200977,
"reward_std": 0.7209791392087936,
"rewards/cosine_scaled_reward": -0.15650581319641788,
"rewards/format_reward": 0.4791666753590107,
"step": 207
},
{
"completion_length": 2615.5833587646484,
"epoch": 0.2377142857142857,
"grad_norm": 0.590552568435669,
"kl": 0.137298583984375,
"learning_rate": 7.528948933102438e-07,
"loss": 0.0517,
"reward": 0.10035054851323366,
"reward_std": 0.6904583033174276,
"rewards/cosine_scaled_reward": -0.06608795002102852,
"rewards/format_reward": 0.541666679084301,
"step": 208
},
{
"completion_length": 2907.0625610351562,
"epoch": 0.23885714285714285,
"grad_norm": 0.3516550660133362,
"kl": 0.1842041015625,
"learning_rate": 7.500858306332172e-07,
"loss": 0.0368,
"reward": 0.19722715765237808,
"reward_std": 0.7542031668126583,
"rewards/cosine_scaled_reward": 0.045978354290127754,
"rewards/format_reward": 0.43750000558793545,
"step": 209
},
{
"completion_length": 2872.8333740234375,
"epoch": 0.24,
"grad_norm": 0.31704995036125183,
"kl": 0.17333984375,
"learning_rate": 7.472670160550848e-07,
"loss": 0.0472,
"reward": 0.005523813422769308,
"reward_std": 0.6337867602705956,
"rewards/cosine_scaled_reward": -0.05793424462899566,
"rewards/format_reward": 0.3958333432674408,
"step": 210
},
{
"completion_length": 2413.104202270508,
"epoch": 0.24114285714285713,
"grad_norm": 0.30833762884140015,
"kl": 0.18377685546875,
"learning_rate": 7.444385869608921e-07,
"loss": 0.0023,
"reward": 0.17130140773952007,
"reward_std": 0.6732404362410307,
"rewards/cosine_scaled_reward": -0.004216981120407581,
"rewards/format_reward": 0.5000000074505806,
"step": 211
},
{
"completion_length": 2611.0000762939453,
"epoch": 0.2422857142857143,
"grad_norm": 0.36040380597114563,
"kl": 0.195404052734375,
"learning_rate": 7.416006812042827e-07,
"loss": 0.0443,
"reward": 0.207231349311769,
"reward_std": 0.8025440499186516,
"rewards/cosine_scaled_reward": 0.004049690440297127,
"rewards/format_reward": 0.520833345130086,
"step": 212
},
{
"completion_length": 2825.479217529297,
"epoch": 0.24342857142857144,
"grad_norm": 0.4717042148113251,
"kl": 0.2646484375,
"learning_rate": 7.387534371007797e-07,
"loss": 0.0311,
"reward": 0.24717780202627182,
"reward_std": 0.9448489658534527,
"rewards/cosine_scaled_reward": 0.05484255403280258,
"rewards/format_reward": 0.45833334140479565,
"step": 213
},
{
"completion_length": 2770.4375762939453,
"epoch": 0.24457142857142858,
"grad_norm": 0.5870217084884644,
"kl": 0.2110595703125,
"learning_rate": 7.358969934210438e-07,
"loss": 0.0885,
"reward": 0.2060818038880825,
"reward_std": 0.8542167469859123,
"rewards/cosine_scaled_reward": 0.04777835123240948,
"rewards/format_reward": 0.4375000074505806,
"step": 214
},
{
"completion_length": 2372.354217529297,
"epoch": 0.24571428571428572,
"grad_norm": 0.2540161609649658,
"kl": 0.204833984375,
"learning_rate": 7.330314893841101e-07,
"loss": 0.0192,
"reward": -0.08897280413657427,
"reward_std": 0.5330733954906464,
"rewards/cosine_scaled_reward": -0.18874458596110344,
"rewards/format_reward": 0.541666679084301,
"step": 215
},
{
"completion_length": 2626.2083892822266,
"epoch": 0.24685714285714286,
"grad_norm": 0.38485583662986755,
"kl": 0.2821044921875,
"learning_rate": 7.301570646506027e-07,
"loss": 0.0632,
"reward": 0.3148540537804365,
"reward_std": 0.8362242169678211,
"rewards/cosine_scaled_reward": 0.04583857045508921,
"rewards/format_reward": 0.5833333469927311,
"step": 216
},
{
"completion_length": 3083.104217529297,
"epoch": 0.248,
"grad_norm": 0.35301655530929565,
"kl": 0.3206787109375,
"learning_rate": 7.27273859315928e-07,
"loss": 0.0247,
"reward": 0.1059049442410469,
"reward_std": 0.8536710906773806,
"rewards/cosine_scaled_reward": -0.003996940446086228,
"rewards/format_reward": 0.3958333358168602,
"step": 217
},
{
"completion_length": 2729.8125610351562,
"epoch": 0.24914285714285714,
"grad_norm": 0.3946671783924103,
"kl": 0.2547607421875,
"learning_rate": 7.243820139034464e-07,
"loss": 0.0153,
"reward": -0.2474758685566485,
"reward_std": 0.6462593208998442,
"rewards/cosine_scaled_reward": -0.22353118157479912,
"rewards/format_reward": 0.3750000149011612,
"step": 218
},
{
"completion_length": 2773.6250610351562,
"epoch": 0.2502857142857143,
"grad_norm": 0.5205709338188171,
"kl": 0.28680419921875,
"learning_rate": 7.214816693576234e-07,
"loss": 0.0678,
"reward": 0.17035892861895263,
"reward_std": 0.8409441113471985,
"rewards/cosine_scaled_reward": -0.03153566690161824,
"rewards/format_reward": 0.5416666772216558,
"step": 219
},
{
"completion_length": 3064.9792404174805,
"epoch": 0.25142857142857145,
"grad_norm": 0.3793766498565674,
"kl": 0.3310546875,
"learning_rate": 7.185729670371604e-07,
"loss": 0.0536,
"reward": -0.40579412039369345,
"reward_std": 0.4997274577617645,
"rewards/cosine_scaled_reward": -0.2578846197575331,
"rewards/format_reward": 0.25000000186264515,
"step": 220
},
{
"completion_length": 2607.375045776367,
"epoch": 0.25257142857142856,
"grad_norm": 0.4927677810192108,
"kl": 0.2911376953125,
"learning_rate": 7.156560487081051e-07,
"loss": 0.0538,
"reward": 0.29882943257689476,
"reward_std": 0.8470067903399467,
"rewards/cosine_scaled_reward": 0.015899650752544403,
"rewards/format_reward": 0.625000013038516,
"step": 221
},
{
"completion_length": 2744.041748046875,
"epoch": 0.2537142857142857,
"grad_norm": 0.3646686375141144,
"kl": 0.331298828125,
"learning_rate": 7.127310565369415e-07,
"loss": 0.0247,
"reward": 0.22724982630461454,
"reward_std": 0.6808402426540852,
"rewards/cosine_scaled_reward": -0.022449948824942112,
"rewards/format_reward": 0.6250000074505806,
"step": 222
},
{
"completion_length": 2798.2083740234375,
"epoch": 0.25485714285714284,
"grad_norm": 0.397216260433197,
"kl": 0.3436279296875,
"learning_rate": 7.097981330836616e-07,
"loss": 0.0215,
"reward": 0.16939541138708591,
"reward_std": 0.5345852337777615,
"rewards/cosine_scaled_reward": 0.007962611503899097,
"rewards/format_reward": 0.5,
"step": 223
},
{
"completion_length": 3249.1666870117188,
"epoch": 0.256,
"grad_norm": 0.7151598930358887,
"kl": 0.38037109375,
"learning_rate": 7.068574212948169e-07,
"loss": 0.0902,
"reward": -0.05321236699819565,
"reward_std": 0.8325384929776192,
"rewards/cosine_scaled_reward": -0.046919144690036774,
"rewards/format_reward": 0.2708333432674408,
"step": 224
},
{
"completion_length": 2881.2709350585938,
"epoch": 0.2571428571428571,
"grad_norm": 0.412383109331131,
"kl": 0.35223388671875,
"learning_rate": 7.039090644965509e-07,
"loss": 0.0437,
"reward": 0.012765285558998585,
"reward_std": 0.6845123060047626,
"rewards/cosine_scaled_reward": -0.08529938757419586,
"rewards/format_reward": 0.45833333395421505,
"step": 225
},
{
"completion_length": 2802.041717529297,
"epoch": 0.2582857142857143,
"grad_norm": 0.4051804840564728,
"kl": 0.37164306640625,
"learning_rate": 7.009532063876148e-07,
"loss": 0.0577,
"reward": 0.17664484679698944,
"reward_std": 0.7071616277098656,
"rewards/cosine_scaled_reward": 0.03299903869628906,
"rewards/format_reward": 0.43750000558793545,
"step": 226
},
{
"completion_length": 2718.291732788086,
"epoch": 0.25942857142857145,
"grad_norm": 0.4118206799030304,
"kl": 0.400390625,
"learning_rate": 6.979899910323624e-07,
"loss": 0.0558,
"reward": 0.059418462216854095,
"reward_std": 0.6883127726614475,
"rewards/cosine_scaled_reward": -0.16530904080718756,
"rewards/format_reward": 0.6666666734963655,
"step": 227
},
{
"completion_length": 2551.270866394043,
"epoch": 0.26057142857142856,
"grad_norm": 0.5437552332878113,
"kl": 0.3446044921875,
"learning_rate": 6.950195628537299e-07,
"loss": 0.0212,
"reward": 0.16873644525185227,
"reward_std": 0.6643664948642254,
"rewards/cosine_scaled_reward": -0.00022228434681892395,
"rewards/format_reward": 0.5000000037252903,
"step": 228
},
{
"completion_length": 3142.5208740234375,
"epoch": 0.26171428571428573,
"grad_norm": 0.824421226978302,
"kl": 0.490966796875,
"learning_rate": 6.920420666261961e-07,
"loss": 0.022,
"reward": -0.009648986160755157,
"reward_std": 0.523324097506702,
"rewards/cosine_scaled_reward": -0.06068984791636467,
"rewards/format_reward": 0.3958333395421505,
"step": 229
},
{
"completion_length": 3071.4583740234375,
"epoch": 0.26285714285714284,
"grad_norm": 0.5922899842262268,
"kl": 0.43035888671875,
"learning_rate": 6.890576474687263e-07,
"loss": 0.0518,
"reward": -0.29550562985241413,
"reward_std": 0.6693998202681541,
"rewards/cosine_scaled_reward": -0.22350211441516876,
"rewards/format_reward": 0.31250000931322575,
"step": 230
},
{
"completion_length": 2912.7083740234375,
"epoch": 0.264,
"grad_norm": 0.5167642831802368,
"kl": 0.3560791015625,
"learning_rate": 6.860664508377001e-07,
"loss": 0.0283,
"reward": 0.1596711277961731,
"reward_std": 0.7965347096323967,
"rewards/cosine_scaled_reward": -0.01580745540559292,
"rewards/format_reward": 0.5000000074505806,
"step": 231
},
{
"completion_length": 3158.1459045410156,
"epoch": 0.2651428571428571,
"grad_norm": 0.6749157905578613,
"kl": 0.4017333984375,
"learning_rate": 6.83068622519821e-07,
"loss": 0.0837,
"reward": -0.297568422742188,
"reward_std": 0.6748274452984333,
"rewards/cosine_scaled_reward": -0.18697709869593382,
"rewards/format_reward": 0.22916666977107525,
"step": 232
},
{
"completion_length": 2762.1666870117188,
"epoch": 0.2662857142857143,
"grad_norm": 0.3311655819416046,
"kl": 0.317626953125,
"learning_rate": 6.800643086250121e-07,
"loss": 0.0416,
"reward": -0.08082350715994835,
"reward_std": 0.6287317145615816,
"rewards/cosine_scaled_reward": -0.16008687764406204,
"rewards/format_reward": 0.47916666977107525,
"step": 233
},
{
"completion_length": 2734.8750648498535,
"epoch": 0.2674285714285714,
"grad_norm": 0.2763339579105377,
"kl": 0.291839599609375,
"learning_rate": 6.770536555792944e-07,
"loss": 0.0202,
"reward": -0.14287907630205154,
"reward_std": 0.5912873484194279,
"rewards/cosine_scaled_reward": -0.13689745217561722,
"rewards/format_reward": 0.35416666977107525,
"step": 234
},
{
"completion_length": 2442.187515258789,
"epoch": 0.26857142857142857,
"grad_norm": 0.324736088514328,
"kl": 0.25311279296875,
"learning_rate": 6.740368101176495e-07,
"loss": 0.0321,
"reward": -0.05069837532937527,
"reward_std": 0.6721667312085629,
"rewards/cosine_scaled_reward": -0.173017387278378,
"rewards/format_reward": 0.5416666716337204,
"step": 235
},
{
"completion_length": 2971.2083740234375,
"epoch": 0.26971428571428574,
"grad_norm": 0.6796517968177795,
"kl": 0.234130859375,
"learning_rate": 6.710139192768694e-07,
"loss": 0.0103,
"reward": 0.2688827021047473,
"reward_std": 0.9795111119747162,
"rewards/cosine_scaled_reward": -0.004769146267790347,
"rewards/format_reward": 0.604166679084301,
"step": 236
},
{
"completion_length": 2812.8750610351562,
"epoch": 0.27085714285714285,
"grad_norm": 0.9011460542678833,
"kl": 0.2601318359375,
"learning_rate": 6.679851303883891e-07,
"loss": 0.0936,
"reward": 0.06308133527636528,
"reward_std": 0.7535826116800308,
"rewards/cosine_scaled_reward": -0.068068141117692,
"rewards/format_reward": 0.47916667349636555,
"step": 237
},
{
"completion_length": 2691.7084045410156,
"epoch": 0.272,
"grad_norm": 1.501541018486023,
"kl": 0.273681640625,
"learning_rate": 6.649505910711058e-07,
"loss": 0.0986,
"reward": 0.32719678059220314,
"reward_std": 1.0173063725233078,
"rewards/cosine_scaled_reward": 0.05737040005624294,
"rewards/format_reward": 0.5625000074505806,
"step": 238
},
{
"completion_length": 2189.7708587646484,
"epoch": 0.27314285714285713,
"grad_norm": 0.20339642465114594,
"kl": 0.24578857421875,
"learning_rate": 6.619104492241847e-07,
"loss": 0.0259,
"reward": 0.3876400392036885,
"reward_std": 0.6854967623949051,
"rewards/cosine_scaled_reward": 0.07331139780580997,
"rewards/format_reward": 0.6458333395421505,
"step": 239
},
{
"completion_length": 2928.354232788086,
"epoch": 0.2742857142857143,
"grad_norm": 0.46207401156425476,
"kl": 0.510498046875,
"learning_rate": 6.588648530198504e-07,
"loss": 0.0666,
"reward": -0.031538160517811775,
"reward_std": 0.6143174581229687,
"rewards/cosine_scaled_reward": -0.12371946685016155,
"rewards/format_reward": 0.47916667349636555,
"step": 240
},
{
"completion_length": 3160.937545776367,
"epoch": 0.2754285714285714,
"grad_norm": 0.4524831175804138,
"kl": 0.5111083984375,
"learning_rate": 6.558139508961654e-07,
"loss": 0.0504,
"reward": -0.24879638850688934,
"reward_std": 0.6495300643146038,
"rewards/cosine_scaled_reward": -0.22295443713665009,
"rewards/format_reward": 0.37500000558793545,
"step": 241
},
{
"completion_length": 2566.5000610351562,
"epoch": 0.2765714285714286,
"grad_norm": 0.578352153301239,
"kl": 0.4998779296875,
"learning_rate": 6.527578915497951e-07,
"loss": 0.0285,
"reward": 0.15099805174395442,
"reward_std": 0.515793077647686,
"rewards/cosine_scaled_reward": -0.11855055205523968,
"rewards/format_reward": 0.7291666809469461,
"step": 242
},
{
"completion_length": 2909.3958587646484,
"epoch": 0.2777142857142857,
"grad_norm": 0.8137231469154358,
"kl": 0.51025390625,
"learning_rate": 6.496968239287603e-07,
"loss": 0.0269,
"reward": 0.334348788484931,
"reward_std": 0.7821071408689022,
"rewards/cosine_scaled_reward": 0.07382349669933319,
"rewards/format_reward": 0.5625000037252903,
"step": 243
},
{
"completion_length": 2917.541748046875,
"epoch": 0.27885714285714286,
"grad_norm": 0.6295875906944275,
"kl": 0.50042724609375,
"learning_rate": 6.466308972251785e-07,
"loss": 0.0786,
"reward": 0.2729534022510052,
"reward_std": 0.830750398337841,
"rewards/cosine_scaled_reward": 0.08813249412924051,
"rewards/format_reward": 0.43750000558793545,
"step": 244
},
{
"completion_length": 3121.8958740234375,
"epoch": 0.28,
"grad_norm": 1.1297352313995361,
"kl": 0.5673828125,
"learning_rate": 6.435602608679916e-07,
"loss": 0.12,
"reward": 0.08920413255691528,
"reward_std": 1.0364465415477753,
"rewards/cosine_scaled_reward": -0.05938760610297322,
"rewards/format_reward": 0.4583333432674408,
"step": 245
},
{
"completion_length": 3070.0833740234375,
"epoch": 0.28114285714285714,
"grad_norm": 0.941576361656189,
"kl": 0.61669921875,
"learning_rate": 6.404850645156841e-07,
"loss": 0.1055,
"reward": 0.07732813712209463,
"reward_std": 0.9496094807982445,
"rewards/cosine_scaled_reward": -0.05677332216873765,
"rewards/format_reward": 0.45833334513008595,
"step": 246
},
{
"completion_length": 3249.375045776367,
"epoch": 0.2822857142857143,
"grad_norm": 0.7210645079612732,
"kl": 0.67041015625,
"learning_rate": 6.374054580489873e-07,
"loss": 0.0552,
"reward": -0.3232395015656948,
"reward_std": 0.6331962943077087,
"rewards/cosine_scaled_reward": -0.21088325325399637,
"rewards/format_reward": 0.2500000037252903,
"step": 247
},
{
"completion_length": 2833.7083892822266,
"epoch": 0.2834285714285714,
"grad_norm": 0.5416091680526733,
"kl": 0.570068359375,
"learning_rate": 6.343215915635761e-07,
"loss": 0.0338,
"reward": 0.2699696607887745,
"reward_std": 0.769625548273325,
"rewards/cosine_scaled_reward": 0.08322756737470627,
"rewards/format_reward": 0.4583333395421505,
"step": 248
},
{
"completion_length": 2345.041702270508,
"epoch": 0.2845714285714286,
"grad_norm": 0.4412481486797333,
"kl": 0.44451904296875,
"learning_rate": 6.31233615362752e-07,
"loss": 0.0447,
"reward": 0.4360041692852974,
"reward_std": 0.8868755213916302,
"rewards/cosine_scaled_reward": 0.08689466118812561,
"rewards/format_reward": 0.6666666772216558,
"step": 249
},
{
"completion_length": 2516.8333587646484,
"epoch": 0.2857142857142857,
"grad_norm": 0.5790634155273438,
"kl": 0.5372314453125,
"learning_rate": 6.281416799501187e-07,
"loss": 0.0766,
"reward": 0.1508978575002402,
"reward_std": 0.7845667004585266,
"rewards/cosine_scaled_reward": -0.10766792530193925,
"rewards/format_reward": 0.666666679084301,
"step": 250
},
{
"completion_length": 2367.6875534057617,
"epoch": 0.28685714285714287,
"grad_norm": 0.465372771024704,
"kl": 0.53857421875,
"learning_rate": 6.25045936022246e-07,
"loss": 0.045,
"reward": 0.2600698294118047,
"reward_std": 0.7807004749774933,
"rewards/cosine_scaled_reward": -0.08423476293683052,
"rewards/format_reward": 0.7708333395421505,
"step": 251
},
{
"completion_length": 2789.3750915527344,
"epoch": 0.288,
"grad_norm": 0.5998123288154602,
"kl": 0.5885009765625,
"learning_rate": 6.219465344613258e-07,
"loss": 0.0437,
"reward": 0.0935278192628175,
"reward_std": 0.7258482351899147,
"rewards/cosine_scaled_reward": -0.09951683203689754,
"rewards/format_reward": 0.583333345130086,
"step": 252
},
{
"completion_length": 2500.1875915527344,
"epoch": 0.28914285714285715,
"grad_norm": 0.6858162879943848,
"kl": 0.5001220703125,
"learning_rate": 6.188436263278172e-07,
"loss": 0.0555,
"reward": 0.2763403048738837,
"reward_std": 0.9541792422533035,
"rewards/cosine_scaled_reward": -0.020189424976706505,
"rewards/format_reward": 0.645833345130086,
"step": 253
},
{
"completion_length": 3007.7500610351562,
"epoch": 0.29028571428571426,
"grad_norm": 0.7101835012435913,
"kl": 0.715576171875,
"learning_rate": 6.157373628530852e-07,
"loss": 0.0531,
"reward": 0.011770043522119522,
"reward_std": 0.6751254117116332,
"rewards/cosine_scaled_reward": -0.06386499013751745,
"rewards/format_reward": 0.4166666716337204,
"step": 254
},
{
"completion_length": 2951.0625534057617,
"epoch": 0.2914285714285714,
"grad_norm": 0.8391396403312683,
"kl": 0.64337158203125,
"learning_rate": 6.126278954320294e-07,
"loss": 0.0387,
"reward": -0.2175804078578949,
"reward_std": 0.633871290832758,
"rewards/cosine_scaled_reward": -0.2334041576832533,
"rewards/format_reward": 0.4375000111758709,
"step": 255
},
{
"completion_length": 2761.041702270508,
"epoch": 0.2925714285714286,
"grad_norm": 1.0026954412460327,
"kl": 0.5166015625,
"learning_rate": 6.095153756157051e-07,
"loss": 0.0119,
"reward": 0.14736154675483704,
"reward_std": 0.7076679766178131,
"rewards/cosine_scaled_reward": -0.052284312434494495,
"rewards/format_reward": 0.5625000093132257,
"step": 256
},
{
"completion_length": 3124.2709350585938,
"epoch": 0.2937142857142857,
"grad_norm": 0.9888486266136169,
"kl": 0.50518798828125,
"learning_rate": 6.06399955103937e-07,
"loss": 0.0677,
"reward": 0.3817547345533967,
"reward_std": 0.9725684300065041,
"rewards/cosine_scaled_reward": 0.10540459351614118,
"rewards/format_reward": 0.541666679084301,
"step": 257
},
{
"completion_length": 3022.979217529297,
"epoch": 0.2948571428571429,
"grad_norm": 1.060575246810913,
"kl": 0.49462890625,
"learning_rate": 6.032817857379256e-07,
"loss": 0.0775,
"reward": 0.11859412118792534,
"reward_std": 0.8907319009304047,
"rewards/cosine_scaled_reward": -0.01629660092294216,
"rewards/format_reward": 0.4375000074505806,
"step": 258
},
{
"completion_length": 2550.833366394043,
"epoch": 0.296,
"grad_norm": 0.4294055998325348,
"kl": 0.43841552734375,
"learning_rate": 6.001610194928464e-07,
"loss": 0.052,
"reward": 0.30166149651631713,
"reward_std": 0.831500705331564,
"rewards/cosine_scaled_reward": 0.02446294855326414,
"rewards/format_reward": 0.6041666828095913,
"step": 259
},
{
"completion_length": 2399.1250610351562,
"epoch": 0.29714285714285715,
"grad_norm": 0.3650396168231964,
"kl": 0.36968994140625,
"learning_rate": 5.97037808470444e-07,
"loss": 0.0443,
"reward": 0.403296634554863,
"reward_std": 0.926733735948801,
"rewards/cosine_scaled_reward": 0.07911694049835205,
"rewards/format_reward": 0.6250000149011612,
"step": 260
},
{
"completion_length": 2862.2708892822266,
"epoch": 0.29828571428571427,
"grad_norm": 1.088972568511963,
"kl": 0.4329833984375,
"learning_rate": 5.939123048916173e-07,
"loss": 0.068,
"reward": -0.11121963523328304,
"reward_std": 0.6940420866012573,
"rewards/cosine_scaled_reward": -0.19372293539345264,
"rewards/format_reward": 0.5000000074505806,
"step": 261
},
{
"completion_length": 2654.479217529297,
"epoch": 0.29942857142857143,
"grad_norm": 0.9074969291687012,
"kl": 0.4139404296875,
"learning_rate": 5.907846610890011e-07,
"loss": 0.0371,
"reward": 0.052452532574534416,
"reward_std": 0.6277891993522644,
"rewards/cosine_scaled_reward": -0.18083440139889717,
"rewards/format_reward": 0.7083333395421505,
"step": 262
},
{
"completion_length": 2793.729202270508,
"epoch": 0.30057142857142854,
"grad_norm": 2.0290629863739014,
"kl": 0.50604248046875,
"learning_rate": 5.87655029499542e-07,
"loss": 0.0191,
"reward": -0.09252774063497782,
"reward_std": 0.5941142216324806,
"rewards/cosine_scaled_reward": -0.2058687130920589,
"rewards/format_reward": 0.5625000074505806,
"step": 263
},
{
"completion_length": 2861.9584350585938,
"epoch": 0.3017142857142857,
"grad_norm": 0.7169579863548279,
"kl": 0.4697265625,
"learning_rate": 5.845235626570683e-07,
"loss": 0.0771,
"reward": 0.00036056432873010635,
"reward_std": 0.7790909744799137,
"rewards/cosine_scaled_reward": -0.16522826021537185,
"rewards/format_reward": 0.5833333469927311,
"step": 264
},
{
"completion_length": 2631.666732788086,
"epoch": 0.3028571428571429,
"grad_norm": 0.7453656196594238,
"kl": 0.45709228515625,
"learning_rate": 5.813904131848564e-07,
"loss": 0.0216,
"reward": 0.3167956112883985,
"reward_std": 0.8776490315794945,
"rewards/cosine_scaled_reward": -0.05162257980555296,
"rewards/format_reward": 0.7708333414047956,
"step": 265
},
{
"completion_length": 2901.4583892822266,
"epoch": 0.304,
"grad_norm": 1.1252886056900024,
"kl": 0.4161376953125,
"learning_rate": 5.78255733788191e-07,
"loss": -0.0038,
"reward": 0.05045348312705755,
"reward_std": 0.8374455273151398,
"rewards/cosine_scaled_reward": -0.1557634025812149,
"rewards/format_reward": 0.6250000055879354,
"step": 266
},
{
"completion_length": 3027.8750762939453,
"epoch": 0.30514285714285716,
"grad_norm": 0.44741290807724,
"kl": 0.4798583984375,
"learning_rate": 5.751196772469237e-07,
"loss": 0.0489,
"reward": 0.042676386423408985,
"reward_std": 0.7426804676651955,
"rewards/cosine_scaled_reward": -0.14299316331744194,
"rewards/format_reward": 0.604166679084301,
"step": 267
},
{
"completion_length": 2573.0208740234375,
"epoch": 0.3062857142857143,
"grad_norm": 0.42620187997817993,
"kl": 0.3807373046875,
"learning_rate": 5.71982396408026e-07,
"loss": 0.0182,
"reward": 0.010210057254880667,
"reward_std": 0.6326607689261436,
"rewards/cosine_scaled_reward": -0.11539698392152786,
"rewards/format_reward": 0.5208333395421505,
"step": 268
},
{
"completion_length": 2693.4584350585938,
"epoch": 0.30742857142857144,
"grad_norm": 0.8198751211166382,
"kl": 0.38409423828125,
"learning_rate": 5.688440441781398e-07,
"loss": 0.0175,
"reward": 0.04538612812757492,
"reward_std": 0.8305856361985207,
"rewards/cosine_scaled_reward": -0.16101938486099243,
"rewards/format_reward": 0.6250000093132257,
"step": 269
},
{
"completion_length": 2817.5834197998047,
"epoch": 0.30857142857142855,
"grad_norm": 0.6884281039237976,
"kl": 0.3497314453125,
"learning_rate": 5.657047735161255e-07,
"loss": 0.0278,
"reward": 0.19405698496848345,
"reward_std": 0.9544314742088318,
"rewards/cosine_scaled_reward": -0.05997245345497504,
"rewards/format_reward": 0.6041666828095913,
"step": 270
},
{
"completion_length": 2640.4584045410156,
"epoch": 0.3097142857142857,
"grad_norm": 0.48852798342704773,
"kl": 0.32904052734375,
"learning_rate": 5.625647374256061e-07,
"loss": 0.0483,
"reward": 0.3868153728544712,
"reward_std": 0.8188424855470657,
"rewards/cosine_scaled_reward": 0.05347882490605116,
"rewards/format_reward": 0.6666666697710752,
"step": 271
},
{
"completion_length": 3057.0000915527344,
"epoch": 0.31085714285714283,
"grad_norm": 0.805952787399292,
"kl": 0.4097900390625,
"learning_rate": 5.594240889475106e-07,
"loss": 0.0648,
"reward": -0.030033869668841362,
"reward_std": 0.788214236497879,
"rewards/cosine_scaled_reward": -0.1403372660279274,
"rewards/format_reward": 0.5000000111758709,
"step": 272
},
{
"completion_length": 2664.3125610351562,
"epoch": 0.312,
"grad_norm": 0.6616376638412476,
"kl": 0.384765625,
"learning_rate": 5.562829811526154e-07,
"loss": 0.0021,
"reward": 0.062399049289524555,
"reward_std": 0.7391597256064415,
"rewards/cosine_scaled_reward": -0.06876735016703606,
"rewards/format_reward": 0.47916667349636555,
"step": 273
},
{
"completion_length": 2088.8541870117188,
"epoch": 0.31314285714285717,
"grad_norm": 1.1467257738113403,
"kl": 0.261138916015625,
"learning_rate": 5.531415671340826e-07,
"loss": -0.0011,
"reward": 0.5514285732060671,
"reward_std": 0.6735243201255798,
"rewards/cosine_scaled_reward": 0.1613447144627571,
"rewards/format_reward": 0.6875,
"step": 274
},
{
"completion_length": 2230.0625381469727,
"epoch": 0.3142857142857143,
"grad_norm": 0.48795756697654724,
"kl": 0.2476806640625,
"learning_rate": 5.5e-07,
"loss": 0.0153,
"reward": 0.4183642081916332,
"reward_std": 0.9099490307271481,
"rewards/cosine_scaled_reward": 0.07764563540695235,
"rewards/format_reward": 0.6458333432674408,
"step": 275
},
{
"completion_length": 2390.729248046875,
"epoch": 0.31542857142857145,
"grad_norm": 0.5403981804847717,
"kl": 0.2706298828125,
"learning_rate": 5.468584328659172e-07,
"loss": 0.0278,
"reward": 0.5140420459210873,
"reward_std": 0.9649418145418167,
"rewards/cosine_scaled_reward": 0.08755906065925956,
"rewards/format_reward": 0.7500000204890966,
"step": 276
},
{
"completion_length": 2273.7917404174805,
"epoch": 0.31657142857142856,
"grad_norm": 0.681844174861908,
"kl": 0.2330322265625,
"learning_rate": 5.437170188473847e-07,
"loss": 0.0456,
"reward": 0.39711499866098166,
"reward_std": 0.8123867064714432,
"rewards/cosine_scaled_reward": 0.028049522661603987,
"rewards/format_reward": 0.7291666772216558,
"step": 277
},
{
"completion_length": 2178.3333740234375,
"epoch": 0.3177142857142857,
"grad_norm": 0.31549885869026184,
"kl": 0.264892578125,
"learning_rate": 5.405759110524894e-07,
"loss": 0.0333,
"reward": 0.46570797031745315,
"reward_std": 0.5852809324860573,
"rewards/cosine_scaled_reward": 0.054337045177817345,
"rewards/format_reward": 0.791666679084301,
"step": 278
},
{
"completion_length": 2957.604248046875,
"epoch": 0.31885714285714284,
"grad_norm": 0.6503915786743164,
"kl": 0.3946533203125,
"learning_rate": 5.37435262574394e-07,
"loss": 0.0216,
"reward": -0.10753144230693579,
"reward_std": 0.821341261267662,
"rewards/cosine_scaled_reward": -0.20013360609300435,
"rewards/format_reward": 0.5000000149011612,
"step": 279
},
{
"completion_length": 2075.479217529297,
"epoch": 0.32,
"grad_norm": 1.738971471786499,
"kl": 0.265228271484375,
"learning_rate": 5.342952264838747e-07,
"loss": 0.0882,
"reward": 0.510447891894728,
"reward_std": 0.914945088326931,
"rewards/cosine_scaled_reward": 0.14110325649380684,
"rewards/format_reward": 0.6458333488553762,
"step": 280
},
{
"completion_length": 3364.3333740234375,
"epoch": 0.3211428571428571,
"grad_norm": 0.8055136203765869,
"kl": 0.486083984375,
"learning_rate": 5.311559558218603e-07,
"loss": 0.0224,
"reward": -0.18871852289885283,
"reward_std": 0.6349410191178322,
"rewards/cosine_scaled_reward": -0.18654117919504642,
"rewards/format_reward": 0.39583334140479565,
"step": 281
},
{
"completion_length": 2701.916748046875,
"epoch": 0.3222857142857143,
"grad_norm": 0.4943215548992157,
"kl": 0.34820556640625,
"learning_rate": 5.28017603591974e-07,
"loss": 0.0322,
"reward": 0.3567043347284198,
"reward_std": 0.687431275844574,
"rewards/cosine_scaled_reward": 0.010191468521952629,
"rewards/format_reward": 0.7291666865348816,
"step": 282
},
{
"completion_length": 2955.5000610351562,
"epoch": 0.32342857142857145,
"grad_norm": 2.2376575469970703,
"kl": 0.4698486328125,
"learning_rate": 5.248803227530763e-07,
"loss": 0.1237,
"reward": 0.17788631992880255,
"reward_std": 0.9274262189865112,
"rewards/cosine_scaled_reward": -0.03141580242663622,
"rewards/format_reward": 0.5416666828095913,
"step": 283
},
{
"completion_length": 2488.4584350585938,
"epoch": 0.32457142857142857,
"grad_norm": 0.9498338103294373,
"kl": 0.3389892578125,
"learning_rate": 5.21744266211809e-07,
"loss": 0.0732,
"reward": 0.36859262455254793,
"reward_std": 0.7699229158461094,
"rewards/cosine_scaled_reward": -0.00418412871658802,
"rewards/format_reward": 0.7708333507180214,
"step": 284
},
{
"completion_length": 2094.3750762939453,
"epoch": 0.32571428571428573,
"grad_norm": 0.24566514790058136,
"kl": 0.274169921875,
"learning_rate": 5.186095868151436e-07,
"loss": 0.0373,
"reward": 0.21163646131753922,
"reward_std": 0.6311895027756691,
"rewards/cosine_scaled_reward": -0.09609892021398991,
"rewards/format_reward": 0.7500000111758709,
"step": 285
},
{
"completion_length": 2660.3333740234375,
"epoch": 0.32685714285714285,
"grad_norm": 0.4989126920700073,
"kl": 0.52783203125,
"learning_rate": 5.154764373429315e-07,
"loss": 0.0441,
"reward": 0.1898297774605453,
"reward_std": 0.8184352703392506,
"rewards/cosine_scaled_reward": -0.08057594299316406,
"rewards/format_reward": 0.6666666846722364,
"step": 286
},
{
"completion_length": 1924.020881652832,
"epoch": 0.328,
"grad_norm": 1.0765957832336426,
"kl": 0.527130126953125,
"learning_rate": 5.123449705004581e-07,
"loss": -0.0004,
"reward": 0.320914643118158,
"reward_std": 0.6506007239222527,
"rewards/cosine_scaled_reward": -0.0552152032032609,
"rewards/format_reward": 0.8125000149011612,
"step": 287
},
{
"completion_length": 2751.916748046875,
"epoch": 0.3291428571428571,
"grad_norm": 0.525725781917572,
"kl": 0.47174072265625,
"learning_rate": 5.09215338910999e-07,
"loss": 0.0345,
"reward": 0.28691791370511055,
"reward_std": 0.925068948417902,
"rewards/cosine_scaled_reward": -0.05298358201980591,
"rewards/format_reward": 0.7291666716337204,
"step": 288
},
{
"completion_length": 2081.791702270508,
"epoch": 0.3302857142857143,
"grad_norm": 0.4456733167171478,
"kl": 0.426513671875,
"learning_rate": 5.060876951083828e-07,
"loss": 0.0375,
"reward": 0.2169907259522006,
"reward_std": 0.6403507255017757,
"rewards/cosine_scaled_reward": -0.062007103115320206,
"rewards/format_reward": 0.6875000055879354,
"step": 289
},
{
"completion_length": 2641.5625915527344,
"epoch": 0.3314285714285714,
"grad_norm": 0.49116048216819763,
"kl": 0.56640625,
"learning_rate": 5.02962191529556e-07,
"loss": 0.0674,
"reward": 0.32489653676748276,
"reward_std": 0.7462290413677692,
"rewards/cosine_scaled_reward": -0.07674999348819256,
"rewards/format_reward": 0.854166679084301,
"step": 290
},
{
"completion_length": 2844.3959045410156,
"epoch": 0.3325714285714286,
"grad_norm": 0.9147380590438843,
"kl": 0.5509033203125,
"learning_rate": 4.998389805071536e-07,
"loss": 0.0864,
"reward": 0.2941320105455816,
"reward_std": 0.8829877898097038,
"rewards/cosine_scaled_reward": -0.09763280488550663,
"rewards/format_reward": 0.8333333544433117,
"step": 291
},
{
"completion_length": 2980.3126220703125,
"epoch": 0.33371428571428574,
"grad_norm": 1.0147628784179688,
"kl": 0.628662109375,
"learning_rate": 4.967182142620745e-07,
"loss": 0.0385,
"reward": 0.16018317895941436,
"reward_std": 0.6120474711060524,
"rewards/cosine_scaled_reward": -0.09870209451764822,
"rewards/format_reward": 0.6875000111758709,
"step": 292
},
{
"completion_length": 2222.3125610351562,
"epoch": 0.33485714285714285,
"grad_norm": 1.3469599485397339,
"kl": 0.42333984375,
"learning_rate": 4.93600044896063e-07,
"loss": -0.0092,
"reward": 0.35159002151340246,
"reward_std": 0.6072813756763935,
"rewards/cosine_scaled_reward": -0.03080323152244091,
"rewards/format_reward": 0.8125000074505806,
"step": 293
},
{
"completion_length": 3145.1041870117188,
"epoch": 0.336,
"grad_norm": 1.195894718170166,
"kl": 0.774658203125,
"learning_rate": 4.904846243842949e-07,
"loss": 0.0366,
"reward": -0.011713245883584023,
"reward_std": 0.686446376144886,
"rewards/cosine_scaled_reward": -0.1353329624980688,
"rewards/format_reward": 0.5208333414047956,
"step": 294
},
{
"completion_length": 2874.3126068115234,
"epoch": 0.33714285714285713,
"grad_norm": 0.6826883554458618,
"kl": 0.62451171875,
"learning_rate": 4.873721045679706e-07,
"loss": 0.0648,
"reward": 0.4660371467471123,
"reward_std": 0.8896083161234856,
"rewards/cosine_scaled_reward": 0.10532690212130547,
"rewards/format_reward": 0.6666666865348816,
"step": 295
},
{
"completion_length": 3109.916748046875,
"epoch": 0.3382857142857143,
"grad_norm": 0.6580361723899841,
"kl": 0.54638671875,
"learning_rate": 4.842626371469149e-07,
"loss": 0.0593,
"reward": 0.1299741494731279,
"reward_std": 0.7981628403067589,
"rewards/cosine_scaled_reward": -0.1501350817270577,
"rewards/format_reward": 0.7291666977107525,
"step": 296
},
{
"completion_length": 3412.6875915527344,
"epoch": 0.3394285714285714,
"grad_norm": 0.5886114835739136,
"kl": 0.65771484375,
"learning_rate": 4.811563736721829e-07,
"loss": 0.0604,
"reward": -0.13705510459840298,
"reward_std": 0.7897306978702545,
"rewards/cosine_scaled_reward": -0.1760294260457158,
"rewards/format_reward": 0.41666666977107525,
"step": 297
},
{
"completion_length": 2516.0000610351562,
"epoch": 0.3405714285714286,
"grad_norm": 0.9390971660614014,
"kl": 0.326019287109375,
"learning_rate": 4.780534655386743e-07,
"loss": -0.005,
"reward": 0.2075387438526377,
"reward_std": 0.678107738494873,
"rewards/cosine_scaled_reward": -0.06827879883348942,
"rewards/format_reward": 0.687500013038516,
"step": 298
},
{
"completion_length": 2798.750045776367,
"epoch": 0.3417142857142857,
"grad_norm": 0.32884588837623596,
"kl": 0.3729248046875,
"learning_rate": 4.749540639777539e-07,
"loss": 0.029,
"reward": 0.3677529713604599,
"reward_std": 0.581198662519455,
"rewards/cosine_scaled_reward": 0.07487833127379417,
"rewards/format_reward": 0.6250000149011612,
"step": 299
},
{
"completion_length": 3013.9584350585938,
"epoch": 0.34285714285714286,
"grad_norm": 0.45819181203842163,
"kl": 0.378662109375,
"learning_rate": 4.7185832004988133e-07,
"loss": 0.0371,
"reward": 0.11145258648321033,
"reward_std": 0.8130291737616062,
"rewards/cosine_scaled_reward": -0.12973832013085485,
"rewards/format_reward": 0.6666666865348816,
"step": 300
},
{
"completion_length": 2654.791717529297,
"epoch": 0.344,
"grad_norm": 0.4394110143184662,
"kl": 0.3597412109375,
"learning_rate": 4.68766384637248e-07,
"loss": 0.0512,
"reward": 0.15728470450267196,
"reward_std": 0.8084782175719738,
"rewards/cosine_scaled_reward": -0.07348441705107689,
"rewards/format_reward": 0.6041666753590107,
"step": 301
},
{
"completion_length": 2718.750030517578,
"epoch": 0.34514285714285714,
"grad_norm": 1.5091288089752197,
"kl": 0.374420166015625,
"learning_rate": 4.656784084364238e-07,
"loss": -0.0067,
"reward": -0.010899038054049015,
"reward_std": 0.6495076231658459,
"rewards/cosine_scaled_reward": -0.1349128014408052,
"rewards/format_reward": 0.5208333358168602,
"step": 302
},
{
"completion_length": 2785.0833740234375,
"epoch": 0.3462857142857143,
"grad_norm": 0.6161960363388062,
"kl": 0.3387451171875,
"learning_rate": 4.6259454195101267e-07,
"loss": 0.0533,
"reward": 0.19499525055289268,
"reward_std": 0.9087162502110004,
"rewards/cosine_scaled_reward": -0.10367321688681841,
"rewards/format_reward": 0.7083333469927311,
"step": 303
},
{
"completion_length": 2953.541778564453,
"epoch": 0.3474285714285714,
"grad_norm": 0.4118628203868866,
"kl": 0.3302001953125,
"learning_rate": 4.59514935484316e-07,
"loss": 0.05,
"reward": 0.17398711014539003,
"reward_std": 0.7196006700396538,
"rewards/cosine_scaled_reward": -0.13903226237744093,
"rewards/format_reward": 0.7708333488553762,
"step": 304
},
{
"completion_length": 3122.791748046875,
"epoch": 0.3485714285714286,
"grad_norm": 0.9769808650016785,
"kl": 0.315673828125,
"learning_rate": 4.5643973913200837e-07,
"loss": 0.0601,
"reward": 0.1985210245475173,
"reward_std": 0.7755458503961563,
"rewards/cosine_scaled_reward": -0.06837073154747486,
"rewards/format_reward": 0.6666666865348816,
"step": 305
},
{
"completion_length": 2781.9584350585938,
"epoch": 0.3497142857142857,
"grad_norm": 0.8336344957351685,
"kl": 0.295135498046875,
"learning_rate": 4.5336910277482155e-07,
"loss": -0.0088,
"reward": 0.34240800654515624,
"reward_std": 0.7838171310722828,
"rewards/cosine_scaled_reward": 0.005472499451570911,
"rewards/format_reward": 0.7083333432674408,
"step": 306
},
{
"completion_length": 2925.625045776367,
"epoch": 0.35085714285714287,
"grad_norm": 1.742169737815857,
"kl": 0.31103515625,
"learning_rate": 4.503031760712397e-07,
"loss": 0.0824,
"reward": 0.24728509783744812,
"reward_std": 1.0561788342893124,
"rewards/cosine_scaled_reward": -0.03675899375230074,
"rewards/format_reward": 0.625000013038516,
"step": 307
},
{
"completion_length": 3307.916748046875,
"epoch": 0.352,
"grad_norm": 0.5225728154182434,
"kl": 0.390625,
"learning_rate": 4.4724210845020494e-07,
"loss": 0.0089,
"reward": 0.06603662599809468,
"reward_std": 0.6128693893551826,
"rewards/cosine_scaled_reward": -0.17313291411846876,
"rewards/format_reward": 0.7083333414047956,
"step": 308
},
{
"completion_length": 2979.479217529297,
"epoch": 0.35314285714285715,
"grad_norm": 0.4211791157722473,
"kl": 0.259429931640625,
"learning_rate": 4.441860491038345e-07,
"loss": 0.0147,
"reward": 0.2866139570251107,
"reward_std": 0.7773850671947002,
"rewards/cosine_scaled_reward": -0.009212411940097809,
"rewards/format_reward": 0.6666666753590107,
"step": 309
},
{
"completion_length": 2679.1667404174805,
"epoch": 0.35428571428571426,
"grad_norm": 0.5345349907875061,
"kl": 0.33343505859375,
"learning_rate": 4.4113514698014953e-07,
"loss": 0.0432,
"reward": 0.24587753415107727,
"reward_std": 0.8905804492533207,
"rewards/cosine_scaled_reward": -0.016845128498971462,
"rewards/format_reward": 0.6041666809469461,
"step": 310
},
{
"completion_length": 2646.2500610351562,
"epoch": 0.3554285714285714,
"grad_norm": 0.32929864525794983,
"kl": 0.261383056640625,
"learning_rate": 4.3808955077581546e-07,
"loss": 0.0258,
"reward": 0.1976035200059414,
"reward_std": 0.8284550718963146,
"rewards/cosine_scaled_reward": -0.04515882022678852,
"rewards/format_reward": 0.6041666716337204,
"step": 311
},
{
"completion_length": 2335.729217529297,
"epoch": 0.3565714285714286,
"grad_norm": 0.296840637922287,
"kl": 0.254730224609375,
"learning_rate": 4.350494089288943e-07,
"loss": 0.0257,
"reward": 0.5823600944131613,
"reward_std": 0.691787600517273,
"rewards/cosine_scaled_reward": 0.1733840461820364,
"rewards/format_reward": 0.708333333954215,
"step": 312
},
{
"completion_length": 2918.68758392334,
"epoch": 0.3577142857142857,
"grad_norm": 0.8261905312538147,
"kl": 0.3216552734375,
"learning_rate": 4.3201486961161093e-07,
"loss": 0.0615,
"reward": 0.32681620866060257,
"reward_std": 0.8109251782298088,
"rewards/cosine_scaled_reward": 0.06859785690903664,
"rewards/format_reward": 0.5625000149011612,
"step": 313
},
{
"completion_length": 2584.812545776367,
"epoch": 0.3588571428571429,
"grad_norm": 0.28141534328460693,
"kl": 0.24951171875,
"learning_rate": 4.2898608072313045e-07,
"loss": -0.0017,
"reward": 0.5454013433773071,
"reward_std": 0.7297746650874615,
"rewards/cosine_scaled_reward": 0.13757295534014702,
"rewards/format_reward": 0.729166679084301,
"step": 314
},
{
"completion_length": 3015.229232788086,
"epoch": 0.36,
"grad_norm": 0.40665552020072937,
"kl": 0.357452392578125,
"learning_rate": 4.2596318988235037e-07,
"loss": 0.0237,
"reward": 0.010891908779740334,
"reward_std": 0.7054996266961098,
"rewards/cosine_scaled_reward": -0.11506736988667399,
"rewards/format_reward": 0.5000000074505806,
"step": 315
},
{
"completion_length": 3297.6875610351562,
"epoch": 0.36114285714285715,
"grad_norm": 0.38543111085891724,
"kl": 0.36572265625,
"learning_rate": 4.2294634442070553e-07,
"loss": 0.0324,
"reward": -0.13673021260183305,
"reward_std": 0.5753876939415932,
"rewards/cosine_scaled_reward": -0.2870428040623665,
"rewards/format_reward": 0.6666666865348816,
"step": 316
},
{
"completion_length": 3066.8958892822266,
"epoch": 0.36228571428571427,
"grad_norm": 0.5159828066825867,
"kl": 0.4085693359375,
"learning_rate": 4.1993569137498776e-07,
"loss": 0.0244,
"reward": 0.13786310516297817,
"reward_std": 0.6435285620391369,
"rewards/cosine_scaled_reward": -0.11537853349000216,
"rewards/format_reward": 0.6875000279396772,
"step": 317
},
{
"completion_length": 2283.9375610351562,
"epoch": 0.36342857142857143,
"grad_norm": 0.4553980827331543,
"kl": 0.24249267578125,
"learning_rate": 4.1693137748017915e-07,
"loss": -0.0024,
"reward": 0.41944058798253536,
"reward_std": 0.9410093426704407,
"rewards/cosine_scaled_reward": -0.0460501410998404,
"rewards/format_reward": 0.8958333507180214,
"step": 318
},
{
"completion_length": 3003.750030517578,
"epoch": 0.36457142857142855,
"grad_norm": 0.3188877999782562,
"kl": 0.35693359375,
"learning_rate": 4.1393354916230005e-07,
"loss": 0.0374,
"reward": -0.02601405733730644,
"reward_std": 0.7571947351098061,
"rewards/cosine_scaled_reward": -0.16905247140675783,
"rewards/format_reward": 0.5625000111758709,
"step": 319
},
{
"completion_length": 2351.791748046875,
"epoch": 0.3657142857142857,
"grad_norm": 0.28264015913009644,
"kl": 0.226318359375,
"learning_rate": 4.1094235253127374e-07,
"loss": 0.0313,
"reward": 0.41428154334425926,
"reward_std": 0.6797180473804474,
"rewards/cosine_scaled_reward": -0.014672968536615372,
"rewards/format_reward": 0.8541666753590107,
"step": 320
},
{
"completion_length": 2420.500099182129,
"epoch": 0.3668571428571429,
"grad_norm": 0.3555741012096405,
"kl": 0.2991943359375,
"learning_rate": 4.079579333738039e-07,
"loss": 0.0064,
"reward": 0.3246429590508342,
"reward_std": 0.84641108289361,
"rewards/cosine_scaled_reward": -0.021552213234826922,
"rewards/format_reward": 0.7291666902601719,
"step": 321
},
{
"completion_length": 2859.395950317383,
"epoch": 0.368,
"grad_norm": 0.3224516808986664,
"kl": 0.27166748046875,
"learning_rate": 4.0498043714627006e-07,
"loss": 0.013,
"reward": 0.41165721049765125,
"reward_std": 0.8220065757632256,
"rewards/cosine_scaled_reward": 0.0005064834840595722,
"rewards/format_reward": 0.8125000149011612,
"step": 322
},
{
"completion_length": 2920.3125610351562,
"epoch": 0.36914285714285716,
"grad_norm": 0.7611133456230164,
"kl": 0.2857666015625,
"learning_rate": 4.020100089676376e-07,
"loss": 0.0486,
"reward": 0.29905556747689843,
"reward_std": 0.7838596850633621,
"rewards/cosine_scaled_reward": -0.0440771235153079,
"rewards/format_reward": 0.7500000223517418,
"step": 323
},
{
"completion_length": 3198.3334045410156,
"epoch": 0.3702857142857143,
"grad_norm": 0.5829113125801086,
"kl": 0.34033203125,
"learning_rate": 3.9904679361238526e-07,
"loss": 0.035,
"reward": 0.06515605933964252,
"reward_std": 0.6232591606676579,
"rewards/cosine_scaled_reward": -0.12263932824134827,
"rewards/format_reward": 0.6041666734963655,
"step": 324
},
{
"completion_length": 2887.104202270508,
"epoch": 0.37142857142857144,
"grad_norm": 0.3115164339542389,
"kl": 0.286163330078125,
"learning_rate": 3.9609093550344907e-07,
"loss": 0.0308,
"reward": 0.04865055438131094,
"reward_std": 0.769499409943819,
"rewards/cosine_scaled_reward": -0.14275423251092434,
"rewards/format_reward": 0.6041666716337204,
"step": 325
},
{
"completion_length": 2667.5834045410156,
"epoch": 0.37257142857142855,
"grad_norm": 0.2706601023674011,
"kl": 0.24713134765625,
"learning_rate": 3.931425787051832e-07,
"loss": 0.0352,
"reward": 0.3725670697167516,
"reward_std": 0.7936568856239319,
"rewards/cosine_scaled_reward": -0.0039613001281395555,
"rewards/format_reward": 0.7708333395421505,
"step": 326
},
{
"completion_length": 2546.291702270508,
"epoch": 0.3737142857142857,
"grad_norm": 0.2914826273918152,
"kl": 0.26849365234375,
"learning_rate": 3.902018669163384e-07,
"loss": 0.0326,
"reward": 0.36802724679000676,
"reward_std": 0.7584006376564503,
"rewards/cosine_scaled_reward": 0.046540172887034714,
"rewards/format_reward": 0.6666666772216558,
"step": 327
},
{
"completion_length": 3366.104217529297,
"epoch": 0.37485714285714283,
"grad_norm": 0.42033851146698,
"kl": 0.346923828125,
"learning_rate": 3.872689434630585e-07,
"loss": 0.04,
"reward": -0.22989960946142673,
"reward_std": 0.757425207644701,
"rewards/cosine_scaled_reward": -0.22673101257532835,
"rewards/format_reward": 0.3958333469927311,
"step": 328
},
{
"completion_length": 2347.7500534057617,
"epoch": 0.376,
"grad_norm": 0.8453741073608398,
"kl": 0.227630615234375,
"learning_rate": 3.843439512918949e-07,
"loss": 0.0358,
"reward": 0.4706056764116511,
"reward_std": 0.8056343197822571,
"rewards/cosine_scaled_reward": 0.021218265290372074,
"rewards/format_reward": 0.8541666865348816,
"step": 329
},
{
"completion_length": 2407.020866394043,
"epoch": 0.37714285714285717,
"grad_norm": 0.7580931782722473,
"kl": 0.237945556640625,
"learning_rate": 3.8142703296283953e-07,
"loss": -0.0101,
"reward": 0.12094018794596195,
"reward_std": 0.6862670034170151,
"rewards/cosine_scaled_reward": -0.19688204117119312,
"rewards/format_reward": 0.8125000018626451,
"step": 330
},
{
"completion_length": 2858.2083740234375,
"epoch": 0.3782857142857143,
"grad_norm": 0.528272271156311,
"kl": 0.30035400390625,
"learning_rate": 3.785183306423767e-07,
"loss": 0.0324,
"reward": -0.01362695125862956,
"reward_std": 0.7233411334455013,
"rewards/cosine_scaled_reward": -0.1397905834019184,
"rewards/format_reward": 0.5208333414047956,
"step": 331
},
{
"completion_length": 2584.729217529297,
"epoch": 0.37942857142857145,
"grad_norm": 0.6263169050216675,
"kl": 0.2864990234375,
"learning_rate": 3.7561798609655373e-07,
"loss": 0.0382,
"reward": 0.33939856104552746,
"reward_std": 0.8283688835799694,
"rewards/cosine_scaled_reward": 0.012080416432581842,
"rewards/format_reward": 0.6875000149011612,
"step": 332
},
{
"completion_length": 2398.562545776367,
"epoch": 0.38057142857142856,
"grad_norm": 0.2747509777545929,
"kl": 0.20452880859375,
"learning_rate": 3.72726140684072e-07,
"loss": 0.0178,
"reward": 0.3634027219377458,
"reward_std": 0.7900962755084038,
"rewards/cosine_scaled_reward": -0.07603580364957452,
"rewards/format_reward": 0.8958333507180214,
"step": 333
},
{
"completion_length": 3390.7500915527344,
"epoch": 0.38171428571428573,
"grad_norm": 0.6542767882347107,
"kl": 0.46240234375,
"learning_rate": 3.6984293534939737e-07,
"loss": 0.0728,
"reward": -0.052195049822330475,
"reward_std": 0.894780658185482,
"rewards/cosine_scaled_reward": -0.17814880423247814,
"rewards/format_reward": 0.5208333414047956,
"step": 334
},
{
"completion_length": 2516.604217529297,
"epoch": 0.38285714285714284,
"grad_norm": 0.2772899866104126,
"kl": 0.257080078125,
"learning_rate": 3.6696851061588994e-07,
"loss": 0.0318,
"reward": 0.1776880531979259,
"reward_std": 0.7250074371695518,
"rewards/cosine_scaled_reward": -0.06235098314937204,
"rewards/format_reward": 0.6250000093132257,
"step": 335
},
{
"completion_length": 3047.0000915527344,
"epoch": 0.384,
"grad_norm": 1.0677460432052612,
"kl": 0.34222412109375,
"learning_rate": 3.641030065789562e-07,
"loss": 0.0384,
"reward": 0.3056840244680643,
"reward_std": 0.8023335263133049,
"rewards/cosine_scaled_reward": -0.030411685816943645,
"rewards/format_reward": 0.729166692122817,
"step": 336
},
{
"completion_length": 2612.916748046875,
"epoch": 0.3851428571428571,
"grad_norm": 1.6512010097503662,
"kl": 0.316650390625,
"learning_rate": 3.612465628992203e-07,
"loss": 0.0747,
"reward": 0.456050219014287,
"reward_std": 1.0556022450327873,
"rewards/cosine_scaled_reward": 0.01445878017693758,
"rewards/format_reward": 0.8125000223517418,
"step": 337
},
{
"completion_length": 2305.4791870117188,
"epoch": 0.3862857142857143,
"grad_norm": 0.9437749981880188,
"kl": 0.30230712890625,
"learning_rate": 3.5839931879571725e-07,
"loss": -0.0135,
"reward": 0.5471778312930837,
"reward_std": 0.794790405780077,
"rewards/cosine_scaled_reward": 0.12062014266848564,
"rewards/format_reward": 0.75,
"step": 338
},
{
"completion_length": 3010.1250534057617,
"epoch": 0.38742857142857146,
"grad_norm": 0.5747052431106567,
"kl": 0.360382080078125,
"learning_rate": 3.555614130391079e-07,
"loss": 0.0191,
"reward": -0.03356679296121001,
"reward_std": 0.7060465253889561,
"rewards/cosine_scaled_reward": -0.17198694869875908,
"rewards/format_reward": 0.5625000111758709,
"step": 339
},
{
"completion_length": 2871.2709045410156,
"epoch": 0.38857142857142857,
"grad_norm": 0.5294919610023499,
"kl": 0.3682861328125,
"learning_rate": 3.5273298394491515e-07,
"loss": 0.0492,
"reward": 0.29094321094453335,
"reward_std": 0.7570115067064762,
"rewards/cosine_scaled_reward": -0.11235274095088243,
"rewards/format_reward": 0.8750000074505806,
"step": 340
},
{
"completion_length": 2639.062545776367,
"epoch": 0.38971428571428574,
"grad_norm": 0.45286688208580017,
"kl": 0.29937744140625,
"learning_rate": 3.4991416936678276e-07,
"loss": 0.0171,
"reward": 0.6249844692647457,
"reward_std": 0.815760787576437,
"rewards/cosine_scaled_reward": 0.17363815288990736,
"rewards/format_reward": 0.7500000093132257,
"step": 341
},
{
"completion_length": 2894.0208587646484,
"epoch": 0.39085714285714285,
"grad_norm": 0.3460341691970825,
"kl": 0.4681396484375,
"learning_rate": 3.471051066897562e-07,
"loss": 0.0579,
"reward": 0.08103763521648943,
"reward_std": 0.7632105089724064,
"rewards/cosine_scaled_reward": -0.1950855739414692,
"rewards/format_reward": 0.7500000074505806,
"step": 342
},
{
"completion_length": 2808.6250610351562,
"epoch": 0.392,
"grad_norm": 1.3520926237106323,
"kl": 0.3250732421875,
"learning_rate": 3.4430593282358777e-07,
"loss": 0.0791,
"reward": 0.48628374096006155,
"reward_std": 0.8851627111434937,
"rewards/cosine_scaled_reward": 0.07822982332436368,
"rewards/format_reward": 0.7500000167638063,
"step": 343
},
{
"completion_length": 2434.958366394043,
"epoch": 0.3931428571428571,
"grad_norm": 0.38433942198753357,
"kl": 0.2788543701171875,
"learning_rate": 3.4151678419606233e-07,
"loss": 0.0233,
"reward": 0.6341738551855087,
"reward_std": 0.7020405307412148,
"rewards/cosine_scaled_reward": 0.1561431773006916,
"rewards/format_reward": 0.8125000037252903,
"step": 344
},
{
"completion_length": 2946.3958892822266,
"epoch": 0.3942857142857143,
"grad_norm": 0.9588437080383301,
"kl": 0.4776611328125,
"learning_rate": 3.387377967463493e-07,
"loss": 0.0089,
"reward": 0.4796811621636152,
"reward_std": 0.8463685475289822,
"rewards/cosine_scaled_reward": 0.07671219296753407,
"rewards/format_reward": 0.7500000037252903,
"step": 345
},
{
"completion_length": 2810.916717529297,
"epoch": 0.3954285714285714,
"grad_norm": 0.7113232612609863,
"kl": 0.45458984375,
"learning_rate": 3.359691059183761e-07,
"loss": 0.0365,
"reward": 0.29816207382827997,
"reward_std": 0.8527140244841576,
"rewards/cosine_scaled_reward": -0.028450995916500688,
"rewards/format_reward": 0.7083333507180214,
"step": 346
},
{
"completion_length": 3077.9584045410156,
"epoch": 0.3965714285714286,
"grad_norm": 1.0087529420852661,
"kl": 0.5037841796875,
"learning_rate": 3.3321084665422803e-07,
"loss": 0.027,
"reward": -0.07710191514343023,
"reward_std": 0.6414642743766308,
"rewards/cosine_scaled_reward": -0.2412154171615839,
"rewards/format_reward": 0.6458333469927311,
"step": 347
},
{
"completion_length": 2701.1042709350586,
"epoch": 0.3977142857142857,
"grad_norm": 0.7987737059593201,
"kl": 0.3995208740234375,
"learning_rate": 3.3046315338757026e-07,
"loss": 0.0226,
"reward": 0.299939407967031,
"reward_std": 0.602581400424242,
"rewards/cosine_scaled_reward": -0.08414606470614672,
"rewards/format_reward": 0.8541666772216558,
"step": 348
},
{
"completion_length": 2987.229278564453,
"epoch": 0.39885714285714285,
"grad_norm": 1.26498544216156,
"kl": 0.451416015625,
"learning_rate": 3.2772616003709616e-07,
"loss": 0.0201,
"reward": 0.11506683845072985,
"reward_std": 0.7807445377111435,
"rewards/cosine_scaled_reward": -0.13993432931602,
"rewards/format_reward": 0.687500013038516,
"step": 349
},
{
"completion_length": 2590.395950317383,
"epoch": 0.4,
"grad_norm": 0.5970591902732849,
"kl": 0.4644775390625,
"learning_rate": 3.250000000000001e-07,
"loss": 0.0338,
"reward": 0.18042142933700234,
"reward_std": 0.8871416859328747,
"rewards/cosine_scaled_reward": -0.14772336441092193,
"rewards/format_reward": 0.7708333544433117,
"step": 350
},
{
"completion_length": 2721.6250762939453,
"epoch": 0.40114285714285713,
"grad_norm": 0.43472933769226074,
"kl": 0.25958251953125,
"learning_rate": 3.222848061454764e-07,
"loss": 0.0395,
"reward": 0.42602336849085987,
"reward_std": 0.8064659312367439,
"rewards/cosine_scaled_reward": 0.042032238095998764,
"rewards/format_reward": 0.7500000149011612,
"step": 351
},
{
"completion_length": 2447.8334045410156,
"epoch": 0.4022857142857143,
"grad_norm": 1.039471983909607,
"kl": 0.2555999755859375,
"learning_rate": 3.195807108082429e-07,
"loss": 0.0413,
"reward": 0.24063719739206135,
"reward_std": 0.8650816380977631,
"rewards/cosine_scaled_reward": -0.0003559635952115059,
"rewards/format_reward": 0.5625000093132257,
"step": 352
},
{
"completion_length": 2217.14591217041,
"epoch": 0.4034285714285714,
"grad_norm": 0.41167160868644714,
"kl": 0.203216552734375,
"learning_rate": 3.168878457820915e-07,
"loss": 0.0232,
"reward": 0.5322136869654059,
"reward_std": 0.666268203407526,
"rewards/cosine_scaled_reward": 0.04740488715469837,
"rewards/format_reward": 0.8958333507180214,
"step": 353
},
{
"completion_length": 2258.625045776367,
"epoch": 0.4045714285714286,
"grad_norm": 0.22829324007034302,
"kl": 0.180816650390625,
"learning_rate": 3.142063423134644e-07,
"loss": 0.0048,
"reward": 0.5726254731416702,
"reward_std": 0.7559525668621063,
"rewards/cosine_scaled_reward": 0.12823878531344235,
"rewards/format_reward": 0.7708333414047956,
"step": 354
},
{
"completion_length": 2250.791778564453,
"epoch": 0.4057142857142857,
"grad_norm": 0.602950394153595,
"kl": 0.2418212890625,
"learning_rate": 3.115363310950578e-07,
"loss": 0.0347,
"reward": 0.6632253341376781,
"reward_std": 0.8946371823549271,
"rewards/cosine_scaled_reward": 0.15655125584453344,
"rewards/format_reward": 0.8333333544433117,
"step": 355
},
{
"completion_length": 2813.9583892822266,
"epoch": 0.40685714285714286,
"grad_norm": 0.49410441517829895,
"kl": 0.27777099609375,
"learning_rate": 3.0887794225945143e-07,
"loss": 0.0194,
"reward": 0.21525360643863678,
"reward_std": 0.9108888022601604,
"rewards/cosine_scaled_reward": -0.058567093685269356,
"rewards/format_reward": 0.6458333469927311,
"step": 356
},
{
"completion_length": 3031.3751220703125,
"epoch": 0.408,
"grad_norm": 0.46776533126831055,
"kl": 0.290283203125,
"learning_rate": 3.062313053727671e-07,
"loss": 0.0666,
"reward": 0.11198169272392988,
"reward_std": 0.7659357041120529,
"rewards/cosine_scaled_reward": -0.17274215212091804,
"rewards/format_reward": 0.7500000149011612,
"step": 357
},
{
"completion_length": 2567.7709197998047,
"epoch": 0.40914285714285714,
"grad_norm": 0.9789031744003296,
"kl": 0.2848052978515625,
"learning_rate": 3.0359654942835247e-07,
"loss": 0.082,
"reward": 0.4794031195342541,
"reward_std": 0.9315466657280922,
"rewards/cosine_scaled_reward": 0.028890431160107255,
"rewards/format_reward": 0.8333333507180214,
"step": 358
},
{
"completion_length": 2519.666702270508,
"epoch": 0.4102857142857143,
"grad_norm": 0.3785220980644226,
"kl": 0.329376220703125,
"learning_rate": 3.0097380284049523e-07,
"loss": 0.0228,
"reward": 0.32482871878892183,
"reward_std": 0.804768543690443,
"rewards/cosine_scaled_reward": -0.028755411505699158,
"rewards/format_reward": 0.7500000186264515,
"step": 359
},
{
"completion_length": 2652.604263305664,
"epoch": 0.4114285714285714,
"grad_norm": 0.5460115075111389,
"kl": 0.22186279296875,
"learning_rate": 2.9836319343816397e-07,
"loss": 0.0358,
"reward": 0.5476951543241739,
"reward_std": 0.8413685485720634,
"rewards/cosine_scaled_reward": 0.047556648729369044,
"rewards/format_reward": 0.8958333432674408,
"step": 360
},
{
"completion_length": 2775.6251220703125,
"epoch": 0.4125714285714286,
"grad_norm": 0.6770034432411194,
"kl": 0.27569580078125,
"learning_rate": 2.9576484845877793e-07,
"loss": 0.0547,
"reward": 0.21503734902944416,
"reward_std": 0.7295170053839684,
"rewards/cosine_scaled_reward": -0.1001668400131166,
"rewards/format_reward": 0.7500000186264515,
"step": 361
},
{
"completion_length": 1685.1875305175781,
"epoch": 0.4137142857142857,
"grad_norm": 0.2777215838432312,
"kl": 0.158172607421875,
"learning_rate": 2.931788945420058e-07,
"loss": 0.0005,
"reward": 0.4802695903927088,
"reward_std": 0.6514372602105141,
"rewards/cosine_scaled_reward": 0.03301592729985714,
"rewards/format_reward": 0.8541666679084301,
"step": 362
},
{
"completion_length": 1911.145896911621,
"epoch": 0.41485714285714287,
"grad_norm": 0.5700949430465698,
"kl": 0.2712554931640625,
"learning_rate": 2.9060545772359305e-07,
"loss": 0.0486,
"reward": 0.3370086522772908,
"reward_std": 0.7742916233837605,
"rewards/cosine_scaled_reward": 0.0005665780045092106,
"rewards/format_reward": 0.7083333395421505,
"step": 363
},
{
"completion_length": 2883.229248046875,
"epoch": 0.416,
"grad_norm": 0.5325201749801636,
"kl": 0.36102294921875,
"learning_rate": 2.8804466342921987e-07,
"loss": 0.0226,
"reward": -0.025897093466483057,
"reward_std": 0.6142625138163567,
"rewards/cosine_scaled_reward": -0.23417398636229336,
"rewards/format_reward": 0.7083333469927311,
"step": 364
},
{
"completion_length": 2721.312545776367,
"epoch": 0.41714285714285715,
"grad_norm": 0.7317801713943481,
"kl": 0.36126708984375,
"learning_rate": 2.854966364683872e-07,
"loss": 0.0568,
"reward": 0.3746628388762474,
"reward_std": 0.8317583128809929,
"rewards/cosine_scaled_reward": 0.022838744800537825,
"rewards/format_reward": 0.7083333395421505,
"step": 365
},
{
"completion_length": 2259.645854949951,
"epoch": 0.41828571428571426,
"grad_norm": 0.5810126662254333,
"kl": 0.244354248046875,
"learning_rate": 2.829615010283344e-07,
"loss": 0.0253,
"reward": 0.4160766340792179,
"reward_std": 0.7030897587537766,
"rewards/cosine_scaled_reward": 0.050911818630993366,
"rewards/format_reward": 0.7291666828095913,
"step": 366
},
{
"completion_length": 3181.5209350585938,
"epoch": 0.41942857142857143,
"grad_norm": 0.4794647991657257,
"kl": 0.4725341796875,
"learning_rate": 2.8043938066798645e-07,
"loss": 0.0514,
"reward": 0.09267363836988807,
"reward_std": 0.7625463083386421,
"rewards/cosine_scaled_reward": -0.12190989265218377,
"rewards/format_reward": 0.6250000204890966,
"step": 367
},
{
"completion_length": 2849.666732788086,
"epoch": 0.4205714285714286,
"grad_norm": 1.0089714527130127,
"kl": 0.32720947265625,
"learning_rate": 2.7793039831193133e-07,
"loss": -0.0027,
"reward": 0.1751155611127615,
"reward_std": 0.549890723079443,
"rewards/cosine_scaled_reward": -0.07263889070600271,
"rewards/format_reward": 0.666666679084301,
"step": 368
},
{
"completion_length": 2855.750045776367,
"epoch": 0.4217142857142857,
"grad_norm": 0.9463785290718079,
"kl": 0.40081787109375,
"learning_rate": 2.7543467624442956e-07,
"loss": 0.0535,
"reward": 0.1500562410801649,
"reward_std": 0.7940640598535538,
"rewards/cosine_scaled_reward": -0.10133596323430538,
"rewards/format_reward": 0.666666679084301,
"step": 369
},
{
"completion_length": 2896.3959197998047,
"epoch": 0.4228571428571429,
"grad_norm": 0.8330612778663635,
"kl": 0.416717529296875,
"learning_rate": 2.729523361034538e-07,
"loss": 0.0286,
"reward": 0.183703294955194,
"reward_std": 0.45754858292639256,
"rewards/cosine_scaled_reward": -0.09295342303812504,
"rewards/format_reward": 0.7291666772216558,
"step": 370
},
{
"completion_length": 1955.479263305664,
"epoch": 0.424,
"grad_norm": 0.3391231894493103,
"kl": 0.262359619140625,
"learning_rate": 2.7048349887476037e-07,
"loss": 0.0184,
"reward": 0.6992178884102032,
"reward_std": 0.6993260197341442,
"rewards/cosine_scaled_reward": 0.16803688369691372,
"rewards/format_reward": 0.8750000149011612,
"step": 371
},
{
"completion_length": 3005.6458740234375,
"epoch": 0.42514285714285716,
"grad_norm": 0.5167466402053833,
"kl": 0.4355926513671875,
"learning_rate": 2.6802828488599294e-07,
"loss": 0.0432,
"reward": 0.19448892027139664,
"reward_std": 0.7890233434736729,
"rewards/cosine_scaled_reward": -0.021921467036008835,
"rewards/format_reward": 0.5625000093132257,
"step": 372
},
{
"completion_length": 1942.3333892822266,
"epoch": 0.42628571428571427,
"grad_norm": 0.3325159549713135,
"kl": 0.2429656982421875,
"learning_rate": 2.655868138008171e-07,
"loss": 0.0212,
"reward": 0.3014055141247809,
"reward_std": 0.8629297837615013,
"rewards/cosine_scaled_reward": -0.07299621542915702,
"rewards/format_reward": 0.791666679084301,
"step": 373
},
{
"completion_length": 2335.020866394043,
"epoch": 0.42742857142857144,
"grad_norm": 0.8563280701637268,
"kl": 0.3255615234375,
"learning_rate": 2.631592046130896e-07,
"loss": 0.0616,
"reward": 0.2297945898026228,
"reward_std": 0.9343635328114033,
"rewards/cosine_scaled_reward": -0.047990256920456886,
"rewards/format_reward": 0.6458333469927311,
"step": 374
},
{
"completion_length": 2628.291748046875,
"epoch": 0.42857142857142855,
"grad_norm": 0.9893895387649536,
"kl": 0.34588623046875,
"learning_rate": 2.6074557564105724e-07,
"loss": 0.0665,
"reward": 0.35648069988383213,
"reward_std": 0.7849867083132267,
"rewards/cosine_scaled_reward": 0.0042181313037872314,
"rewards/format_reward": 0.729166679084301,
"step": 375
},
{
"completion_length": 2365.6459045410156,
"epoch": 0.4297142857142857,
"grad_norm": 0.37569403648376465,
"kl": 0.351654052734375,
"learning_rate": 2.583460445215911e-07,
"loss": 0.0279,
"reward": 0.13466822169721127,
"reward_std": 0.7210676558315754,
"rewards/cosine_scaled_reward": -0.1441901307553053,
"rewards/format_reward": 0.7291666753590107,
"step": 376
},
{
"completion_length": 3061.979278564453,
"epoch": 0.4308571428571429,
"grad_norm": 0.7083445191383362,
"kl": 0.4854736328125,
"learning_rate": 2.5596072820445254e-07,
"loss": 0.0347,
"reward": 0.34864536579698324,
"reward_std": 1.0110743790864944,
"rewards/cosine_scaled_reward": 0.04069389193318784,
"rewards/format_reward": 0.6250000186264515,
"step": 377
},
{
"completion_length": 2377.3542709350586,
"epoch": 0.432,
"grad_norm": 0.49458223581314087,
"kl": 0.311187744140625,
"learning_rate": 2.5358974294659373e-07,
"loss": 0.0331,
"reward": 0.5616534340661019,
"reward_std": 0.7573869004845619,
"rewards/cosine_scaled_reward": 0.08385843969881535,
"rewards/format_reward": 0.8541666939854622,
"step": 378
},
{
"completion_length": 2865.729248046875,
"epoch": 0.43314285714285716,
"grad_norm": 1.2933725118637085,
"kl": 0.4803466796875,
"learning_rate": 2.512332043064913e-07,
"loss": 0.0196,
"reward": 0.3152011390775442,
"reward_std": 0.7045796066522598,
"rewards/cosine_scaled_reward": -0.010164887178689241,
"rewards/format_reward": 0.7083333507180214,
"step": 379
},
{
"completion_length": 2530.7708892822266,
"epoch": 0.4342857142857143,
"grad_norm": 0.6860670447349548,
"kl": 0.32769775390625,
"learning_rate": 2.488912271385139e-07,
"loss": 0.0408,
"reward": 0.2659956905990839,
"reward_std": 0.727360412478447,
"rewards/cosine_scaled_reward": -0.08539670892059803,
"rewards/format_reward": 0.7916666753590107,
"step": 380
},
{
"completion_length": 2727.3333587646484,
"epoch": 0.43542857142857144,
"grad_norm": 0.8145273923873901,
"kl": 0.439697265625,
"learning_rate": 2.465639255873246e-07,
"loss": 0.0334,
"reward": 0.08977367201441666,
"reward_std": 0.808791808784008,
"rewards/cosine_scaled_reward": -0.1487837778404355,
"rewards/format_reward": 0.6666666753590107,
"step": 381
},
{
"completion_length": 2468.104248046875,
"epoch": 0.43657142857142855,
"grad_norm": 0.42541611194610596,
"kl": 0.37420654296875,
"learning_rate": 2.4425141308231765e-07,
"loss": 0.0623,
"reward": 0.18184729665517807,
"reward_std": 0.8058411814272404,
"rewards/cosine_scaled_reward": -0.1598250768147409,
"rewards/format_reward": 0.8125000111758709,
"step": 382
},
{
"completion_length": 2856.6458892822266,
"epoch": 0.4377142857142857,
"grad_norm": 0.6056004762649536,
"kl": 0.4031982421875,
"learning_rate": 2.4195380233209006e-07,
"loss": 0.044,
"reward": 0.46744489343836904,
"reward_std": 0.9438849911093712,
"rewards/cosine_scaled_reward": 0.08236491866409779,
"rewards/format_reward": 0.7083333488553762,
"step": 383
},
{
"completion_length": 2167.7500762939453,
"epoch": 0.43885714285714283,
"grad_norm": 0.6828803420066833,
"kl": 0.2847900390625,
"learning_rate": 2.3967120531894857e-07,
"loss": 0.0052,
"reward": 0.8705862760543823,
"reward_std": 0.8881346955895424,
"rewards/cosine_scaled_reward": 0.32432539528235793,
"rewards/format_reward": 0.7708333432674408,
"step": 384
},
{
"completion_length": 2416.9167098999023,
"epoch": 0.44,
"grad_norm": 0.6516284942626953,
"kl": 0.360748291015625,
"learning_rate": 2.374037332934512e-07,
"loss": 0.0588,
"reward": 0.24877717718482018,
"reward_std": 0.735356081277132,
"rewards/cosine_scaled_reward": -0.08876009099185467,
"rewards/format_reward": 0.7708333432674408,
"step": 385
},
{
"completion_length": 2890.8959350585938,
"epoch": 0.44114285714285717,
"grad_norm": 0.6051266193389893,
"kl": 0.4339599609375,
"learning_rate": 2.3515149676898552e-07,
"loss": 0.0433,
"reward": 0.37795007787644863,
"reward_std": 0.8708282820880413,
"rewards/cosine_scaled_reward": 0.02761521004140377,
"rewards/format_reward": 0.708333358168602,
"step": 386
},
{
"completion_length": 2727.416778564453,
"epoch": 0.4422857142857143,
"grad_norm": 0.4502924680709839,
"kl": 0.392852783203125,
"learning_rate": 2.3291460551638237e-07,
"loss": 0.0409,
"reward": 0.40206454193685204,
"reward_std": 0.8078175410628319,
"rewards/cosine_scaled_reward": 0.03367285244166851,
"rewards/format_reward": 0.7291666716337204,
"step": 387
},
{
"completion_length": 2498.291732788086,
"epoch": 0.44342857142857145,
"grad_norm": 0.7455759048461914,
"kl": 0.421722412109375,
"learning_rate": 2.306931685585657e-07,
"loss": 0.0377,
"reward": 0.4461521580815315,
"reward_std": 0.6339214891195297,
"rewards/cosine_scaled_reward": 0.06488732434809208,
"rewards/format_reward": 0.7500000074505806,
"step": 388
},
{
"completion_length": 2519.2500610351562,
"epoch": 0.44457142857142856,
"grad_norm": 0.3315029740333557,
"kl": 0.347686767578125,
"learning_rate": 2.2848729416523859e-07,
"loss": 0.0373,
"reward": 0.41832172160502523,
"reward_std": 0.8497539162635803,
"rewards/cosine_scaled_reward": 0.04175483621656895,
"rewards/format_reward": 0.7291666846722364,
"step": 389
},
{
"completion_length": 2520.270881652832,
"epoch": 0.44571428571428573,
"grad_norm": 0.6555607914924622,
"kl": 0.3785400390625,
"learning_rate": 2.2629708984760706e-07,
"loss": 0.0494,
"reward": 0.36745146568864584,
"reward_std": 0.7271608784794807,
"rewards/cosine_scaled_reward": 0.005871989764273167,
"rewards/format_reward": 0.7500000223517418,
"step": 390
},
{
"completion_length": 2606.7916870117188,
"epoch": 0.44685714285714284,
"grad_norm": 0.5517706871032715,
"kl": 0.4510498046875,
"learning_rate": 2.2412266235313973e-07,
"loss": 0.0531,
"reward": 0.30924548767507076,
"reward_std": 0.7186228446662426,
"rewards/cosine_scaled_reward": 0.02813367173075676,
"rewards/format_reward": 0.6250000037252903,
"step": 391
},
{
"completion_length": 2368.2500762939453,
"epoch": 0.448,
"grad_norm": 1.1368578672409058,
"kl": 0.367889404296875,
"learning_rate": 2.2196411766036487e-07,
"loss": 0.0712,
"reward": 0.43708939105272293,
"reward_std": 0.9549101106822491,
"rewards/cosine_scaled_reward": -0.014195648953318596,
"rewards/format_reward": 0.8541666865348816,
"step": 392
},
{
"completion_length": 2708.354217529297,
"epoch": 0.4491428571428571,
"grad_norm": 1.338770866394043,
"kl": 0.411865234375,
"learning_rate": 2.1982156097370557e-07,
"loss": 0.0669,
"reward": 0.5190371284261346,
"reward_std": 1.0145683512091637,
"rewards/cosine_scaled_reward": 0.11480109271360561,
"rewards/format_reward": 0.708333358168602,
"step": 393
},
{
"completion_length": 3059.1250610351562,
"epoch": 0.4502857142857143,
"grad_norm": 0.698340117931366,
"kl": 0.5703125,
"learning_rate": 2.1769509671835223e-07,
"loss": 0.0558,
"reward": 0.08016422716900706,
"reward_std": 0.9071974717080593,
"rewards/cosine_scaled_reward": -0.10081264981999993,
"rewards/format_reward": 0.541666679084301,
"step": 394
},
{
"completion_length": 2238.604232788086,
"epoch": 0.4514285714285714,
"grad_norm": 0.527318000793457,
"kl": 0.3275146484375,
"learning_rate": 2.1558482853517253e-07,
"loss": 0.0224,
"reward": 0.518818385200575,
"reward_std": 0.812805999070406,
"rewards/cosine_scaled_reward": 0.07729056139942259,
"rewards/format_reward": 0.7916666902601719,
"step": 395
},
{
"completion_length": 2808.7084045410156,
"epoch": 0.45257142857142857,
"grad_norm": 1.1238527297973633,
"kl": 0.48876953125,
"learning_rate": 2.134908592756607e-07,
"loss": 0.1152,
"reward": 0.2978093853453174,
"reward_std": 0.9074007868766785,
"rewards/cosine_scaled_reward": -0.03470429126173258,
"rewards/format_reward": 0.7083333507180214,
"step": 396
},
{
"completion_length": 2351.8126068115234,
"epoch": 0.45371428571428574,
"grad_norm": 0.3706618845462799,
"kl": 0.246002197265625,
"learning_rate": 2.1141329099692406e-07,
"loss": 0.0129,
"reward": 0.37504480965435505,
"reward_std": 0.5897877439856529,
"rewards/cosine_scaled_reward": -0.05014793388545513,
"rewards/format_reward": 0.8750000149011612,
"step": 397
},
{
"completion_length": 2328.37508392334,
"epoch": 0.45485714285714285,
"grad_norm": 0.7165157198905945,
"kl": 0.3564453125,
"learning_rate": 2.0935222495670968e-07,
"loss": 0.0288,
"reward": 0.3573523070663214,
"reward_std": 0.7765154354274273,
"rewards/cosine_scaled_reward": 0.0030241278000175953,
"rewards/format_reward": 0.729166679084301,
"step": 398
},
{
"completion_length": 2353.854202270508,
"epoch": 0.456,
"grad_norm": 1.2241944074630737,
"kl": 0.3509521484375,
"learning_rate": 2.0730776160846853e-07,
"loss": 0.0858,
"reward": 0.7341470178216696,
"reward_std": 1.0539017170667648,
"rewards/cosine_scaled_reward": 0.18186382483690977,
"rewards/format_reward": 0.8541666939854622,
"step": 399
},
{
"completion_length": 1577.7917098999023,
"epoch": 0.45714285714285713,
"grad_norm": 1.081485390663147,
"kl": 0.142333984375,
"learning_rate": 2.0528000059645995e-07,
"loss": -0.0309,
"reward": 0.879038143903017,
"reward_std": 0.9596639424562454,
"rewards/cosine_scaled_reward": 0.2600486520677805,
"rewards/format_reward": 0.8958333395421505,
"step": 400
},
{
"completion_length": 2888.479217529297,
"epoch": 0.4582857142857143,
"grad_norm": 0.948920726776123,
"kl": 0.559173583984375,
"learning_rate": 2.032690407508949e-07,
"loss": 0.0508,
"reward": 0.2996121197938919,
"reward_std": 0.6879791766405106,
"rewards/cosine_scaled_reward": -0.01863069087266922,
"rewards/format_reward": 0.7083333488553762,
"step": 401
},
{
"completion_length": 2353.104217529297,
"epoch": 0.4594285714285714,
"grad_norm": 0.6571682095527649,
"kl": 0.3966827392578125,
"learning_rate": 2.0127498008311922e-07,
"loss": 0.0259,
"reward": 0.2608861066401005,
"reward_std": 0.6378191784024239,
"rewards/cosine_scaled_reward": -0.031162479892373085,
"rewards/format_reward": 0.6875000167638063,
"step": 402
},
{
"completion_length": 2268.104232788086,
"epoch": 0.4605714285714286,
"grad_norm": 0.6540146470069885,
"kl": 0.3348388671875,
"learning_rate": 1.9929791578083655e-07,
"loss": 0.0243,
"reward": 0.3493347900584922,
"reward_std": 0.7477662637829781,
"rewards/cosine_scaled_reward": -0.0498207900673151,
"rewards/format_reward": 0.8333333432674408,
"step": 403
},
{
"completion_length": 2252.0625610351562,
"epoch": 0.4617142857142857,
"grad_norm": 0.5845898985862732,
"kl": 0.31449127197265625,
"learning_rate": 1.9733794420337213e-07,
"loss": 0.0223,
"reward": 0.3203935632482171,
"reward_std": 0.5589652694761753,
"rewards/cosine_scaled_reward": -0.0724391471594572,
"rewards/format_reward": 0.8541666716337204,
"step": 404
},
{
"completion_length": 2176.437557220459,
"epoch": 0.46285714285714286,
"grad_norm": 0.5805149078369141,
"kl": 0.4361114501953125,
"learning_rate": 1.9539516087697517e-07,
"loss": 0.0716,
"reward": 0.5914140390232205,
"reward_std": 0.9176547713577747,
"rewards/cosine_scaled_reward": 0.1673335493542254,
"rewards/format_reward": 0.7083333469927311,
"step": 405
},
{
"completion_length": 2222.3125762939453,
"epoch": 0.464,
"grad_norm": 0.714565098285675,
"kl": 0.348388671875,
"learning_rate": 1.934696604901642e-07,
"loss": 0.053,
"reward": 0.5401954464614391,
"reward_std": 0.9346184208989143,
"rewards/cosine_scaled_reward": 0.03536188416182995,
"rewards/format_reward": 0.8958333432674408,
"step": 406
},
{
"completion_length": 2257.3542404174805,
"epoch": 0.46514285714285714,
"grad_norm": 0.7260815501213074,
"kl": 0.316375732421875,
"learning_rate": 1.915615368891117e-07,
"loss": 0.0606,
"reward": 0.6055942573584616,
"reward_std": 0.6878091357648373,
"rewards/cosine_scaled_reward": 0.12408588983817026,
"rewards/format_reward": 0.8333333507180214,
"step": 407
},
{
"completion_length": 2682.041717529297,
"epoch": 0.4662857142857143,
"grad_norm": 0.7148492336273193,
"kl": 0.502105712890625,
"learning_rate": 1.8967088307307e-07,
"loss": 0.0441,
"reward": 0.5293791117146611,
"reward_std": 0.863475501537323,
"rewards/cosine_scaled_reward": 0.14691544696688652,
"rewards/format_reward": 0.6666666716337204,
"step": 408
},
{
"completion_length": 3076.0834045410156,
"epoch": 0.4674285714285714,
"grad_norm": 0.8249404430389404,
"kl": 0.6727294921875,
"learning_rate": 1.8779779118983867e-07,
"loss": 0.0616,
"reward": 0.002814117819070816,
"reward_std": 0.8114083893597126,
"rewards/cosine_scaled_reward": -0.10041821748018265,
"rewards/format_reward": 0.4583333395421505,
"step": 409
},
{
"completion_length": 2465.000087738037,
"epoch": 0.4685714285714286,
"grad_norm": 0.6067022681236267,
"kl": 0.5772857666015625,
"learning_rate": 1.8594235253127372e-07,
"loss": 0.0528,
"reward": 0.12211680319160223,
"reward_std": 0.7772498056292534,
"rewards/cosine_scaled_reward": -0.13637491036206484,
"rewards/format_reward": 0.6875000093132257,
"step": 410
},
{
"completion_length": 2848.7708740234375,
"epoch": 0.4697142857142857,
"grad_norm": 0.5674929022789001,
"kl": 0.5439453125,
"learning_rate": 1.8410465752883758e-07,
"loss": 0.0797,
"reward": 0.23797017033211887,
"reward_std": 0.8856572322547436,
"rewards/cosine_scaled_reward": -0.07174723839852959,
"rewards/format_reward": 0.7083333507180214,
"step": 411
},
{
"completion_length": 2702.9375610351562,
"epoch": 0.47085714285714286,
"grad_norm": 0.9189234375953674,
"kl": 0.4500732421875,
"learning_rate": 1.822847957491922e-07,
"loss": 0.025,
"reward": 0.47001307643949986,
"reward_std": 0.9889923110604286,
"rewards/cosine_scaled_reward": 0.04870818182826042,
"rewards/format_reward": 0.770833358168602,
"step": 412
},
{
"completion_length": 2750.1250762939453,
"epoch": 0.472,
"grad_norm": 0.7404162287712097,
"kl": 0.49786376953125,
"learning_rate": 1.804828558898332e-07,
"loss": 0.0398,
"reward": 0.37627727701328695,
"reward_std": 0.7015467956662178,
"rewards/cosine_scaled_reward": -0.021446891129016876,
"rewards/format_reward": 0.8125000298023224,
"step": 413
},
{
"completion_length": 3040.6459350585938,
"epoch": 0.47314285714285714,
"grad_norm": 1.531003475189209,
"kl": 0.66845703125,
"learning_rate": 1.7869892577476722e-07,
"loss": 0.0372,
"reward": -0.023312292993068695,
"reward_std": 0.6657886579632759,
"rewards/cosine_scaled_reward": -0.20822268491610885,
"rewards/format_reward": 0.6458333507180214,
"step": 414
},
{
"completion_length": 3178.354248046875,
"epoch": 0.4742857142857143,
"grad_norm": 1.0874366760253906,
"kl": 0.808349609375,
"learning_rate": 1.7693309235023127e-07,
"loss": 0.0745,
"reward": -0.028731117257848382,
"reward_std": 0.8625498786568642,
"rewards/cosine_scaled_reward": -0.1594004575163126,
"rewards/format_reward": 0.5208333395421505,
"step": 415
},
{
"completion_length": 2240.5625610351562,
"epoch": 0.4754285714285714,
"grad_norm": 1.3929944038391113,
"kl": 0.28546142578125,
"learning_rate": 1.7518544168045524e-07,
"loss": 0.0775,
"reward": 0.5215831075329334,
"reward_std": 0.8785289078950882,
"rewards/cosine_scaled_reward": 0.049687013030052185,
"rewards/format_reward": 0.8541666939854622,
"step": 416
},
{
"completion_length": 3151.9375915527344,
"epoch": 0.4765714285714286,
"grad_norm": 0.6912859678268433,
"kl": 0.676025390625,
"learning_rate": 1.7345605894346726e-07,
"loss": 0.0716,
"reward": -0.13146568089723587,
"reward_std": 0.7075865548104048,
"rewards/cosine_scaled_reward": -0.1975945346057415,
"rewards/format_reward": 0.47916668467223644,
"step": 417
},
{
"completion_length": 2078.3542289733887,
"epoch": 0.4777142857142857,
"grad_norm": 1.4748533964157104,
"kl": 0.241912841796875,
"learning_rate": 1.7174502842694212e-07,
"loss": -0.0234,
"reward": 0.6464557002764195,
"reward_std": 0.8331911526620388,
"rewards/cosine_scaled_reward": 0.1137741282582283,
"rewards/format_reward": 0.8958333432674408,
"step": 418
},
{
"completion_length": 2601.4584197998047,
"epoch": 0.47885714285714287,
"grad_norm": 0.4618014395236969,
"kl": 0.39984130859375,
"learning_rate": 1.7005243352409333e-07,
"loss": 0.0497,
"reward": 0.6187925288686529,
"reward_std": 0.9556396976113319,
"rewards/cosine_scaled_reward": 0.1003096466884017,
"rewards/format_reward": 0.8750000223517418,
"step": 419
},
{
"completion_length": 2392.33341217041,
"epoch": 0.48,
"grad_norm": 0.7824205160140991,
"kl": 0.47357177734375,
"learning_rate": 1.6837835672960831e-07,
"loss": 0.0699,
"reward": 0.2953487314807717,
"reward_std": 0.6610889099538326,
"rewards/cosine_scaled_reward": -0.03334581479430199,
"rewards/format_reward": 0.729166679084301,
"step": 420
},
{
"completion_length": 3098.0834045410156,
"epoch": 0.48114285714285715,
"grad_norm": 1.4201315641403198,
"kl": 0.530029296875,
"learning_rate": 1.6672287963562852e-07,
"loss": 0.0068,
"reward": -0.07355257519520819,
"reward_std": 0.6699985489249229,
"rewards/cosine_scaled_reward": -0.20630090683698654,
"rewards/format_reward": 0.583333345130086,
"step": 421
},
{
"completion_length": 2704.5833892822266,
"epoch": 0.48228571428571426,
"grad_norm": 0.31716275215148926,
"kl": 0.3623046875,
"learning_rate": 1.6508608292777203e-07,
"loss": 0.0337,
"reward": 0.30903656780719757,
"reward_std": 0.791595920920372,
"rewards/cosine_scaled_reward": -0.028064551996067166,
"rewards/format_reward": 0.7291666865348816,
"step": 422
},
{
"completion_length": 2378.3334197998047,
"epoch": 0.48342857142857143,
"grad_norm": 0.561937153339386,
"kl": 0.29791259765625,
"learning_rate": 1.6346804638120098e-07,
"loss": -0.0015,
"reward": 0.24677492817863822,
"reward_std": 0.749111071228981,
"rewards/cosine_scaled_reward": -0.10133864358067513,
"rewards/format_reward": 0.7916666753590107,
"step": 423
},
{
"completion_length": 2921.854202270508,
"epoch": 0.4845714285714286,
"grad_norm": 0.8485268950462341,
"kl": 0.461669921875,
"learning_rate": 1.6186884885673413e-07,
"loss": 0.0307,
"reward": 0.025166813982650638,
"reward_std": 0.8601982519030571,
"rewards/cosine_scaled_reward": -0.17313844989985228,
"rewards/format_reward": 0.6250000204890966,
"step": 424
},
{
"completion_length": 2087.3125762939453,
"epoch": 0.4857142857142857,
"grad_norm": 1.4564896821975708,
"kl": 0.282867431640625,
"learning_rate": 1.6028856829700258e-07,
"loss": 0.0946,
"reward": 0.7901451410725713,
"reward_std": 1.1320747658610344,
"rewards/cosine_scaled_reward": 0.2687780649284832,
"rewards/format_reward": 0.7500000111758709,
"step": 425
},
{
"completion_length": 2330.562599182129,
"epoch": 0.4868571428571429,
"grad_norm": 0.6082496047019958,
"kl": 0.316375732421875,
"learning_rate": 1.5872728172265146e-07,
"loss": 0.0174,
"reward": 0.3098383769392967,
"reward_std": 0.7229567170143127,
"rewards/cosine_scaled_reward": -0.0652941414155066,
"rewards/format_reward": 0.8125000186264515,
"step": 426
},
{
"completion_length": 2724.479217529297,
"epoch": 0.488,
"grad_norm": 0.6990248560905457,
"kl": 0.45989990234375,
"learning_rate": 1.5718506522858572e-07,
"loss": 0.0519,
"reward": 0.12407399946823716,
"reward_std": 0.8184213675558567,
"rewards/cosine_scaled_reward": -0.0571793733688537,
"rewards/format_reward": 0.541666679084301,
"step": 427
},
{
"completion_length": 2934.5834045410156,
"epoch": 0.48914285714285716,
"grad_norm": 0.8835780620574951,
"kl": 0.461181640625,
"learning_rate": 1.5566199398026147e-07,
"loss": 0.0673,
"reward": 0.10110859386622906,
"reward_std": 0.9170898050069809,
"rewards/cosine_scaled_reward": -0.1494620693847537,
"rewards/format_reward": 0.6666666902601719,
"step": 428
},
{
"completion_length": 2098.0208892822266,
"epoch": 0.49028571428571427,
"grad_norm": 0.4786301255226135,
"kl": 0.250732421875,
"learning_rate": 1.5415814221002265e-07,
"loss": 0.0115,
"reward": 0.21683326549828053,
"reward_std": 0.8974504359066486,
"rewards/cosine_scaled_reward": -0.14522514073178172,
"rewards/format_reward": 0.8125000149011612,
"step": 429
},
{
"completion_length": 2277.187530517578,
"epoch": 0.49142857142857144,
"grad_norm": 0.34657105803489685,
"kl": 0.2308349609375,
"learning_rate": 1.5267358321348285e-07,
"loss": 0.0271,
"reward": 0.20404995302669704,
"reward_std": 0.6838134452700615,
"rewards/cosine_scaled_reward": -0.08210572600364685,
"rewards/format_reward": 0.7083333469927311,
"step": 430
},
{
"completion_length": 2592.3125915527344,
"epoch": 0.49257142857142855,
"grad_norm": 0.40773648023605347,
"kl": 0.3669281005859375,
"learning_rate": 1.5120838934595337e-07,
"loss": 0.0348,
"reward": -0.10748124029487371,
"reward_std": 0.5141724869608879,
"rewards/cosine_scaled_reward": -0.23057417757809162,
"rewards/format_reward": 0.6041666697710752,
"step": 431
},
{
"completion_length": 2727.416748046875,
"epoch": 0.4937142857142857,
"grad_norm": 0.3802461326122284,
"kl": 0.365966796875,
"learning_rate": 1.4976263201891613e-07,
"loss": 0.0608,
"reward": 0.1675979122519493,
"reward_std": 0.601953960955143,
"rewards/cosine_scaled_reward": -0.07371543161571026,
"rewards/format_reward": 0.6458333376795053,
"step": 432
},
{
"completion_length": 2820.9791870117188,
"epoch": 0.4948571428571429,
"grad_norm": 0.9824896454811096,
"kl": 0.38507080078125,
"learning_rate": 1.483363816965435e-07,
"loss": 0.0532,
"reward": 0.24698512954637408,
"reward_std": 0.9584920853376389,
"rewards/cosine_scaled_reward": -0.006244649179279804,
"rewards/format_reward": 0.5833333507180214,
"step": 433
},
{
"completion_length": 2987.625045776367,
"epoch": 0.496,
"grad_norm": 1.5248559713363647,
"kl": 0.41156005859375,
"learning_rate": 1.469297078922642e-07,
"loss": 0.0052,
"reward": -0.08143354021012783,
"reward_std": 0.48334776237607,
"rewards/cosine_scaled_reward": -0.25419083051383495,
"rewards/format_reward": 0.6875000037252903,
"step": 434
},
{
"completion_length": 2147.1250228881836,
"epoch": 0.49714285714285716,
"grad_norm": 0.3896443545818329,
"kl": 0.2529144287109375,
"learning_rate": 1.4554267916537495e-07,
"loss": 0.0041,
"reward": 0.21179522573947906,
"reward_std": 0.6268462352454662,
"rewards/cosine_scaled_reward": -0.12806864827871323,
"rewards/format_reward": 0.8125000111758709,
"step": 435
},
{
"completion_length": 2244.6042251586914,
"epoch": 0.4982857142857143,
"grad_norm": 1.1815028190612793,
"kl": 0.34857177734375,
"learning_rate": 1.4417536311769885e-07,
"loss": -0.0171,
"reward": 0.31779580656439066,
"reward_std": 0.8181377947330475,
"rewards/cosine_scaled_reward": -0.02664483431726694,
"rewards/format_reward": 0.7291666772216558,
"step": 436
},
{
"completion_length": 2895.5834045410156,
"epoch": 0.49942857142857144,
"grad_norm": 0.6824524402618408,
"kl": 0.3638916015625,
"learning_rate": 1.4282782639029128e-07,
"loss": 0.053,
"reward": 0.1753413761034608,
"reward_std": 0.7372228689491749,
"rewards/cosine_scaled_reward": -0.11190539970993996,
"rewards/format_reward": 0.7291666828095913,
"step": 437
},
{
"completion_length": 2927.729232788086,
"epoch": 0.5005714285714286,
"grad_norm": 0.5597608685493469,
"kl": 0.362335205078125,
"learning_rate": 1.4150013466019114e-07,
"loss": 0.0319,
"reward": 0.1679457863792777,
"reward_std": 0.7294919807463884,
"rewards/cosine_scaled_reward": -0.09090281999669969,
"rewards/format_reward": 0.666666679084301,
"step": 438
},
{
"completion_length": 2691.5833740234375,
"epoch": 0.5017142857142857,
"grad_norm": 0.8128771781921387,
"kl": 0.296539306640625,
"learning_rate": 1.4019235263722034e-07,
"loss": 0.0088,
"reward": -0.08755548892077059,
"reward_std": 0.49066366255283356,
"rewards/cosine_scaled_reward": -0.1950508914887905,
"rewards/format_reward": 0.5625000074505806,
"step": 439
},
{
"completion_length": 3066.7292404174805,
"epoch": 0.5028571428571429,
"grad_norm": 1.3051230907440186,
"kl": 0.438079833984375,
"learning_rate": 1.3890454406082956e-07,
"loss": 0.0168,
"reward": -0.09066830575466156,
"reward_std": 0.5652249306440353,
"rewards/cosine_scaled_reward": -0.1592588908970356,
"rewards/format_reward": 0.4791666716337204,
"step": 440
},
{
"completion_length": 2851.2500610351562,
"epoch": 0.504,
"grad_norm": 0.5024367570877075,
"kl": 0.3612060546875,
"learning_rate": 1.3763677169699217e-07,
"loss": 0.0408,
"reward": 0.41364969685673714,
"reward_std": 0.8325144313275814,
"rewards/cosine_scaled_reward": 0.01983257569372654,
"rewards/format_reward": 0.7708333469927311,
"step": 441
},
{
"completion_length": 2756.0833892822266,
"epoch": 0.5051428571428571,
"grad_norm": 0.634526789188385,
"kl": 0.311614990234375,
"learning_rate": 1.3638909733514452e-07,
"loss": 0.0151,
"reward": 0.31202031567227095,
"reward_std": 0.8749045357108116,
"rewards/cosine_scaled_reward": -0.0003691236488521099,
"rewards/format_reward": 0.666666679084301,
"step": 442
},
{
"completion_length": 3059.8125534057617,
"epoch": 0.5062857142857143,
"grad_norm": 0.5289911031723022,
"kl": 0.453643798828125,
"learning_rate": 1.351615817851748e-07,
"loss": 0.0414,
"reward": 0.029406734742224216,
"reward_std": 0.8291610702872276,
"rewards/cosine_scaled_reward": -0.1101669161580503,
"rewards/format_reward": 0.500000013038516,
"step": 443
},
{
"completion_length": 2669.6876068115234,
"epoch": 0.5074285714285715,
"grad_norm": 1.0344839096069336,
"kl": 0.39227294921875,
"learning_rate": 1.3395428487445914e-07,
"loss": 0.0173,
"reward": 0.18073289468884468,
"reward_std": 0.8002647534012794,
"rewards/cosine_scaled_reward": -0.05391998961567879,
"rewards/format_reward": 0.6041666809469461,
"step": 444
},
{
"completion_length": 2744.3959045410156,
"epoch": 0.5085714285714286,
"grad_norm": 1.1817784309387207,
"kl": 0.28948974609375,
"learning_rate": 1.3276726544494571e-07,
"loss": 0.0885,
"reward": 0.130654014647007,
"reward_std": 0.9245680868625641,
"rewards/cosine_scaled_reward": -0.07616318017244339,
"rewards/format_reward": 0.562500013038516,
"step": 445
},
{
"completion_length": 2821.354217529297,
"epoch": 0.5097142857142857,
"grad_norm": 0.5576800107955933,
"kl": 0.2890625,
"learning_rate": 1.316005813502869e-07,
"loss": 0.0212,
"reward": 0.15429426170885563,
"reward_std": 0.7137273997068405,
"rewards/cosine_scaled_reward": -0.09584356285631657,
"rewards/format_reward": 0.6666666679084301,
"step": 446
},
{
"completion_length": 2405.583396911621,
"epoch": 0.5108571428571429,
"grad_norm": 0.211807519197464,
"kl": 0.246002197265625,
"learning_rate": 1.3045428945301953e-07,
"loss": 0.0362,
"reward": -0.02742259856313467,
"reward_std": 0.662248931825161,
"rewards/cosine_scaled_reward": -0.2382231242954731,
"rewards/format_reward": 0.708333345130086,
"step": 447
},
{
"completion_length": 2420.5834045410156,
"epoch": 0.512,
"grad_norm": 0.43571776151657104,
"kl": 0.217529296875,
"learning_rate": 1.2932844562179352e-07,
"loss": 0.0407,
"reward": 0.2491408372297883,
"reward_std": 0.7298646531999111,
"rewards/cosine_scaled_reward": -0.05479799164459109,
"rewards/format_reward": 0.7083333563059568,
"step": 448
},
{
"completion_length": 2219.5000610351562,
"epoch": 0.5131428571428571,
"grad_norm": 0.36097243428230286,
"kl": 0.170135498046875,
"learning_rate": 1.2822310472864885e-07,
"loss": 0.0368,
"reward": 0.2152203669102164,
"reward_std": 0.7272625118494034,
"rewards/cosine_scaled_reward": -0.09725791588425636,
"rewards/format_reward": 0.750000013038516,
"step": 449
},
{
"completion_length": 2734.604217529297,
"epoch": 0.5142857142857142,
"grad_norm": 0.21459351480007172,
"kl": 0.228668212890625,
"learning_rate": 1.2713832064634125e-07,
"loss": 0.0388,
"reward": 0.31540712295100093,
"reward_std": 0.6263695955276489,
"rewards/cosine_scaled_reward": -0.024456078186631203,
"rewards/format_reward": 0.7500000074505806,
"step": 450
},
{
"completion_length": 2182.0000610351562,
"epoch": 0.5154285714285715,
"grad_norm": 1.156898021697998,
"kl": 0.1932373046875,
"learning_rate": 1.260741462457165e-07,
"loss": 0.0423,
"reward": 0.4187639909796417,
"reward_std": 0.828635673969984,
"rewards/cosine_scaled_reward": -0.02044131373986602,
"rewards/format_reward": 0.854166679084301,
"step": 451
},
{
"completion_length": 2950.0833892822266,
"epoch": 0.5165714285714286,
"grad_norm": 1.0129191875457764,
"kl": 0.2197265625,
"learning_rate": 1.2503063339313356e-07,
"loss": 0.0467,
"reward": 0.22497872763779014,
"reward_std": 0.7840702906250954,
"rewards/cosine_scaled_reward": -9.69860702753067e-05,
"rewards/format_reward": 0.5625000111758709,
"step": 452
},
{
"completion_length": 2666.812545776367,
"epoch": 0.5177142857142857,
"grad_norm": 0.9255622029304504,
"kl": 0.2947998046875,
"learning_rate": 1.2400783294793668e-07,
"loss": 0.0568,
"reward": 0.23635170864872634,
"reward_std": 0.9474616125226021,
"rewards/cosine_scaled_reward": 0.014322125818580389,
"rewards/format_reward": 0.5208333432674408,
"step": 453
},
{
"completion_length": 2574.166732788086,
"epoch": 0.5188571428571429,
"grad_norm": 0.32647454738616943,
"kl": 0.2278289794921875,
"learning_rate": 1.2300579475997657e-07,
"loss": 0.0413,
"reward": -0.001333402469754219,
"reward_std": 0.6261133253574371,
"rewards/cosine_scaled_reward": -0.1682842280715704,
"rewards/format_reward": 0.6041666753590107,
"step": 454
},
{
"completion_length": 3117.4375915527344,
"epoch": 0.52,
"grad_norm": 0.5304046869277954,
"kl": 0.443359375,
"learning_rate": 1.220245676671809e-07,
"loss": 0.0768,
"reward": -0.15152850991580635,
"reward_std": 0.6764501072466373,
"rewards/cosine_scaled_reward": -0.2339008767157793,
"rewards/format_reward": 0.5208333469927311,
"step": 455
},
{
"completion_length": 2992.0000610351562,
"epoch": 0.5211428571428571,
"grad_norm": 0.8478217720985413,
"kl": 0.326171875,
"learning_rate": 1.2106419949317388e-07,
"loss": 0.0617,
"reward": 0.046928441151976585,
"reward_std": 0.8120874315500259,
"rewards/cosine_scaled_reward": -0.0800070259720087,
"rewards/format_reward": 0.479166679084301,
"step": 456
},
{
"completion_length": 2832.1875610351562,
"epoch": 0.5222857142857142,
"grad_norm": 0.8208078145980835,
"kl": 0.300048828125,
"learning_rate": 1.2012473704494537e-07,
"loss": 0.0135,
"reward": 0.3024125434458256,
"reward_std": 0.7796786315739155,
"rewards/cosine_scaled_reward": 0.018373452126979828,
"rewards/format_reward": 0.6250000149011612,
"step": 457
},
{
"completion_length": 2559.0834045410156,
"epoch": 0.5234285714285715,
"grad_norm": 1.155639886856079,
"kl": 0.3125,
"learning_rate": 1.1920622611056974e-07,
"loss": 0.0601,
"reward": -0.02342590008629486,
"reward_std": 0.6464426964521408,
"rewards/cosine_scaled_reward": -0.1930369706824422,
"rewards/format_reward": 0.6250000167638063,
"step": 458
},
{
"completion_length": 2455.3125228881836,
"epoch": 0.5245714285714286,
"grad_norm": 0.5656007528305054,
"kl": 0.265380859375,
"learning_rate": 1.1830871145697412e-07,
"loss": 0.0203,
"reward": 0.17724867910146713,
"reward_std": 0.7673604302108288,
"rewards/cosine_scaled_reward": -0.0773997250944376,
"rewards/format_reward": 0.6458333358168602,
"step": 459
},
{
"completion_length": 2945.979248046875,
"epoch": 0.5257142857142857,
"grad_norm": 0.9485638737678528,
"kl": 0.40087890625,
"learning_rate": 1.1743223682775649e-07,
"loss": 0.0091,
"reward": -0.21174129098653793,
"reward_std": 0.6470157653093338,
"rewards/cosine_scaled_reward": -0.2377923596650362,
"rewards/format_reward": 0.45833334140479565,
"step": 460
},
{
"completion_length": 2731.500045776367,
"epoch": 0.5268571428571428,
"grad_norm": 1.1951149702072144,
"kl": 0.3209228515625,
"learning_rate": 1.1657684494105386e-07,
"loss": 0.0507,
"reward": 0.395471319090575,
"reward_std": 1.0323380753397942,
"rewards/cosine_scaled_reward": 0.06153442489448935,
"rewards/format_reward": 0.6250000204890966,
"step": 461
},
{
"completion_length": 2310.6875610351562,
"epoch": 0.528,
"grad_norm": 0.764068603515625,
"kl": 0.26934814453125,
"learning_rate": 1.1574257748745986e-07,
"loss": 0.0606,
"reward": 0.05179802142083645,
"reward_std": 0.5869512222707272,
"rewards/cosine_scaled_reward": -0.16689125541597605,
"rewards/format_reward": 0.6875000149011612,
"step": 462
},
{
"completion_length": 2988.8959197998047,
"epoch": 0.5291428571428571,
"grad_norm": 0.6601057052612305,
"kl": 0.36944580078125,
"learning_rate": 1.1492947512799328e-07,
"loss": 0.0391,
"reward": 0.3766809217631817,
"reward_std": 1.0442971400916576,
"rewards/cosine_scaled_reward": 0.0020326152443885803,
"rewards/format_reward": 0.7291666902601719,
"step": 463
},
{
"completion_length": 2086.9584045410156,
"epoch": 0.5302857142857142,
"grad_norm": 0.5062448382377625,
"kl": 0.2507781982421875,
"learning_rate": 1.1413757749211602e-07,
"loss": 0.0197,
"reward": 0.4759128368459642,
"reward_std": 0.610221728682518,
"rewards/cosine_scaled_reward": 0.0023420676589012146,
"rewards/format_reward": 0.916666679084301,
"step": 464
},
{
"completion_length": 2885.041732788086,
"epoch": 0.5314285714285715,
"grad_norm": 0.7739923596382141,
"kl": 0.41680908203125,
"learning_rate": 1.1336692317580158e-07,
"loss": 0.0037,
"reward": 0.2254231304395944,
"reward_std": 0.9453909136354923,
"rewards/cosine_scaled_reward": -0.08590294234454632,
"rewards/format_reward": 0.7083333563059568,
"step": 465
},
{
"completion_length": 2756.520896911621,
"epoch": 0.5325714285714286,
"grad_norm": 1.5458998680114746,
"kl": 0.32745361328125,
"learning_rate": 1.1261754973965422e-07,
"loss": 0.1051,
"reward": 0.3231991082429886,
"reward_std": 0.8955418467521667,
"rewards/cosine_scaled_reward": 0.024495400488376617,
"rewards/format_reward": 0.625000013038516,
"step": 466
},
{
"completion_length": 2878.729248046875,
"epoch": 0.5337142857142857,
"grad_norm": 0.46346038579940796,
"kl": 0.31219482421875,
"learning_rate": 1.1188949370707787e-07,
"loss": 0.0296,
"reward": 0.08808497712016106,
"reward_std": 0.6584981977939606,
"rewards/cosine_scaled_reward": -0.19319903245195746,
"rewards/format_reward": 0.7708333432674408,
"step": 467
},
{
"completion_length": 2945.479248046875,
"epoch": 0.5348571428571428,
"grad_norm": 0.5377303957939148,
"kl": 0.395263671875,
"learning_rate": 1.1118279056249653e-07,
"loss": 0.0565,
"reward": 0.3272525854408741,
"reward_std": 0.7806177064776421,
"rewards/cosine_scaled_reward": 0.046021029353141785,
"rewards/format_reward": 0.6041666772216558,
"step": 468
},
{
"completion_length": 2482.9167251586914,
"epoch": 0.536,
"grad_norm": 0.6872317790985107,
"kl": 0.414398193359375,
"learning_rate": 1.1049747474962444e-07,
"loss": 0.05,
"reward": 0.3257849495857954,
"reward_std": 0.8201456405222416,
"rewards/cosine_scaled_reward": -0.05236888420768082,
"rewards/format_reward": 0.791666679084301,
"step": 469
},
{
"completion_length": 3126.6458892822266,
"epoch": 0.5371428571428571,
"grad_norm": 1.1381498575210571,
"kl": 0.4842529296875,
"learning_rate": 1.0983357966978745e-07,
"loss": 0.0271,
"reward": 0.055752304033376276,
"reward_std": 0.8491252809762955,
"rewards/cosine_scaled_reward": -0.12250774865970016,
"rewards/format_reward": 0.5625000111758709,
"step": 470
},
{
"completion_length": 2754.291778564453,
"epoch": 0.5382857142857143,
"grad_norm": 0.48720309138298035,
"kl": 0.3826904296875,
"learning_rate": 1.0919113768029517e-07,
"loss": 0.0524,
"reward": 0.5262258416041732,
"reward_std": 0.9357166737318039,
"rewards/cosine_scaled_reward": 0.09029024560004473,
"rewards/format_reward": 0.7708333544433117,
"step": 471
},
{
"completion_length": 2644.812545776367,
"epoch": 0.5394285714285715,
"grad_norm": 0.3475642800331116,
"kl": 0.34210205078125,
"learning_rate": 1.0857018009286381e-07,
"loss": 0.0365,
"reward": 0.23837368440581486,
"reward_std": 0.8128412887454033,
"rewards/cosine_scaled_reward": -0.07736600749194622,
"rewards/format_reward": 0.729166679084301,
"step": 472
},
{
"completion_length": 2967.229232788086,
"epoch": 0.5405714285714286,
"grad_norm": 0.6972779631614685,
"kl": 0.3824462890625,
"learning_rate": 1.0797073717209013e-07,
"loss": 0.0307,
"reward": 0.10205891542136669,
"reward_std": 0.7758991979062557,
"rewards/cosine_scaled_reward": -0.09507068432867527,
"rewards/format_reward": 0.5833333432674408,
"step": 473
},
{
"completion_length": 2673.6875610351562,
"epoch": 0.5417142857142857,
"grad_norm": 0.45455074310302734,
"kl": 0.395111083984375,
"learning_rate": 1.0739283813397639e-07,
"loss": 0.039,
"reward": 0.46767894667573273,
"reward_std": 0.8645204231142998,
"rewards/cosine_scaled_reward": 0.09637415563338436,
"rewards/format_reward": 0.687500013038516,
"step": 474
},
{
"completion_length": 2484.291717529297,
"epoch": 0.5428571428571428,
"grad_norm": 0.712739109992981,
"kl": 0.348419189453125,
"learning_rate": 1.068365111445064e-07,
"loss": 0.0527,
"reward": 0.3686076030135155,
"reward_std": 0.916143324226141,
"rewards/cosine_scaled_reward": 0.012207330204546452,
"rewards/format_reward": 0.7083333469927311,
"step": 475
},
{
"completion_length": 3011.8959045410156,
"epoch": 0.544,
"grad_norm": 1.0881688594818115,
"kl": 0.4395751953125,
"learning_rate": 1.063017833182728e-07,
"loss": 0.0524,
"reward": 0.3218140173703432,
"reward_std": 1.0662804022431374,
"rewards/cosine_scaled_reward": -0.025444235419854522,
"rewards/format_reward": 0.7083333507180214,
"step": 476
},
{
"completion_length": 2656.3751068115234,
"epoch": 0.5451428571428572,
"grad_norm": 2.0791175365448,
"kl": 0.4376220703125,
"learning_rate": 1.0578868071715544e-07,
"loss": 0.0965,
"reward": 0.2924130540341139,
"reward_std": 0.976725772023201,
"rewards/cosine_scaled_reward": 0.013521750457584858,
"rewards/format_reward": 0.6041666865348816,
"step": 477
},
{
"completion_length": 2819.812545776367,
"epoch": 0.5462857142857143,
"grad_norm": 0.5398616194725037,
"kl": 0.363677978515625,
"learning_rate": 1.0529722834905125e-07,
"loss": 0.0443,
"reward": -0.004572154954075813,
"reward_std": 0.6425527259707451,
"rewards/cosine_scaled_reward": -0.13902169838547707,
"rewards/format_reward": 0.5416666753590107,
"step": 478
},
{
"completion_length": 2858.562545776367,
"epoch": 0.5474285714285714,
"grad_norm": 0.6435410976409912,
"kl": 0.378143310546875,
"learning_rate": 1.0482745016665526e-07,
"loss": 0.0261,
"reward": 0.18186998274177313,
"reward_std": 0.8897733464837074,
"rewards/cosine_scaled_reward": -0.12430705223232508,
"rewards/format_reward": 0.7291666846722364,
"step": 479
},
{
"completion_length": 2607.62508392334,
"epoch": 0.5485714285714286,
"grad_norm": 0.8115129470825195,
"kl": 0.40704345703125,
"learning_rate": 1.0437936906629334e-07,
"loss": 0.043,
"reward": 0.002752909902483225,
"reward_std": 0.6639882102608681,
"rewards/cosine_scaled_reward": -0.1971071765292436,
"rewards/format_reward": 0.6666666716337204,
"step": 480
},
{
"completion_length": 3112.166748046875,
"epoch": 0.5497142857142857,
"grad_norm": 0.5421264171600342,
"kl": 0.4429931640625,
"learning_rate": 1.0395300688680625e-07,
"loss": 0.0528,
"reward": 0.04194536246359348,
"reward_std": 0.6818186715245247,
"rewards/cosine_scaled_reward": -0.13939355686306953,
"rewards/format_reward": 0.6041666902601719,
"step": 481
},
{
"completion_length": 2706.7084045410156,
"epoch": 0.5508571428571428,
"grad_norm": 0.4543374478816986,
"kl": 0.368194580078125,
"learning_rate": 1.0354838440848501e-07,
"loss": 0.0506,
"reward": 0.4122487809509039,
"reward_std": 0.7574451714754105,
"rewards/cosine_scaled_reward": 0.056680090725421906,
"rewards/format_reward": 0.7083333395421505,
"step": 482
},
{
"completion_length": 2878.229217529297,
"epoch": 0.552,
"grad_norm": 0.6022698879241943,
"kl": 0.40606689453125,
"learning_rate": 1.0316552135205837e-07,
"loss": 0.0356,
"reward": -0.09259207546710968,
"reward_std": 0.6557772308588028,
"rewards/cosine_scaled_reward": -0.19897521962411702,
"rewards/format_reward": 0.5416666772216558,
"step": 483
},
{
"completion_length": 2701.9375610351562,
"epoch": 0.5531428571428572,
"grad_norm": 1.3102695941925049,
"kl": 0.356781005859375,
"learning_rate": 1.0280443637773163e-07,
"loss": 0.0726,
"reward": 0.3319150470197201,
"reward_std": 1.0350141674280167,
"rewards/cosine_scaled_reward": -0.02834635879844427,
"rewards/format_reward": 0.729166679084301,
"step": 484
},
{
"completion_length": 2762.916717529297,
"epoch": 0.5542857142857143,
"grad_norm": 0.8539823293685913,
"kl": 0.47900390625,
"learning_rate": 1.0246514708427701e-07,
"loss": 0.08,
"reward": 0.18864674912765622,
"reward_std": 0.7356266789138317,
"rewards/cosine_scaled_reward": -0.11939917271956801,
"rewards/format_reward": 0.7500000204890966,
"step": 485
},
{
"completion_length": 2239.6667404174805,
"epoch": 0.5554285714285714,
"grad_norm": 0.3320426940917969,
"kl": 0.263824462890625,
"learning_rate": 1.0214767000817596e-07,
"loss": 0.0167,
"reward": 0.2528488418611232,
"reward_std": 0.7695866264402866,
"rewards/cosine_scaled_reward": -0.10691238380968571,
"rewards/format_reward": 0.812500013038516,
"step": 486
},
{
"completion_length": 2169.6666984558105,
"epoch": 0.5565714285714286,
"grad_norm": 0.562497615814209,
"kl": 0.274017333984375,
"learning_rate": 1.0185202062281336e-07,
"loss": 0.0039,
"reward": 0.44721357547678053,
"reward_std": 0.8235045485198498,
"rewards/cosine_scaled_reward": 0.010554181411862373,
"rewards/format_reward": 0.8333333395421505,
"step": 487
},
{
"completion_length": 2426.8125534057617,
"epoch": 0.5577142857142857,
"grad_norm": 0.2765026092529297,
"kl": 0.3238372802734375,
"learning_rate": 1.0157821333772304e-07,
"loss": 0.0469,
"reward": 0.08554835570976138,
"reward_std": 0.5648987516760826,
"rewards/cosine_scaled_reward": -0.19786089658737183,
"rewards/format_reward": 0.7916666753590107,
"step": 488
},
{
"completion_length": 3204.875030517578,
"epoch": 0.5588571428571428,
"grad_norm": 1.2936737537384033,
"kl": 0.4969482421875,
"learning_rate": 1.013262614978859e-07,
"loss": 0.0295,
"reward": -0.17606779746711254,
"reward_std": 0.6055882386863232,
"rewards/cosine_scaled_reward": -0.21996455593034625,
"rewards/format_reward": 0.47916668094694614,
"step": 489
},
{
"completion_length": 2257.1458892822266,
"epoch": 0.56,
"grad_norm": 0.4323062598705292,
"kl": 0.3099365234375,
"learning_rate": 1.0109617738307911e-07,
"loss": 0.0338,
"reward": 0.3065086267888546,
"reward_std": 0.7493411414325237,
"rewards/cosine_scaled_reward": -0.061892845667898655,
"rewards/format_reward": 0.7916666734963655,
"step": 490
},
{
"completion_length": 2736.312545776367,
"epoch": 0.5611428571428572,
"grad_norm": 0.5150272846221924,
"kl": 0.3674163818359375,
"learning_rate": 1.0088797220727779e-07,
"loss": 0.0312,
"reward": 0.5249021016061306,
"reward_std": 1.010607898235321,
"rewards/cosine_scaled_reward": 0.1508561042137444,
"rewards/format_reward": 0.6250000204890966,
"step": 491
},
{
"completion_length": 2490.8958740234375,
"epoch": 0.5622857142857143,
"grad_norm": 1.4419853687286377,
"kl": 0.42706298828125,
"learning_rate": 1.0070165611810855e-07,
"loss": 0.0032,
"reward": 0.1767475767992437,
"reward_std": 0.777750164270401,
"rewards/cosine_scaled_reward": -0.07436079788021743,
"rewards/format_reward": 0.6458333469927311,
"step": 492
},
{
"completion_length": 2550.9375762939453,
"epoch": 0.5634285714285714,
"grad_norm": 0.4972594380378723,
"kl": 0.450469970703125,
"learning_rate": 1.005372381963547e-07,
"loss": 0.0431,
"reward": 0.6753305066376925,
"reward_std": 0.8794010616838932,
"rewards/cosine_scaled_reward": 0.15572454407811165,
"rewards/format_reward": 0.8541666716337204,
"step": 493
},
{
"completion_length": 2323.479232788086,
"epoch": 0.5645714285714286,
"grad_norm": 0.5891183614730835,
"kl": 0.3460693359375,
"learning_rate": 1.0039472645551372e-07,
"loss": 0.0271,
"reward": 0.6445839628577232,
"reward_std": 1.0191630199551582,
"rewards/cosine_scaled_reward": 0.12392374624687363,
"rewards/format_reward": 0.8541666753590107,
"step": 494
},
{
"completion_length": 2892.916717529297,
"epoch": 0.5657142857142857,
"grad_norm": 0.4543853998184204,
"kl": 0.45623779296875,
"learning_rate": 1.002741278414069e-07,
"loss": 0.0451,
"reward": 0.18997809663414955,
"reward_std": 0.8969040662050247,
"rewards/cosine_scaled_reward": -0.05571722239255905,
"rewards/format_reward": 0.6041666828095913,
"step": 495
},
{
"completion_length": 2114.500045776367,
"epoch": 0.5668571428571428,
"grad_norm": 1.2619980573654175,
"kl": 0.32281494140625,
"learning_rate": 1.0017544823184055e-07,
"loss": 0.0048,
"reward": 0.488508015871048,
"reward_std": 0.7361417338252068,
"rewards/cosine_scaled_reward": 0.10424756724387407,
"rewards/format_reward": 0.7083333358168602,
"step": 496
},
{
"completion_length": 2322.916679382324,
"epoch": 0.568,
"grad_norm": 0.9984971284866333,
"kl": 0.2513885498046875,
"learning_rate": 1.0009869243631952e-07,
"loss": -0.0203,
"reward": 0.6254424216458574,
"reward_std": 0.7260546460747719,
"rewards/cosine_scaled_reward": 0.1703429389744997,
"rewards/format_reward": 0.7708333488553762,
"step": 497
},
{
"completion_length": 2773.166748046875,
"epoch": 0.5691428571428572,
"grad_norm": 0.5907127857208252,
"kl": 0.4683837890625,
"learning_rate": 1.000438641958131e-07,
"loss": 0.0642,
"reward": 0.36813890002667904,
"reward_std": 0.9811634942889214,
"rewards/cosine_scaled_reward": -0.0031640082597732544,
"rewards/format_reward": 0.7291666828095913,
"step": 498
},
{
"completion_length": 2583.5834350585938,
"epoch": 0.5702857142857143,
"grad_norm": 0.35024699568748474,
"kl": 0.358978271484375,
"learning_rate": 1.0001096618257236e-07,
"loss": 0.0401,
"reward": 0.21864662691950798,
"reward_std": 0.8943019956350327,
"rewards/cosine_scaled_reward": -0.07930864673107862,
"rewards/format_reward": 0.6875000037252903,
"step": 499
},
{
"completion_length": 3055.6666870117188,
"epoch": 0.5714285714285714,
"grad_norm": 1.0981477499008179,
"kl": 0.4345703125,
"learning_rate": 1e-07,
"loss": 0.023,
"reward": 0.14122199080884457,
"reward_std": 0.8392615914344788,
"rewards/cosine_scaled_reward": -0.031122979940846562,
"rewards/format_reward": 0.5000000055879354,
"step": 500
},
{
"epoch": 0.5714285714285714,
"step": 500,
"total_flos": 0.0,
"train_loss": 0.038267850877775345,
"train_runtime": 72125.5591,
"train_samples_per_second": 0.333,
"train_steps_per_second": 0.007
}
],
"logging_steps": 1,
"max_steps": 500,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 50,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 6,
"trial_name": null,
"trial_params": null
}