mamba-2.8b-instruct-openhermes / trainer_state.json
mrm8488's picture
Upload folder using huggingface_hub
bcc4960
raw
history blame
55.2 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.4825064241945047,
"eval_steps": 500,
"global_step": 22500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.989018712114557e-05,
"loss": 0.6189,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 4.9780374242291136e-05,
"loss": 0.4018,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 4.967056136343671e-05,
"loss": 0.4266,
"step": 150
},
{
"epoch": 0.01,
"learning_rate": 4.9560748484582276e-05,
"loss": 0.4151,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 4.945093560572784e-05,
"loss": 0.3993,
"step": 250
},
{
"epoch": 0.02,
"learning_rate": 4.934112272687341e-05,
"loss": 0.3948,
"step": 300
},
{
"epoch": 0.02,
"learning_rate": 4.9231309848018984e-05,
"loss": 0.4181,
"step": 350
},
{
"epoch": 0.03,
"learning_rate": 4.9121496969164544e-05,
"loss": 0.3979,
"step": 400
},
{
"epoch": 0.03,
"learning_rate": 4.901168409031011e-05,
"loss": 0.4006,
"step": 450
},
{
"epoch": 0.03,
"learning_rate": 4.890187121145568e-05,
"loss": 0.4029,
"step": 500
},
{
"epoch": 0.04,
"learning_rate": 4.879205833260125e-05,
"loss": 0.378,
"step": 550
},
{
"epoch": 0.04,
"learning_rate": 4.868224545374682e-05,
"loss": 0.3894,
"step": 600
},
{
"epoch": 0.04,
"learning_rate": 4.8572432574892385e-05,
"loss": 0.3784,
"step": 650
},
{
"epoch": 0.05,
"learning_rate": 4.846261969603795e-05,
"loss": 0.3743,
"step": 700
},
{
"epoch": 0.05,
"learning_rate": 4.835280681718352e-05,
"loss": 0.3816,
"step": 750
},
{
"epoch": 0.05,
"learning_rate": 4.824299393832909e-05,
"loss": 0.3861,
"step": 800
},
{
"epoch": 0.06,
"learning_rate": 4.813318105947466e-05,
"loss": 0.3885,
"step": 850
},
{
"epoch": 0.06,
"learning_rate": 4.8023368180620226e-05,
"loss": 0.3948,
"step": 900
},
{
"epoch": 0.06,
"learning_rate": 4.791355530176579e-05,
"loss": 0.3545,
"step": 950
},
{
"epoch": 0.07,
"learning_rate": 4.7803742422911366e-05,
"loss": 0.3743,
"step": 1000
},
{
"epoch": 0.07,
"learning_rate": 4.769392954405693e-05,
"loss": 0.3843,
"step": 1050
},
{
"epoch": 0.07,
"learning_rate": 4.75841166652025e-05,
"loss": 0.3489,
"step": 1100
},
{
"epoch": 0.08,
"learning_rate": 4.7474303786348067e-05,
"loss": 0.3727,
"step": 1150
},
{
"epoch": 0.08,
"learning_rate": 4.7364490907493633e-05,
"loss": 0.3713,
"step": 1200
},
{
"epoch": 0.08,
"learning_rate": 4.72546780286392e-05,
"loss": 0.3662,
"step": 1250
},
{
"epoch": 0.09,
"learning_rate": 4.714486514978477e-05,
"loss": 0.3777,
"step": 1300
},
{
"epoch": 0.09,
"learning_rate": 4.7035052270930334e-05,
"loss": 0.3819,
"step": 1350
},
{
"epoch": 0.09,
"learning_rate": 4.69252393920759e-05,
"loss": 0.362,
"step": 1400
},
{
"epoch": 0.1,
"learning_rate": 4.6815426513221474e-05,
"loss": 0.3577,
"step": 1450
},
{
"epoch": 0.1,
"learning_rate": 4.670561363436704e-05,
"loss": 0.3879,
"step": 1500
},
{
"epoch": 0.1,
"learning_rate": 4.659580075551261e-05,
"loss": 0.3678,
"step": 1550
},
{
"epoch": 0.11,
"learning_rate": 4.6485987876658175e-05,
"loss": 0.3491,
"step": 1600
},
{
"epoch": 0.11,
"learning_rate": 4.637617499780375e-05,
"loss": 0.3904,
"step": 1650
},
{
"epoch": 0.11,
"learning_rate": 4.6266362118949315e-05,
"loss": 0.3495,
"step": 1700
},
{
"epoch": 0.12,
"learning_rate": 4.615654924009488e-05,
"loss": 0.3743,
"step": 1750
},
{
"epoch": 0.12,
"learning_rate": 4.604673636124045e-05,
"loss": 0.3814,
"step": 1800
},
{
"epoch": 0.12,
"learning_rate": 4.5936923482386016e-05,
"loss": 0.3495,
"step": 1850
},
{
"epoch": 0.13,
"learning_rate": 4.582711060353159e-05,
"loss": 0.3688,
"step": 1900
},
{
"epoch": 0.13,
"learning_rate": 4.5717297724677156e-05,
"loss": 0.3401,
"step": 1950
},
{
"epoch": 0.13,
"learning_rate": 4.5607484845822716e-05,
"loss": 0.3647,
"step": 2000
},
{
"epoch": 0.14,
"learning_rate": 4.549767196696828e-05,
"loss": 0.3716,
"step": 2050
},
{
"epoch": 0.14,
"learning_rate": 4.538785908811386e-05,
"loss": 0.373,
"step": 2100
},
{
"epoch": 0.14,
"learning_rate": 4.5278046209259424e-05,
"loss": 0.3471,
"step": 2150
},
{
"epoch": 0.14,
"learning_rate": 4.516823333040499e-05,
"loss": 0.3843,
"step": 2200
},
{
"epoch": 0.15,
"learning_rate": 4.505842045155056e-05,
"loss": 0.3629,
"step": 2250
},
{
"epoch": 0.15,
"learning_rate": 4.494860757269613e-05,
"loss": 0.3579,
"step": 2300
},
{
"epoch": 0.15,
"learning_rate": 4.48387946938417e-05,
"loss": 0.3373,
"step": 2350
},
{
"epoch": 0.16,
"learning_rate": 4.4728981814987264e-05,
"loss": 0.3509,
"step": 2400
},
{
"epoch": 0.16,
"learning_rate": 4.461916893613283e-05,
"loss": 0.3753,
"step": 2450
},
{
"epoch": 0.16,
"learning_rate": 4.45093560572784e-05,
"loss": 0.3421,
"step": 2500
},
{
"epoch": 0.17,
"learning_rate": 4.439954317842397e-05,
"loss": 0.3686,
"step": 2550
},
{
"epoch": 0.17,
"learning_rate": 4.428973029956954e-05,
"loss": 0.355,
"step": 2600
},
{
"epoch": 0.17,
"learning_rate": 4.4179917420715105e-05,
"loss": 0.3725,
"step": 2650
},
{
"epoch": 0.18,
"learning_rate": 4.407010454186067e-05,
"loss": 0.3495,
"step": 2700
},
{
"epoch": 0.18,
"learning_rate": 4.396029166300624e-05,
"loss": 0.3606,
"step": 2750
},
{
"epoch": 0.18,
"learning_rate": 4.3850478784151806e-05,
"loss": 0.3496,
"step": 2800
},
{
"epoch": 0.19,
"learning_rate": 4.374066590529737e-05,
"loss": 0.3606,
"step": 2850
},
{
"epoch": 0.19,
"learning_rate": 4.363085302644294e-05,
"loss": 0.3453,
"step": 2900
},
{
"epoch": 0.19,
"learning_rate": 4.352104014758851e-05,
"loss": 0.3617,
"step": 2950
},
{
"epoch": 0.2,
"learning_rate": 4.341122726873408e-05,
"loss": 0.3297,
"step": 3000
},
{
"epoch": 0.2,
"learning_rate": 4.330141438987965e-05,
"loss": 0.3448,
"step": 3050
},
{
"epoch": 0.2,
"learning_rate": 4.3191601511025214e-05,
"loss": 0.3631,
"step": 3100
},
{
"epoch": 0.21,
"learning_rate": 4.308178863217078e-05,
"loss": 0.352,
"step": 3150
},
{
"epoch": 0.21,
"learning_rate": 4.2971975753316354e-05,
"loss": 0.3482,
"step": 3200
},
{
"epoch": 0.21,
"learning_rate": 4.286216287446192e-05,
"loss": 0.3343,
"step": 3250
},
{
"epoch": 0.22,
"learning_rate": 4.275234999560749e-05,
"loss": 0.3204,
"step": 3300
},
{
"epoch": 0.22,
"learning_rate": 4.2642537116753055e-05,
"loss": 0.3442,
"step": 3350
},
{
"epoch": 0.22,
"learning_rate": 4.253272423789862e-05,
"loss": 0.3519,
"step": 3400
},
{
"epoch": 0.23,
"learning_rate": 4.2422911359044195e-05,
"loss": 0.3504,
"step": 3450
},
{
"epoch": 0.23,
"learning_rate": 4.231309848018976e-05,
"loss": 0.3612,
"step": 3500
},
{
"epoch": 0.23,
"learning_rate": 4.220328560133532e-05,
"loss": 0.3299,
"step": 3550
},
{
"epoch": 0.24,
"learning_rate": 4.2093472722480896e-05,
"loss": 0.3466,
"step": 3600
},
{
"epoch": 0.24,
"learning_rate": 4.198365984362646e-05,
"loss": 0.3459,
"step": 3650
},
{
"epoch": 0.24,
"learning_rate": 4.187384696477203e-05,
"loss": 0.3687,
"step": 3700
},
{
"epoch": 0.25,
"learning_rate": 4.1764034085917596e-05,
"loss": 0.3496,
"step": 3750
},
{
"epoch": 0.25,
"learning_rate": 4.165422120706316e-05,
"loss": 0.3558,
"step": 3800
},
{
"epoch": 0.25,
"learning_rate": 4.1544408328208736e-05,
"loss": 0.3649,
"step": 3850
},
{
"epoch": 0.26,
"learning_rate": 4.14345954493543e-05,
"loss": 0.3542,
"step": 3900
},
{
"epoch": 0.26,
"learning_rate": 4.132478257049987e-05,
"loss": 0.3415,
"step": 3950
},
{
"epoch": 0.26,
"learning_rate": 4.121496969164544e-05,
"loss": 0.364,
"step": 4000
},
{
"epoch": 0.27,
"learning_rate": 4.1105156812791004e-05,
"loss": 0.3532,
"step": 4050
},
{
"epoch": 0.27,
"learning_rate": 4.099534393393658e-05,
"loss": 0.3625,
"step": 4100
},
{
"epoch": 0.27,
"learning_rate": 4.0885531055082144e-05,
"loss": 0.3439,
"step": 4150
},
{
"epoch": 0.28,
"learning_rate": 4.077571817622771e-05,
"loss": 0.349,
"step": 4200
},
{
"epoch": 0.28,
"learning_rate": 4.066590529737328e-05,
"loss": 0.3468,
"step": 4250
},
{
"epoch": 0.28,
"learning_rate": 4.055609241851885e-05,
"loss": 0.3442,
"step": 4300
},
{
"epoch": 0.29,
"learning_rate": 4.044627953966441e-05,
"loss": 0.3383,
"step": 4350
},
{
"epoch": 0.29,
"learning_rate": 4.033646666080998e-05,
"loss": 0.3433,
"step": 4400
},
{
"epoch": 0.29,
"learning_rate": 4.0226653781955545e-05,
"loss": 0.3393,
"step": 4450
},
{
"epoch": 0.3,
"learning_rate": 4.011684090310112e-05,
"loss": 0.3616,
"step": 4500
},
{
"epoch": 0.3,
"learning_rate": 4.0007028024246686e-05,
"loss": 0.352,
"step": 4550
},
{
"epoch": 0.3,
"learning_rate": 3.989721514539225e-05,
"loss": 0.3269,
"step": 4600
},
{
"epoch": 0.31,
"learning_rate": 3.978740226653782e-05,
"loss": 0.3261,
"step": 4650
},
{
"epoch": 0.31,
"learning_rate": 3.9677589387683386e-05,
"loss": 0.3421,
"step": 4700
},
{
"epoch": 0.31,
"learning_rate": 3.956777650882896e-05,
"loss": 0.3286,
"step": 4750
},
{
"epoch": 0.32,
"learning_rate": 3.9457963629974527e-05,
"loss": 0.3317,
"step": 4800
},
{
"epoch": 0.32,
"learning_rate": 3.9348150751120093e-05,
"loss": 0.3489,
"step": 4850
},
{
"epoch": 0.32,
"learning_rate": 3.923833787226566e-05,
"loss": 0.3536,
"step": 4900
},
{
"epoch": 0.33,
"learning_rate": 3.9128524993411234e-05,
"loss": 0.3633,
"step": 4950
},
{
"epoch": 0.33,
"learning_rate": 3.90187121145568e-05,
"loss": 0.335,
"step": 5000
},
{
"epoch": 0.33,
"learning_rate": 3.890889923570237e-05,
"loss": 0.3373,
"step": 5050
},
{
"epoch": 0.34,
"learning_rate": 3.8799086356847934e-05,
"loss": 0.3477,
"step": 5100
},
{
"epoch": 0.34,
"learning_rate": 3.86892734779935e-05,
"loss": 0.3373,
"step": 5150
},
{
"epoch": 0.34,
"learning_rate": 3.857946059913907e-05,
"loss": 0.3318,
"step": 5200
},
{
"epoch": 0.35,
"learning_rate": 3.8469647720284635e-05,
"loss": 0.3324,
"step": 5250
},
{
"epoch": 0.35,
"learning_rate": 3.83598348414302e-05,
"loss": 0.3376,
"step": 5300
},
{
"epoch": 0.35,
"learning_rate": 3.825002196257577e-05,
"loss": 0.3623,
"step": 5350
},
{
"epoch": 0.36,
"learning_rate": 3.814020908372134e-05,
"loss": 0.3498,
"step": 5400
},
{
"epoch": 0.36,
"learning_rate": 3.803039620486691e-05,
"loss": 0.3599,
"step": 5450
},
{
"epoch": 0.36,
"learning_rate": 3.7920583326012476e-05,
"loss": 0.336,
"step": 5500
},
{
"epoch": 0.37,
"learning_rate": 3.781077044715804e-05,
"loss": 0.3426,
"step": 5550
},
{
"epoch": 0.37,
"learning_rate": 3.7700957568303616e-05,
"loss": 0.3327,
"step": 5600
},
{
"epoch": 0.37,
"learning_rate": 3.759114468944918e-05,
"loss": 0.3352,
"step": 5650
},
{
"epoch": 0.38,
"learning_rate": 3.748133181059475e-05,
"loss": 0.3165,
"step": 5700
},
{
"epoch": 0.38,
"learning_rate": 3.737151893174032e-05,
"loss": 0.3335,
"step": 5750
},
{
"epoch": 0.38,
"learning_rate": 3.7261706052885884e-05,
"loss": 0.3475,
"step": 5800
},
{
"epoch": 0.39,
"learning_rate": 3.715189317403146e-05,
"loss": 0.3335,
"step": 5850
},
{
"epoch": 0.39,
"learning_rate": 3.704208029517702e-05,
"loss": 0.3429,
"step": 5900
},
{
"epoch": 0.39,
"learning_rate": 3.6932267416322584e-05,
"loss": 0.3355,
"step": 5950
},
{
"epoch": 0.4,
"learning_rate": 3.682245453746815e-05,
"loss": 0.3102,
"step": 6000
},
{
"epoch": 0.4,
"learning_rate": 3.6712641658613724e-05,
"loss": 0.3491,
"step": 6050
},
{
"epoch": 0.4,
"learning_rate": 3.660282877975929e-05,
"loss": 0.3349,
"step": 6100
},
{
"epoch": 0.41,
"learning_rate": 3.649301590090486e-05,
"loss": 0.3509,
"step": 6150
},
{
"epoch": 0.41,
"learning_rate": 3.6383203022050425e-05,
"loss": 0.3374,
"step": 6200
},
{
"epoch": 0.41,
"learning_rate": 3.6273390143196e-05,
"loss": 0.3242,
"step": 6250
},
{
"epoch": 0.42,
"learning_rate": 3.6163577264341565e-05,
"loss": 0.3425,
"step": 6300
},
{
"epoch": 0.42,
"learning_rate": 3.605376438548713e-05,
"loss": 0.3274,
"step": 6350
},
{
"epoch": 0.42,
"learning_rate": 3.59439515066327e-05,
"loss": 0.3619,
"step": 6400
},
{
"epoch": 0.42,
"learning_rate": 3.5834138627778266e-05,
"loss": 0.3385,
"step": 6450
},
{
"epoch": 0.43,
"learning_rate": 3.572432574892384e-05,
"loss": 0.3236,
"step": 6500
},
{
"epoch": 0.43,
"learning_rate": 3.5614512870069406e-05,
"loss": 0.331,
"step": 6550
},
{
"epoch": 0.43,
"learning_rate": 3.550469999121497e-05,
"loss": 0.3379,
"step": 6600
},
{
"epoch": 0.44,
"learning_rate": 3.539488711236054e-05,
"loss": 0.3284,
"step": 6650
},
{
"epoch": 0.44,
"learning_rate": 3.528507423350611e-05,
"loss": 0.3315,
"step": 6700
},
{
"epoch": 0.44,
"learning_rate": 3.5175261354651674e-05,
"loss": 0.3268,
"step": 6750
},
{
"epoch": 0.45,
"learning_rate": 3.506544847579724e-05,
"loss": 0.3439,
"step": 6800
},
{
"epoch": 0.45,
"learning_rate": 3.495563559694281e-05,
"loss": 0.3399,
"step": 6850
},
{
"epoch": 0.45,
"learning_rate": 3.484582271808838e-05,
"loss": 0.3374,
"step": 6900
},
{
"epoch": 0.46,
"learning_rate": 3.473600983923395e-05,
"loss": 0.3409,
"step": 6950
},
{
"epoch": 0.46,
"learning_rate": 3.4626196960379515e-05,
"loss": 0.3196,
"step": 7000
},
{
"epoch": 0.46,
"learning_rate": 3.451638408152508e-05,
"loss": 0.3247,
"step": 7050
},
{
"epoch": 0.47,
"learning_rate": 3.440657120267065e-05,
"loss": 0.3469,
"step": 7100
},
{
"epoch": 0.47,
"learning_rate": 3.429675832381622e-05,
"loss": 0.3149,
"step": 7150
},
{
"epoch": 0.47,
"learning_rate": 3.418694544496179e-05,
"loss": 0.3413,
"step": 7200
},
{
"epoch": 0.48,
"learning_rate": 3.4077132566107355e-05,
"loss": 0.3285,
"step": 7250
},
{
"epoch": 0.48,
"learning_rate": 3.396731968725292e-05,
"loss": 0.312,
"step": 7300
},
{
"epoch": 0.48,
"learning_rate": 3.3857506808398496e-05,
"loss": 0.3355,
"step": 7350
},
{
"epoch": 0.49,
"learning_rate": 3.374769392954406e-05,
"loss": 0.3402,
"step": 7400
},
{
"epoch": 0.49,
"learning_rate": 3.363788105068963e-05,
"loss": 0.3373,
"step": 7450
},
{
"epoch": 0.49,
"learning_rate": 3.352806817183519e-05,
"loss": 0.3311,
"step": 7500
},
{
"epoch": 0.5,
"learning_rate": 3.341825529298076e-05,
"loss": 0.3213,
"step": 7550
},
{
"epoch": 0.5,
"learning_rate": 3.330844241412633e-05,
"loss": 0.3323,
"step": 7600
},
{
"epoch": 0.5,
"learning_rate": 3.31986295352719e-05,
"loss": 0.3307,
"step": 7650
},
{
"epoch": 0.51,
"learning_rate": 3.3088816656417464e-05,
"loss": 0.3168,
"step": 7700
},
{
"epoch": 0.51,
"learning_rate": 3.297900377756303e-05,
"loss": 0.3287,
"step": 7750
},
{
"epoch": 0.51,
"learning_rate": 3.2869190898708604e-05,
"loss": 0.3267,
"step": 7800
},
{
"epoch": 0.52,
"learning_rate": 3.275937801985417e-05,
"loss": 0.331,
"step": 7850
},
{
"epoch": 0.52,
"learning_rate": 3.264956514099974e-05,
"loss": 0.335,
"step": 7900
},
{
"epoch": 0.52,
"learning_rate": 3.2539752262145305e-05,
"loss": 0.3332,
"step": 7950
},
{
"epoch": 0.53,
"learning_rate": 3.242993938329087e-05,
"loss": 0.3442,
"step": 8000
},
{
"epoch": 0.53,
"learning_rate": 3.2320126504436445e-05,
"loss": 0.317,
"step": 8050
},
{
"epoch": 0.53,
"learning_rate": 3.221031362558201e-05,
"loss": 0.3243,
"step": 8100
},
{
"epoch": 0.54,
"learning_rate": 3.210050074672758e-05,
"loss": 0.3328,
"step": 8150
},
{
"epoch": 0.54,
"learning_rate": 3.1990687867873146e-05,
"loss": 0.3376,
"step": 8200
},
{
"epoch": 0.54,
"learning_rate": 3.188087498901872e-05,
"loss": 0.3387,
"step": 8250
},
{
"epoch": 0.55,
"learning_rate": 3.177106211016428e-05,
"loss": 0.3092,
"step": 8300
},
{
"epoch": 0.55,
"learning_rate": 3.1661249231309846e-05,
"loss": 0.3241,
"step": 8350
},
{
"epoch": 0.55,
"learning_rate": 3.155143635245541e-05,
"loss": 0.3219,
"step": 8400
},
{
"epoch": 0.56,
"learning_rate": 3.1441623473600987e-05,
"loss": 0.3256,
"step": 8450
},
{
"epoch": 0.56,
"learning_rate": 3.133181059474655e-05,
"loss": 0.3249,
"step": 8500
},
{
"epoch": 0.56,
"learning_rate": 3.122199771589212e-05,
"loss": 0.3265,
"step": 8550
},
{
"epoch": 0.57,
"learning_rate": 3.111218483703769e-05,
"loss": 0.3242,
"step": 8600
},
{
"epoch": 0.57,
"learning_rate": 3.1002371958183254e-05,
"loss": 0.3568,
"step": 8650
},
{
"epoch": 0.57,
"learning_rate": 3.089255907932883e-05,
"loss": 0.3261,
"step": 8700
},
{
"epoch": 0.58,
"learning_rate": 3.0782746200474394e-05,
"loss": 0.3197,
"step": 8750
},
{
"epoch": 0.58,
"learning_rate": 3.067293332161996e-05,
"loss": 0.3419,
"step": 8800
},
{
"epoch": 0.58,
"learning_rate": 3.056312044276553e-05,
"loss": 0.3448,
"step": 8850
},
{
"epoch": 0.59,
"learning_rate": 3.0453307563911098e-05,
"loss": 0.3506,
"step": 8900
},
{
"epoch": 0.59,
"learning_rate": 3.034349468505667e-05,
"loss": 0.3337,
"step": 8950
},
{
"epoch": 0.59,
"learning_rate": 3.0233681806202235e-05,
"loss": 0.3214,
"step": 9000
},
{
"epoch": 0.6,
"learning_rate": 3.01238689273478e-05,
"loss": 0.3358,
"step": 9050
},
{
"epoch": 0.6,
"learning_rate": 3.0014056048493365e-05,
"loss": 0.3267,
"step": 9100
},
{
"epoch": 0.6,
"learning_rate": 2.9904243169638936e-05,
"loss": 0.3231,
"step": 9150
},
{
"epoch": 0.61,
"learning_rate": 2.9794430290784503e-05,
"loss": 0.3071,
"step": 9200
},
{
"epoch": 0.61,
"learning_rate": 2.9684617411930073e-05,
"loss": 0.3306,
"step": 9250
},
{
"epoch": 0.61,
"learning_rate": 2.957480453307564e-05,
"loss": 0.3086,
"step": 9300
},
{
"epoch": 0.62,
"learning_rate": 2.9464991654221206e-05,
"loss": 0.3199,
"step": 9350
},
{
"epoch": 0.62,
"learning_rate": 2.9355178775366777e-05,
"loss": 0.3297,
"step": 9400
},
{
"epoch": 0.62,
"learning_rate": 2.9245365896512343e-05,
"loss": 0.3245,
"step": 9450
},
{
"epoch": 0.63,
"learning_rate": 2.9135553017657914e-05,
"loss": 0.3161,
"step": 9500
},
{
"epoch": 0.63,
"learning_rate": 2.902574013880348e-05,
"loss": 0.3199,
"step": 9550
},
{
"epoch": 0.63,
"learning_rate": 2.891592725994905e-05,
"loss": 0.3292,
"step": 9600
},
{
"epoch": 0.64,
"learning_rate": 2.8806114381094618e-05,
"loss": 0.3242,
"step": 9650
},
{
"epoch": 0.64,
"learning_rate": 2.8696301502240184e-05,
"loss": 0.3254,
"step": 9700
},
{
"epoch": 0.64,
"learning_rate": 2.8586488623385755e-05,
"loss": 0.341,
"step": 9750
},
{
"epoch": 0.65,
"learning_rate": 2.847667574453132e-05,
"loss": 0.3195,
"step": 9800
},
{
"epoch": 0.65,
"learning_rate": 2.8366862865676885e-05,
"loss": 0.3351,
"step": 9850
},
{
"epoch": 0.65,
"learning_rate": 2.8257049986822455e-05,
"loss": 0.333,
"step": 9900
},
{
"epoch": 0.66,
"learning_rate": 2.8147237107968022e-05,
"loss": 0.3332,
"step": 9950
},
{
"epoch": 0.66,
"learning_rate": 2.803742422911359e-05,
"loss": 0.3279,
"step": 10000
},
{
"epoch": 0.66,
"learning_rate": 2.792761135025916e-05,
"loss": 0.3566,
"step": 10050
},
{
"epoch": 0.67,
"learning_rate": 2.7817798471404726e-05,
"loss": 0.3266,
"step": 10100
},
{
"epoch": 0.67,
"learning_rate": 2.7707985592550296e-05,
"loss": 0.3283,
"step": 10150
},
{
"epoch": 0.67,
"learning_rate": 2.7598172713695863e-05,
"loss": 0.3302,
"step": 10200
},
{
"epoch": 0.68,
"learning_rate": 2.7488359834841433e-05,
"loss": 0.3278,
"step": 10250
},
{
"epoch": 0.68,
"learning_rate": 2.7378546955987e-05,
"loss": 0.3144,
"step": 10300
},
{
"epoch": 0.68,
"learning_rate": 2.7268734077132567e-05,
"loss": 0.3418,
"step": 10350
},
{
"epoch": 0.69,
"learning_rate": 2.7158921198278137e-05,
"loss": 0.3218,
"step": 10400
},
{
"epoch": 0.69,
"learning_rate": 2.7049108319423704e-05,
"loss": 0.3113,
"step": 10450
},
{
"epoch": 0.69,
"learning_rate": 2.6939295440569274e-05,
"loss": 0.3078,
"step": 10500
},
{
"epoch": 0.7,
"learning_rate": 2.682948256171484e-05,
"loss": 0.3112,
"step": 10550
},
{
"epoch": 0.7,
"learning_rate": 2.671966968286041e-05,
"loss": 0.3261,
"step": 10600
},
{
"epoch": 0.7,
"learning_rate": 2.660985680400597e-05,
"loss": 0.3324,
"step": 10650
},
{
"epoch": 0.71,
"learning_rate": 2.650004392515154e-05,
"loss": 0.3323,
"step": 10700
},
{
"epoch": 0.71,
"learning_rate": 2.6390231046297108e-05,
"loss": 0.337,
"step": 10750
},
{
"epoch": 0.71,
"learning_rate": 2.628041816744268e-05,
"loss": 0.3037,
"step": 10800
},
{
"epoch": 0.71,
"learning_rate": 2.6170605288588245e-05,
"loss": 0.3261,
"step": 10850
},
{
"epoch": 0.72,
"learning_rate": 2.6060792409733815e-05,
"loss": 0.3229,
"step": 10900
},
{
"epoch": 0.72,
"learning_rate": 2.5950979530879382e-05,
"loss": 0.3178,
"step": 10950
},
{
"epoch": 0.72,
"learning_rate": 2.584116665202495e-05,
"loss": 0.3231,
"step": 11000
},
{
"epoch": 0.73,
"learning_rate": 2.573135377317052e-05,
"loss": 0.332,
"step": 11050
},
{
"epoch": 0.73,
"learning_rate": 2.5621540894316086e-05,
"loss": 0.3495,
"step": 11100
},
{
"epoch": 0.73,
"learning_rate": 2.5511728015461656e-05,
"loss": 0.3164,
"step": 11150
},
{
"epoch": 0.74,
"learning_rate": 2.5401915136607223e-05,
"loss": 0.3304,
"step": 11200
},
{
"epoch": 0.74,
"learning_rate": 2.5292102257752793e-05,
"loss": 0.3075,
"step": 11250
},
{
"epoch": 0.74,
"learning_rate": 2.518228937889836e-05,
"loss": 0.3179,
"step": 11300
},
{
"epoch": 0.75,
"learning_rate": 2.507247650004393e-05,
"loss": 0.3306,
"step": 11350
},
{
"epoch": 0.75,
"learning_rate": 2.4962663621189494e-05,
"loss": 0.3231,
"step": 11400
},
{
"epoch": 0.75,
"learning_rate": 2.4852850742335064e-05,
"loss": 0.3179,
"step": 11450
},
{
"epoch": 0.76,
"learning_rate": 2.474303786348063e-05,
"loss": 0.3164,
"step": 11500
},
{
"epoch": 0.76,
"learning_rate": 2.4633224984626198e-05,
"loss": 0.3158,
"step": 11550
},
{
"epoch": 0.76,
"learning_rate": 2.4523412105771765e-05,
"loss": 0.306,
"step": 11600
},
{
"epoch": 0.77,
"learning_rate": 2.441359922691733e-05,
"loss": 0.3362,
"step": 11650
},
{
"epoch": 0.77,
"learning_rate": 2.43037863480629e-05,
"loss": 0.3257,
"step": 11700
},
{
"epoch": 0.77,
"learning_rate": 2.419397346920847e-05,
"loss": 0.3358,
"step": 11750
},
{
"epoch": 0.78,
"learning_rate": 2.408416059035404e-05,
"loss": 0.32,
"step": 11800
},
{
"epoch": 0.78,
"learning_rate": 2.3974347711499606e-05,
"loss": 0.3149,
"step": 11850
},
{
"epoch": 0.78,
"learning_rate": 2.3864534832645176e-05,
"loss": 0.3087,
"step": 11900
},
{
"epoch": 0.79,
"learning_rate": 2.3754721953790743e-05,
"loss": 0.3213,
"step": 11950
},
{
"epoch": 0.79,
"learning_rate": 2.364490907493631e-05,
"loss": 0.3196,
"step": 12000
},
{
"epoch": 0.79,
"learning_rate": 2.3535096196081876e-05,
"loss": 0.2998,
"step": 12050
},
{
"epoch": 0.8,
"learning_rate": 2.3425283317227446e-05,
"loss": 0.3216,
"step": 12100
},
{
"epoch": 0.8,
"learning_rate": 2.3315470438373013e-05,
"loss": 0.3236,
"step": 12150
},
{
"epoch": 0.8,
"learning_rate": 2.320565755951858e-05,
"loss": 0.344,
"step": 12200
},
{
"epoch": 0.81,
"learning_rate": 2.309584468066415e-05,
"loss": 0.3096,
"step": 12250
},
{
"epoch": 0.81,
"learning_rate": 2.2986031801809717e-05,
"loss": 0.3119,
"step": 12300
},
{
"epoch": 0.81,
"learning_rate": 2.2876218922955287e-05,
"loss": 0.3272,
"step": 12350
},
{
"epoch": 0.82,
"learning_rate": 2.276640604410085e-05,
"loss": 0.3193,
"step": 12400
},
{
"epoch": 0.82,
"learning_rate": 2.265659316524642e-05,
"loss": 0.31,
"step": 12450
},
{
"epoch": 0.82,
"learning_rate": 2.2546780286391988e-05,
"loss": 0.3113,
"step": 12500
},
{
"epoch": 0.83,
"learning_rate": 2.2436967407537558e-05,
"loss": 0.3177,
"step": 12550
},
{
"epoch": 0.83,
"learning_rate": 2.2327154528683125e-05,
"loss": 0.3284,
"step": 12600
},
{
"epoch": 0.83,
"learning_rate": 2.2217341649828692e-05,
"loss": 0.3174,
"step": 12650
},
{
"epoch": 0.84,
"learning_rate": 2.2107528770974262e-05,
"loss": 0.3145,
"step": 12700
},
{
"epoch": 0.84,
"learning_rate": 2.199771589211983e-05,
"loss": 0.3258,
"step": 12750
},
{
"epoch": 0.84,
"learning_rate": 2.1887903013265396e-05,
"loss": 0.3307,
"step": 12800
},
{
"epoch": 0.85,
"learning_rate": 2.1778090134410963e-05,
"loss": 0.319,
"step": 12850
},
{
"epoch": 0.85,
"learning_rate": 2.1668277255556533e-05,
"loss": 0.3349,
"step": 12900
},
{
"epoch": 0.85,
"learning_rate": 2.15584643767021e-05,
"loss": 0.3216,
"step": 12950
},
{
"epoch": 0.86,
"learning_rate": 2.144865149784767e-05,
"loss": 0.3324,
"step": 13000
},
{
"epoch": 0.86,
"learning_rate": 2.1338838618993237e-05,
"loss": 0.3192,
"step": 13050
},
{
"epoch": 0.86,
"learning_rate": 2.1229025740138807e-05,
"loss": 0.3233,
"step": 13100
},
{
"epoch": 0.87,
"learning_rate": 2.1119212861284374e-05,
"loss": 0.3202,
"step": 13150
},
{
"epoch": 0.87,
"learning_rate": 2.100939998242994e-05,
"loss": 0.3126,
"step": 13200
},
{
"epoch": 0.87,
"learning_rate": 2.0899587103575507e-05,
"loss": 0.3187,
"step": 13250
},
{
"epoch": 0.88,
"learning_rate": 2.0789774224721074e-05,
"loss": 0.3324,
"step": 13300
},
{
"epoch": 0.88,
"learning_rate": 2.0679961345866644e-05,
"loss": 0.3228,
"step": 13350
},
{
"epoch": 0.88,
"learning_rate": 2.057014846701221e-05,
"loss": 0.3118,
"step": 13400
},
{
"epoch": 0.89,
"learning_rate": 2.046033558815778e-05,
"loss": 0.3324,
"step": 13450
},
{
"epoch": 0.89,
"learning_rate": 2.0350522709303348e-05,
"loss": 0.3133,
"step": 13500
},
{
"epoch": 0.89,
"learning_rate": 2.024070983044892e-05,
"loss": 0.3205,
"step": 13550
},
{
"epoch": 0.9,
"learning_rate": 2.0130896951594482e-05,
"loss": 0.3024,
"step": 13600
},
{
"epoch": 0.9,
"learning_rate": 2.0021084072740052e-05,
"loss": 0.3093,
"step": 13650
},
{
"epoch": 0.9,
"learning_rate": 1.991127119388562e-05,
"loss": 0.322,
"step": 13700
},
{
"epoch": 0.91,
"learning_rate": 1.980145831503119e-05,
"loss": 0.3094,
"step": 13750
},
{
"epoch": 0.91,
"learning_rate": 1.9691645436176756e-05,
"loss": 0.3242,
"step": 13800
},
{
"epoch": 0.91,
"learning_rate": 1.9581832557322323e-05,
"loss": 0.3485,
"step": 13850
},
{
"epoch": 0.92,
"learning_rate": 1.9472019678467893e-05,
"loss": 0.3369,
"step": 13900
},
{
"epoch": 0.92,
"learning_rate": 1.936220679961346e-05,
"loss": 0.3137,
"step": 13950
},
{
"epoch": 0.92,
"learning_rate": 1.9252393920759027e-05,
"loss": 0.3176,
"step": 14000
},
{
"epoch": 0.93,
"learning_rate": 1.9142581041904594e-05,
"loss": 0.3194,
"step": 14050
},
{
"epoch": 0.93,
"learning_rate": 1.9032768163050164e-05,
"loss": 0.3208,
"step": 14100
},
{
"epoch": 0.93,
"learning_rate": 1.892295528419573e-05,
"loss": 0.3322,
"step": 14150
},
{
"epoch": 0.94,
"learning_rate": 1.88131424053413e-05,
"loss": 0.3092,
"step": 14200
},
{
"epoch": 0.94,
"learning_rate": 1.8703329526486868e-05,
"loss": 0.3264,
"step": 14250
},
{
"epoch": 0.94,
"learning_rate": 1.8593516647632438e-05,
"loss": 0.3188,
"step": 14300
},
{
"epoch": 0.95,
"learning_rate": 1.8483703768778e-05,
"loss": 0.2941,
"step": 14350
},
{
"epoch": 0.95,
"learning_rate": 1.837389088992357e-05,
"loss": 0.3129,
"step": 14400
},
{
"epoch": 0.95,
"learning_rate": 1.826407801106914e-05,
"loss": 0.3155,
"step": 14450
},
{
"epoch": 0.96,
"learning_rate": 1.8154265132214705e-05,
"loss": 0.3256,
"step": 14500
},
{
"epoch": 0.96,
"learning_rate": 1.8044452253360275e-05,
"loss": 0.3219,
"step": 14550
},
{
"epoch": 0.96,
"learning_rate": 1.7934639374505842e-05,
"loss": 0.3102,
"step": 14600
},
{
"epoch": 0.97,
"learning_rate": 1.7824826495651412e-05,
"loss": 0.3404,
"step": 14650
},
{
"epoch": 0.97,
"learning_rate": 1.771501361679698e-05,
"loss": 0.3211,
"step": 14700
},
{
"epoch": 0.97,
"learning_rate": 1.7605200737942546e-05,
"loss": 0.3272,
"step": 14750
},
{
"epoch": 0.98,
"learning_rate": 1.7495387859088113e-05,
"loss": 0.3323,
"step": 14800
},
{
"epoch": 0.98,
"learning_rate": 1.7385574980233683e-05,
"loss": 0.3059,
"step": 14850
},
{
"epoch": 0.98,
"learning_rate": 1.727576210137925e-05,
"loss": 0.3086,
"step": 14900
},
{
"epoch": 0.99,
"learning_rate": 1.716594922252482e-05,
"loss": 0.3152,
"step": 14950
},
{
"epoch": 0.99,
"learning_rate": 1.7056136343670387e-05,
"loss": 0.3097,
"step": 15000
},
{
"epoch": 0.99,
"learning_rate": 1.6946323464815954e-05,
"loss": 0.3306,
"step": 15050
},
{
"epoch": 0.99,
"learning_rate": 1.6836510585961524e-05,
"loss": 0.3157,
"step": 15100
},
{
"epoch": 1.0,
"learning_rate": 1.6726697707107088e-05,
"loss": 0.3206,
"step": 15150
},
{
"epoch": 1.0,
"learning_rate": 1.6616884828252658e-05,
"loss": 0.2961,
"step": 15200
},
{
"epoch": 1.0,
"learning_rate": 1.6507071949398225e-05,
"loss": 0.274,
"step": 15250
},
{
"epoch": 1.01,
"learning_rate": 1.6397259070543795e-05,
"loss": 0.2576,
"step": 15300
},
{
"epoch": 1.01,
"learning_rate": 1.628744619168936e-05,
"loss": 0.2774,
"step": 15350
},
{
"epoch": 1.01,
"learning_rate": 1.6177633312834932e-05,
"loss": 0.2734,
"step": 15400
},
{
"epoch": 1.02,
"learning_rate": 1.60678204339805e-05,
"loss": 0.2656,
"step": 15450
},
{
"epoch": 1.02,
"learning_rate": 1.5958007555126066e-05,
"loss": 0.2605,
"step": 15500
},
{
"epoch": 1.02,
"learning_rate": 1.5848194676271632e-05,
"loss": 0.2601,
"step": 15550
},
{
"epoch": 1.03,
"learning_rate": 1.5738381797417203e-05,
"loss": 0.2673,
"step": 15600
},
{
"epoch": 1.03,
"learning_rate": 1.562856891856277e-05,
"loss": 0.2604,
"step": 15650
},
{
"epoch": 1.03,
"learning_rate": 1.5518756039708336e-05,
"loss": 0.2779,
"step": 15700
},
{
"epoch": 1.04,
"learning_rate": 1.5408943160853906e-05,
"loss": 0.2599,
"step": 15750
},
{
"epoch": 1.04,
"learning_rate": 1.5299130281999473e-05,
"loss": 0.2726,
"step": 15800
},
{
"epoch": 1.04,
"learning_rate": 1.5189317403145042e-05,
"loss": 0.2679,
"step": 15850
},
{
"epoch": 1.05,
"learning_rate": 1.507950452429061e-05,
"loss": 0.2743,
"step": 15900
},
{
"epoch": 1.05,
"learning_rate": 1.4969691645436175e-05,
"loss": 0.2561,
"step": 15950
},
{
"epoch": 1.05,
"learning_rate": 1.4859878766581744e-05,
"loss": 0.2685,
"step": 16000
},
{
"epoch": 1.06,
"learning_rate": 1.4750065887727313e-05,
"loss": 0.2598,
"step": 16050
},
{
"epoch": 1.06,
"learning_rate": 1.4640253008872881e-05,
"loss": 0.2625,
"step": 16100
},
{
"epoch": 1.06,
"learning_rate": 1.453044013001845e-05,
"loss": 0.2811,
"step": 16150
},
{
"epoch": 1.07,
"learning_rate": 1.4420627251164018e-05,
"loss": 0.2613,
"step": 16200
},
{
"epoch": 1.07,
"learning_rate": 1.4310814372309587e-05,
"loss": 0.26,
"step": 16250
},
{
"epoch": 1.07,
"learning_rate": 1.4201001493455155e-05,
"loss": 0.2716,
"step": 16300
},
{
"epoch": 1.08,
"learning_rate": 1.409118861460072e-05,
"loss": 0.2665,
"step": 16350
},
{
"epoch": 1.08,
"learning_rate": 1.3981375735746289e-05,
"loss": 0.2632,
"step": 16400
},
{
"epoch": 1.08,
"learning_rate": 1.3871562856891857e-05,
"loss": 0.2613,
"step": 16450
},
{
"epoch": 1.09,
"learning_rate": 1.3761749978037424e-05,
"loss": 0.2654,
"step": 16500
},
{
"epoch": 1.09,
"learning_rate": 1.3651937099182993e-05,
"loss": 0.2674,
"step": 16550
},
{
"epoch": 1.09,
"learning_rate": 1.3542124220328561e-05,
"loss": 0.2626,
"step": 16600
},
{
"epoch": 1.1,
"learning_rate": 1.343231134147413e-05,
"loss": 0.2727,
"step": 16650
},
{
"epoch": 1.1,
"learning_rate": 1.3322498462619698e-05,
"loss": 0.2634,
"step": 16700
},
{
"epoch": 1.1,
"learning_rate": 1.3212685583765263e-05,
"loss": 0.2635,
"step": 16750
},
{
"epoch": 1.11,
"learning_rate": 1.3102872704910832e-05,
"loss": 0.2658,
"step": 16800
},
{
"epoch": 1.11,
"learning_rate": 1.29930598260564e-05,
"loss": 0.2598,
"step": 16850
},
{
"epoch": 1.11,
"learning_rate": 1.2883246947201969e-05,
"loss": 0.255,
"step": 16900
},
{
"epoch": 1.12,
"learning_rate": 1.2773434068347537e-05,
"loss": 0.261,
"step": 16950
},
{
"epoch": 1.12,
"learning_rate": 1.2663621189493104e-05,
"loss": 0.2717,
"step": 17000
},
{
"epoch": 1.12,
"learning_rate": 1.2553808310638673e-05,
"loss": 0.2642,
"step": 17050
},
{
"epoch": 1.13,
"learning_rate": 1.244399543178424e-05,
"loss": 0.2796,
"step": 17100
},
{
"epoch": 1.13,
"learning_rate": 1.2334182552929808e-05,
"loss": 0.2584,
"step": 17150
},
{
"epoch": 1.13,
"learning_rate": 1.2224369674075375e-05,
"loss": 0.2646,
"step": 17200
},
{
"epoch": 1.14,
"learning_rate": 1.2114556795220944e-05,
"loss": 0.2682,
"step": 17250
},
{
"epoch": 1.14,
"learning_rate": 1.2004743916366512e-05,
"loss": 0.2564,
"step": 17300
},
{
"epoch": 1.14,
"learning_rate": 1.189493103751208e-05,
"loss": 0.2637,
"step": 17350
},
{
"epoch": 1.15,
"learning_rate": 1.1785118158657647e-05,
"loss": 0.2753,
"step": 17400
},
{
"epoch": 1.15,
"learning_rate": 1.1675305279803216e-05,
"loss": 0.253,
"step": 17450
},
{
"epoch": 1.15,
"learning_rate": 1.1565492400948784e-05,
"loss": 0.2607,
"step": 17500
},
{
"epoch": 1.16,
"learning_rate": 1.1455679522094353e-05,
"loss": 0.2691,
"step": 17550
},
{
"epoch": 1.16,
"learning_rate": 1.134586664323992e-05,
"loss": 0.2659,
"step": 17600
},
{
"epoch": 1.16,
"learning_rate": 1.1236053764385487e-05,
"loss": 0.2757,
"step": 17650
},
{
"epoch": 1.17,
"learning_rate": 1.1126240885531055e-05,
"loss": 0.2654,
"step": 17700
},
{
"epoch": 1.17,
"learning_rate": 1.1016428006676624e-05,
"loss": 0.2575,
"step": 17750
},
{
"epoch": 1.17,
"learning_rate": 1.090661512782219e-05,
"loss": 0.262,
"step": 17800
},
{
"epoch": 1.18,
"learning_rate": 1.0796802248967759e-05,
"loss": 0.2686,
"step": 17850
},
{
"epoch": 1.18,
"learning_rate": 1.0686989370113328e-05,
"loss": 0.2808,
"step": 17900
},
{
"epoch": 1.18,
"learning_rate": 1.0577176491258896e-05,
"loss": 0.286,
"step": 17950
},
{
"epoch": 1.19,
"learning_rate": 1.0467363612404463e-05,
"loss": 0.2578,
"step": 18000
},
{
"epoch": 1.19,
"learning_rate": 1.0357550733550031e-05,
"loss": 0.2699,
"step": 18050
},
{
"epoch": 1.19,
"learning_rate": 1.02477378546956e-05,
"loss": 0.2848,
"step": 18100
},
{
"epoch": 1.2,
"learning_rate": 1.0137924975841167e-05,
"loss": 0.272,
"step": 18150
},
{
"epoch": 1.2,
"learning_rate": 1.0028112096986735e-05,
"loss": 0.2658,
"step": 18200
},
{
"epoch": 1.2,
"learning_rate": 9.918299218132302e-06,
"loss": 0.2617,
"step": 18250
},
{
"epoch": 1.21,
"learning_rate": 9.80848633927787e-06,
"loss": 0.248,
"step": 18300
},
{
"epoch": 1.21,
"learning_rate": 9.69867346042344e-06,
"loss": 0.2762,
"step": 18350
},
{
"epoch": 1.21,
"learning_rate": 9.588860581569006e-06,
"loss": 0.2617,
"step": 18400
},
{
"epoch": 1.22,
"learning_rate": 9.479047702714575e-06,
"loss": 0.2676,
"step": 18450
},
{
"epoch": 1.22,
"learning_rate": 9.369234823860143e-06,
"loss": 0.2782,
"step": 18500
},
{
"epoch": 1.22,
"learning_rate": 9.259421945005712e-06,
"loss": 0.2547,
"step": 18550
},
{
"epoch": 1.23,
"learning_rate": 9.149609066151278e-06,
"loss": 0.2564,
"step": 18600
},
{
"epoch": 1.23,
"learning_rate": 9.039796187296847e-06,
"loss": 0.2579,
"step": 18650
},
{
"epoch": 1.23,
"learning_rate": 8.929983308442416e-06,
"loss": 0.2592,
"step": 18700
},
{
"epoch": 1.24,
"learning_rate": 8.820170429587982e-06,
"loss": 0.2512,
"step": 18750
},
{
"epoch": 1.24,
"learning_rate": 8.71035755073355e-06,
"loss": 0.2632,
"step": 18800
},
{
"epoch": 1.24,
"learning_rate": 8.600544671879118e-06,
"loss": 0.2584,
"step": 18850
},
{
"epoch": 1.25,
"learning_rate": 8.490731793024686e-06,
"loss": 0.2586,
"step": 18900
},
{
"epoch": 1.25,
"learning_rate": 8.380918914170255e-06,
"loss": 0.2536,
"step": 18950
},
{
"epoch": 1.25,
"learning_rate": 8.271106035315822e-06,
"loss": 0.282,
"step": 19000
},
{
"epoch": 1.26,
"learning_rate": 8.16129315646139e-06,
"loss": 0.2616,
"step": 19050
},
{
"epoch": 1.26,
"learning_rate": 8.051480277606959e-06,
"loss": 0.2584,
"step": 19100
},
{
"epoch": 1.26,
"learning_rate": 7.941667398752527e-06,
"loss": 0.2528,
"step": 19150
},
{
"epoch": 1.27,
"learning_rate": 7.831854519898094e-06,
"loss": 0.267,
"step": 19200
},
{
"epoch": 1.27,
"learning_rate": 7.722041641043663e-06,
"loss": 0.2669,
"step": 19250
},
{
"epoch": 1.27,
"learning_rate": 7.61222876218923e-06,
"loss": 0.2728,
"step": 19300
},
{
"epoch": 1.27,
"learning_rate": 7.502415883334799e-06,
"loss": 0.2457,
"step": 19350
},
{
"epoch": 1.28,
"learning_rate": 7.3926030044803656e-06,
"loss": 0.2678,
"step": 19400
},
{
"epoch": 1.28,
"learning_rate": 7.282790125625934e-06,
"loss": 0.2707,
"step": 19450
},
{
"epoch": 1.28,
"learning_rate": 7.172977246771502e-06,
"loss": 0.2645,
"step": 19500
},
{
"epoch": 1.29,
"learning_rate": 7.06316436791707e-06,
"loss": 0.2557,
"step": 19550
},
{
"epoch": 1.29,
"learning_rate": 6.953351489062637e-06,
"loss": 0.2571,
"step": 19600
},
{
"epoch": 1.29,
"learning_rate": 6.843538610208206e-06,
"loss": 0.2709,
"step": 19650
},
{
"epoch": 1.3,
"learning_rate": 6.733725731353774e-06,
"loss": 0.2571,
"step": 19700
},
{
"epoch": 1.3,
"learning_rate": 6.623912852499342e-06,
"loss": 0.2688,
"step": 19750
},
{
"epoch": 1.3,
"learning_rate": 6.514099973644909e-06,
"loss": 0.2516,
"step": 19800
},
{
"epoch": 1.31,
"learning_rate": 6.404287094790477e-06,
"loss": 0.2723,
"step": 19850
},
{
"epoch": 1.31,
"learning_rate": 6.294474215936046e-06,
"loss": 0.2799,
"step": 19900
},
{
"epoch": 1.31,
"learning_rate": 6.184661337081613e-06,
"loss": 0.2623,
"step": 19950
},
{
"epoch": 1.32,
"learning_rate": 6.074848458227181e-06,
"loss": 0.2735,
"step": 20000
},
{
"epoch": 1.32,
"learning_rate": 5.965035579372749e-06,
"loss": 0.279,
"step": 20050
},
{
"epoch": 1.32,
"learning_rate": 5.8552227005183164e-06,
"loss": 0.2725,
"step": 20100
},
{
"epoch": 1.33,
"learning_rate": 5.745409821663885e-06,
"loss": 0.2624,
"step": 20150
},
{
"epoch": 1.33,
"learning_rate": 5.635596942809453e-06,
"loss": 0.275,
"step": 20200
},
{
"epoch": 1.33,
"learning_rate": 5.525784063955021e-06,
"loss": 0.2655,
"step": 20250
},
{
"epoch": 1.34,
"learning_rate": 5.415971185100589e-06,
"loss": 0.2715,
"step": 20300
},
{
"epoch": 1.34,
"learning_rate": 5.3061583062461565e-06,
"loss": 0.2715,
"step": 20350
},
{
"epoch": 1.34,
"learning_rate": 5.196345427391724e-06,
"loss": 0.2634,
"step": 20400
},
{
"epoch": 1.35,
"learning_rate": 5.086532548537293e-06,
"loss": 0.2636,
"step": 20450
},
{
"epoch": 1.35,
"learning_rate": 4.97671966968286e-06,
"loss": 0.2615,
"step": 20500
},
{
"epoch": 1.35,
"learning_rate": 4.866906790828429e-06,
"loss": 0.2644,
"step": 20550
},
{
"epoch": 1.36,
"learning_rate": 4.757093911973997e-06,
"loss": 0.2813,
"step": 20600
},
{
"epoch": 1.36,
"learning_rate": 4.647281033119564e-06,
"loss": 0.2662,
"step": 20650
},
{
"epoch": 1.36,
"learning_rate": 4.537468154265132e-06,
"loss": 0.2644,
"step": 20700
},
{
"epoch": 1.37,
"learning_rate": 4.4276552754107005e-06,
"loss": 0.2724,
"step": 20750
},
{
"epoch": 1.37,
"learning_rate": 4.317842396556268e-06,
"loss": 0.2706,
"step": 20800
},
{
"epoch": 1.37,
"learning_rate": 4.208029517701837e-06,
"loss": 0.2662,
"step": 20850
},
{
"epoch": 1.38,
"learning_rate": 4.098216638847404e-06,
"loss": 0.261,
"step": 20900
},
{
"epoch": 1.38,
"learning_rate": 3.988403759992972e-06,
"loss": 0.247,
"step": 20950
},
{
"epoch": 1.38,
"learning_rate": 3.87859088113854e-06,
"loss": 0.272,
"step": 21000
},
{
"epoch": 1.39,
"learning_rate": 3.7687780022841083e-06,
"loss": 0.2712,
"step": 21050
},
{
"epoch": 1.39,
"learning_rate": 3.658965123429676e-06,
"loss": 0.2762,
"step": 21100
},
{
"epoch": 1.39,
"learning_rate": 3.549152244575244e-06,
"loss": 0.2703,
"step": 21150
},
{
"epoch": 1.4,
"learning_rate": 3.4393393657208117e-06,
"loss": 0.2565,
"step": 21200
},
{
"epoch": 1.4,
"learning_rate": 3.3295264868663802e-06,
"loss": 0.2615,
"step": 21250
},
{
"epoch": 1.4,
"learning_rate": 3.2197136080119475e-06,
"loss": 0.2644,
"step": 21300
},
{
"epoch": 1.41,
"learning_rate": 3.1099007291575156e-06,
"loss": 0.267,
"step": 21350
},
{
"epoch": 1.41,
"learning_rate": 3.0000878503030837e-06,
"loss": 0.264,
"step": 21400
},
{
"epoch": 1.41,
"learning_rate": 2.8902749714486514e-06,
"loss": 0.2754,
"step": 21450
},
{
"epoch": 1.42,
"learning_rate": 2.7804620925942195e-06,
"loss": 0.2544,
"step": 21500
},
{
"epoch": 1.42,
"learning_rate": 2.6706492137397876e-06,
"loss": 0.262,
"step": 21550
},
{
"epoch": 1.42,
"learning_rate": 2.5608363348853552e-06,
"loss": 0.2513,
"step": 21600
},
{
"epoch": 1.43,
"learning_rate": 2.4510234560309233e-06,
"loss": 0.26,
"step": 21650
},
{
"epoch": 1.43,
"learning_rate": 2.3412105771764915e-06,
"loss": 0.2577,
"step": 21700
},
{
"epoch": 1.43,
"learning_rate": 2.231397698322059e-06,
"loss": 0.2552,
"step": 21750
},
{
"epoch": 1.44,
"learning_rate": 2.1215848194676272e-06,
"loss": 0.268,
"step": 21800
},
{
"epoch": 1.44,
"learning_rate": 2.011771940613195e-06,
"loss": 0.267,
"step": 21850
},
{
"epoch": 1.44,
"learning_rate": 1.901959061758763e-06,
"loss": 0.2688,
"step": 21900
},
{
"epoch": 1.45,
"learning_rate": 1.7921461829043311e-06,
"loss": 0.2548,
"step": 21950
},
{
"epoch": 1.45,
"learning_rate": 1.682333304049899e-06,
"loss": 0.2638,
"step": 22000
},
{
"epoch": 1.45,
"learning_rate": 1.5725204251954669e-06,
"loss": 0.2742,
"step": 22050
},
{
"epoch": 1.46,
"learning_rate": 1.4627075463410348e-06,
"loss": 0.2684,
"step": 22100
},
{
"epoch": 1.46,
"learning_rate": 1.3528946674866029e-06,
"loss": 0.26,
"step": 22150
},
{
"epoch": 1.46,
"learning_rate": 1.2430817886321708e-06,
"loss": 0.2573,
"step": 22200
},
{
"epoch": 1.47,
"learning_rate": 1.1332689097777387e-06,
"loss": 0.2736,
"step": 22250
},
{
"epoch": 1.47,
"learning_rate": 1.0234560309233068e-06,
"loss": 0.2732,
"step": 22300
},
{
"epoch": 1.47,
"learning_rate": 9.136431520688747e-07,
"loss": 0.2625,
"step": 22350
},
{
"epoch": 1.48,
"learning_rate": 8.038302732144426e-07,
"loss": 0.2649,
"step": 22400
},
{
"epoch": 1.48,
"learning_rate": 6.940173943600105e-07,
"loss": 0.2646,
"step": 22450
},
{
"epoch": 1.48,
"learning_rate": 5.842045155055785e-07,
"loss": 0.2688,
"step": 22500
}
],
"logging_steps": 50,
"max_steps": 22766,
"num_train_epochs": 2,
"save_steps": 500,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}