autoprogrammer's picture
Upload sdar_1.7b_random_mask_math-checkpoint-537
f665667 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 537,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.027932960893854747,
"grad_norm": 6.217156410217285,
"learning_rate": 2.3529411764705885e-06,
"loss": 0.4419,
"step": 5
},
{
"epoch": 0.055865921787709494,
"grad_norm": 5.510581970214844,
"learning_rate": 5.294117647058824e-06,
"loss": 0.4127,
"step": 10
},
{
"epoch": 0.08379888268156424,
"grad_norm": 5.5799174308776855,
"learning_rate": 8.23529411764706e-06,
"loss": 0.4641,
"step": 15
},
{
"epoch": 0.11173184357541899,
"grad_norm": 7.177088737487793,
"learning_rate": 9.999635004278054e-06,
"loss": 0.4635,
"step": 20
},
{
"epoch": 0.13966480446927373,
"grad_norm": 6.450545310974121,
"learning_rate": 9.995529414369582e-06,
"loss": 0.4839,
"step": 25
},
{
"epoch": 0.16759776536312848,
"grad_norm": 5.527288913726807,
"learning_rate": 9.986865748457457e-06,
"loss": 0.4175,
"step": 30
},
{
"epoch": 0.19553072625698323,
"grad_norm": 5.916306972503662,
"learning_rate": 9.973651911539209e-06,
"loss": 0.4671,
"step": 35
},
{
"epoch": 0.22346368715083798,
"grad_norm": 5.527400493621826,
"learning_rate": 9.955899960329546e-06,
"loss": 0.4643,
"step": 40
},
{
"epoch": 0.25139664804469275,
"grad_norm": 5.982852458953857,
"learning_rate": 9.933626092259439e-06,
"loss": 0.4754,
"step": 45
},
{
"epoch": 0.27932960893854747,
"grad_norm": 5.4895710945129395,
"learning_rate": 9.906850630697068e-06,
"loss": 0.4728,
"step": 50
},
{
"epoch": 0.30726256983240224,
"grad_norm": 6.2167463302612305,
"learning_rate": 9.875598006404164e-06,
"loss": 0.4711,
"step": 55
},
{
"epoch": 0.33519553072625696,
"grad_norm": 5.332173824310303,
"learning_rate": 9.839896735244615e-06,
"loss": 0.4728,
"step": 60
},
{
"epoch": 0.36312849162011174,
"grad_norm": 6.088379383087158,
"learning_rate": 9.799779392165716e-06,
"loss": 0.5057,
"step": 65
},
{
"epoch": 0.39106145251396646,
"grad_norm": 6.17716121673584,
"learning_rate": 9.755282581475769e-06,
"loss": 0.4768,
"step": 70
},
{
"epoch": 0.41899441340782123,
"grad_norm": 7.213939666748047,
"learning_rate": 9.706446903445179e-06,
"loss": 0.5043,
"step": 75
},
{
"epoch": 0.44692737430167595,
"grad_norm": 5.0204691886901855,
"learning_rate": 9.653316917261511e-06,
"loss": 0.4568,
"step": 80
},
{
"epoch": 0.4748603351955307,
"grad_norm": 6.00094747543335,
"learning_rate": 9.595941100372301e-06,
"loss": 0.4883,
"step": 85
},
{
"epoch": 0.5027932960893855,
"grad_norm": 5.573389530181885,
"learning_rate": 9.534371804252727e-06,
"loss": 0.489,
"step": 90
},
{
"epoch": 0.5307262569832403,
"grad_norm": 4.436511039733887,
"learning_rate": 9.46866520663851e-06,
"loss": 0.4552,
"step": 95
},
{
"epoch": 0.5586592178770949,
"grad_norm": 5.606967926025391,
"learning_rate": 9.398881260267589e-06,
"loss": 0.4961,
"step": 100
},
{
"epoch": 0.5865921787709497,
"grad_norm": 6.480439186096191,
"learning_rate": 9.325083638177401e-06,
"loss": 0.5207,
"step": 105
},
{
"epoch": 0.6145251396648045,
"grad_norm": 5.892640590667725,
"learning_rate": 9.247339675607606e-06,
"loss": 0.4952,
"step": 110
},
{
"epoch": 0.6424581005586593,
"grad_norm": 5.1815314292907715,
"learning_rate": 9.165720308561347e-06,
"loss": 0.499,
"step": 115
},
{
"epoch": 0.6703910614525139,
"grad_norm": 5.886406421661377,
"learning_rate": 9.080300009081025e-06,
"loss": 0.4907,
"step": 120
},
{
"epoch": 0.6983240223463687,
"grad_norm": 5.219560146331787,
"learning_rate": 8.991156717297702e-06,
"loss": 0.4764,
"step": 125
},
{
"epoch": 0.7262569832402235,
"grad_norm": 5.498112678527832,
"learning_rate": 8.898371770316113e-06,
"loss": 0.4684,
"step": 130
},
{
"epoch": 0.7541899441340782,
"grad_norm": 5.503330707550049,
"learning_rate": 8.802029828000157e-06,
"loss": 0.4854,
"step": 135
},
{
"epoch": 0.7821229050279329,
"grad_norm": 5.640200614929199,
"learning_rate": 8.702218795726619e-06,
"loss": 0.5091,
"step": 140
},
{
"epoch": 0.8100558659217877,
"grad_norm": 5.910928726196289,
"learning_rate": 8.599029744177577e-06,
"loss": 0.4581,
"step": 145
},
{
"epoch": 0.8379888268156425,
"grad_norm": 7.0951738357543945,
"learning_rate": 8.492556826244687e-06,
"loss": 0.4986,
"step": 150
},
{
"epoch": 0.8659217877094972,
"grad_norm": 4.941314697265625,
"learning_rate": 8.382897191121157e-06,
"loss": 0.4898,
"step": 155
},
{
"epoch": 0.8938547486033519,
"grad_norm": 5.991004467010498,
"learning_rate": 8.270150895659824e-06,
"loss": 0.4634,
"step": 160
},
{
"epoch": 0.9217877094972067,
"grad_norm": 5.34398889541626,
"learning_rate": 8.154420813078155e-06,
"loss": 0.505,
"step": 165
},
{
"epoch": 0.9497206703910615,
"grad_norm": 5.032632350921631,
"learning_rate": 8.035812539093557e-06,
"loss": 0.4554,
"step": 170
},
{
"epoch": 0.9776536312849162,
"grad_norm": 5.789834022521973,
"learning_rate": 7.914434295574552e-06,
"loss": 0.4456,
"step": 175
},
{
"epoch": 1.005586592178771,
"grad_norm": 3.7482521533966064,
"learning_rate": 7.790396831795792e-06,
"loss": 0.3991,
"step": 180
},
{
"epoch": 1.0335195530726258,
"grad_norm": 3.6354920864105225,
"learning_rate": 7.663813323386988e-06,
"loss": 0.2124,
"step": 185
},
{
"epoch": 1.0614525139664805,
"grad_norm": 4.032209873199463,
"learning_rate": 7.534799269067952e-06,
"loss": 0.1957,
"step": 190
},
{
"epoch": 1.089385474860335,
"grad_norm": 4.484015941619873,
"learning_rate": 7.403472385263979e-06,
"loss": 0.2009,
"step": 195
},
{
"epoch": 1.1173184357541899,
"grad_norm": 4.480172157287598,
"learning_rate": 7.269952498697734e-06,
"loss": 0.1663,
"step": 200
},
{
"epoch": 1.1452513966480447,
"grad_norm": 4.166629314422607,
"learning_rate": 7.134361437055633e-06,
"loss": 0.2036,
"step": 205
},
{
"epoch": 1.1731843575418994,
"grad_norm": 4.196361064910889,
"learning_rate": 6.9968229178284775e-06,
"loss": 0.1817,
"step": 210
},
{
"epoch": 1.2011173184357542,
"grad_norm": 4.6725077629089355,
"learning_rate": 6.8574624354277866e-06,
"loss": 0.1827,
"step": 215
},
{
"epoch": 1.229050279329609,
"grad_norm": 3.796790838241577,
"learning_rate": 6.716407146680793e-06,
"loss": 0.1671,
"step": 220
},
{
"epoch": 1.2569832402234637,
"grad_norm": 4.3832173347473145,
"learning_rate": 6.573785754808615e-06,
"loss": 0.2045,
"step": 225
},
{
"epoch": 1.2849162011173183,
"grad_norm": 4.749040603637695,
"learning_rate": 6.429728391993446e-06,
"loss": 0.2029,
"step": 230
},
{
"epoch": 1.3128491620111733,
"grad_norm": 4.857269287109375,
"learning_rate": 6.284366500641914e-06,
"loss": 0.2017,
"step": 235
},
{
"epoch": 1.3407821229050279,
"grad_norm": 4.255931377410889,
"learning_rate": 6.13783271345295e-06,
"loss": 0.1941,
"step": 240
},
{
"epoch": 1.3687150837988826,
"grad_norm": 4.656070232391357,
"learning_rate": 5.990260732399598e-06,
"loss": 0.2023,
"step": 245
},
{
"epoch": 1.3966480446927374,
"grad_norm": 3.9086103439331055,
"learning_rate": 5.841785206735192e-06,
"loss": 0.1979,
"step": 250
},
{
"epoch": 1.4245810055865922,
"grad_norm": 3.5551209449768066,
"learning_rate": 5.6925416101352e-06,
"loss": 0.1684,
"step": 255
},
{
"epoch": 1.452513966480447,
"grad_norm": 3.7401092052459717,
"learning_rate": 5.542666117086832e-06,
"loss": 0.1843,
"step": 260
},
{
"epoch": 1.4804469273743017,
"grad_norm": 3.9582438468933105,
"learning_rate": 5.392295478639226e-06,
"loss": 0.1844,
"step": 265
},
{
"epoch": 1.5083798882681565,
"grad_norm": 4.367541790008545,
"learning_rate": 5.241566897627536e-06,
"loss": 0.1896,
"step": 270
},
{
"epoch": 1.536312849162011,
"grad_norm": 3.8938488960266113,
"learning_rate": 5.09061790348482e-06,
"loss": 0.2109,
"step": 275
},
{
"epoch": 1.564245810055866,
"grad_norm": 4.649727821350098,
"learning_rate": 4.939586226755913e-06,
"loss": 0.1998,
"step": 280
},
{
"epoch": 1.5921787709497206,
"grad_norm": 4.463720798492432,
"learning_rate": 4.788609673427807e-06,
"loss": 0.1952,
"step": 285
},
{
"epoch": 1.6201117318435754,
"grad_norm": 4.8687424659729,
"learning_rate": 4.637825999191189e-06,
"loss": 0.2003,
"step": 290
},
{
"epoch": 1.6480446927374302,
"grad_norm": 4.026172637939453,
"learning_rate": 4.487372783747884e-06,
"loss": 0.1791,
"step": 295
},
{
"epoch": 1.675977653631285,
"grad_norm": 4.026905536651611,
"learning_rate": 4.337387305278864e-06,
"loss": 0.2009,
"step": 300
},
{
"epoch": 1.7039106145251397,
"grad_norm": 4.886240482330322,
"learning_rate": 4.18800641518737e-06,
"loss": 0.1724,
"step": 305
},
{
"epoch": 1.7318435754189943,
"grad_norm": 3.9127626419067383,
"learning_rate": 4.039366413231458e-06,
"loss": 0.161,
"step": 310
},
{
"epoch": 1.7597765363128492,
"grad_norm": 4.5372724533081055,
"learning_rate": 3.8916029231598655e-06,
"loss": 0.2068,
"step": 315
},
{
"epoch": 1.7877094972067038,
"grad_norm": 4.877954959869385,
"learning_rate": 3.744850768964692e-06,
"loss": 0.1865,
"step": 320
},
{
"epoch": 1.8156424581005588,
"grad_norm": 4.188896179199219,
"learning_rate": 3.599243851863816e-06,
"loss": 0.1953,
"step": 325
},
{
"epoch": 1.8435754189944134,
"grad_norm": 4.291688919067383,
"learning_rate": 3.4549150281252635e-06,
"loss": 0.2032,
"step": 330
},
{
"epoch": 1.8715083798882681,
"grad_norm": 3.2976162433624268,
"learning_rate": 3.3119959878450257e-06,
"loss": 0.1829,
"step": 335
},
{
"epoch": 1.899441340782123,
"grad_norm": 3.9580321311950684,
"learning_rate": 3.170617134788939e-06,
"loss": 0.1744,
"step": 340
},
{
"epoch": 1.9273743016759777,
"grad_norm": 4.327995300292969,
"learning_rate": 3.030907467408235e-06,
"loss": 0.1849,
"step": 345
},
{
"epoch": 1.9553072625698324,
"grad_norm": 4.522080898284912,
"learning_rate": 2.8929944611373555e-06,
"loss": 0.1985,
"step": 350
},
{
"epoch": 1.983240223463687,
"grad_norm": 5.031853675842285,
"learning_rate": 2.757003952081411e-06,
"loss": 0.1873,
"step": 355
},
{
"epoch": 2.011173184357542,
"grad_norm": 2.2975807189941406,
"learning_rate": 2.6230600221994195e-06,
"loss": 0.1476,
"step": 360
},
{
"epoch": 2.0391061452513966,
"grad_norm": 2.7804503440856934,
"learning_rate": 2.49128488608807e-06,
"loss": 0.0766,
"step": 365
},
{
"epoch": 2.0670391061452515,
"grad_norm": 2.3072023391723633,
"learning_rate": 2.3617987794693358e-06,
"loss": 0.0677,
"step": 370
},
{
"epoch": 2.094972067039106,
"grad_norm": 2.175658941268921,
"learning_rate": 2.23471984948368e-06,
"loss": 0.0612,
"step": 375
},
{
"epoch": 2.122905027932961,
"grad_norm": 2.7598915100097656,
"learning_rate": 2.1101640468889255e-06,
"loss": 0.0788,
"step": 380
},
{
"epoch": 2.1508379888268156,
"grad_norm": 3.056736946105957,
"learning_rate": 1.988245020263197e-06,
"loss": 0.065,
"step": 385
},
{
"epoch": 2.17877094972067,
"grad_norm": 3.000897169113159,
"learning_rate": 1.8690740123084316e-06,
"loss": 0.0674,
"step": 390
},
{
"epoch": 2.206703910614525,
"grad_norm": 4.054080963134766,
"learning_rate": 1.7527597583490825e-06,
"loss": 0.0639,
"step": 395
},
{
"epoch": 2.2346368715083798,
"grad_norm": 2.4640097618103027,
"learning_rate": 1.6394083871186362e-06,
"loss": 0.0779,
"step": 400
},
{
"epoch": 2.2625698324022347,
"grad_norm": 3.62778639793396,
"learning_rate": 1.5291233239244728e-06,
"loss": 0.0585,
"step": 405
},
{
"epoch": 2.2905027932960893,
"grad_norm": 3.1452324390411377,
"learning_rate": 1.4220051962793952e-06,
"loss": 0.0579,
"step": 410
},
{
"epoch": 2.3184357541899443,
"grad_norm": 2.2052979469299316,
"learning_rate": 1.3181517420859696e-06,
"loss": 0.0656,
"step": 415
},
{
"epoch": 2.346368715083799,
"grad_norm": 2.898425817489624,
"learning_rate": 1.2176577204574318e-06,
"loss": 0.0652,
"step": 420
},
{
"epoch": 2.3743016759776534,
"grad_norm": 3.3313210010528564,
"learning_rate": 1.1206148252565385e-06,
"loss": 0.0685,
"step": 425
},
{
"epoch": 2.4022346368715084,
"grad_norm": 2.530761480331421,
"learning_rate": 1.0271116014312293e-06,
"loss": 0.073,
"step": 430
},
{
"epoch": 2.430167597765363,
"grad_norm": 2.3693220615386963,
"learning_rate": 9.372333642234787e-07,
"loss": 0.0626,
"step": 435
},
{
"epoch": 2.458100558659218,
"grad_norm": 3.0379860401153564,
"learning_rate": 8.510621213250248e-07,
"loss": 0.0574,
"step": 440
},
{
"epoch": 2.4860335195530725,
"grad_norm": 2.477531909942627,
"learning_rate": 7.686764980509986e-07,
"loss": 0.0572,
"step": 445
},
{
"epoch": 2.5139664804469275,
"grad_norm": 4.053262710571289,
"learning_rate": 6.901516655997536e-07,
"loss": 0.0596,
"step": 450
},
{
"epoch": 2.541899441340782,
"grad_norm": 2.117450714111328,
"learning_rate": 6.155592724643339e-07,
"loss": 0.0602,
"step": 455
},
{
"epoch": 2.5698324022346366,
"grad_norm": 2.347970724105835,
"learning_rate": 5.449673790581611e-07,
"loss": 0.053,
"step": 460
},
{
"epoch": 2.5977653631284916,
"grad_norm": 3.6320173740386963,
"learning_rate": 4.784403956146039e-07,
"loss": 0.0657,
"step": 465
},
{
"epoch": 2.6256983240223466,
"grad_norm": 2.684598207473755,
"learning_rate": 4.1603902341708804e-07,
"loss": 0.0624,
"step": 470
},
{
"epoch": 2.653631284916201,
"grad_norm": 3.334993362426758,
"learning_rate": 3.5782019941335345e-07,
"loss": 0.0691,
"step": 475
},
{
"epoch": 2.6815642458100557,
"grad_norm": 3.143279552459717,
"learning_rate": 3.0383704426442396e-07,
"loss": 0.0642,
"step": 480
},
{
"epoch": 2.7094972067039107,
"grad_norm": 3.094574451446533,
"learning_rate": 2.541388138756645e-07,
"loss": 0.0659,
"step": 485
},
{
"epoch": 2.7374301675977653,
"grad_norm": 2.291473865509033,
"learning_rate": 2.0877085445416889e-07,
"loss": 0.0628,
"step": 490
},
{
"epoch": 2.7653631284916202,
"grad_norm": 2.65637469291687,
"learning_rate": 1.6777456113346857e-07,
"loss": 0.0636,
"step": 495
},
{
"epoch": 2.793296089385475,
"grad_norm": 2.951915740966797,
"learning_rate": 1.3118734020333257e-07,
"loss": 0.0564,
"step": 500
},
{
"epoch": 2.82122905027933,
"grad_norm": 2.7770986557006836,
"learning_rate": 9.904257497910796e-08,
"loss": 0.0645,
"step": 505
},
{
"epoch": 2.8491620111731844,
"grad_norm": 2.533130407333374,
"learning_rate": 7.136959534174592e-08,
"loss": 0.0669,
"step": 510
},
{
"epoch": 2.877094972067039,
"grad_norm": 2.74575138092041,
"learning_rate": 4.8193650976308124e-08,
"loss": 0.064,
"step": 515
},
{
"epoch": 2.905027932960894,
"grad_norm": 2.7461798191070557,
"learning_rate": 2.9535888333374064e-08,
"loss": 0.0619,
"step": 520
},
{
"epoch": 2.9329608938547485,
"grad_norm": 2.249274253845215,
"learning_rate": 1.541333133436018e-08,
"loss": 0.0628,
"step": 525
},
{
"epoch": 2.9608938547486034,
"grad_norm": 2.458946466445923,
"learning_rate": 5.838865838366792e-09,
"loss": 0.0625,
"step": 530
},
{
"epoch": 2.988826815642458,
"grad_norm": 2.7815520763397217,
"learning_rate": 8.212278847224975e-10,
"loss": 0.0568,
"step": 535
}
],
"logging_steps": 5,
"max_steps": 537,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 7.266229010709873e+17,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}