{
  "best_global_step": 900,
  "best_metric": 1.4535424709320068,
  "best_model_checkpoint": "outputs/Qwen3.5-9B-cpt/checkpoint-900",
  "epoch": 0.5689450809956539,
  "eval_steps": 50,
  "global_step": 900,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006321612011062821,
      "grad_norm": 2.2357523441314697,
      "learning_rate": 1.8947368421052634e-06,
      "loss": 1.8448415756225587,
      "step": 10
    },
    {
      "epoch": 0.012643224022125641,
      "grad_norm": 1.1725322008132935,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.728614044189453,
      "step": 20
    },
    {
      "epoch": 0.018964836033188465,
      "grad_norm": 1.0519399642944336,
      "learning_rate": 6.105263157894738e-06,
      "loss": 1.6825117111206054,
      "step": 30
    },
    {
      "epoch": 0.025286448044251283,
      "grad_norm": 0.9476595520973206,
      "learning_rate": 8.210526315789475e-06,
      "loss": 1.627689743041992,
      "step": 40
    },
    {
      "epoch": 0.03160806005531411,
      "grad_norm": 1.026922345161438,
      "learning_rate": 1.0315789473684213e-05,
      "loss": 1.625675582885742,
      "step": 50
    },
    {
      "epoch": 0.03160806005531411,
      "eval_loss": 1.6307754516601562,
      "eval_runtime": 1436.5207,
      "eval_samples_per_second": 3.71,
      "eval_steps_per_second": 1.855,
      "step": 50
    },
    {
      "epoch": 0.03792967206637693,
      "grad_norm": 0.9994499683380127,
      "learning_rate": 1.2421052631578949e-05,
      "loss": 1.6042352676391602,
      "step": 60
    },
    {
      "epoch": 0.044251284077439744,
      "grad_norm": 1.2228177785873413,
      "learning_rate": 1.4526315789473687e-05,
      "loss": 1.60198917388916,
      "step": 70
    },
    {
      "epoch": 0.050572896088502566,
      "grad_norm": 0.9573417901992798,
      "learning_rate": 1.6631578947368423e-05,
      "loss": 1.6050102233886718,
      "step": 80
    },
    {
      "epoch": 0.05689450809956539,
      "grad_norm": 0.9371166825294495,
      "learning_rate": 1.873684210526316e-05,
      "loss": 1.5779685020446776,
      "step": 90
    },
    {
      "epoch": 0.06321612011062822,
      "grad_norm": 0.9192007780075073,
      "learning_rate": 1.9999916170794767e-05,
      "loss": 1.5790796279907227,
      "step": 100
    },
    {
      "epoch": 0.06321612011062822,
      "eval_loss": 1.5887949466705322,
      "eval_runtime": 1428.4275,
      "eval_samples_per_second": 3.731,
      "eval_steps_per_second": 1.866,
      "step": 100
    },
    {
      "epoch": 0.06953773212169104,
      "grad_norm": 0.9729642271995544,
      "learning_rate": 1.99989731083767e-05,
      "loss": 1.54850435256958,
      "step": 110
    },
    {
      "epoch": 0.07585934413275386,
      "grad_norm": 0.9361611008644104,
      "learning_rate": 1.9996982296182915e-05,
      "loss": 1.5509929656982422,
      "step": 120
    },
    {
      "epoch": 0.08218095614381667,
      "grad_norm": 1.1141622066497803,
      "learning_rate": 1.999394394282214e-05,
      "loss": 1.5608986854553222,
      "step": 130
    },
    {
      "epoch": 0.08850256815487949,
      "grad_norm": 1.1783406734466553,
      "learning_rate": 1.9989858366670475e-05,
      "loss": 1.5713417053222656,
      "step": 140
    },
    {
      "epoch": 0.09482418016594231,
      "grad_norm": 0.8572059273719788,
      "learning_rate": 1.9984725995838032e-05,
      "loss": 1.5512660026550293,
      "step": 150
    },
    {
      "epoch": 0.09482418016594231,
      "eval_loss": 1.5633909702301025,
      "eval_runtime": 1434.0069,
      "eval_samples_per_second": 3.716,
      "eval_steps_per_second": 1.858,
      "step": 150
    },
    {
      "epoch": 0.10114579217700513,
      "grad_norm": 1.1298227310180664,
      "learning_rate": 1.9978547368124073e-05,
      "loss": 1.5835878372192382,
      "step": 160
    },
    {
      "epoch": 0.10746740418806795,
      "grad_norm": 0.85079425573349,
      "learning_rate": 1.997132313096066e-05,
      "loss": 1.5457270622253418,
      "step": 170
    },
    {
      "epoch": 0.11378901619913077,
      "grad_norm": 0.7577758431434631,
      "learning_rate": 1.9963054041344826e-05,
      "loss": 1.5363761901855468,
      "step": 180
    },
    {
      "epoch": 0.1201106282101936,
      "grad_norm": 0.9953492879867554,
      "learning_rate": 1.9953740965759222e-05,
      "loss": 1.5482823371887207,
      "step": 190
    },
    {
      "epoch": 0.12643224022125643,
      "grad_norm": 1.0341110229492188,
      "learning_rate": 1.994338488008134e-05,
      "loss": 1.533665084838867,
      "step": 200
    },
    {
      "epoch": 0.12643224022125643,
      "eval_loss": 1.5468075275421143,
      "eval_runtime": 1432.7951,
      "eval_samples_per_second": 3.719,
      "eval_steps_per_second": 1.86,
      "step": 200
    },
    {
      "epoch": 0.13275385223231925,
      "grad_norm": 1.2680150270462036,
      "learning_rate": 1.993198686948127e-05,
      "loss": 1.5316329956054688,
      "step": 210
    },
    {
      "epoch": 0.13907546424338207,
      "grad_norm": 0.7387447953224182,
      "learning_rate": 1.9919548128307954e-05,
      "loss": 1.5236242294311524,
      "step": 220
    },
    {
      "epoch": 0.1453970762544449,
      "grad_norm": 1.4287664890289307,
      "learning_rate": 1.9906069959964073e-05,
      "loss": 1.5394685745239258,
      "step": 230
    },
    {
      "epoch": 0.15171868826550772,
      "grad_norm": 1.0277458429336548,
      "learning_rate": 1.989155377676944e-05,
      "loss": 1.52191104888916,
      "step": 240
    },
    {
      "epoch": 0.15804030027657054,
      "grad_norm": 0.7789772152900696,
      "learning_rate": 1.9876001099813018e-05,
      "loss": 1.546618366241455,
      "step": 250
    },
    {
      "epoch": 0.15804030027657054,
      "eval_loss": 1.533789873123169,
      "eval_runtime": 1436.9068,
      "eval_samples_per_second": 3.709,
      "eval_steps_per_second": 1.855,
      "step": 250
    },
    {
      "epoch": 0.16436191228763333,
      "grad_norm": 1.4293335676193237,
      "learning_rate": 1.9859413558793545e-05,
      "loss": 1.540808868408203,
      "step": 260
    },
    {
      "epoch": 0.17068352429869615,
      "grad_norm": 0.757379949092865,
      "learning_rate": 1.9841792891848744e-05,
      "loss": 1.521357822418213,
      "step": 270
    },
    {
      "epoch": 0.17700513630975898,
      "grad_norm": 1.5915902853012085,
      "learning_rate": 1.982314094537321e-05,
      "loss": 1.5167811393737793,
      "step": 280
    },
    {
      "epoch": 0.1833267483208218,
      "grad_norm": 0.8537394404411316,
      "learning_rate": 1.980345967382491e-05,
      "loss": 1.5079852104187013,
      "step": 290
    },
    {
      "epoch": 0.18964836033188462,
      "grad_norm": 0.7207210063934326,
      "learning_rate": 1.978275113952041e-05,
      "loss": 1.504821491241455,
      "step": 300
    },
    {
      "epoch": 0.18964836033188462,
      "eval_loss": 1.5223760604858398,
      "eval_runtime": 1434.9006,
      "eval_samples_per_second": 3.714,
      "eval_steps_per_second": 1.857,
      "step": 300
    },
    {
      "epoch": 0.19596997234294744,
      "grad_norm": 0.761333167552948,
      "learning_rate": 1.9761017512418757e-05,
      "loss": 1.511648941040039,
      "step": 310
    },
    {
      "epoch": 0.20229158435401026,
      "grad_norm": 1.4120904207229614,
      "learning_rate": 1.973826106989411e-05,
      "loss": 1.4993165016174317,
      "step": 320
    },
    {
      "epoch": 0.20861319636507308,
      "grad_norm": 1.1191859245300293,
      "learning_rate": 1.9714484196497087e-05,
      "loss": 1.5109681129455566,
      "step": 330
    },
    {
      "epoch": 0.2149348083761359,
      "grad_norm": 0.7707664370536804,
      "learning_rate": 1.9689689383704907e-05,
      "loss": 1.5163365364074708,
      "step": 340
    },
    {
      "epoch": 0.22125642038719873,
      "grad_norm": 1.249089241027832,
      "learning_rate": 1.9663879229660338e-05,
      "loss": 1.5154297828674317,
      "step": 350
    },
    {
      "epoch": 0.22125642038719873,
      "eval_loss": 1.5136882066726685,
      "eval_runtime": 1434.1581,
      "eval_samples_per_second": 3.716,
      "eval_steps_per_second": 1.858,
      "step": 350
    },
    {
      "epoch": 0.22757803239826155,
      "grad_norm": 0.8356915712356567,
      "learning_rate": 1.963705643889941e-05,
      "loss": 1.500742244720459,
      "step": 360
    },
    {
      "epoch": 0.23389964440932437,
      "grad_norm": 0.9244163036346436,
      "learning_rate": 1.9609223822068045e-05,
      "loss": 1.5091425895690918,
      "step": 370
    },
    {
      "epoch": 0.2402212564203872,
      "grad_norm": 2.6756114959716797,
      "learning_rate": 1.9580384295627546e-05,
      "loss": 1.5230279922485352,
      "step": 380
    },
    {
      "epoch": 0.24654286843145,
      "grad_norm": 0.8224273324012756,
      "learning_rate": 1.9550540881548962e-05,
      "loss": 1.4959823608398437,
      "step": 390
    },
    {
      "epoch": 0.25286448044251286,
      "grad_norm": 2.7606499195098877,
      "learning_rate": 1.9519696706996457e-05,
      "loss": 1.504225730895996,
      "step": 400
    },
    {
      "epoch": 0.25286448044251286,
      "eval_loss": 1.5063496828079224,
      "eval_runtime": 1442.4518,
      "eval_samples_per_second": 3.694,
      "eval_steps_per_second": 1.848,
      "step": 400
    },
    {
      "epoch": 0.25918609245357566,
      "grad_norm": 1.9451311826705933,
      "learning_rate": 1.9487855003999632e-05,
      "loss": 1.5024944305419923,
      "step": 410
    },
    {
      "epoch": 0.2655077044646385,
      "grad_norm": 0.7222695350646973,
      "learning_rate": 1.9455019109114833e-05,
      "loss": 1.5029170036315918,
      "step": 420
    },
    {
      "epoch": 0.2718293164757013,
      "grad_norm": 1.9110443592071533,
      "learning_rate": 1.9421192463075532e-05,
      "loss": 1.511544895172119,
      "step": 430
    },
    {
      "epoch": 0.27815092848676415,
      "grad_norm": 0.8087928295135498,
      "learning_rate": 1.938637861043179e-05,
      "loss": 1.4900238037109375,
      "step": 440
    },
    {
      "epoch": 0.28447254049782694,
      "grad_norm": 1.2225767374038696,
      "learning_rate": 1.9350581199178858e-05,
      "loss": 1.479935646057129,
      "step": 450
    },
    {
      "epoch": 0.28447254049782694,
      "eval_loss": 1.4983773231506348,
      "eval_runtime": 1432.1372,
      "eval_samples_per_second": 3.721,
      "eval_steps_per_second": 1.861,
      "step": 450
    },
    {
      "epoch": 0.2907941525088898,
      "grad_norm": 0.7390967011451721,
      "learning_rate": 1.9313803980374882e-05,
      "loss": 1.507608699798584,
      "step": 460
    },
    {
      "epoch": 0.2971157645199526,
      "grad_norm": 1.7765216827392578,
      "learning_rate": 1.9276050807747878e-05,
      "loss": 1.486252784729004,
      "step": 470
    },
    {
      "epoch": 0.30343737653101543,
      "grad_norm": 1.1083087921142578,
      "learning_rate": 1.9237325637291897e-05,
      "loss": 1.4911764144897461,
      "step": 480
    },
    {
      "epoch": 0.30975898854207823,
      "grad_norm": 1.495973825454712,
      "learning_rate": 1.9197632526852502e-05,
      "loss": 1.4877856254577637,
      "step": 490
    },
    {
      "epoch": 0.3160806005531411,
      "grad_norm": 0.844947338104248,
      "learning_rate": 1.915697563570156e-05,
      "loss": 1.479910182952881,
      "step": 500
    },
    {
      "epoch": 0.3160806005531411,
      "eval_loss": 1.4914230108261108,
      "eval_runtime": 1432.4776,
      "eval_samples_per_second": 3.72,
      "eval_steps_per_second": 1.86,
      "step": 500
    },
    {
      "epoch": 0.32240221256420387,
      "grad_norm": 1.77666437625885,
      "learning_rate": 1.9115359224101416e-05,
      "loss": 1.490778350830078,
      "step": 510
    },
    {
      "epoch": 0.32872382457526667,
      "grad_norm": 2.99625301361084,
      "learning_rate": 1.9072787652858463e-05,
      "loss": 1.494775390625,
      "step": 520
    },
    {
      "epoch": 0.3350454365863295,
      "grad_norm": 1.031172752380371,
      "learning_rate": 1.9029265382866216e-05,
      "loss": 1.4774039268493653,
      "step": 530
    },
    {
      "epoch": 0.3413670485973923,
      "grad_norm": 0.9397842288017273,
      "learning_rate": 1.8984796974637845e-05,
      "loss": 1.4831487655639648,
      "step": 540
    },
    {
      "epoch": 0.34768866060845516,
      "grad_norm": 0.822282075881958,
      "learning_rate": 1.8939387087828326e-05,
      "loss": 1.4724056243896484,
      "step": 550
    },
    {
      "epoch": 0.34768866060845516,
      "eval_loss": 1.4864764213562012,
      "eval_runtime": 1435.983,
      "eval_samples_per_second": 3.711,
      "eval_steps_per_second": 1.856,
      "step": 550
    },
    {
      "epoch": 0.35401027261951795,
      "grad_norm": 1.08564293384552,
      "learning_rate": 1.889304048074616e-05,
      "loss": 1.4828974723815918,
      "step": 560
    },
    {
      "epoch": 0.3603318846305808,
      "grad_norm": 1.6417145729064941,
      "learning_rate": 1.8845762009854765e-05,
      "loss": 1.4577611923217773,
      "step": 570
    },
    {
      "epoch": 0.3666534966416436,
      "grad_norm": 1.2441695928573608,
      "learning_rate": 1.8797556629263603e-05,
      "loss": 1.4795442581176759,
      "step": 580
    },
    {
      "epoch": 0.37297510865270644,
      "grad_norm": 0.7002350687980652,
      "learning_rate": 1.8748429390209052e-05,
      "loss": 1.4860575675964356,
      "step": 590
    },
    {
      "epoch": 0.37929672066376924,
      "grad_norm": 0.7346900701522827,
      "learning_rate": 1.869838544052511e-05,
      "loss": 1.4883034706115723,
      "step": 600
    },
    {
      "epoch": 0.37929672066376924,
      "eval_loss": 1.4805960655212402,
      "eval_runtime": 1434.6251,
      "eval_samples_per_second": 3.715,
      "eval_steps_per_second": 1.858,
      "step": 600
    },
    {
      "epoch": 0.3856183326748321,
      "grad_norm": 1.2719956636428833,
      "learning_rate": 1.8647430024103975e-05,
      "loss": 1.4628222465515137,
      "step": 610
    },
    {
      "epoch": 0.3919399446858949,
      "grad_norm": 3.929159641265869,
      "learning_rate": 1.8595568480346555e-05,
      "loss": 1.4697487831115723,
      "step": 620
    },
    {
      "epoch": 0.39826155669695773,
      "grad_norm": 1.7072134017944336,
      "learning_rate": 1.8542806243602984e-05,
      "loss": 1.466939926147461,
      "step": 630
    },
    {
      "epoch": 0.4045831687080205,
      "grad_norm": 0.8139464855194092,
      "learning_rate": 1.8489148842603172e-05,
      "loss": 1.47437744140625,
      "step": 640
    },
    {
      "epoch": 0.4109047807190834,
      "grad_norm": 0.8353721499443054,
      "learning_rate": 1.8434601899877484e-05,
      "loss": 1.4795381546020507,
      "step": 650
    },
    {
      "epoch": 0.4109047807190834,
      "eval_loss": 1.4759365320205688,
      "eval_runtime": 1440.6676,
      "eval_samples_per_second": 3.699,
      "eval_steps_per_second": 1.85,
      "step": 650
    },
    {
      "epoch": 0.41722639273014617,
      "grad_norm": 1.203122854232788,
      "learning_rate": 1.8379171131167557e-05,
      "loss": 1.4443695068359375,
      "step": 660
    },
    {
      "epoch": 0.423548004741209,
      "grad_norm": 0.7811446785926819,
      "learning_rate": 1.8322862344827404e-05,
      "loss": 1.453621006011963,
      "step": 670
    },
    {
      "epoch": 0.4298696167522718,
      "grad_norm": 0.7570304274559021,
      "learning_rate": 1.8265681441214744e-05,
      "loss": 1.4751152992248535,
      "step": 680
    },
    {
      "epoch": 0.43619122876333466,
      "grad_norm": 1.1196154356002808,
      "learning_rate": 1.8207634412072765e-05,
      "loss": 1.4665375709533692,
      "step": 690
    },
    {
      "epoch": 0.44251284077439745,
      "grad_norm": 1.452694296836853,
      "learning_rate": 1.8148727339902245e-05,
      "loss": 1.4367003440856934,
      "step": 700
    },
    {
      "epoch": 0.44251284077439745,
      "eval_loss": 1.4716598987579346,
      "eval_runtime": 1433.735,
      "eval_samples_per_second": 3.717,
      "eval_steps_per_second": 1.859,
      "step": 700
    },
    {
      "epoch": 0.4488344527854603,
      "grad_norm": 0.9854422807693481,
      "learning_rate": 1.8088966397324215e-05,
      "loss": 1.4711213111877441,
      "step": 710
    },
    {
      "epoch": 0.4551560647965231,
      "grad_norm": 0.8611396551132202,
      "learning_rate": 1.8028357846433144e-05,
      "loss": 1.447887134552002,
      "step": 720
    },
    {
      "epoch": 0.46147767680758595,
      "grad_norm": 0.9595324397087097,
      "learning_rate": 1.796690803814076e-05,
      "loss": 1.4460906982421875,
      "step": 730
    },
    {
      "epoch": 0.46779928881864874,
      "grad_norm": 0.995147705078125,
      "learning_rate": 1.7904623411510565e-05,
      "loss": 1.4776107788085937,
      "step": 740
    },
    {
      "epoch": 0.4741209008297116,
      "grad_norm": 1.3566371202468872,
      "learning_rate": 1.7841510493083126e-05,
      "loss": 1.4614599227905274,
      "step": 750
    },
    {
      "epoch": 0.4741209008297116,
      "eval_loss": 1.4664231538772583,
      "eval_runtime": 1435.7161,
      "eval_samples_per_second": 3.712,
      "eval_steps_per_second": 1.856,
      "step": 750
    },
    {
      "epoch": 0.4804425128407744,
      "grad_norm": 1.0786532163619995,
      "learning_rate": 1.7777575896192177e-05,
      "loss": 1.4631671905517578,
      "step": 760
    },
    {
      "epoch": 0.48676412485183723,
      "grad_norm": 1.5780668258666992,
      "learning_rate": 1.7712826320271633e-05,
      "loss": 1.4548497200012207,
      "step": 770
    },
    {
      "epoch": 0.4930857368629,
      "grad_norm": 1.075210690498352,
      "learning_rate": 1.7647268550153583e-05,
      "loss": 1.4418771743774415,
      "step": 780
    },
    {
      "epoch": 0.4994073488739629,
      "grad_norm": 0.7408574819564819,
      "learning_rate": 1.7580909455357344e-05,
      "loss": 1.4451427459716797,
      "step": 790
    },
    {
      "epoch": 0.5057289608850257,
      "grad_norm": 0.9548968076705933,
      "learning_rate": 1.7513755989369636e-05,
      "loss": 1.4488779067993165,
      "step": 800
    },
    {
      "epoch": 0.5057289608850257,
      "eval_loss": 1.461513876914978,
      "eval_runtime": 1439.1351,
      "eval_samples_per_second": 3.703,
      "eval_steps_per_second": 1.852,
      "step": 800
    },
    {
      "epoch": 0.5120505728960885,
      "grad_norm": 0.8898774981498718,
      "learning_rate": 1.744581518891595e-05,
      "loss": 1.4661118507385253,
      "step": 810
    },
    {
      "epoch": 0.5183721849071513,
      "grad_norm": 2.370990514755249,
      "learning_rate": 1.737709417322319e-05,
      "loss": 1.4475753784179688,
      "step": 820
    },
    {
      "epoch": 0.5246937969182142,
      "grad_norm": 0.7134878039360046,
      "learning_rate": 1.730760014327371e-05,
      "loss": 1.4453336715698242,
      "step": 830
    },
    {
      "epoch": 0.531015408929277,
      "grad_norm": 0.7120240330696106,
      "learning_rate": 1.72373403810507e-05,
      "loss": 1.4352140426635742,
      "step": 840
    },
    {
      "epoch": 0.5373370209403397,
      "grad_norm": 1.5345282554626465,
      "learning_rate": 1.7166322248775212e-05,
      "loss": 1.434824275970459,
      "step": 850
    },
    {
      "epoch": 0.5373370209403397,
      "eval_loss": 1.4574466943740845,
      "eval_runtime": 1435.0764,
      "eval_samples_per_second": 3.713,
      "eval_steps_per_second": 1.857,
      "step": 850
    },
    {
      "epoch": 0.5436586329514026,
      "grad_norm": 0.7333837151527405,
      "learning_rate": 1.709455318813463e-05,
      "loss": 1.4499679565429688,
      "step": 860
    },
    {
      "epoch": 0.5499802449624654,
      "grad_norm": 1.3881059885025024,
      "learning_rate": 1.7022040719502935e-05,
      "loss": 1.4426145553588867,
      "step": 870
    },
    {
      "epoch": 0.5563018569735283,
      "grad_norm": 2.6665091514587402,
      "learning_rate": 1.6948792441152666e-05,
      "loss": 1.4703943252563476,
      "step": 880
    },
    {
      "epoch": 0.562623468984591,
      "grad_norm": 0.7150371074676514,
      "learning_rate": 1.687481602845872e-05,
      "loss": 1.441333293914795,
      "step": 890
    },
    {
      "epoch": 0.5689450809956539,
      "grad_norm": 3.045902729034424,
      "learning_rate": 1.6800119233094093e-05,
      "loss": 1.4491585731506347,
      "step": 900
    },
    {
      "epoch": 0.5689450809956539,
      "eval_loss": 1.4535424709320068,
      "eval_runtime": 1434.0825,
      "eval_samples_per_second": 3.716,
      "eval_steps_per_second": 1.858,
      "step": 900
    }
  ],
  "logging_steps": 10,
  "max_steps": 3164,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.005
      },
      "attributes": {
        "early_stopping_patience_counter": 3
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 6.183432293135155e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}