n52eea6fef943 / trainer_state.json
insagur's picture
Add files using upload-large-folder tool
a539414 verified
Raw
History Blame Contribute Delete
89.1 kB
{
"best_global_step": 2000,
"best_metric": 1.13275814,
"best_model_checkpoint": "/NHNHOME/data/sanghyeok/qwen-cua/runs/probe2000_idm_code_type_text/v1-20260621-231852/checkpoint-2000",
"epoch": 0.46126126126126127,
"eval_steps": 500,
"global_step": 2000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.00023063063063063062,
"grad_norm": 1594.41796875,
"learning_rate": 1.0000000000000001e-07,
"loss": 78.32112121582031,
"loss_idm": 4.875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 1.8660085201263428,
"loss_idm_type": 3.003211200237274,
"loss_sft": 0.013178251218050718,
"step": 1,
"token_acc": 0.855064946292334
},
{
"epoch": 0.0023063063063063064,
"grad_norm": 1612.195068359375,
"learning_rate": 1.0000000000000002e-06,
"loss": 75.12723795572917,
"loss_idm": 4.256944444444445,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 1.771455658806695,
"loss_idm_type": 2.897422472635905,
"loss_sft": 0.013165258698993258,
"step": 10,
"token_acc": 0.8557998765011616
},
{
"epoch": 0.004612612612612613,
"grad_norm": 353.6252136230469,
"learning_rate": 2.0000000000000003e-06,
"loss": 49.47017822265625,
"loss_idm": 2.946875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 1.6457603931427003,
"loss_idm_type": 1.4199507236480713,
"loss_sft": 0.01303936094045639,
"step": 20,
"token_acc": 0.8574650963417557
},
{
"epoch": 0.006918918918918919,
"grad_norm": 126.47821044921875,
"learning_rate": 3e-06,
"loss": 43.140435791015626,
"loss_idm": 2.525,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 1.6730948686599731,
"loss_idm_type": 0.9972076416015625,
"loss_sft": 0.013039363175630569,
"step": 30,
"token_acc": 0.8573986811865872
},
{
"epoch": 0.009225225225225226,
"grad_norm": 162.51278686523438,
"learning_rate": 4.000000000000001e-06,
"loss": 39.47804260253906,
"loss_idm": 2.31875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 1.5029781103134154,
"loss_idm_type": 0.9386286258697509,
"loss_sft": 0.013111228495836258,
"step": 40,
"token_acc": 0.856925661335508
},
{
"epoch": 0.011531531531531532,
"grad_norm": 356.9111633300781,
"learning_rate": 5e-06,
"loss": 36.0298828125,
"loss_idm": 2.125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 1.4550552129745484,
"loss_idm_type": 0.7688171148300171,
"loss_sft": 0.014152135699987411,
"step": 50,
"token_acc": 0.8486634001116418
},
{
"epoch": 0.013837837837837838,
"grad_norm": 323.67437744140625,
"learning_rate": 6e-06,
"loss": 27.128363037109374,
"loss_idm": 1.6203125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 1.0920132637023925,
"loss_idm_type": 0.574056351184845,
"loss_sft": 0.01448214016854763,
"step": 60,
"token_acc": 0.8464492900583126
},
{
"epoch": 0.016144144144144144,
"grad_norm": 259.5957336425781,
"learning_rate": 7e-06,
"loss": 28.27000732421875,
"loss_idm": 1.675,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 1.0847449064254762,
"loss_idm_type": 0.6541952967643738,
"loss_sft": 0.014033125713467598,
"step": 70,
"token_acc": 0.8503391363046473
},
{
"epoch": 0.01845045045045045,
"grad_norm": 144.2830810546875,
"learning_rate": 8.000000000000001e-06,
"loss": 24.161322021484374,
"loss_idm": 1.415625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.8819304466247558,
"loss_idm_type": 0.600779139995575,
"loss_sft": 0.013908731937408447,
"step": 80,
"token_acc": 0.8504072708357134
},
{
"epoch": 0.020756756756756756,
"grad_norm": 273.4071350097656,
"learning_rate": 9e-06,
"loss": 23.594862365722655,
"loss_idm": 1.390625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.8310215711593628,
"loss_idm_type": 0.6153789043426514,
"loss_sft": 0.013951633498072625,
"step": 90,
"token_acc": 0.8492400086679256
},
{
"epoch": 0.023063063063063063,
"grad_norm": 251.73202514648438,
"learning_rate": 1e-05,
"loss": 25.476950073242186,
"loss_idm": 1.49375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.9382863879203797,
"loss_idm_type": 0.6257401466369629,
"loss_sft": 0.014301823824644089,
"step": 100,
"token_acc": 0.8458359752497978
},
{
"epoch": 0.025369369369369368,
"grad_norm": 286.3401184082031,
"learning_rate": 9.999316524962347e-06,
"loss": 24.417092895507814,
"loss_idm": 1.428125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.904452109336853,
"loss_idm_type": 0.593038022518158,
"loss_sft": 0.014249381050467491,
"step": 110,
"token_acc": 0.8465379122197904
},
{
"epoch": 0.027675675675675675,
"grad_norm": 222.7222442626953,
"learning_rate": 9.99726628670463e-06,
"loss": 26.060919189453124,
"loss_idm": 1.5484375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.973819124698639,
"loss_idm_type": 0.6266206264495849,
"loss_sft": 0.014172710105776786,
"step": 120,
"token_acc": 0.847854247216184
},
{
"epoch": 0.029981981981981983,
"grad_norm": 275.38079833984375,
"learning_rate": 9.993849845741525e-06,
"loss": 25.397323608398438,
"loss_idm": 1.484375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.9372030019760131,
"loss_idm_type": 0.6214419364929199,
"loss_sft": 0.014417377859354019,
"step": 130,
"token_acc": 0.8455794621991264
},
{
"epoch": 0.03228828828828829,
"grad_norm": 193.24574279785156,
"learning_rate": 9.989068136093873e-06,
"loss": 23.078701782226563,
"loss_idm": 1.3359375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7943235158920288,
"loss_idm_type": 0.6200331211090088,
"loss_sft": 0.014127673953771592,
"step": 140,
"token_acc": 0.8480571648523011
},
{
"epoch": 0.034594594594594595,
"grad_norm": 252.88157653808594,
"learning_rate": 9.98292246503335e-06,
"loss": 25.396533203125,
"loss_idm": 1.50625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.9365179538726807,
"loss_idm_type": 0.6222358584403992,
"loss_sft": 0.014179562777280807,
"step": 150,
"token_acc": 0.846938125590989
},
{
"epoch": 0.0369009009009009,
"grad_norm": 132.69137573242188,
"learning_rate": 9.975414512725058e-06,
"loss": 21.090823364257812,
"loss_idm": 1.2234375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7004317998886108,
"loss_idm_type": 0.5901555061340332,
"loss_sft": 0.013755682110786437,
"step": 160,
"token_acc": 0.8513433717153823
},
{
"epoch": 0.03920720720720721,
"grad_norm": 180.37062072753906,
"learning_rate": 9.966546331768192e-06,
"loss": 21.250894165039064,
"loss_idm": 1.25,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6869279980659485,
"loss_idm_type": 0.6129321694374085,
"loss_sft": 0.014121741428971291,
"step": 170,
"token_acc": 0.8468387733683489
},
{
"epoch": 0.04151351351351351,
"grad_norm": 185.7732696533203,
"learning_rate": 9.956320346634877e-06,
"loss": 22.12333984375,
"loss_idm": 1.2890625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7296592950820923,
"loss_idm_type": 0.6256211042404175,
"loss_sft": 0.013716945424675941,
"step": 180,
"token_acc": 0.8514388900899068
},
{
"epoch": 0.04381981981981982,
"grad_norm": 192.45997619628906,
"learning_rate": 9.944739353007344e-06,
"loss": 20.918809509277345,
"loss_idm": 1.2453125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6708066940307618,
"loss_idm_type": 0.6089411020278931,
"loss_sft": 0.01396913044154644,
"step": 190,
"token_acc": 0.8494491495904698
},
{
"epoch": 0.04612612612612613,
"grad_norm": 138.6709442138672,
"learning_rate": 9.931806517013612e-06,
"loss": 21.51440887451172,
"loss_idm": 1.2484375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6464977025985718,
"loss_idm_type": 0.6701341271400452,
"loss_sft": 0.01424136757850647,
"step": 200,
"token_acc": 0.8475712431229523
},
{
"epoch": 0.048432432432432435,
"grad_norm": 156.5552520751953,
"learning_rate": 9.917525374361913e-06,
"loss": 21.93793182373047,
"loss_idm": 1.2875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7565288722515107,
"loss_idm_type": 0.5858924746513366,
"loss_sft": 0.014323525503277779,
"step": 210,
"token_acc": 0.8463558121278266
},
{
"epoch": 0.050738738738738735,
"grad_norm": 189.6932830810547,
"learning_rate": 9.901899829374048e-06,
"loss": 20.054583740234374,
"loss_idm": 1.159375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6576752543449402,
"loss_idm_type": 0.567785918712616,
"loss_sft": 0.014022772386670112,
"step": 220,
"token_acc": 0.8487998119313549
},
{
"epoch": 0.05304504504504504,
"grad_norm": 140.5413360595703,
"learning_rate": 9.884934153917998e-06,
"loss": 22.073431396484374,
"loss_idm": 1.3,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7526954412460327,
"loss_idm_type": 0.5985090017318726,
"loss_sft": 0.014082278311252593,
"step": 230,
"token_acc": 0.8485526712222985
},
{
"epoch": 0.05535135135135135,
"grad_norm": 133.97390747070312,
"learning_rate": 9.86663298624003e-06,
"loss": 21.144790649414062,
"loss_idm": 1.2375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7176072120666503,
"loss_idm_type": 0.5748069643974304,
"loss_sft": 0.014453838020563126,
"step": 240,
"token_acc": 0.8452394567722186
},
{
"epoch": 0.05765765765765766,
"grad_norm": 142.247314453125,
"learning_rate": 9.847001329696653e-06,
"loss": 20.1635498046875,
"loss_idm": 1.178125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6437483191490173,
"loss_idm_type": 0.5870923161506653,
"loss_sft": 0.014495516940951348,
"step": 250,
"token_acc": 0.8446375732200728
},
{
"epoch": 0.059963963963963966,
"grad_norm": 202.28245544433594,
"learning_rate": 9.826044551386743e-06,
"loss": 21.010342407226563,
"loss_idm": 1.2265625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6738935589790345,
"loss_idm_type": 0.6098474025726318,
"loss_sft": 0.014558565244078637,
"step": 260,
"token_acc": 0.8439074729940742
},
{
"epoch": 0.06227027027027027,
"grad_norm": 180.11004638671875,
"learning_rate": 9.803768380684242e-06,
"loss": 20.15380096435547,
"loss_idm": 1.178125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6457672238349914,
"loss_idm_type": 0.5851505637168884,
"loss_sft": 0.01434638984501362,
"step": 270,
"token_acc": 0.8461026195507937
},
{
"epoch": 0.06457657657657657,
"grad_norm": 108.87934875488281,
"learning_rate": 9.780178907671788e-06,
"loss": 20.92729797363281,
"loss_idm": 1.2203125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6868312954902649,
"loss_idm_type": 0.5920520305633545,
"loss_sft": 0.014423593878746033,
"step": 280,
"token_acc": 0.8456991679616043
},
{
"epoch": 0.06688288288288288,
"grad_norm": 149.56564331054688,
"learning_rate": 9.755282581475769e-06,
"loss": 18.62049102783203,
"loss_idm": 1.09375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.58313547372818,
"loss_idm_type": 0.5516607403755188,
"loss_sft": 0.014439313858747482,
"step": 290,
"token_acc": 0.8454596672959642
},
{
"epoch": 0.06918918918918919,
"grad_norm": 153.041015625,
"learning_rate": 9.729086208503174e-06,
"loss": 20.677386474609374,
"loss_idm": 1.215625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7001219868659974,
"loss_idm_type": 0.5628572702407837,
"loss_sft": 0.014919180795550347,
"step": 300,
"token_acc": 0.84088942131384
},
{
"epoch": 0.0714954954954955,
"grad_norm": 177.866943359375,
"learning_rate": 9.701596950580807e-06,
"loss": 19.88404235839844,
"loss_idm": 1.1453125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6421271324157715,
"loss_idm_type": 0.5717909574508667,
"loss_sft": 0.014424886554479599,
"step": 310,
"token_acc": 0.846033168982519
},
{
"epoch": 0.0738018018018018,
"grad_norm": 129.16363525390625,
"learning_rate": 9.672822322997305e-06,
"loss": 17.68323974609375,
"loss_idm": 1.0296875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5472238302230835,
"loss_idm_type": 0.5287622213363647,
"loss_sft": 0.014649244770407677,
"step": 320,
"token_acc": 0.8431312749319898
},
{
"epoch": 0.07610810810810811,
"grad_norm": 156.39739990234375,
"learning_rate": 9.642770192448537e-06,
"loss": 21.36103515625,
"loss_idm": 1.253125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7405102610588074,
"loss_idm_type": 0.5659091114997864,
"loss_sft": 0.014557573571801185,
"step": 330,
"token_acc": 0.8441445527139338
},
{
"epoch": 0.07841441441441442,
"grad_norm": 129.96697998046875,
"learning_rate": 9.611448774886925e-06,
"loss": 21.416160583496094,
"loss_idm": 1.234375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7018738508224487,
"loss_idm_type": 0.607703685760498,
"loss_sft": 0.014446442201733588,
"step": 340,
"token_acc": 0.8443760364686431
},
{
"epoch": 0.08072072072072072,
"grad_norm": 180.76304626464844,
"learning_rate": 9.578866633275289e-06,
"loss": 20.02213134765625,
"loss_idm": 1.1671875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6722601294517517,
"loss_idm_type": 0.5498591780662536,
"loss_sft": 0.014624126255512238,
"step": 350,
"token_acc": 0.8432171463399417
},
{
"epoch": 0.08302702702702702,
"grad_norm": 118.01971435546875,
"learning_rate": 9.545032675245814e-06,
"loss": 18.858502197265626,
"loss_idm": 1.109375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6102901577949524,
"loss_idm_type": 0.5399273037910461,
"loss_sft": 0.01428530141711235,
"step": 360,
"token_acc": 0.846329725225022
},
{
"epoch": 0.08533333333333333,
"grad_norm": 125.42524719238281,
"learning_rate": 9.509956150664796e-06,
"loss": 18.46220703125,
"loss_idm": 1.0765625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6047607123851776,
"loss_idm_type": 0.5198914885520936,
"loss_sft": 0.014666610956192016,
"step": 370,
"token_acc": 0.8430305689629954
},
{
"epoch": 0.08763963963963964,
"grad_norm": 180.96270751953125,
"learning_rate": 9.473646649103819e-06,
"loss": 20.59235534667969,
"loss_idm": 1.1921875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6622613787651062,
"loss_idm_type": 0.59524484872818,
"loss_sft": 0.014829005673527717,
"step": 380,
"token_acc": 0.8417713702410807
},
{
"epoch": 0.08994594594594595,
"grad_norm": 144.25331115722656,
"learning_rate": 9.43611409721806e-06,
"loss": 18.71649627685547,
"loss_idm": 1.0890625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6037861943244934,
"loss_idm_type": 0.5369527459144592,
"loss_sft": 0.014459025114774704,
"step": 390,
"token_acc": 0.8447289389919099
},
{
"epoch": 0.09225225225225225,
"grad_norm": 184.02232360839844,
"learning_rate": 9.397368756032445e-06,
"loss": 18.94854278564453,
"loss_idm": 1.1,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6332925677299499,
"loss_idm_type": 0.5221684217453003,
"loss_sft": 0.014429151266813278,
"step": 400,
"token_acc": 0.8452291937983728
},
{
"epoch": 0.09455855855855856,
"grad_norm": 173.45257568359375,
"learning_rate": 9.357421218136387e-06,
"loss": 18.99466552734375,
"loss_idm": 1.096875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6210426688194275,
"loss_idm_type": 0.5382487654685975,
"loss_sft": 0.014015758782625199,
"step": 410,
"token_acc": 0.8495117177798309
},
{
"epoch": 0.09686486486486487,
"grad_norm": 165.1482391357422,
"learning_rate": 9.31628240478787e-06,
"loss": 19.396931457519532,
"loss_idm": 1.1328125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6425803899765015,
"loss_idm_type": 0.5403215169906617,
"loss_sft": 0.01475544199347496,
"step": 420,
"token_acc": 0.8424114958179343
},
{
"epoch": 0.09917117117117118,
"grad_norm": 179.56707763671875,
"learning_rate": 9.273963562927695e-06,
"loss": 19.1131591796875,
"loss_idm": 1.121875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5948790431022644,
"loss_idm_type": 0.570659625530243,
"loss_sft": 0.014644772931933402,
"step": 430,
"token_acc": 0.8437164287906691
},
{
"epoch": 0.10147747747747747,
"grad_norm": 141.6302032470703,
"learning_rate": 9.230476262104678e-06,
"loss": 20.165437316894533,
"loss_idm": 1.175,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6892072916030884,
"loss_idm_type": 0.5419347524642945,
"loss_sft": 0.014781763404607772,
"step": 440,
"token_acc": 0.8412593497029246
},
{
"epoch": 0.10378378378378378,
"grad_norm": 84.81742095947266,
"learning_rate": 9.185832391312644e-06,
"loss": 19.339076232910156,
"loss_idm": 1.13125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7115584373474121,
"loss_idm_type": 0.4681084334850311,
"loss_sft": 0.014591620862483978,
"step": 450,
"token_acc": 0.8433540866857198
},
{
"epoch": 0.10609009009009009,
"grad_norm": 164.00550842285156,
"learning_rate": 9.140044155740102e-06,
"loss": 18.864373779296876,
"loss_idm": 1.09375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6536977291107178,
"loss_idm_type": 0.4957777142524719,
"loss_sft": 0.014747689291834832,
"step": 460,
"token_acc": 0.841443148203543
},
{
"epoch": 0.1083963963963964,
"grad_norm": 188.7095184326172,
"learning_rate": 9.093124073433464e-06,
"loss": 20.679414367675783,
"loss_idm": 1.2171875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7044301986694336,
"loss_idm_type": 0.5580654740333557,
"loss_sft": 0.014930276200175285,
"step": 470,
"token_acc": 0.8409072937658298
},
{
"epoch": 0.1107027027027027,
"grad_norm": 124.54925537109375,
"learning_rate": 9.045084971874738e-06,
"loss": 18.53998260498047,
"loss_idm": 1.0703125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6069076418876648,
"loss_idm_type": 0.521586561203003,
"loss_sft": 0.015061390027403832,
"step": 480,
"token_acc": 0.8401639957878136
},
{
"epoch": 0.11300900900900901,
"grad_norm": 117.37905883789062,
"learning_rate": 8.995939984474624e-06,
"loss": 19.989923095703126,
"loss_idm": 1.1703125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6740325570106507,
"loss_idm_type": 0.5453832864761352,
"loss_sft": 0.014970069751143456,
"step": 490,
"token_acc": 0.8400487229410242
},
{
"epoch": 0.11531531531531532,
"grad_norm": 132.42762756347656,
"learning_rate": 8.94570254698197e-06,
"loss": 19.590615844726564,
"loss_idm": 1.15625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6635529398918152,
"loss_idm_type": 0.5308433651924134,
"loss_sft": 0.014947501569986343,
"step": 500,
"token_acc": 0.84046714392237
},
{
"epoch": 0.11531531531531532,
"eval_loss": 1.3233741521835327,
"eval_runtime": 170.0148,
"eval_samples_per_second": 10.264,
"eval_steps_per_second": 2.57,
"eval_token_acc": 0.828006241506015,
"step": 500
},
{
"epoch": 0.11762162162162162,
"grad_norm": 123.0207290649414,
"learning_rate": 8.894386393810563e-06,
"loss": 18.979844665527345,
"loss_idm": 1.1046875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6254270136356354,
"loss_idm_type": 0.5312720656394958,
"loss_sft": 0.014834708720445632,
"step": 510,
"token_acc": 0.8414443277826296
},
{
"epoch": 0.11992792792792793,
"grad_norm": 128.770751953125,
"learning_rate": 8.842005554284296e-06,
"loss": 21.127740478515626,
"loss_idm": 1.2484375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7544469356536865,
"loss_idm_type": 0.5359104752540589,
"loss_sft": 0.014924750849604606,
"step": 520,
"token_acc": 0.8416575605712348
},
{
"epoch": 0.12223423423423424,
"grad_norm": 114.88973236083984,
"learning_rate": 8.788574348801676e-06,
"loss": 19.248388671875,
"loss_idm": 1.11875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6367936968803406,
"loss_idm_type": 0.5363446712493897,
"loss_sft": 0.014853022247552871,
"step": 530,
"token_acc": 0.8422921276881885
},
{
"epoch": 0.12454054054054053,
"grad_norm": 134.62933349609375,
"learning_rate": 8.734107384920771e-06,
"loss": 19.299974060058595,
"loss_idm": 1.1296875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6587535738945007,
"loss_idm_type": 0.5175979852676391,
"loss_sft": 0.014881648868322373,
"step": 540,
"token_acc": 0.8417766816577203
},
{
"epoch": 0.12684684684684686,
"grad_norm": 142.13165283203125,
"learning_rate": 8.67861955336566e-06,
"loss": 18.453768920898437,
"loss_idm": 1.0765625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6252750515937805,
"loss_idm_type": 0.4980467677116394,
"loss_sft": 0.014992362260818482,
"step": 550,
"token_acc": 0.8397899195340869
},
{
"epoch": 0.12915315315315315,
"grad_norm": 137.8123016357422,
"learning_rate": 8.622126023955446e-06,
"loss": 17.314250183105468,
"loss_idm": 1.0171875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5588197946548462,
"loss_idm_type": 0.49371981620788574,
"loss_sft": 0.014778397232294082,
"step": 560,
"token_acc": 0.8418601053031333
},
{
"epoch": 0.13145945945945947,
"grad_norm": 133.22518920898438,
"learning_rate": 8.564642241456986e-06,
"loss": 19.21575164794922,
"loss_idm": 1.1234375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6260247945785522,
"loss_idm_type": 0.5450846672058105,
"loss_sft": 0.014973315969109535,
"step": 570,
"token_acc": 0.8413558847724713
},
{
"epoch": 0.13376576576576577,
"grad_norm": 127.0101547241211,
"learning_rate": 8.506183921362443e-06,
"loss": 17.988644409179688,
"loss_idm": 1.0375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5716003537178039,
"loss_idm_type": 0.5227356672286987,
"loss_sft": 0.01492447629570961,
"step": 580,
"token_acc": 0.8405641411639163
},
{
"epoch": 0.13607207207207206,
"grad_norm": 171.4197540283203,
"learning_rate": 8.446767045592829e-06,
"loss": 19.189237976074217,
"loss_idm": 1.103125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6561702966690064,
"loss_idm_type": 0.5143969297409058,
"loss_sft": 0.014564141258597374,
"step": 590,
"token_acc": 0.8446573910446276
},
{
"epoch": 0.13837837837837838,
"grad_norm": 102.463623046875,
"learning_rate": 8.386407858128707e-06,
"loss": 16.7314208984375,
"loss_idm": 0.9640625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.536580491065979,
"loss_idm_type": 0.4802058100700378,
"loss_sft": 0.014577708020806313,
"step": 600,
"token_acc": 0.8437585025098785
},
{
"epoch": 0.14068468468468467,
"grad_norm": 129.72825622558594,
"learning_rate": 8.325122860569241e-06,
"loss": 15.771197509765624,
"loss_idm": 0.921875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5012216567993164,
"loss_idm_type": 0.4551218390464783,
"loss_sft": 0.014644627273082734,
"step": 610,
"token_acc": 0.8428215900612409
},
{
"epoch": 0.142990990990991,
"grad_norm": 123.2136001586914,
"learning_rate": 8.262928807620843e-06,
"loss": 17.49943389892578,
"loss_idm": 1.009375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5229299306869507,
"loss_idm_type": 0.5412803411483764,
"loss_sft": 0.01478283479809761,
"step": 620,
"token_acc": 0.8411515174521229
},
{
"epoch": 0.1452972972972973,
"grad_norm": 178.2458038330078,
"learning_rate": 8.199842702516584e-06,
"loss": 17.452848815917967,
"loss_idm": 1.009375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5550838947296143,
"loss_idm_type": 0.505655574798584,
"loss_sft": 0.015062566101551055,
"step": 630,
"token_acc": 0.8393233091772877
},
{
"epoch": 0.1476036036036036,
"grad_norm": 154.72146606445312,
"learning_rate": 8.135881792367686e-06,
"loss": 19.142218017578124,
"loss_idm": 1.134375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6731308341026306,
"loss_idm_type": 0.4940042018890381,
"loss_sft": 0.014690161868929862,
"step": 640,
"token_acc": 0.8424485284630235
},
{
"epoch": 0.1499099099099099,
"grad_norm": 138.528076171875,
"learning_rate": 8.071063563448341e-06,
"loss": 17.799058532714845,
"loss_idm": 1.0421875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6064084529876709,
"loss_idm_type": 0.47745844125747683,
"loss_sft": 0.014413310214877129,
"step": 650,
"token_acc": 0.8453410276633372
},
{
"epoch": 0.15221621621621623,
"grad_norm": 113.0783462524414,
"learning_rate": 8.005405736415127e-06,
"loss": 17.12696075439453,
"loss_idm": 0.99375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5633171319961547,
"loss_idm_type": 0.4775223731994629,
"loss_sft": 0.014819654077291489,
"step": 660,
"token_acc": 0.8410728177518738
},
{
"epoch": 0.15452252252252252,
"grad_norm": 183.07449340820312,
"learning_rate": 7.938926261462366e-06,
"loss": 16.33360137939453,
"loss_idm": 0.9453125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5148284912109375,
"loss_idm_type": 0.4764027833938599,
"loss_sft": 0.01487119197845459,
"step": 670,
"token_acc": 0.8408539980403075
},
{
"epoch": 0.15682882882882884,
"grad_norm": 162.03778076171875,
"learning_rate": 7.871643313414718e-06,
"loss": 17.195193481445312,
"loss_idm": 0.99375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5480063557624817,
"loss_idm_type": 0.496666419506073,
"loss_sft": 0.01497359536588192,
"step": 680,
"token_acc": 0.8398047298658003
},
{
"epoch": 0.15913513513513514,
"grad_norm": 156.19992065429688,
"learning_rate": 7.803575286758365e-06,
"loss": 16.83672180175781,
"loss_idm": 0.9796875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5627113699913024,
"loss_idm_type": 0.46034821271896365,
"loss_sft": 0.014724036678671837,
"step": 690,
"token_acc": 0.8421856731610026
},
{
"epoch": 0.16144144144144143,
"grad_norm": 136.38833618164062,
"learning_rate": 7.734740790612137e-06,
"loss": 16.986585998535155,
"loss_idm": 0.984375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5753911018371582,
"loss_idm_type": 0.45663111209869384,
"loss_sft": 0.014708232507109642,
"step": 700,
"token_acc": 0.8426992953605216
},
{
"epoch": 0.16374774774774775,
"grad_norm": 165.4947052001953,
"learning_rate": 7.66515864363997e-06,
"loss": 18.781927490234374,
"loss_idm": 1.1046875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6780822038650512,
"loss_idm_type": 0.4660151481628418,
"loss_sft": 0.014779145643115044,
"step": 710,
"token_acc": 0.8423064202890983
},
{
"epoch": 0.16605405405405405,
"grad_norm": 172.88607788085938,
"learning_rate": 7.594847868906076e-06,
"loss": 18.374440002441407,
"loss_idm": 1.0796875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.648744547367096,
"loss_idm_type": 0.4703775644302368,
"loss_sft": 0.014710108563303948,
"step": 720,
"token_acc": 0.8429784964811741
},
{
"epoch": 0.16836036036036037,
"grad_norm": 143.69039916992188,
"learning_rate": 7.52382768867422e-06,
"loss": 18.16466827392578,
"loss_idm": 1.05625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6015788435935974,
"loss_idm_type": 0.504117202758789,
"loss_sft": 0.014878561347723007,
"step": 730,
"token_acc": 0.8414000430536017
},
{
"epoch": 0.17066666666666666,
"grad_norm": 202.4527587890625,
"learning_rate": 7.452117519152542e-06,
"loss": 16.627297973632814,
"loss_idm": 0.9640625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5718691349029541,
"loss_idm_type": 0.4376717388629913,
"loss_sft": 0.01475221961736679,
"step": 740,
"token_acc": 0.84257142439324
},
{
"epoch": 0.17297297297297298,
"grad_norm": 144.1239013671875,
"learning_rate": 7.379736965185369e-06,
"loss": 17.226173400878906,
"loss_idm": 1.003125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5690211534500123,
"loss_idm_type": 0.4779587149620056,
"loss_sft": 0.014824393019080163,
"step": 750,
"token_acc": 0.8413612833037625
},
{
"epoch": 0.17527927927927928,
"grad_norm": 125.00702667236328,
"learning_rate": 7.30670581489344e-06,
"loss": 17.1047119140625,
"loss_idm": 0.9984375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5596216320991516,
"loss_idm_type": 0.47991371154785156,
"loss_sft": 0.014669803529977798,
"step": 760,
"token_acc": 0.842292145323193
},
{
"epoch": 0.1775855855855856,
"grad_norm": 178.8215789794922,
"learning_rate": 7.233044034264034e-06,
"loss": 17.40855255126953,
"loss_idm": 1.0140625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6001033067703248,
"loss_idm_type": 0.4585636019706726,
"loss_sft": 0.014787257090210915,
"step": 770,
"token_acc": 0.8423987697626192
},
{
"epoch": 0.1798918918918919,
"grad_norm": 121.97682189941406,
"learning_rate": 7.158771761692464e-06,
"loss": 16.357644653320314,
"loss_idm": 0.9375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5047407984733582,
"loss_idm_type": 0.487633216381073,
"loss_sft": 0.014906040206551553,
"step": 780,
"token_acc": 0.8404371211807341
},
{
"epoch": 0.18219819819819819,
"grad_norm": 174.59080505371094,
"learning_rate": 7.083909302476453e-06,
"loss": 20.202664184570313,
"loss_idm": 1.1828125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6652926445007324,
"loss_idm_type": 0.5675289869308472,
"loss_sft": 0.014899690076708794,
"step": 790,
"token_acc": 0.8409781183129702
},
{
"epoch": 0.1845045045045045,
"grad_norm": 129.4432830810547,
"learning_rate": 7.008477123264849e-06,
"loss": 17.296986389160157,
"loss_idm": 1.0171875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5993288516998291,
"loss_idm_type": 0.452558171749115,
"loss_sft": 0.014659978821873665,
"step": 800,
"token_acc": 0.8428556157974678
},
{
"epoch": 0.1868108108108108,
"grad_norm": 156.69631958007812,
"learning_rate": 6.932495846462262e-06,
"loss": 20.33173828125,
"loss_idm": 1.1921875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7339765548706054,
"loss_idm_type": 0.5067543745040893,
"loss_sft": 0.014883234724402427,
"step": 810,
"token_acc": 0.8409235431467038
},
{
"epoch": 0.18911711711711712,
"grad_norm": 202.87460327148438,
"learning_rate": 6.855986244591104e-06,
"loss": 16.402493286132813,
"loss_idm": 0.959375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5709848403930664,
"loss_idm_type": 0.4241690397262573,
"loss_sft": 0.014913064613938331,
"step": 820,
"token_acc": 0.8403605935424822
},
{
"epoch": 0.19142342342342342,
"grad_norm": 134.71864318847656,
"learning_rate": 6.778969234612583e-06,
"loss": 18.09036865234375,
"loss_idm": 1.05625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6172916054725647,
"loss_idm_type": 0.4832586348056793,
"loss_sft": 0.014967308938503265,
"step": 830,
"token_acc": 0.8397193120522027
},
{
"epoch": 0.19372972972972974,
"grad_norm": 163.35415649414062,
"learning_rate": 6.701465872208216e-06,
"loss": 17.737567138671874,
"loss_idm": 1.040625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.610646378993988,
"loss_idm_type": 0.4680565595626831,
"loss_sft": 0.015027177706360816,
"step": 840,
"token_acc": 0.8388462775444171
},
{
"epoch": 0.19603603603603603,
"grad_norm": 131.97361755371094,
"learning_rate": 6.6234973460234184e-06,
"loss": 13.829145812988282,
"loss_idm": 0.80546875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4581299006938934,
"loss_idm_type": 0.3765411376953125,
"loss_sft": 0.014915959164500236,
"step": 850,
"token_acc": 0.8408314009627786
},
{
"epoch": 0.19834234234234235,
"grad_norm": 154.64279174804688,
"learning_rate": 6.545084971874738e-06,
"loss": 19.65125427246094,
"loss_idm": 1.128125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6232832789421081,
"loss_idm_type": 0.5748029708862304,
"loss_sft": 0.01495017521083355,
"step": 860,
"token_acc": 0.8412365453675464
},
{
"epoch": 0.20064864864864865,
"grad_norm": 115.81620788574219,
"learning_rate": 6.466250186922325e-06,
"loss": 16.905009460449218,
"loss_idm": 1.0,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5850194334983826,
"loss_idm_type": 0.44217710494995116,
"loss_sft": 0.01471223458647728,
"step": 870,
"token_acc": 0.8420623604083333
},
{
"epoch": 0.20295495495495494,
"grad_norm": 121.02389526367188,
"learning_rate": 6.387014543809224e-06,
"loss": 17.58789978027344,
"loss_idm": 1.0234375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6387038111686707,
"loss_idm_type": 0.43074326515197753,
"loss_sft": 0.014781225100159645,
"step": 880,
"token_acc": 0.8414889055632973
},
{
"epoch": 0.20526126126126126,
"grad_norm": 169.18638610839844,
"learning_rate": 6.3073997047691e-06,
"loss": 16.101837158203125,
"loss_idm": 0.94609375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5324645042419434,
"loss_idm_type": 0.4441194772720337,
"loss_sft": 0.014909639954566956,
"step": 890,
"token_acc": 0.8407675658086115
},
{
"epoch": 0.20756756756756756,
"grad_norm": 131.3477325439453,
"learning_rate": 6.227427435703997e-06,
"loss": 16.599580383300783,
"loss_idm": 0.96328125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5476276993751525,
"loss_idm_type": 0.46061848402023314,
"loss_sft": 0.014580646157264709,
"step": 900,
"token_acc": 0.8437034651583426
},
{
"epoch": 0.20987387387387388,
"grad_norm": 125.8461685180664,
"learning_rate": 6.147119600233758e-06,
"loss": 17.687442016601562,
"loss_idm": 1.0234375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5852304697036743,
"loss_idm_type": 0.4907158613204956,
"loss_sft": 0.014707296714186668,
"step": 910,
"token_acc": 0.842660486829581
},
{
"epoch": 0.21218018018018017,
"grad_norm": 182.08822631835938,
"learning_rate": 6.066498153718735e-06,
"loss": 16.377294921875,
"loss_idm": 0.9609375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.562787675857544,
"loss_idm_type": 0.4312257766723633,
"loss_sft": 0.014806845411658286,
"step": 920,
"token_acc": 0.8419050583960423
},
{
"epoch": 0.2144864864864865,
"grad_norm": 122.23482513427734,
"learning_rate": 5.985585137257401e-06,
"loss": 17.473541259765625,
"loss_idm": 1.0140625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.631916093826294,
"loss_idm_type": 0.4307005345821381,
"loss_sft": 0.014695174247026443,
"step": 930,
"token_acc": 0.8427138438412396
},
{
"epoch": 0.2167927927927928,
"grad_norm": 127.09993743896484,
"learning_rate": 5.904402671660551e-06,
"loss": 17.38238525390625,
"loss_idm": 1.015625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6432816028594971,
"loss_idm_type": 0.41384764313697814,
"loss_sft": 0.014650922641158104,
"step": 940,
"token_acc": 0.8433863297046639
},
{
"epoch": 0.2190990990990991,
"grad_norm": 124.23971557617188,
"learning_rate": 5.82297295140367e-06,
"loss": 16.113180541992186,
"loss_idm": 0.94765625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5540828824043273,
"loss_idm_type": 0.4236065447330475,
"loss_sft": 0.014640064164996146,
"step": 950,
"token_acc": 0.8430864303493263
},
{
"epoch": 0.2214054054054054,
"grad_norm": 140.92527770996094,
"learning_rate": 5.74131823855921e-06,
"loss": 18.509353637695312,
"loss_idm": 1.0890625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6742399454116821,
"loss_idm_type": 0.4530523955821991,
"loss_sft": 0.014544900879263878,
"step": 960,
"token_acc": 0.8436570668828863
},
{
"epoch": 0.22371171171171172,
"grad_norm": 142.63829040527344,
"learning_rate": 5.659460856710346e-06,
"loss": 18.395001220703126,
"loss_idm": 1.0765625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.7047812342643738,
"loss_idm_type": 0.41602908372879027,
"loss_sft": 0.014448409527540207,
"step": 970,
"token_acc": 0.8444647711579308
},
{
"epoch": 0.22601801801801802,
"grad_norm": 158.4767608642578,
"learning_rate": 5.577423184847932e-06,
"loss": 18.995347595214845,
"loss_idm": 1.103125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6868046164512634,
"loss_idm_type": 0.4712188720703125,
"loss_sft": 0.014675500243902207,
"step": 980,
"token_acc": 0.8425279346161687
},
{
"epoch": 0.2283243243243243,
"grad_norm": 152.37652587890625,
"learning_rate": 5.495227651252315e-06,
"loss": 16.80406951904297,
"loss_idm": 0.96953125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.587271648645401,
"loss_idm_type": 0.4330908119678497,
"loss_sft": 0.014900797232985496,
"step": 990,
"token_acc": 0.8411618261296613
},
{
"epoch": 0.23063063063063063,
"grad_norm": 140.89053344726562,
"learning_rate": 5.412896727361663e-06,
"loss": 15.126055908203124,
"loss_idm": 0.8828125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5050670087337494,
"loss_idm_type": 0.41060460805892945,
"loss_sft": 0.01481594294309616,
"step": 1000,
"token_acc": 0.8417271846712852
},
{
"epoch": 0.23063063063063063,
"eval_loss": 1.2197141647338867,
"eval_runtime": 167.8304,
"eval_samples_per_second": 10.397,
"eval_steps_per_second": 2.604,
"eval_token_acc": 0.8293519552080156,
"step": 1000
},
{
"epoch": 0.23293693693693693,
"grad_norm": 130.3173828125,
"learning_rate": 5.3304529216284974e-06,
"loss": 16.93656768798828,
"loss_idm": 0.9765625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5961753606796265,
"loss_idm_type": 0.43301868438720703,
"loss_sft": 0.01461159810423851,
"step": 1010,
"token_acc": 0.8440814841392973
},
{
"epoch": 0.23524324324324325,
"grad_norm": 151.72671508789062,
"learning_rate": 5.247918773366112e-06,
"loss": 13.777294921875,
"loss_idm": 0.79609375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4060074031352997,
"loss_idm_type": 0.42571691870689393,
"loss_sft": 0.014732184261083603,
"step": 1020,
"token_acc": 0.8413653295862882
},
{
"epoch": 0.23754954954954954,
"grad_norm": 104.45248413085938,
"learning_rate": 5.165316846586541e-06,
"loss": 15.454080200195312,
"loss_idm": 0.9,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5253732204437256,
"loss_idm_type": 0.4112230122089386,
"loss_sft": 0.014539775997400283,
"step": 1030,
"token_acc": 0.8432018444851538
},
{
"epoch": 0.23985585585585587,
"grad_norm": 147.8674774169922,
"learning_rate": 5.082669723831793e-06,
"loss": 16.801179504394533,
"loss_idm": 0.9625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5981590151786804,
"loss_idm_type": 0.4219686686992645,
"loss_sft": 0.015009181201457977,
"step": 1040,
"token_acc": 0.8393066142930146
},
{
"epoch": 0.24216216216216216,
"grad_norm": 150.60118103027344,
"learning_rate": 5e-06,
"loss": 14.039901733398438,
"loss_idm": 0.81484375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4564467430114746,
"loss_idm_type": 0.39142499566078187,
"loss_sft": 0.014857906848192215,
"step": 1050,
"token_acc": 0.8403285527654876
},
{
"epoch": 0.24446846846846848,
"grad_norm": 171.65626525878906,
"learning_rate": 4.917330276168208e-06,
"loss": 15.874789428710937,
"loss_idm": 0.9296875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5143399000167846,
"loss_idm_type": 0.44793660640716554,
"loss_sft": 0.014889773353934288,
"step": 1060,
"token_acc": 0.8406433811941442
},
{
"epoch": 0.24677477477477477,
"grad_norm": 132.1097869873047,
"learning_rate": 4.8346831534134595e-06,
"loss": 17.117350769042968,
"loss_idm": 1.01015625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5567643463611602,
"loss_idm_type": 0.4839338779449463,
"loss_sft": 0.014736819267272949,
"step": 1070,
"token_acc": 0.8418920880835263
},
{
"epoch": 0.24908108108108107,
"grad_norm": 169.35418701171875,
"learning_rate": 4.752081226633888e-06,
"loss": 16.461138916015624,
"loss_idm": 0.9640625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5909170269966125,
"loss_idm_type": 0.40844297409057617,
"loss_sft": 0.014634986221790314,
"step": 1080,
"token_acc": 0.8421809887688029
},
{
"epoch": 0.25138738738738736,
"grad_norm": 175.02035522460938,
"learning_rate": 4.669547078371503e-06,
"loss": 17.491482543945313,
"loss_idm": 1.0109375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5998989224433899,
"loss_idm_type": 0.464590322971344,
"loss_sft": 0.014502325654029846,
"step": 1090,
"token_acc": 0.8444604404318916
},
{
"epoch": 0.2536936936936937,
"grad_norm": 122.71571350097656,
"learning_rate": 4.587103272638339e-06,
"loss": 15.66458740234375,
"loss_idm": 0.9109375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5785394430160522,
"loss_idm_type": 0.371947318315506,
"loss_sft": 0.014418132230639457,
"step": 1100,
"token_acc": 0.8450031826861871
},
{
"epoch": 0.256,
"grad_norm": 118.75171661376953,
"learning_rate": 4.504772348747687e-06,
"loss": 16.25250701904297,
"loss_idm": 0.946875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5665743708610534,
"loss_idm_type": 0.420313960313797,
"loss_sft": 0.014555140212178231,
"step": 1110,
"token_acc": 0.8431390480333372
},
{
"epoch": 0.2583063063063063,
"grad_norm": 121.8458480834961,
"learning_rate": 4.42257681515207e-06,
"loss": 17.592311096191406,
"loss_idm": 1.03203125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6627050995826721,
"loss_idm_type": 0.4078234493732452,
"loss_sft": 0.014561154693365098,
"step": 1120,
"token_acc": 0.8442745710600775
},
{
"epoch": 0.2606126126126126,
"grad_norm": 112.1263656616211,
"learning_rate": 4.340539143289655e-06,
"loss": 17.300210571289064,
"loss_idm": 1.0015625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6403586149215699,
"loss_idm_type": 0.41174074411392214,
"loss_sft": 0.014503993466496468,
"step": 1130,
"token_acc": 0.8440554077281796
},
{
"epoch": 0.26291891891891894,
"grad_norm": 163.71958923339844,
"learning_rate": 4.25868176144079e-06,
"loss": 16.03717041015625,
"loss_idm": 0.92890625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5435431241989136,
"loss_idm_type": 0.42921730875968933,
"loss_sft": 0.014790638536214828,
"step": 1140,
"token_acc": 0.8415522579531016
},
{
"epoch": 0.26522522522522524,
"grad_norm": 163.40823364257812,
"learning_rate": 4.17702704859633e-06,
"loss": 16.0930908203125,
"loss_idm": 0.9375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5482388496398926,
"loss_idm_type": 0.4281274676322937,
"loss_sft": 0.014842434972524642,
"step": 1150,
"token_acc": 0.8399775713902174
},
{
"epoch": 0.26753153153153153,
"grad_norm": 196.81509399414062,
"learning_rate": 4.0955973283394525e-06,
"loss": 15.240176391601562,
"loss_idm": 0.8859375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5218775153160096,
"loss_idm_type": 0.4013541877269745,
"loss_sft": 0.014632915332913398,
"step": 1160,
"token_acc": 0.8420310842684929
},
{
"epoch": 0.2698378378378378,
"grad_norm": 149.12844848632812,
"learning_rate": 4.0144148627426e-06,
"loss": 15.899664306640625,
"loss_idm": 0.921875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5356297016143798,
"loss_idm_type": 0.42870389223098754,
"loss_sft": 0.014795275405049324,
"step": 1170,
"token_acc": 0.8410206589805115
},
{
"epoch": 0.2721441441441441,
"grad_norm": 128.20870971679688,
"learning_rate": 3.9335018462812664e-06,
"loss": 15.448335266113281,
"loss_idm": 0.890625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.45652764439582827,
"loss_idm_type": 0.4790655612945557,
"loss_sft": 0.01491394303739071,
"step": 1180,
"token_acc": 0.840088487985511
},
{
"epoch": 0.27445045045045047,
"grad_norm": 144.2740020751953,
"learning_rate": 3.852880399766243e-06,
"loss": 17.782185363769532,
"loss_idm": 1.04375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6208782076835633,
"loss_idm_type": 0.4611518859863281,
"loss_sft": 0.01475386880338192,
"step": 1190,
"token_acc": 0.8415199212985736
},
{
"epoch": 0.27675675675675676,
"grad_norm": 124.68944549560547,
"learning_rate": 3.7725725642960047e-06,
"loss": 15.25560302734375,
"loss_idm": 0.88359375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5610093712806702,
"loss_idm_type": 0.36283376812934875,
"loss_sft": 0.014852567762136459,
"step": 1200,
"token_acc": 0.8405048860315332
},
{
"epoch": 0.27906306306306305,
"grad_norm": 131.37930297851562,
"learning_rate": 3.6926002952309015e-06,
"loss": 14.997708129882813,
"loss_idm": 0.88125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5199843943119049,
"loss_idm_type": 0.38818838000297545,
"loss_sft": 0.014678910747170449,
"step": 1210,
"token_acc": 0.842534823846304
},
{
"epoch": 0.28136936936936935,
"grad_norm": 172.28526306152344,
"learning_rate": 3.6129854561907786e-06,
"loss": 12.263350677490234,
"loss_idm": 0.7109375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.3645629286766052,
"loss_idm_type": 0.3720847427845001,
"loss_sft": 0.014872964471578598,
"step": 1220,
"token_acc": 0.840045987605899
},
{
"epoch": 0.2836756756756757,
"grad_norm": 175.21762084960938,
"learning_rate": 3.533749813077677e-06,
"loss": 16.723817443847658,
"loss_idm": 0.97109375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6298807859420776,
"loss_idm_type": 0.38632739186286924,
"loss_sft": 0.014517011120915413,
"step": 1230,
"token_acc": 0.844244322443512
},
{
"epoch": 0.285981981981982,
"grad_norm": 168.99330139160156,
"learning_rate": 3.4549150281252635e-06,
"loss": 14.8901123046875,
"loss_idm": 0.86640625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5231156468391418,
"loss_idm_type": 0.3788294196128845,
"loss_sft": 0.01444997638463974,
"step": 1240,
"token_acc": 0.8439816397285024
},
{
"epoch": 0.2882882882882883,
"grad_norm": 138.8653106689453,
"learning_rate": 3.3765026539765832e-06,
"loss": 16.886776733398438,
"loss_idm": 0.9796875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6324859857559204,
"loss_idm_type": 0.3942205488681793,
"loss_sft": 0.01420472450554371,
"step": 1250,
"token_acc": 0.8471226070215119
},
{
"epoch": 0.2905945945945946,
"grad_norm": 169.21224975585938,
"learning_rate": 3.298534127791785e-06,
"loss": 15.787973022460937,
"loss_idm": 0.940625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5526616215705872,
"loss_idm_type": 0.4051926553249359,
"loss_sft": 0.014482183754444123,
"step": 1260,
"token_acc": 0.844163385654321
},
{
"epoch": 0.2929009009009009,
"grad_norm": 143.0062255859375,
"learning_rate": 3.2210307653874175e-06,
"loss": 14.827607727050781,
"loss_idm": 0.86640625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5377736687660217,
"loss_idm_type": 0.3600290656089783,
"loss_sft": 0.014454721286892892,
"step": 1270,
"token_acc": 0.8450970398554747
},
{
"epoch": 0.2952072072072072,
"grad_norm": 151.57211303710938,
"learning_rate": 3.1440137554088957e-06,
"loss": 15.278717041015625,
"loss_idm": 0.88515625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5193838059902192,
"loss_idm_type": 0.4066817402839661,
"loss_sft": 0.014557655900716782,
"step": 1280,
"token_acc": 0.8426562450570764
},
{
"epoch": 0.2975135135135135,
"grad_norm": 154.91415405273438,
"learning_rate": 3.06750415353774e-06,
"loss": 14.918099975585937,
"loss_idm": 0.86875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5365460634231567,
"loss_idm_type": 0.36664303541183474,
"loss_sft": 0.014581486210227012,
"step": 1290,
"token_acc": 0.8425958329841925
},
{
"epoch": 0.2998198198198198,
"grad_norm": 184.6160125732422,
"learning_rate": 2.991522876735154e-06,
"loss": 15.444512939453125,
"loss_idm": 0.9015625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5625595211982727,
"loss_idm_type": 0.37340242266654966,
"loss_sft": 0.014611029624938964,
"step": 1300,
"token_acc": 0.8428093884884362
},
{
"epoch": 0.3021261261261261,
"grad_norm": 153.15501403808594,
"learning_rate": 2.9160906975235493e-06,
"loss": 13.48157958984375,
"loss_idm": 0.77890625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4528303623199463,
"loss_idm_type": 0.3603403568267822,
"loss_sft": 0.014692678675055503,
"step": 1310,
"token_acc": 0.8424510106746506
},
{
"epoch": 0.30443243243243245,
"grad_norm": 190.70094299316406,
"learning_rate": 2.8412282383075362e-06,
"loss": 16.24449005126953,
"loss_idm": 0.953125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6082704544067383,
"loss_idm_type": 0.37772817611694337,
"loss_sft": 0.014619594439864159,
"step": 1320,
"token_acc": 0.8426580787448936
},
{
"epoch": 0.30673873873873875,
"grad_norm": 168.48741149902344,
"learning_rate": 2.766955965735968e-06,
"loss": 14.77047119140625,
"loss_idm": 0.8625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5200425565242768,
"loss_idm_type": 0.3733003795146942,
"loss_sft": 0.014853813871741294,
"step": 1330,
"token_acc": 0.8407578943217082
},
{
"epoch": 0.30904504504504504,
"grad_norm": 170.83592224121094,
"learning_rate": 2.693294185106562e-06,
"loss": 13.439047241210938,
"loss_idm": 0.778125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.42561874985694886,
"loss_idm_type": 0.3851174831390381,
"loss_sft": 0.014666719362139702,
"step": 1340,
"token_acc": 0.8422743491962282
},
{
"epoch": 0.31135135135135134,
"grad_norm": 153.26455688476562,
"learning_rate": 2.6202630348146323e-06,
"loss": 14.712338256835938,
"loss_idm": 0.85546875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5100657343864441,
"loss_idm_type": 0.38050331473350524,
"loss_sft": 0.014491959661245345,
"step": 1350,
"token_acc": 0.8438898168585576
},
{
"epoch": 0.3136576576576577,
"grad_norm": 143.23974609375,
"learning_rate": 2.5478824808474613e-06,
"loss": 15.174507141113281,
"loss_idm": 0.88203125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5404232978820801,
"loss_idm_type": 0.3794264256954193,
"loss_sft": 0.014417991414666176,
"step": 1360,
"token_acc": 0.8445792418646086
},
{
"epoch": 0.315963963963964,
"grad_norm": 162.9635467529297,
"learning_rate": 2.476172311325783e-06,
"loss": 15.683633422851562,
"loss_idm": 0.925,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5593078851699829,
"loss_idm_type": 0.39195163249969484,
"loss_sft": 0.014683765918016433,
"step": 1370,
"token_acc": 0.8423480887239503
},
{
"epoch": 0.3182702702702703,
"grad_norm": 200.83192443847656,
"learning_rate": 2.4051521310939258e-06,
"loss": 15.915536499023437,
"loss_idm": 0.9296875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.56450035572052,
"loss_idm_type": 0.40067616701126096,
"loss_sft": 0.014767180755734443,
"step": 1380,
"token_acc": 0.8416730217876071
},
{
"epoch": 0.32057657657657657,
"grad_norm": 177.85284423828125,
"learning_rate": 2.3348413563600324e-06,
"loss": 14.628662109375,
"loss_idm": 0.8421875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.44927034378051756,
"loss_idm_type": 0.43579859733581544,
"loss_sft": 0.01455475315451622,
"step": 1390,
"token_acc": 0.843599916985094
},
{
"epoch": 0.32288288288288286,
"grad_norm": 119.86055755615234,
"learning_rate": 2.265259209387867e-06,
"loss": 14.170437622070313,
"loss_idm": 0.8203125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4732060968875885,
"loss_idm_type": 0.38350951075553896,
"loss_sft": 0.014363614469766616,
"step": 1400,
"token_acc": 0.8447947852357484
},
{
"epoch": 0.3251891891891892,
"grad_norm": 168.4610137939453,
"learning_rate": 2.1964247132416373e-06,
"loss": 14.9684814453125,
"loss_idm": 0.86953125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.550681185722351,
"loss_idm_type": 0.3555972933769226,
"loss_sft": 0.014621422439813615,
"step": 1410,
"token_acc": 0.8420637028465194
},
{
"epoch": 0.3274954954954955,
"grad_norm": 139.091064453125,
"learning_rate": 2.1283566865852824e-06,
"loss": 13.624758911132812,
"loss_idm": 0.78515625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4843453407287598,
"loss_idm_type": 0.33815571665763855,
"loss_sft": 0.014555397629737853,
"step": 1420,
"token_acc": 0.8435839194998046
},
{
"epoch": 0.3298018018018018,
"grad_norm": 188.8422088623047,
"learning_rate": 2.061073738537635e-06,
"loss": 14.195480346679688,
"loss_idm": 0.83203125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.50216104388237,
"loss_idm_type": 0.3559306800365448,
"loss_sft": 0.014532771334052086,
"step": 1430,
"token_acc": 0.844069125990222
},
{
"epoch": 0.3321081081081081,
"grad_norm": 164.5511016845703,
"learning_rate": 1.9945942635848745e-06,
"loss": 15.28197021484375,
"loss_idm": 0.8921875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5497690439224243,
"loss_idm_type": 0.37624791264533997,
"loss_sft": 0.014516519382596016,
"step": 1440,
"token_acc": 0.8445253276742052
},
{
"epoch": 0.33441441441441444,
"grad_norm": 158.5382537841797,
"learning_rate": 1.928936436551661e-06,
"loss": 16.544921875,
"loss_idm": 0.965625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6344911813735962,
"loss_idm_type": 0.3706976056098938,
"loss_sft": 0.014539649710059166,
"step": 1450,
"token_acc": 0.8431598369407435
},
{
"epoch": 0.33672072072072073,
"grad_norm": 148.32135009765625,
"learning_rate": 1.864118207632315e-06,
"loss": 14.539396667480469,
"loss_idm": 0.8421875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.508997893333435,
"loss_idm_type": 0.37080259919166564,
"loss_sft": 0.014571281149983406,
"step": 1460,
"token_acc": 0.8433491618937913
},
{
"epoch": 0.33902702702702703,
"grad_norm": 173.5018768310547,
"learning_rate": 1.8001572974834169e-06,
"loss": 14.692778015136719,
"loss_idm": 0.85625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5423828899860382,
"loss_idm_type": 0.34737696051597594,
"loss_sft": 0.014310755580663682,
"step": 1470,
"token_acc": 0.8453443239566244
},
{
"epoch": 0.3413333333333333,
"grad_norm": 172.0346221923828,
"learning_rate": 1.7370711923791567e-06,
"loss": 12.533441162109375,
"loss_idm": 0.7171875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4120396256446838,
"loss_idm_type": 0.3425567626953125,
"loss_sft": 0.014442760124802589,
"step": 1480,
"token_acc": 0.8447839272919999
},
{
"epoch": 0.3436396396396396,
"grad_norm": 119.68669891357422,
"learning_rate": 1.6748771394307584e-06,
"loss": 14.060650634765626,
"loss_idm": 0.8171875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.47142205834388734,
"loss_idm_type": 0.3782303512096405,
"loss_sft": 0.014632508531212807,
"step": 1490,
"token_acc": 0.8420826182134532
},
{
"epoch": 0.34594594594594597,
"grad_norm": 164.32711791992188,
"learning_rate": 1.6135921418712959e-06,
"loss": 14.183511352539062,
"loss_idm": 0.82890625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.49015386700630187,
"loss_idm_type": 0.3677518665790558,
"loss_sft": 0.014323790743947028,
"step": 1500,
"token_acc": 0.8457848994458402
},
{
"epoch": 0.34594594594594597,
"eval_loss": 1.1410207748413086,
"eval_runtime": 170.3105,
"eval_samples_per_second": 10.246,
"eval_steps_per_second": 2.566,
"eval_token_acc": 0.831800591581073,
"step": 1500
},
{
"epoch": 0.34825225225225226,
"grad_norm": 134.0631866455078,
"learning_rate": 1.5532329544071712e-06,
"loss": 14.652560424804687,
"loss_idm": 0.85,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5261287689208984,
"loss_idm_type": 0.36060481071472167,
"loss_sft": 0.014461614564061165,
"step": 1510,
"token_acc": 0.844176361115255
},
{
"epoch": 0.35055855855855855,
"grad_norm": 172.90817260742188,
"learning_rate": 1.4938160786375571e-06,
"loss": 13.671232604980469,
"loss_idm": 0.7875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.46637305021286013,
"loss_idm_type": 0.35906325578689574,
"loss_sft": 0.014588062092661858,
"step": 1520,
"token_acc": 0.8432830672186084
},
{
"epoch": 0.35286486486486485,
"grad_norm": 151.86465454101562,
"learning_rate": 1.4353577585430152e-06,
"loss": 14.50615997314453,
"loss_idm": 0.84375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5052099466323853,
"loss_idm_type": 0.3723944365978241,
"loss_sft": 0.014470845833420754,
"step": 1530,
"token_acc": 0.8444796349526804
},
{
"epoch": 0.3551711711711712,
"grad_norm": 156.16677856445312,
"learning_rate": 1.3778739760445552e-06,
"loss": 14.215463256835937,
"loss_idm": 0.8265625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5262757420539856,
"loss_idm_type": 0.33302531838417054,
"loss_sft": 0.014604969695210458,
"step": 1540,
"token_acc": 0.8422479843334597
},
{
"epoch": 0.3574774774774775,
"grad_norm": 169.1072235107422,
"learning_rate": 1.321380446634342e-06,
"loss": 15.241757202148438,
"loss_idm": 0.88828125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5481937170028687,
"loss_idm_type": 0.3758387088775635,
"loss_sft": 0.014357226341962815,
"step": 1550,
"token_acc": 0.8454274612361345
},
{
"epoch": 0.3597837837837838,
"grad_norm": 192.34034729003906,
"learning_rate": 1.2658926150792321e-06,
"loss": 12.882757568359375,
"loss_idm": 0.74296875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.43231621384620667,
"loss_idm_type": 0.34373490810394286,
"loss_sft": 0.014632303640246392,
"step": 1560,
"token_acc": 0.842279257949761
},
{
"epoch": 0.3620900900900901,
"grad_norm": 104.9269027709961,
"learning_rate": 1.2114256511983274e-06,
"loss": 13.561306762695313,
"loss_idm": 0.7890625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.48736552000045774,
"loss_idm_type": 0.3312386333942413,
"loss_sft": 0.014400731772184372,
"step": 1570,
"token_acc": 0.8450077725080943
},
{
"epoch": 0.36439639639639637,
"grad_norm": 137.5407257080078,
"learning_rate": 1.157994445715706e-06,
"loss": 12.73305892944336,
"loss_idm": 0.74296875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.45667484998703,
"loss_idm_type": 0.3100732803344727,
"loss_sft": 0.014485693350434304,
"step": 1580,
"token_acc": 0.8445482043989659
},
{
"epoch": 0.3667027027027027,
"grad_norm": 173.36251831054688,
"learning_rate": 1.1056136061894386e-06,
"loss": 15.467356872558593,
"loss_idm": 0.90625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5730970621109008,
"loss_idm_type": 0.3644375681877136,
"loss_sft": 0.014653195813298225,
"step": 1590,
"token_acc": 0.8428073355777439
},
{
"epoch": 0.369009009009009,
"grad_norm": 186.8936004638672,
"learning_rate": 1.0542974530180327e-06,
"loss": 15.348123168945312,
"loss_idm": 0.8921875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5705770015716553,
"loss_idm_type": 0.3599770188331604,
"loss_sft": 0.014523046091198921,
"step": 1600,
"token_acc": 0.8441250875980975
},
{
"epoch": 0.3713153153153153,
"grad_norm": 178.75588989257812,
"learning_rate": 1.0040600155253766e-06,
"loss": 13.724775695800782,
"loss_idm": 0.80390625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.47891607880592346,
"loss_idm_type": 0.34979560375213625,
"loss_sft": 0.014572210982441903,
"step": 1610,
"token_acc": 0.8430812068067953
},
{
"epoch": 0.3736216216216216,
"grad_norm": 159.6516571044922,
"learning_rate": 9.549150281252633e-07,
"loss": 13.920655822753906,
"loss_idm": 0.81796875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4648734211921692,
"loss_idm_type": 0.37584719657897947,
"loss_sft": 0.014640329405665398,
"step": 1620,
"token_acc": 0.8430257319440593
},
{
"epoch": 0.37592792792792795,
"grad_norm": 207.8046112060547,
"learning_rate": 9.068759265665384e-07,
"loss": 14.935466003417968,
"loss_idm": 0.86484375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4873470187187195,
"loss_idm_type": 0.4168549060821533,
"loss_sft": 0.014615200087428093,
"step": 1630,
"token_acc": 0.8436280183299338
},
{
"epoch": 0.37823423423423425,
"grad_norm": 204.06910705566406,
"learning_rate": 8.599558442598998e-07,
"loss": 14.955171203613281,
"loss_idm": 0.878125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5665068745613098,
"loss_idm_type": 0.338987672328949,
"loss_sft": 0.014446334168314933,
"step": 1640,
"token_acc": 0.8444171697580176
},
{
"epoch": 0.38054054054054054,
"grad_norm": 158.568603515625,
"learning_rate": 8.141676086873574e-07,
"loss": 14.716546630859375,
"loss_idm": 0.85625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5281898379325867,
"loss_idm_type": 0.3627839803695679,
"loss_sft": 0.014333575591444969,
"step": 1650,
"token_acc": 0.844941266398204
},
{
"epoch": 0.38284684684684683,
"grad_norm": 135.85830688476562,
"learning_rate": 7.695237378953224e-07,
"loss": 13.48638916015625,
"loss_idm": 0.78203125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.459857702255249,
"loss_idm_type": 0.3540190577507019,
"loss_sft": 0.014460830017924309,
"step": 1660,
"token_acc": 0.8442960665642909
},
{
"epoch": 0.3851531531531531,
"grad_norm": 189.99652099609375,
"learning_rate": 7.260364370723044e-07,
"loss": 13.14095458984375,
"loss_idm": 0.76484375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4547333836555481,
"loss_idm_type": 0.3371516466140747,
"loss_sft": 0.014711619913578033,
"step": 1670,
"token_acc": 0.842195473774804
},
{
"epoch": 0.3874594594594595,
"grad_norm": 157.12652587890625,
"learning_rate": 6.837175952121305e-07,
"loss": 14.613890075683594,
"loss_idm": 0.853125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5038619041442871,
"loss_idm_type": 0.3803012192249298,
"loss_sft": 0.014559409022331238,
"step": 1680,
"token_acc": 0.8442514363439291
},
{
"epoch": 0.38976576576576577,
"grad_norm": 119.22041320800781,
"learning_rate": 6.425787818636131e-07,
"loss": 13.258279418945312,
"loss_idm": 0.771875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5110712051391602,
"loss_idm_type": 0.28840842843055725,
"loss_sft": 0.014597396552562713,
"step": 1690,
"token_acc": 0.8432623564580572
},
{
"epoch": 0.39207207207207206,
"grad_norm": 175.15042114257812,
"learning_rate": 6.026312439675553e-07,
"loss": 12.107723236083984,
"loss_idm": 0.7015625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.42078765034675597,
"loss_idm_type": 0.3071610450744629,
"loss_sft": 0.014557906612753867,
"step": 1700,
"token_acc": 0.8431742125984252
},
{
"epoch": 0.39437837837837836,
"grad_norm": 185.7315216064453,
"learning_rate": 5.63885902781941e-07,
"loss": 12.675588989257813,
"loss_idm": 0.734375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4352951407432556,
"loss_idm_type": 0.32744150757789614,
"loss_sft": 0.014774365723133088,
"step": 1710,
"token_acc": 0.8410658616749658
},
{
"epoch": 0.3966846846846847,
"grad_norm": 172.1112518310547,
"learning_rate": 5.263533508961827e-07,
"loss": 13.636004638671874,
"loss_idm": 0.79921875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4823790669441223,
"loss_idm_type": 0.341190242767334,
"loss_sft": 0.014395377039909363,
"step": 1720,
"token_acc": 0.8447912741887179
},
{
"epoch": 0.398990990990991,
"grad_norm": 194.44125366210938,
"learning_rate": 4.900438493352056e-07,
"loss": 14.4458251953125,
"loss_idm": 0.8421875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5258590936660766,
"loss_idm_type": 0.34829120635986327,
"loss_sft": 0.014413874596357346,
"step": 1730,
"token_acc": 0.8443261634304321
},
{
"epoch": 0.4012972972972973,
"grad_norm": 160.75350952148438,
"learning_rate": 4.549673247541875e-07,
"loss": 13.175152587890626,
"loss_idm": 0.765625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.47388927936553954,
"loss_idm_type": 0.32060781717300413,
"loss_sft": 0.01451372280716896,
"step": 1740,
"token_acc": 0.8440697707998421
},
{
"epoch": 0.4036036036036036,
"grad_norm": 141.08255004882812,
"learning_rate": 4.211333667247125e-07,
"loss": 14.0310546875,
"loss_idm": 0.82109375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5559897541999816,
"loss_idm_type": 0.2918687522411346,
"loss_sft": 0.014575795829296112,
"step": 1750,
"token_acc": 0.8436797661415865
},
{
"epoch": 0.4059099099099099,
"grad_norm": 118.26886749267578,
"learning_rate": 3.885512251130763e-07,
"loss": 13.044111633300782,
"loss_idm": 0.75625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4907976269721985,
"loss_idm_type": 0.2953831195831299,
"loss_sft": 0.014592215046286583,
"step": 1760,
"token_acc": 0.843192495151733
},
{
"epoch": 0.40821621621621623,
"grad_norm": 115.94628143310547,
"learning_rate": 3.572298075514652e-07,
"loss": 12.254039001464843,
"loss_idm": 0.71171875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.45790221095085143,
"loss_idm_type": 0.27874688506126405,
"loss_sft": 0.014564001187682151,
"step": 1770,
"token_acc": 0.8432218035804899
},
{
"epoch": 0.4105225225225225,
"grad_norm": 190.5941162109375,
"learning_rate": 3.271776770026963e-07,
"loss": 13.8462158203125,
"loss_idm": 0.80625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4681180715560913,
"loss_idm_type": 0.36803357005119325,
"loss_sft": 0.014678296074271202,
"step": 1780,
"token_acc": 0.8422379609892503
},
{
"epoch": 0.4128288288288288,
"grad_norm": 174.9483642578125,
"learning_rate": 2.984030494191942e-07,
"loss": 15.327230834960938,
"loss_idm": 0.8953125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.6190297245979309,
"loss_idm_type": 0.3098260760307312,
"loss_sft": 0.014648788422346116,
"step": 1790,
"token_acc": 0.8430502741024195
},
{
"epoch": 0.4151351351351351,
"grad_norm": 230.11903381347656,
"learning_rate": 2.7091379149682683e-07,
"loss": 14.7419921875,
"loss_idm": 0.8640625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5182127594947815,
"loss_idm_type": 0.3741490125656128,
"loss_sft": 0.014429467916488647,
"step": 1800,
"token_acc": 0.8446535756849075
},
{
"epoch": 0.41744144144144146,
"grad_norm": 154.8221435546875,
"learning_rate": 2.447174185242324e-07,
"loss": 14.704022216796876,
"loss_idm": 0.8578125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5479737043380737,
"loss_idm_type": 0.34190752506256106,
"loss_sft": 0.014517124742269516,
"step": 1810,
"token_acc": 0.8442064097203433
},
{
"epoch": 0.41974774774774776,
"grad_norm": 135.4575653076172,
"learning_rate": 2.198210923282118e-07,
"loss": 13.46357421875,
"loss_idm": 0.7828125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.44614426493644715,
"loss_idm_type": 0.36666680574417115,
"loss_sft": 0.014297968521714211,
"step": 1820,
"token_acc": 0.8459873948534788
},
{
"epoch": 0.42205405405405405,
"grad_norm": 165.0076446533203,
"learning_rate": 1.962316193157593e-07,
"loss": 13.562332153320312,
"loss_idm": 0.78359375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4542937457561493,
"loss_idm_type": 0.3639734387397766,
"loss_sft": 0.014790578186511994,
"step": 1830,
"token_acc": 0.8415396272753187
},
{
"epoch": 0.42436036036036034,
"grad_norm": 153.8871612548828,
"learning_rate": 1.7395544861325718e-07,
"loss": 13.68433074951172,
"loss_idm": 0.7921875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5039100408554077,
"loss_idm_type": 0.3223310589790344,
"loss_sft": 0.014507872611284256,
"step": 1840,
"token_acc": 0.8443365310135256
},
{
"epoch": 0.4266666666666667,
"grad_norm": 129.01284790039062,
"learning_rate": 1.5299867030334815e-07,
"loss": 14.797528076171876,
"loss_idm": 0.8671875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.515932697057724,
"loss_idm_type": 0.3804643988609314,
"loss_sft": 0.01432812139391899,
"step": 1850,
"token_acc": 0.8463651711771111
},
{
"epoch": 0.428972972972973,
"grad_norm": 157.6551971435547,
"learning_rate": 1.333670137599713e-07,
"loss": 11.997281646728515,
"loss_idm": 0.69296875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.39791205525398254,
"loss_idm_type": 0.32309883236885073,
"loss_sft": 0.014476843923330308,
"step": 1860,
"token_acc": 0.8449281079602797
},
{
"epoch": 0.4312792792792793,
"grad_norm": 148.4026641845703,
"learning_rate": 1.1506584608200366e-07,
"loss": 13.108309936523437,
"loss_idm": 0.76015625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.45751161575317384,
"loss_idm_type": 0.332503741979599,
"loss_sft": 0.014562473632395267,
"step": 1870,
"token_acc": 0.8440633265189181
},
{
"epoch": 0.4335855855855856,
"grad_norm": 183.43215942382812,
"learning_rate": 9.810017062595322e-08,
"loss": 13.245065307617187,
"loss_idm": 0.76015625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.48887420296669004,
"loss_idm_type": 0.3101705193519592,
"loss_sft": 0.014424510672688485,
"step": 1880,
"token_acc": 0.8447887962399647
},
{
"epoch": 0.43589189189189187,
"grad_norm": 167.6036376953125,
"learning_rate": 8.247462563808816e-08,
"loss": 15.903143310546875,
"loss_idm": 0.93125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5748040556907654,
"loss_idm_type": 0.3895548641681671,
"loss_sft": 0.014693231508135795,
"step": 1890,
"token_acc": 0.8419147841402718
},
{
"epoch": 0.4381981981981982,
"grad_norm": 196.0062255859375,
"learning_rate": 6.819348298638839e-08,
"loss": 15.310247802734375,
"loss_idm": 0.88984375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5848313808441162,
"loss_idm_type": 0.34234494566917417,
"loss_sft": 0.014819808304309845,
"step": 1900,
"token_acc": 0.8410884589302976
},
{
"epoch": 0.4405045045045045,
"grad_norm": 121.38626098632812,
"learning_rate": 5.526064699265754e-08,
"loss": 13.221337890625,
"loss_idm": 0.76484375,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.48920504450798036,
"loss_idm_type": 0.3081607401371002,
"loss_sft": 0.014549450576305389,
"step": 1910,
"token_acc": 0.8433808622624156
},
{
"epoch": 0.4428108108108108,
"grad_norm": 153.52781677246094,
"learning_rate": 4.367965336512403e-08,
"loss": 15.649809265136719,
"loss_idm": 0.9140625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5894887566566467,
"loss_idm_type": 0.35940555334091184,
"loss_sft": 0.014617808908224107,
"step": 1920,
"token_acc": 0.8429643249835993
},
{
"epoch": 0.4451171171171171,
"grad_norm": 205.32659912109375,
"learning_rate": 3.345366823180929e-08,
"loss": 12.812985229492188,
"loss_idm": 0.73828125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.46054392457008364,
"loss_idm_type": 0.31097897291183474,
"loss_sft": 0.014559100568294524,
"step": 1930,
"token_acc": 0.8434137485860117
},
{
"epoch": 0.44742342342342345,
"grad_norm": 190.1567840576172,
"learning_rate": 2.4585487274942922e-08,
"loss": 13.618353271484375,
"loss_idm": 0.790625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.4890588939189911,
"loss_idm_type": 0.3323815166950226,
"loss_sft": 0.01473580189049244,
"step": 1940,
"token_acc": 0.8418634068929035
},
{
"epoch": 0.44972972972972974,
"grad_norm": 127.31536865234375,
"learning_rate": 1.7077534966650767e-08,
"loss": 13.815167236328126,
"loss_idm": 0.8,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.47678937911987307,
"loss_idm_type": 0.3570751488208771,
"loss_sft": 0.014726188406348229,
"step": 1950,
"token_acc": 0.8425276171661491
},
{
"epoch": 0.45203603603603604,
"grad_norm": 227.32894897460938,
"learning_rate": 1.0931863906127327e-08,
"loss": 12.043135070800782,
"loss_idm": 0.696875,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.38389376401901243,
"loss_idm_type": 0.33983545303344725,
"loss_sft": 0.014539222791790962,
"step": 1960,
"token_acc": 0.8433404940923738
},
{
"epoch": 0.45434234234234233,
"grad_norm": 128.14857482910156,
"learning_rate": 6.150154258476315e-09,
"loss": 15.116073608398438,
"loss_idm": 0.87578125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5616528272628785,
"loss_idm_type": 0.35381786823272704,
"loss_sft": 0.014636943861842156,
"step": 1970,
"token_acc": 0.8430521353865116
},
{
"epoch": 0.4566486486486486,
"grad_norm": 151.3446502685547,
"learning_rate": 2.7337132953697555e-09,
"loss": 13.028573608398437,
"loss_idm": 0.75625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.459896320104599,
"loss_idm_type": 0.32475685477256777,
"loss_sft": 0.01476614847779274,
"step": 1980,
"token_acc": 0.8421346361905135
},
{
"epoch": 0.458954954954955,
"grad_norm": 156.91143798828125,
"learning_rate": 6.834750376549793e-10,
"loss": 14.470024108886719,
"loss_idm": 0.84140625,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5725686311721802,
"loss_idm_type": 0.3024349570274353,
"loss_sft": 0.014753464981913567,
"step": 1990,
"token_acc": 0.8424756483786394
},
{
"epoch": 0.46126126126126127,
"grad_norm": 212.70709228515625,
"learning_rate": 0.0,
"loss": 15.546224975585938,
"loss_idm": 0.9125,
"loss_idm_anl": 0.0,
"loss_idm_coord": 0.0,
"loss_idm_text": 0.5768885254859925,
"loss_idm_type": 0.3646038830280304,
"loss_sft": 0.014924455806612969,
"step": 2000,
"token_acc": 0.8399534838453687
},
{
"epoch": 0.46126126126126127,
"eval_loss": 1.1327581405639648,
"eval_runtime": 167.6796,
"eval_samples_per_second": 10.407,
"eval_steps_per_second": 2.606,
"eval_token_acc": 0.8322550845475353,
"step": 2000
}
],
"logging_steps": 10,
"max_steps": 2000,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"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": 2.5018639729226154e+19,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}