groot-pick-place-v4 / trainer_state.json
tolasing's picture
Upload folder using huggingface_hub
c59c10f verified
Raw
History Blame Contribute Delete
70.3 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.8333333333333334,
"eval_steps": 500,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"grad_norm": 1.684694528579712,
"learning_rate": 3e-06,
"loss": 1.1635,
"step": 10
},
{
"grad_norm": 0.908963680267334,
"learning_rate": 6.333333333333334e-06,
"loss": 1.1226,
"step": 20
},
{
"grad_norm": 0.6074573993682861,
"learning_rate": 9.666666666666667e-06,
"loss": 1.0434,
"step": 30
},
{
"grad_norm": 0.28302067518234253,
"learning_rate": 1.3000000000000001e-05,
"loss": 1.0359,
"step": 40
},
{
"grad_norm": 0.6111693382263184,
"learning_rate": 1.6333333333333335e-05,
"loss": 1.0124,
"step": 50
},
{
"grad_norm": 0.40753909945487976,
"learning_rate": 1.9666666666666666e-05,
"loss": 1.0219,
"step": 60
},
{
"grad_norm": 3.659215211868286,
"learning_rate": 2.3000000000000003e-05,
"loss": 0.9772,
"step": 70
},
{
"grad_norm": 3.0602831840515137,
"learning_rate": 2.633333333333333e-05,
"loss": 0.8846,
"step": 80
},
{
"grad_norm": 3.634186029434204,
"learning_rate": 2.9666666666666672e-05,
"loss": 0.7313,
"step": 90
},
{
"grad_norm": 4.549008846282959,
"learning_rate": 3.3e-05,
"loss": 0.544,
"step": 100
},
{
"grad_norm": 3.7457642555236816,
"learning_rate": 3.633333333333333e-05,
"loss": 0.3395,
"step": 110
},
{
"grad_norm": 2.88742733001709,
"learning_rate": 3.966666666666667e-05,
"loss": 0.2137,
"step": 120
},
{
"grad_norm": 2.8573951721191406,
"learning_rate": 4.3e-05,
"loss": 0.1806,
"step": 130
},
{
"grad_norm": 2.8793063163757324,
"learning_rate": 4.633333333333333e-05,
"loss": 0.1788,
"step": 140
},
{
"grad_norm": 1.796878457069397,
"learning_rate": 4.966666666666667e-05,
"loss": 0.1311,
"step": 150
},
{
"grad_norm": 2.40000581741333,
"learning_rate": 5.300000000000001e-05,
"loss": 0.145,
"step": 160
},
{
"grad_norm": 2.7650396823883057,
"learning_rate": 5.633333333333334e-05,
"loss": 0.1297,
"step": 170
},
{
"grad_norm": 2.4167962074279785,
"learning_rate": 5.966666666666667e-05,
"loss": 0.1228,
"step": 180
},
{
"grad_norm": 3.2442455291748047,
"learning_rate": 6.3e-05,
"loss": 0.1229,
"step": 190
},
{
"grad_norm": 1.9736961126327515,
"learning_rate": 6.633333333333334e-05,
"loss": 0.1273,
"step": 200
},
{
"grad_norm": 1.9273964166641235,
"learning_rate": 6.966666666666668e-05,
"loss": 0.1232,
"step": 210
},
{
"grad_norm": 1.9642269611358643,
"learning_rate": 7.3e-05,
"loss": 0.1192,
"step": 220
},
{
"grad_norm": 1.161840796470642,
"learning_rate": 7.633333333333334e-05,
"loss": 0.113,
"step": 230
},
{
"grad_norm": 1.7543864250183105,
"learning_rate": 7.966666666666666e-05,
"loss": 0.0993,
"step": 240
},
{
"grad_norm": 1.0030447244644165,
"learning_rate": 8.3e-05,
"loss": 0.0978,
"step": 250
},
{
"grad_norm": 1.941070795059204,
"learning_rate": 8.633333333333334e-05,
"loss": 0.0954,
"step": 260
},
{
"grad_norm": 1.6439497470855713,
"learning_rate": 8.966666666666666e-05,
"loss": 0.1109,
"step": 270
},
{
"grad_norm": 1.209729552268982,
"learning_rate": 9.300000000000001e-05,
"loss": 0.0908,
"step": 280
},
{
"grad_norm": 1.1890689134597778,
"learning_rate": 9.633333333333335e-05,
"loss": 0.1139,
"step": 290
},
{
"grad_norm": 1.4669991731643677,
"learning_rate": 9.966666666666667e-05,
"loss": 0.0916,
"step": 300
},
{
"grad_norm": 0.5628610849380493,
"learning_rate": 9.999938485971279e-05,
"loss": 0.1066,
"step": 310
},
{
"grad_norm": 0.9388463497161865,
"learning_rate": 9.999725846827562e-05,
"loss": 0.0843,
"step": 320
},
{
"grad_norm": 1.2850499153137207,
"learning_rate": 9.999361329594254e-05,
"loss": 0.0914,
"step": 330
},
{
"grad_norm": 0.8086246252059937,
"learning_rate": 9.998844945344405e-05,
"loss": 0.0871,
"step": 340
},
{
"grad_norm": 0.7838727235794067,
"learning_rate": 9.99817670976436e-05,
"loss": 0.0712,
"step": 350
},
{
"grad_norm": 0.8137271404266357,
"learning_rate": 9.997356643153303e-05,
"loss": 0.0825,
"step": 360
},
{
"grad_norm": 0.7650733590126038,
"learning_rate": 9.996384770422629e-05,
"loss": 0.0812,
"step": 370
},
{
"grad_norm": 1.2997642755508423,
"learning_rate": 9.995261121095194e-05,
"loss": 0.08,
"step": 380
},
{
"grad_norm": 0.6342231035232544,
"learning_rate": 9.993985729304408e-05,
"loss": 0.0773,
"step": 390
},
{
"grad_norm": 0.7299171686172485,
"learning_rate": 9.992558633793212e-05,
"loss": 0.0722,
"step": 400
},
{
"grad_norm": 0.9829517006874084,
"learning_rate": 9.990979877912891e-05,
"loss": 0.0847,
"step": 410
},
{
"grad_norm": 1.0101488828659058,
"learning_rate": 9.989249509621759e-05,
"loss": 0.0683,
"step": 420
},
{
"grad_norm": 0.7241670489311218,
"learning_rate": 9.987367581483705e-05,
"loss": 0.0711,
"step": 430
},
{
"grad_norm": 0.9610762596130371,
"learning_rate": 9.985334150666592e-05,
"loss": 0.0818,
"step": 440
},
{
"grad_norm": 0.8153676986694336,
"learning_rate": 9.983149278940526e-05,
"loss": 0.0798,
"step": 450
},
{
"grad_norm": 1.0012998580932617,
"learning_rate": 9.980813032675974e-05,
"loss": 0.0773,
"step": 460
},
{
"grad_norm": 0.7800810933113098,
"learning_rate": 9.978325482841753e-05,
"loss": 0.0733,
"step": 470
},
{
"grad_norm": 0.8783252835273743,
"learning_rate": 9.975686705002867e-05,
"loss": 0.0722,
"step": 480
},
{
"grad_norm": 1.1053400039672852,
"learning_rate": 9.972896779318219e-05,
"loss": 0.0691,
"step": 490
},
{
"grad_norm": 0.5190101265907288,
"learning_rate": 9.969955790538175e-05,
"loss": 0.0657,
"step": 500
},
{
"grad_norm": 0.6777092814445496,
"learning_rate": 9.966863828001982e-05,
"loss": 0.0699,
"step": 510
},
{
"grad_norm": 0.42357662320137024,
"learning_rate": 9.963620985635065e-05,
"loss": 0.063,
"step": 520
},
{
"grad_norm": 0.48872488737106323,
"learning_rate": 9.960227361946164e-05,
"loss": 0.062,
"step": 530
},
{
"grad_norm": 0.6022244691848755,
"learning_rate": 9.95668306002435e-05,
"loss": 0.0566,
"step": 540
},
{
"grad_norm": 0.5650090575218201,
"learning_rate": 9.952988187535886e-05,
"loss": 0.0547,
"step": 550
},
{
"grad_norm": 1.1828614473342896,
"learning_rate": 9.949142856720961e-05,
"loss": 0.0576,
"step": 560
},
{
"grad_norm": 0.4418780505657196,
"learning_rate": 9.945147184390278e-05,
"loss": 0.0641,
"step": 570
},
{
"grad_norm": 0.7184110879898071,
"learning_rate": 9.941001291921512e-05,
"loss": 0.0664,
"step": 580
},
{
"grad_norm": 0.5856490135192871,
"learning_rate": 9.936705305255612e-05,
"loss": 0.0673,
"step": 590
},
{
"grad_norm": 0.4480701982975006,
"learning_rate": 9.932259354892984e-05,
"loss": 0.0709,
"step": 600
},
{
"grad_norm": 0.3481273055076599,
"learning_rate": 9.927663575889521e-05,
"loss": 0.0672,
"step": 610
},
{
"grad_norm": 0.5294920206069946,
"learning_rate": 9.922918107852504e-05,
"loss": 0.0535,
"step": 620
},
{
"grad_norm": 0.4763725996017456,
"learning_rate": 9.918023094936363e-05,
"loss": 0.0467,
"step": 630
},
{
"grad_norm": 0.5558699369430542,
"learning_rate": 9.912978685838294e-05,
"loss": 0.0522,
"step": 640
},
{
"grad_norm": 0.4730995297431946,
"learning_rate": 9.90778503379374e-05,
"loss": 0.0537,
"step": 650
},
{
"grad_norm": 0.4881953299045563,
"learning_rate": 9.902442296571743e-05,
"loss": 0.0556,
"step": 660
},
{
"grad_norm": 0.613318145275116,
"learning_rate": 9.896950636470147e-05,
"loss": 0.0648,
"step": 670
},
{
"grad_norm": 0.4895165264606476,
"learning_rate": 9.891310220310666e-05,
"loss": 0.0571,
"step": 680
},
{
"grad_norm": 0.6173961162567139,
"learning_rate": 9.885521219433823e-05,
"loss": 0.0544,
"step": 690
},
{
"grad_norm": 0.6740694642066956,
"learning_rate": 9.879583809693738e-05,
"loss": 0.0578,
"step": 700
},
{
"grad_norm": 0.511962890625,
"learning_rate": 9.873498171452789e-05,
"loss": 0.0639,
"step": 710
},
{
"grad_norm": 0.9044255018234253,
"learning_rate": 9.867264489576135e-05,
"loss": 0.0562,
"step": 720
},
{
"grad_norm": 0.7572687864303589,
"learning_rate": 9.860882953426099e-05,
"loss": 0.0574,
"step": 730
},
{
"grad_norm": 0.4606987535953522,
"learning_rate": 9.854353756856412e-05,
"loss": 0.0443,
"step": 740
},
{
"grad_norm": 0.4337668716907501,
"learning_rate": 9.847677098206332e-05,
"loss": 0.0536,
"step": 750
},
{
"grad_norm": 0.36243191361427307,
"learning_rate": 9.840853180294608e-05,
"loss": 0.0419,
"step": 760
},
{
"grad_norm": 0.6487115621566772,
"learning_rate": 9.833882210413332e-05,
"loss": 0.0488,
"step": 770
},
{
"grad_norm": 0.5236570239067078,
"learning_rate": 9.826764400321633e-05,
"loss": 0.0733,
"step": 780
},
{
"grad_norm": 0.6646023988723755,
"learning_rate": 9.819499966239243e-05,
"loss": 0.0486,
"step": 790
},
{
"grad_norm": 0.5606136918067932,
"learning_rate": 9.812089128839938e-05,
"loss": 0.0548,
"step": 800
},
{
"grad_norm": 0.3728213608264923,
"learning_rate": 9.804532113244828e-05,
"loss": 0.0583,
"step": 810
},
{
"grad_norm": 0.5252546668052673,
"learning_rate": 9.796829149015517e-05,
"loss": 0.0529,
"step": 820
},
{
"grad_norm": 0.4646774232387543,
"learning_rate": 9.788980470147132e-05,
"loss": 0.052,
"step": 830
},
{
"grad_norm": 0.3394884169101715,
"learning_rate": 9.780986315061218e-05,
"loss": 0.0442,
"step": 840
},
{
"grad_norm": 0.5599905848503113,
"learning_rate": 9.772846926598491e-05,
"loss": 0.058,
"step": 850
},
{
"grad_norm": 0.46263357996940613,
"learning_rate": 9.76456255201146e-05,
"loss": 0.0475,
"step": 860
},
{
"grad_norm": 0.5478836297988892,
"learning_rate": 9.756133442956923e-05,
"loss": 0.0495,
"step": 870
},
{
"grad_norm": 0.6540849208831787,
"learning_rate": 9.747559855488313e-05,
"loss": 0.0486,
"step": 880
},
{
"grad_norm": 0.5630178451538086,
"learning_rate": 9.73884205004793e-05,
"loss": 0.0452,
"step": 890
},
{
"grad_norm": 0.5691519379615784,
"learning_rate": 9.729980291459019e-05,
"loss": 0.0538,
"step": 900
},
{
"grad_norm": 0.3020646870136261,
"learning_rate": 9.720974848917735e-05,
"loss": 0.0421,
"step": 910
},
{
"grad_norm": 0.524530291557312,
"learning_rate": 9.711825995984957e-05,
"loss": 0.0495,
"step": 920
},
{
"grad_norm": 0.7111333012580872,
"learning_rate": 9.702534010577991e-05,
"loss": 0.0636,
"step": 930
},
{
"grad_norm": 0.6500610113143921,
"learning_rate": 9.693099174962103e-05,
"loss": 0.045,
"step": 940
},
{
"grad_norm": 0.89228355884552,
"learning_rate": 9.683521775741977e-05,
"loss": 0.0549,
"step": 950
},
{
"grad_norm": 0.7165541648864746,
"learning_rate": 9.673802103852979e-05,
"loss": 0.0433,
"step": 960
},
{
"grad_norm": 0.5376599431037903,
"learning_rate": 9.663940454552342e-05,
"loss": 0.0451,
"step": 970
},
{
"grad_norm": 0.7044673562049866,
"learning_rate": 9.65393712741018e-05,
"loss": 0.0573,
"step": 980
},
{
"grad_norm": 0.40622320771217346,
"learning_rate": 9.6437924263004e-05,
"loss": 0.0553,
"step": 990
},
{
"grad_norm": 0.5690650939941406,
"learning_rate": 9.63350665939146e-05,
"loss": 0.0464,
"step": 1000
},
{
"grad_norm": 0.5798953175544739,
"learning_rate": 9.623080139137023e-05,
"loss": 0.041,
"step": 1010
},
{
"grad_norm": 0.477433443069458,
"learning_rate": 9.612513182266447e-05,
"loss": 0.0451,
"step": 1020
},
{
"grad_norm": 0.5034040212631226,
"learning_rate": 9.601806109775179e-05,
"loss": 0.0469,
"step": 1030
},
{
"grad_norm": 0.26716241240501404,
"learning_rate": 9.590959246914995e-05,
"loss": 0.0448,
"step": 1040
},
{
"grad_norm": 0.37224116921424866,
"learning_rate": 9.579972923184122e-05,
"loss": 0.0439,
"step": 1050
},
{
"grad_norm": 0.4705376923084259,
"learning_rate": 9.568847472317232e-05,
"loss": 0.0513,
"step": 1060
},
{
"grad_norm": 0.36860716342926025,
"learning_rate": 9.557583232275303e-05,
"loss": 0.0458,
"step": 1070
},
{
"grad_norm": 0.44712164998054504,
"learning_rate": 9.546180545235344e-05,
"loss": 0.0451,
"step": 1080
},
{
"grad_norm": 0.3662862479686737,
"learning_rate": 9.534639757580013e-05,
"loss": 0.0313,
"step": 1090
},
{
"grad_norm": 0.5598801970481873,
"learning_rate": 9.522961219887092e-05,
"loss": 0.0502,
"step": 1100
},
{
"grad_norm": 0.4979083240032196,
"learning_rate": 9.511145286918828e-05,
"loss": 0.0448,
"step": 1110
},
{
"grad_norm": 0.5371891856193542,
"learning_rate": 9.499192317611167e-05,
"loss": 0.0383,
"step": 1120
},
{
"grad_norm": 0.46349143981933594,
"learning_rate": 9.487102675062851e-05,
"loss": 0.0486,
"step": 1130
},
{
"grad_norm": 0.9978846311569214,
"learning_rate": 9.474876726524374e-05,
"loss": 0.0559,
"step": 1140
},
{
"grad_norm": 0.7520447969436646,
"learning_rate": 9.462514843386845e-05,
"loss": 0.0502,
"step": 1150
},
{
"grad_norm": 0.506645917892456,
"learning_rate": 9.450017401170689e-05,
"loss": 0.0522,
"step": 1160
},
{
"grad_norm": 0.45723941922187805,
"learning_rate": 9.437384779514256e-05,
"loss": 0.0438,
"step": 1170
},
{
"grad_norm": 0.44127294421195984,
"learning_rate": 9.424617362162271e-05,
"loss": 0.0431,
"step": 1180
},
{
"grad_norm": 0.45472055673599243,
"learning_rate": 9.411715536954196e-05,
"loss": 0.0418,
"step": 1190
},
{
"grad_norm": 0.29721713066101074,
"learning_rate": 9.39867969581243e-05,
"loss": 0.0457,
"step": 1200
},
{
"grad_norm": 0.3952992856502533,
"learning_rate": 9.385510234730415e-05,
"loss": 0.0417,
"step": 1210
},
{
"grad_norm": 0.25066861510276794,
"learning_rate": 9.372207553760603e-05,
"loss": 0.038,
"step": 1220
},
{
"grad_norm": 0.3423592746257782,
"learning_rate": 9.358772057002312e-05,
"loss": 0.0417,
"step": 1230
},
{
"grad_norm": 0.37416261434555054,
"learning_rate": 9.345204152589428e-05,
"loss": 0.041,
"step": 1240
},
{
"grad_norm": 0.47650808095932007,
"learning_rate": 9.331504252678037e-05,
"loss": 0.05,
"step": 1250
},
{
"grad_norm": 0.4732288122177124,
"learning_rate": 9.317672773433876e-05,
"loss": 0.0438,
"step": 1260
},
{
"grad_norm": 0.42209315299987793,
"learning_rate": 9.30371013501972e-05,
"loss": 0.0463,
"step": 1270
},
{
"grad_norm": 0.38447305560112,
"learning_rate": 9.289616761582587e-05,
"loss": 0.042,
"step": 1280
},
{
"grad_norm": 0.5152119994163513,
"learning_rate": 9.275393081240882e-05,
"loss": 0.049,
"step": 1290
},
{
"grad_norm": 0.5261250138282776,
"learning_rate": 9.261039526071374e-05,
"loss": 0.0443,
"step": 1300
},
{
"grad_norm": 0.5648900270462036,
"learning_rate": 9.246556532096078e-05,
"loss": 0.0425,
"step": 1310
},
{
"grad_norm": 0.31255903840065,
"learning_rate": 9.231944539269009e-05,
"loss": 0.0501,
"step": 1320
},
{
"grad_norm": 0.4260111153125763,
"learning_rate": 9.217203991462815e-05,
"loss": 0.0452,
"step": 1330
},
{
"grad_norm": 0.5075951218605042,
"learning_rate": 9.202335336455296e-05,
"loss": 0.0439,
"step": 1340
},
{
"grad_norm": 0.4853213429450989,
"learning_rate": 9.187339025915802e-05,
"loss": 0.0357,
"step": 1350
},
{
"grad_norm": 0.41578373312950134,
"learning_rate": 9.17221551539151e-05,
"loss": 0.0413,
"step": 1360
},
{
"grad_norm": 0.3264440894126892,
"learning_rate": 9.156965264293586e-05,
"loss": 0.03,
"step": 1370
},
{
"grad_norm": 0.4507807791233063,
"learning_rate": 9.141588735883232e-05,
"loss": 0.04,
"step": 1380
},
{
"grad_norm": 0.4165174961090088,
"learning_rate": 9.126086397257612e-05,
"loss": 0.0375,
"step": 1390
},
{
"grad_norm": 0.5147427320480347,
"learning_rate": 9.110458719335659e-05,
"loss": 0.045,
"step": 1400
},
{
"grad_norm": 0.3948756456375122,
"learning_rate": 9.094706176843777e-05,
"loss": 0.0352,
"step": 1410
},
{
"grad_norm": 0.5233174562454224,
"learning_rate": 9.078829248301417e-05,
"loss": 0.0445,
"step": 1420
},
{
"grad_norm": 0.3124268352985382,
"learning_rate": 9.062828416006539e-05,
"loss": 0.0414,
"step": 1430
},
{
"grad_norm": 0.48625996708869934,
"learning_rate": 9.046704166020961e-05,
"loss": 0.0413,
"step": 1440
},
{
"grad_norm": 0.5645583271980286,
"learning_rate": 9.030456988155596e-05,
"loss": 0.0466,
"step": 1450
},
{
"grad_norm": 0.410849004983902,
"learning_rate": 9.014087375955573e-05,
"loss": 0.0418,
"step": 1460
},
{
"grad_norm": 0.42126670479774475,
"learning_rate": 8.997595826685243e-05,
"loss": 0.0426,
"step": 1470
},
{
"grad_norm": 0.31017786264419556,
"learning_rate": 8.980982841313074e-05,
"loss": 0.0516,
"step": 1480
},
{
"grad_norm": 0.448800653219223,
"learning_rate": 8.964248924496435e-05,
"loss": 0.0377,
"step": 1490
},
{
"grad_norm": 0.5933466553688049,
"learning_rate": 8.947394584566258e-05,
"loss": 0.0566,
"step": 1500
},
{
"grad_norm": 0.5993877649307251,
"learning_rate": 8.930420333511606e-05,
"loss": 0.0442,
"step": 1510
},
{
"grad_norm": 0.5094465613365173,
"learning_rate": 8.913326686964117e-05,
"loss": 0.0387,
"step": 1520
},
{
"grad_norm": 0.3103977143764496,
"learning_rate": 8.89611416418234e-05,
"loss": 0.0425,
"step": 1530
},
{
"grad_norm": 0.3906828761100769,
"learning_rate": 8.878783288035957e-05,
"loss": 0.0464,
"step": 1540
},
{
"grad_norm": 0.3946266174316406,
"learning_rate": 8.86133458498991e-05,
"loss": 0.0394,
"step": 1550
},
{
"grad_norm": 0.3268502354621887,
"learning_rate": 8.843768585088393e-05,
"loss": 0.0403,
"step": 1560
},
{
"grad_norm": 0.4486427307128906,
"learning_rate": 8.82608582193877e-05,
"loss": 0.0364,
"step": 1570
},
{
"grad_norm": 0.4037582278251648,
"learning_rate": 8.80828683269535e-05,
"loss": 0.0363,
"step": 1580
},
{
"grad_norm": 0.3916078507900238,
"learning_rate": 8.790372158043074e-05,
"loss": 0.0404,
"step": 1590
},
{
"grad_norm": 0.5734459757804871,
"learning_rate": 8.772342342181095e-05,
"loss": 0.0399,
"step": 1600
},
{
"grad_norm": 0.4063715934753418,
"learning_rate": 8.75419793280624e-05,
"loss": 0.0353,
"step": 1610
},
{
"grad_norm": 0.3136364817619324,
"learning_rate": 8.735939481096378e-05,
"loss": 0.0341,
"step": 1620
},
{
"grad_norm": 0.2480340301990509,
"learning_rate": 8.717567541693673e-05,
"loss": 0.0445,
"step": 1630
},
{
"grad_norm": 0.49039986729621887,
"learning_rate": 8.699082672687734e-05,
"loss": 0.038,
"step": 1640
},
{
"grad_norm": 0.38957834243774414,
"learning_rate": 8.680485435598673e-05,
"loss": 0.0458,
"step": 1650
},
{
"grad_norm": 0.6209728717803955,
"learning_rate": 8.661776395360029e-05,
"loss": 0.0318,
"step": 1660
},
{
"grad_norm": 0.3944031596183777,
"learning_rate": 8.642956120301626e-05,
"loss": 0.0393,
"step": 1670
},
{
"grad_norm": 0.5090761780738831,
"learning_rate": 8.624025182132292e-05,
"loss": 0.0384,
"step": 1680
},
{
"grad_norm": 0.339475154876709,
"learning_rate": 8.604984155922506e-05,
"loss": 0.0421,
"step": 1690
},
{
"grad_norm": 0.5172349810600281,
"learning_rate": 8.585833620086918e-05,
"loss": 0.0438,
"step": 1700
},
{
"grad_norm": 0.2904316782951355,
"learning_rate": 8.566574156366784e-05,
"loss": 0.0399,
"step": 1710
},
{
"grad_norm": 0.3730908930301666,
"learning_rate": 8.547206349812298e-05,
"loss": 0.0446,
"step": 1720
},
{
"grad_norm": 0.4513612389564514,
"learning_rate": 8.527730788764805e-05,
"loss": 0.0404,
"step": 1730
},
{
"grad_norm": 0.30281123518943787,
"learning_rate": 8.508148064838948e-05,
"loss": 0.0405,
"step": 1740
},
{
"grad_norm": 0.5456830263137817,
"learning_rate": 8.488458772904684e-05,
"loss": 0.0393,
"step": 1750
},
{
"grad_norm": 0.6011849045753479,
"learning_rate": 8.468663511069217e-05,
"loss": 0.0443,
"step": 1760
},
{
"grad_norm": 0.43627694249153137,
"learning_rate": 8.448762880658825e-05,
"loss": 0.039,
"step": 1770
},
{
"grad_norm": 0.30541056394577026,
"learning_rate": 8.428757486200603e-05,
"loss": 0.0402,
"step": 1780
},
{
"grad_norm": 0.3088529109954834,
"learning_rate": 8.40864793540409e-05,
"loss": 0.0447,
"step": 1790
},
{
"grad_norm": 0.5158669352531433,
"learning_rate": 8.388434839142813e-05,
"loss": 0.0404,
"step": 1800
},
{
"grad_norm": 0.36301717162132263,
"learning_rate": 8.368118811435726e-05,
"loss": 0.0395,
"step": 1810
},
{
"grad_norm": 0.38054007291793823,
"learning_rate": 8.347700469428564e-05,
"loss": 0.0392,
"step": 1820
},
{
"grad_norm": 0.30281463265419006,
"learning_rate": 8.327180433375091e-05,
"loss": 0.0384,
"step": 1830
},
{
"grad_norm": 0.5268439054489136,
"learning_rate": 8.306559326618259e-05,
"loss": 0.0397,
"step": 1840
},
{
"grad_norm": 0.5128210186958313,
"learning_rate": 8.285837775571276e-05,
"loss": 0.0387,
"step": 1850
},
{
"grad_norm": 0.34021538496017456,
"learning_rate": 8.265016409698573e-05,
"loss": 0.0429,
"step": 1860
},
{
"grad_norm": 0.4679824113845825,
"learning_rate": 8.244095861496686e-05,
"loss": 0.0392,
"step": 1870
},
{
"grad_norm": 0.42745229601860046,
"learning_rate": 8.223076766475035e-05,
"loss": 0.0358,
"step": 1880
},
{
"grad_norm": 0.4563518464565277,
"learning_rate": 8.201959763136633e-05,
"loss": 0.0358,
"step": 1890
},
{
"grad_norm": 0.6088390946388245,
"learning_rate": 8.180745492958674e-05,
"loss": 0.0468,
"step": 1900
},
{
"grad_norm": 0.45247316360473633,
"learning_rate": 8.159434600373061e-05,
"loss": 0.041,
"step": 1910
},
{
"grad_norm": 0.477315753698349,
"learning_rate": 8.138027732746818e-05,
"loss": 0.0435,
"step": 1920
},
{
"grad_norm": 0.4624563157558441,
"learning_rate": 8.116525540362434e-05,
"loss": 0.0479,
"step": 1930
},
{
"grad_norm": 0.5414932370185852,
"learning_rate": 8.094928676398101e-05,
"loss": 0.0434,
"step": 1940
},
{
"grad_norm": 0.39883971214294434,
"learning_rate": 8.073237796907882e-05,
"loss": 0.0402,
"step": 1950
},
{
"grad_norm": 0.44839268922805786,
"learning_rate": 8.051453560801772e-05,
"loss": 0.035,
"step": 1960
},
{
"grad_norm": 0.6055172681808472,
"learning_rate": 8.029576629825687e-05,
"loss": 0.0459,
"step": 1970
},
{
"grad_norm": 0.6540765762329102,
"learning_rate": 8.007607668541362e-05,
"loss": 0.05,
"step": 1980
},
{
"grad_norm": 0.35224294662475586,
"learning_rate": 7.985547344306161e-05,
"loss": 0.0369,
"step": 1990
},
{
"grad_norm": 0.36379656195640564,
"learning_rate": 7.963396327252812e-05,
"loss": 0.0344,
"step": 2000
},
{
"grad_norm": 0.34664785861968994,
"learning_rate": 7.941155290269038e-05,
"loss": 0.0296,
"step": 2010
},
{
"grad_norm": 0.28230082988739014,
"learning_rate": 7.918824908977123e-05,
"loss": 0.04,
"step": 2020
},
{
"grad_norm": 0.5089810490608215,
"learning_rate": 7.896405861713394e-05,
"loss": 0.0403,
"step": 2030
},
{
"grad_norm": 0.3840276598930359,
"learning_rate": 7.873898829507606e-05,
"loss": 0.045,
"step": 2040
},
{
"grad_norm": 0.4788753390312195,
"learning_rate": 7.851304496062254e-05,
"loss": 0.04,
"step": 2050
},
{
"grad_norm": 0.4636504054069519,
"learning_rate": 7.828623547731818e-05,
"loss": 0.0345,
"step": 2060
},
{
"grad_norm": 0.38975924253463745,
"learning_rate": 7.80585667350189e-05,
"loss": 0.0327,
"step": 2070
},
{
"grad_norm": 0.5037243366241455,
"learning_rate": 7.783004564968263e-05,
"loss": 0.031,
"step": 2080
},
{
"grad_norm": 0.3715066909790039,
"learning_rate": 7.760067916315921e-05,
"loss": 0.034,
"step": 2090
},
{
"grad_norm": 0.34550508856773376,
"learning_rate": 7.737047424297941e-05,
"loss": 0.0315,
"step": 2100
},
{
"grad_norm": 0.30987003445625305,
"learning_rate": 7.713943788214337e-05,
"loss": 0.0348,
"step": 2110
},
{
"grad_norm": 0.4526568055152893,
"learning_rate": 7.690757709890812e-05,
"loss": 0.0327,
"step": 2120
},
{
"grad_norm": 0.31176066398620605,
"learning_rate": 7.66748989365744e-05,
"loss": 0.0354,
"step": 2130
},
{
"grad_norm": 0.3195038437843323,
"learning_rate": 7.644141046327271e-05,
"loss": 0.0377,
"step": 2140
},
{
"grad_norm": 0.3587583601474762,
"learning_rate": 7.620711877174866e-05,
"loss": 0.0445,
"step": 2150
},
{
"grad_norm": 0.44026920199394226,
"learning_rate": 7.597203097914732e-05,
"loss": 0.036,
"step": 2160
},
{
"grad_norm": 0.2840449810028076,
"learning_rate": 7.573615422679726e-05,
"loss": 0.032,
"step": 2170
},
{
"grad_norm": 0.5214433670043945,
"learning_rate": 7.549949567999345e-05,
"loss": 0.0396,
"step": 2180
},
{
"grad_norm": 0.37318941950798035,
"learning_rate": 7.526206252777968e-05,
"loss": 0.031,
"step": 2190
},
{
"grad_norm": 0.311050683259964,
"learning_rate": 7.50238619827301e-05,
"loss": 0.0339,
"step": 2200
},
{
"grad_norm": 0.4276743233203888,
"learning_rate": 7.478490128073022e-05,
"loss": 0.032,
"step": 2210
},
{
"grad_norm": 0.30517420172691345,
"learning_rate": 7.454518768075704e-05,
"loss": 0.031,
"step": 2220
},
{
"grad_norm": 0.5467589497566223,
"learning_rate": 7.430472846465856e-05,
"loss": 0.0403,
"step": 2230
},
{
"grad_norm": 0.31511110067367554,
"learning_rate": 7.406353093693253e-05,
"loss": 0.0322,
"step": 2240
},
{
"grad_norm": 0.47626039385795593,
"learning_rate": 7.382160242450469e-05,
"loss": 0.0394,
"step": 2250
},
{
"grad_norm": 0.4516342282295227,
"learning_rate": 7.357895027650598e-05,
"loss": 0.0361,
"step": 2260
},
{
"grad_norm": 0.3844388425350189,
"learning_rate": 7.333558186404958e-05,
"loss": 0.0318,
"step": 2270
},
{
"grad_norm": 0.3211086094379425,
"learning_rate": 7.309150458000668e-05,
"loss": 0.0387,
"step": 2280
},
{
"grad_norm": 0.21057334542274475,
"learning_rate": 7.284672583878219e-05,
"loss": 0.033,
"step": 2290
},
{
"grad_norm": 0.33732137084007263,
"learning_rate": 7.260125307608929e-05,
"loss": 0.0302,
"step": 2300
},
{
"grad_norm": 0.34155720472335815,
"learning_rate": 7.235509374872373e-05,
"loss": 0.0416,
"step": 2310
},
{
"grad_norm": 0.23326361179351807,
"learning_rate": 7.210825533433719e-05,
"loss": 0.0272,
"step": 2320
},
{
"grad_norm": 0.2873460352420807,
"learning_rate": 7.186074533121013e-05,
"loss": 0.0442,
"step": 2330
},
{
"grad_norm": 0.5878977179527283,
"learning_rate": 7.161257125802413e-05,
"loss": 0.0367,
"step": 2340
},
{
"grad_norm": 0.8126834034919739,
"learning_rate": 7.136374065363334e-05,
"loss": 0.042,
"step": 2350
},
{
"grad_norm": 0.6949858069419861,
"learning_rate": 7.11142610768356e-05,
"loss": 0.0391,
"step": 2360
},
{
"grad_norm": 0.3497920334339142,
"learning_rate": 7.086414010614276e-05,
"loss": 0.0394,
"step": 2370
},
{
"grad_norm": 0.4741900563240051,
"learning_rate": 7.061338533955043e-05,
"loss": 0.0353,
"step": 2380
},
{
"grad_norm": 0.29987454414367676,
"learning_rate": 7.036200439430725e-05,
"loss": 0.0375,
"step": 2390
},
{
"grad_norm": 0.4544006586074829,
"learning_rate": 7.01100049066835e-05,
"loss": 0.0396,
"step": 2400
},
{
"grad_norm": 0.43514639139175415,
"learning_rate": 6.985739453173903e-05,
"loss": 0.0341,
"step": 2410
},
{
"grad_norm": 0.37087947130203247,
"learning_rate": 6.960418094309085e-05,
"loss": 0.035,
"step": 2420
},
{
"grad_norm": 0.5058620572090149,
"learning_rate": 6.93503718326799e-05,
"loss": 0.0347,
"step": 2430
},
{
"grad_norm": 0.3606584966182709,
"learning_rate": 6.909597491053751e-05,
"loss": 0.039,
"step": 2440
},
{
"grad_norm": 0.25050094723701477,
"learning_rate": 6.884099790455113e-05,
"loss": 0.0323,
"step": 2450
},
{
"grad_norm": 0.2565172612667084,
"learning_rate": 6.858544856022952e-05,
"loss": 0.0347,
"step": 2460
},
{
"grad_norm": 0.3210285007953644,
"learning_rate": 6.83293346404676e-05,
"loss": 0.031,
"step": 2470
},
{
"grad_norm": 0.3485363721847534,
"learning_rate": 6.80726639253105e-05,
"loss": 0.029,
"step": 2480
},
{
"grad_norm": 0.19532524049282074,
"learning_rate": 6.781544421171732e-05,
"loss": 0.0273,
"step": 2490
},
{
"grad_norm": 0.2607985734939575,
"learning_rate": 6.755768331332424e-05,
"loss": 0.0371,
"step": 2500
},
{
"grad_norm": 0.3058387041091919,
"learning_rate": 6.729938906020713e-05,
"loss": 0.0356,
"step": 2510
},
{
"grad_norm": 0.3409954905509949,
"learning_rate": 6.704056929864376e-05,
"loss": 0.037,
"step": 2520
},
{
"grad_norm": 0.4085492193698883,
"learning_rate": 6.67812318908754e-05,
"loss": 0.0292,
"step": 2530
},
{
"grad_norm": 0.5568246245384216,
"learning_rate": 6.6521384714868e-05,
"loss": 0.0444,
"step": 2540
},
{
"grad_norm": 0.35344791412353516,
"learning_rate": 6.626103566407295e-05,
"loss": 0.0389,
"step": 2550
},
{
"grad_norm": 0.29405418038368225,
"learning_rate": 6.600019264718713e-05,
"loss": 0.0351,
"step": 2560
},
{
"grad_norm": 0.45030686259269714,
"learning_rate": 6.573886358791285e-05,
"loss": 0.0305,
"step": 2570
},
{
"grad_norm": 0.3539298474788666,
"learning_rate": 6.547705642471703e-05,
"loss": 0.0356,
"step": 2580
},
{
"grad_norm": 0.22217731177806854,
"learning_rate": 6.521477911059008e-05,
"loss": 0.0356,
"step": 2590
},
{
"grad_norm": 0.30436158180236816,
"learning_rate": 6.495203961280434e-05,
"loss": 0.0338,
"step": 2600
},
{
"grad_norm": 0.42589154839515686,
"learning_rate": 6.468884591267204e-05,
"loss": 0.0302,
"step": 2610
},
{
"grad_norm": 0.23870477080345154,
"learning_rate": 6.44252060053028e-05,
"loss": 0.0343,
"step": 2620
},
{
"grad_norm": 0.5205913186073303,
"learning_rate": 6.416112789936086e-05,
"loss": 0.0332,
"step": 2630
},
{
"grad_norm": 0.3604913353919983,
"learning_rate": 6.389661961682173e-05,
"loss": 0.0301,
"step": 2640
},
{
"grad_norm": 0.37570565938949585,
"learning_rate": 6.363168919272846e-05,
"loss": 0.0333,
"step": 2650
},
{
"grad_norm": 0.31830376386642456,
"learning_rate": 6.336634467494768e-05,
"loss": 0.0436,
"step": 2660
},
{
"grad_norm": 0.3479643762111664,
"learning_rate": 6.310059412392505e-05,
"loss": 0.0362,
"step": 2670
},
{
"grad_norm": 0.4024968445301056,
"learning_rate": 6.283444561244042e-05,
"loss": 0.0366,
"step": 2680
},
{
"grad_norm": 0.3066016137599945,
"learning_rate": 6.256790722536251e-05,
"loss": 0.033,
"step": 2690
},
{
"grad_norm": 0.3896540403366089,
"learning_rate": 6.230098705940354e-05,
"loss": 0.0355,
"step": 2700
},
{
"grad_norm": 0.2699594795703888,
"learning_rate": 6.203369322287306e-05,
"loss": 0.0322,
"step": 2710
},
{
"grad_norm": 0.30261123180389404,
"learning_rate": 6.17660338354317e-05,
"loss": 0.0343,
"step": 2720
},
{
"grad_norm": 0.3030012249946594,
"learning_rate": 6.149801702784456e-05,
"loss": 0.0351,
"step": 2730
},
{
"grad_norm": 0.23485051095485687,
"learning_rate": 6.122965094173424e-05,
"loss": 0.0318,
"step": 2740
},
{
"grad_norm": 0.24681192636489868,
"learning_rate": 6.0960943729333374e-05,
"loss": 0.0276,
"step": 2750
},
{
"grad_norm": 0.32509541511535645,
"learning_rate": 6.069190355323717e-05,
"loss": 0.0334,
"step": 2760
},
{
"grad_norm": 0.31495198607444763,
"learning_rate": 6.042253858615532e-05,
"loss": 0.0357,
"step": 2770
},
{
"grad_norm": 0.3132841885089874,
"learning_rate": 6.015285701066382e-05,
"loss": 0.0376,
"step": 2780
},
{
"grad_norm": 0.32683679461479187,
"learning_rate": 5.988286701895631e-05,
"loss": 0.0298,
"step": 2790
},
{
"grad_norm": 0.26779282093048096,
"learning_rate": 5.961257681259535e-05,
"loss": 0.0383,
"step": 2800
},
{
"grad_norm": 0.3105967938899994,
"learning_rate": 5.934199460226317e-05,
"loss": 0.0294,
"step": 2810
},
{
"grad_norm": 0.22489722073078156,
"learning_rate": 5.9071128607512285e-05,
"loss": 0.0272,
"step": 2820
},
{
"grad_norm": 0.20428676903247833,
"learning_rate": 5.8799987056515804e-05,
"loss": 0.0311,
"step": 2830
},
{
"grad_norm": 0.24433785676956177,
"learning_rate": 5.8528578185817514e-05,
"loss": 0.0318,
"step": 2840
},
{
"grad_norm": 0.4590221047401428,
"learning_rate": 5.825691024008162e-05,
"loss": 0.0303,
"step": 2850
},
{
"grad_norm": 0.26181215047836304,
"learning_rate": 5.798499147184233e-05,
"loss": 0.0296,
"step": 2860
},
{
"grad_norm": 0.2606470584869385,
"learning_rate": 5.771283014125317e-05,
"loss": 0.0279,
"step": 2870
},
{
"grad_norm": 0.4306292235851288,
"learning_rate": 5.7440434515836064e-05,
"loss": 0.0284,
"step": 2880
},
{
"grad_norm": 0.2626531422138214,
"learning_rate": 5.7167812870230094e-05,
"loss": 0.0356,
"step": 2890
},
{
"grad_norm": 0.31425535678863525,
"learning_rate": 5.689497348594035e-05,
"loss": 0.0301,
"step": 2900
},
{
"grad_norm": 0.20697806775569916,
"learning_rate": 5.662192465108613e-05,
"loss": 0.0435,
"step": 2910
},
{
"grad_norm": 0.21303074061870575,
"learning_rate": 5.634867466014932e-05,
"loss": 0.0297,
"step": 2920
},
{
"grad_norm": 0.4053211212158203,
"learning_rate": 5.607523181372234e-05,
"loss": 0.0284,
"step": 2930
},
{
"grad_norm": 0.2496165931224823,
"learning_rate": 5.5801604418256117e-05,
"loss": 0.0281,
"step": 2940
},
{
"grad_norm": 0.23278173804283142,
"learning_rate": 5.552780078580756e-05,
"loss": 0.0276,
"step": 2950
},
{
"grad_norm": 0.20613159239292145,
"learning_rate": 5.525382923378728e-05,
"loss": 0.0326,
"step": 2960
},
{
"grad_norm": 0.21831080317497253,
"learning_rate": 5.49796980847068e-05,
"loss": 0.0319,
"step": 2970
},
{
"grad_norm": 0.3597847521305084,
"learning_rate": 5.470541566592573e-05,
"loss": 0.0276,
"step": 2980
},
{
"grad_norm": 0.3352603614330292,
"learning_rate": 5.443099030939887e-05,
"loss": 0.0288,
"step": 2990
},
{
"grad_norm": 0.26657530665397644,
"learning_rate": 5.415643035142309e-05,
"loss": 0.0359,
"step": 3000
},
{
"grad_norm": 0.2680664360523224,
"learning_rate": 5.3881744132384104e-05,
"loss": 0.0328,
"step": 3010
},
{
"grad_norm": 0.2673185169696808,
"learning_rate": 5.360693999650303e-05,
"loss": 0.0388,
"step": 3020
},
{
"grad_norm": 0.2571691572666168,
"learning_rate": 5.3332026291583016e-05,
"loss": 0.0298,
"step": 3030
},
{
"grad_norm": 0.22435830533504486,
"learning_rate": 5.305701136875566e-05,
"loss": 0.0293,
"step": 3040
},
{
"grad_norm": 0.3176920413970947,
"learning_rate": 5.278190358222721e-05,
"loss": 0.0369,
"step": 3050
},
{
"grad_norm": 0.3078315556049347,
"learning_rate": 5.25067112890249e-05,
"loss": 0.0364,
"step": 3060
},
{
"grad_norm": 0.3844327926635742,
"learning_rate": 5.2231442848743064e-05,
"loss": 0.0388,
"step": 3070
},
{
"grad_norm": 0.3389998972415924,
"learning_rate": 5.1956106623289145e-05,
"loss": 0.0353,
"step": 3080
},
{
"grad_norm": 0.5690343379974365,
"learning_rate": 5.168071097662972e-05,
"loss": 0.0341,
"step": 3090
},
{
"grad_norm": 0.2359319031238556,
"learning_rate": 5.1405264274536445e-05,
"loss": 0.0315,
"step": 3100
},
{
"grad_norm": 0.2650228440761566,
"learning_rate": 5.112977488433188e-05,
"loss": 0.036,
"step": 3110
},
{
"grad_norm": 0.4203425347805023,
"learning_rate": 5.085425117463533e-05,
"loss": 0.0331,
"step": 3120
},
{
"grad_norm": 0.28224822878837585,
"learning_rate": 5.057870151510864e-05,
"loss": 0.0339,
"step": 3130
},
{
"grad_norm": 0.374846875667572,
"learning_rate": 5.030313427620197e-05,
"loss": 0.0323,
"step": 3140
},
{
"grad_norm": 0.31122323870658875,
"learning_rate": 5.0027557828899426e-05,
"loss": 0.0311,
"step": 3150
},
{
"grad_norm": 0.3122676610946655,
"learning_rate": 4.975198054446492e-05,
"loss": 0.0287,
"step": 3160
},
{
"grad_norm": 0.28067177534103394,
"learning_rate": 4.947641079418773e-05,
"loss": 0.0316,
"step": 3170
},
{
"grad_norm": 0.22688500583171844,
"learning_rate": 4.920085694912828e-05,
"loss": 0.0307,
"step": 3180
},
{
"grad_norm": 0.21218730509281158,
"learning_rate": 4.892532737986387e-05,
"loss": 0.03,
"step": 3190
},
{
"grad_norm": 0.26027190685272217,
"learning_rate": 4.864983045623434e-05,
"loss": 0.0351,
"step": 3200
},
{
"grad_norm": 0.2783639430999756,
"learning_rate": 4.837437454708784e-05,
"loss": 0.029,
"step": 3210
},
{
"grad_norm": 0.35585737228393555,
"learning_rate": 4.809896802002662e-05,
"loss": 0.0269,
"step": 3220
},
{
"grad_norm": 0.251881867647171,
"learning_rate": 4.7823619241152854e-05,
"loss": 0.0292,
"step": 3230
},
{
"grad_norm": 0.23976771533489227,
"learning_rate": 4.754833657481445e-05,
"loss": 0.0267,
"step": 3240
},
{
"grad_norm": 0.2299049347639084,
"learning_rate": 4.7273128383351015e-05,
"loss": 0.0225,
"step": 3250
},
{
"grad_norm": 0.27877920866012573,
"learning_rate": 4.699800302683981e-05,
"loss": 0.0363,
"step": 3260
},
{
"grad_norm": 0.31591010093688965,
"learning_rate": 4.6722968862841806e-05,
"loss": 0.0312,
"step": 3270
},
{
"grad_norm": 0.34376806020736694,
"learning_rate": 4.6448034246147754e-05,
"loss": 0.0282,
"step": 3280
},
{
"grad_norm": 0.28944969177246094,
"learning_rate": 4.6173207528524476e-05,
"loss": 0.0303,
"step": 3290
},
{
"grad_norm": 0.24412792921066284,
"learning_rate": 4.58984970584611e-05,
"loss": 0.0248,
"step": 3300
},
{
"grad_norm": 0.21915721893310547,
"learning_rate": 4.562391118091544e-05,
"loss": 0.0256,
"step": 3310
},
{
"grad_norm": 0.28424176573753357,
"learning_rate": 4.534945823706056e-05,
"loss": 0.0256,
"step": 3320
},
{
"grad_norm": 0.2543053925037384,
"learning_rate": 4.507514656403137e-05,
"loss": 0.0338,
"step": 3330
},
{
"grad_norm": 0.26843786239624023,
"learning_rate": 4.480098449467132e-05,
"loss": 0.0286,
"step": 3340
},
{
"grad_norm": 0.32712703943252563,
"learning_rate": 4.452698035727929e-05,
"loss": 0.0377,
"step": 3350
},
{
"grad_norm": 0.3210562467575073,
"learning_rate": 4.425314247535668e-05,
"loss": 0.027,
"step": 3360
},
{
"grad_norm": 0.24497050046920776,
"learning_rate": 4.3979479167354477e-05,
"loss": 0.0363,
"step": 3370
},
{
"grad_norm": 0.27035537362098694,
"learning_rate": 4.370599874642055e-05,
"loss": 0.0334,
"step": 3380
},
{
"grad_norm": 0.46714988350868225,
"learning_rate": 4.3432709520147205e-05,
"loss": 0.0293,
"step": 3390
},
{
"grad_norm": 0.3511967062950134,
"learning_rate": 4.315961979031875e-05,
"loss": 0.0337,
"step": 3400
},
{
"grad_norm": 0.2085649073123932,
"learning_rate": 4.2886737852659325e-05,
"loss": 0.03,
"step": 3410
},
{
"grad_norm": 0.29439303278923035,
"learning_rate": 4.261407199658093e-05,
"loss": 0.0337,
"step": 3420
},
{
"grad_norm": 0.3793385624885559,
"learning_rate": 4.234163050493158e-05,
"loss": 0.0276,
"step": 3430
},
{
"grad_norm": 0.35857319831848145,
"learning_rate": 4.2069421653743706e-05,
"loss": 0.029,
"step": 3440
},
{
"grad_norm": 0.18004851043224335,
"learning_rate": 4.179745371198276e-05,
"loss": 0.0293,
"step": 3450
},
{
"grad_norm": 0.350254625082016,
"learning_rate": 4.1525734941296026e-05,
"loss": 0.0302,
"step": 3460
},
{
"grad_norm": 0.2741129696369171,
"learning_rate": 4.125427359576162e-05,
"loss": 0.0278,
"step": 3470
},
{
"grad_norm": 0.16368260979652405,
"learning_rate": 4.0983077921637815e-05,
"loss": 0.0321,
"step": 3480
},
{
"grad_norm": 0.3167386054992676,
"learning_rate": 4.07121561571125e-05,
"loss": 0.0295,
"step": 3490
},
{
"grad_norm": 0.4044698178768158,
"learning_rate": 4.044151653205292e-05,
"loss": 0.0425,
"step": 3500
},
{
"grad_norm": 0.21528899669647217,
"learning_rate": 4.0171167267755696e-05,
"loss": 0.035,
"step": 3510
},
{
"grad_norm": 0.2003302276134491,
"learning_rate": 3.9901116576697083e-05,
"loss": 0.0269,
"step": 3520
},
{
"grad_norm": 0.23838789761066437,
"learning_rate": 3.963137266228349e-05,
"loss": 0.0276,
"step": 3530
},
{
"grad_norm": 0.24220554530620575,
"learning_rate": 3.93619437186023e-05,
"loss": 0.0306,
"step": 3540
},
{
"grad_norm": 0.28130173683166504,
"learning_rate": 3.9092837930172884e-05,
"loss": 0.0323,
"step": 3550
},
{
"grad_norm": 0.4787818491458893,
"learning_rate": 3.8824063471698105e-05,
"loss": 0.0322,
"step": 3560
},
{
"grad_norm": 0.4143725335597992,
"learning_rate": 3.855562850781589e-05,
"loss": 0.0358,
"step": 3570
},
{
"grad_norm": 0.16623908281326294,
"learning_rate": 3.828754119285123e-05,
"loss": 0.0319,
"step": 3580
},
{
"grad_norm": 0.20105788111686707,
"learning_rate": 3.801980967056851e-05,
"loss": 0.03,
"step": 3590
},
{
"grad_norm": 0.28099361062049866,
"learning_rate": 3.77524420739241e-05,
"loss": 0.0359,
"step": 3600
},
{
"grad_norm": 0.2060209959745407,
"learning_rate": 3.748544652481927e-05,
"loss": 0.0257,
"step": 3610
},
{
"grad_norm": 0.38736292719841003,
"learning_rate": 3.721883113385353e-05,
"loss": 0.0293,
"step": 3620
},
{
"grad_norm": 0.3965328633785248,
"learning_rate": 3.695260400007819e-05,
"loss": 0.0327,
"step": 3630
},
{
"grad_norm": 0.30046403408050537,
"learning_rate": 3.6686773210750385e-05,
"loss": 0.0279,
"step": 3640
},
{
"grad_norm": 0.20698264241218567,
"learning_rate": 3.642134684108737e-05,
"loss": 0.03,
"step": 3650
},
{
"grad_norm": 0.43734461069107056,
"learning_rate": 3.615633295402123e-05,
"loss": 0.0267,
"step": 3660
},
{
"grad_norm": 0.23093822598457336,
"learning_rate": 3.5891739599953945e-05,
"loss": 0.0292,
"step": 3670
},
{
"grad_norm": 0.28527817130088806,
"learning_rate": 3.5627574816512846e-05,
"loss": 0.036,
"step": 3680
},
{
"grad_norm": 0.2417895495891571,
"learning_rate": 3.536384662830648e-05,
"loss": 0.0321,
"step": 3690
},
{
"grad_norm": 0.22468143701553345,
"learning_rate": 3.5100563046680764e-05,
"loss": 0.0272,
"step": 3700
},
{
"grad_norm": 0.2053907960653305,
"learning_rate": 3.483773206947572e-05,
"loss": 0.0344,
"step": 3710
},
{
"grad_norm": 0.2612084150314331,
"learning_rate": 3.457536168078247e-05,
"loss": 0.0276,
"step": 3720
},
{
"grad_norm": 0.29906558990478516,
"learning_rate": 3.431345985070067e-05,
"loss": 0.0243,
"step": 3730
},
{
"grad_norm": 0.3167043626308441,
"learning_rate": 3.40520345350965e-05,
"loss": 0.0321,
"step": 3740
},
{
"grad_norm": 0.20929738879203796,
"learning_rate": 3.379109367536089e-05,
"loss": 0.0306,
"step": 3750
},
{
"grad_norm": 0.20394307374954224,
"learning_rate": 3.3530645198168295e-05,
"loss": 0.0277,
"step": 3760
},
{
"grad_norm": 0.30055829882621765,
"learning_rate": 3.327069701523595e-05,
"loss": 0.0351,
"step": 3770
},
{
"grad_norm": 0.17945753037929535,
"learning_rate": 3.301125702308353e-05,
"loss": 0.0284,
"step": 3780
},
{
"grad_norm": 0.3272077441215515,
"learning_rate": 3.275233310279321e-05,
"loss": 0.0259,
"step": 3790
},
{
"grad_norm": 0.22872713208198547,
"learning_rate": 3.249393311977037e-05,
"loss": 0.0301,
"step": 3800
},
{
"grad_norm": 0.2559140920639038,
"learning_rate": 3.223606492350451e-05,
"loss": 0.0259,
"step": 3810
},
{
"grad_norm": 0.24041908979415894,
"learning_rate": 3.197873634733096e-05,
"loss": 0.0328,
"step": 3820
},
{
"grad_norm": 0.22720403969287872,
"learning_rate": 3.172195520819285e-05,
"loss": 0.0358,
"step": 3830
},
{
"grad_norm": 0.1657092422246933,
"learning_rate": 3.146572930640362e-05,
"loss": 0.0268,
"step": 3840
},
{
"grad_norm": 0.316561758518219,
"learning_rate": 3.121006642541014e-05,
"loss": 0.0279,
"step": 3850
},
{
"grad_norm": 0.24142567813396454,
"learning_rate": 3.095497433155626e-05,
"loss": 0.0285,
"step": 3860
},
{
"grad_norm": 0.3372308015823364,
"learning_rate": 3.070046077384682e-05,
"loss": 0.032,
"step": 3870
},
{
"grad_norm": 0.24744057655334473,
"learning_rate": 3.0446533483712304e-05,
"loss": 0.0297,
"step": 3880
},
{
"grad_norm": 0.2249644249677658,
"learning_rate": 3.0193200174774038e-05,
"loss": 0.0261,
"step": 3890
},
{
"grad_norm": 0.2031847983598709,
"learning_rate": 2.994046854260974e-05,
"loss": 0.0321,
"step": 3900
},
{
"grad_norm": 0.1849215030670166,
"learning_rate": 2.9688346264519866e-05,
"loss": 0.026,
"step": 3910
},
{
"grad_norm": 0.28336453437805176,
"learning_rate": 2.943684099929436e-05,
"loss": 0.0301,
"step": 3920
},
{
"grad_norm": 0.21173270046710968,
"learning_rate": 2.918596038697995e-05,
"loss": 0.0283,
"step": 3930
},
{
"grad_norm": 0.2881726324558258,
"learning_rate": 2.8935712048648112e-05,
"loss": 0.0314,
"step": 3940
},
{
"grad_norm": 0.28386256098747253,
"learning_rate": 2.8686103586163626e-05,
"loss": 0.0358,
"step": 3950
},
{
"grad_norm": 0.24731168150901794,
"learning_rate": 2.843714258195346e-05,
"loss": 0.0317,
"step": 3960
},
{
"grad_norm": 0.47752872109413147,
"learning_rate": 2.8188836598776662e-05,
"loss": 0.0339,
"step": 3970
},
{
"grad_norm": 0.21566864848136902,
"learning_rate": 2.7941193179494484e-05,
"loss": 0.0279,
"step": 3980
},
{
"grad_norm": 0.20382902026176453,
"learning_rate": 2.7694219846841262e-05,
"loss": 0.0273,
"step": 3990
},
{
"grad_norm": 0.15836754441261292,
"learning_rate": 2.7447924103195976e-05,
"loss": 0.0331,
"step": 4000
},
{
"grad_norm": 0.24428705871105194,
"learning_rate": 2.7202313430354253e-05,
"loss": 0.0268,
"step": 4010
},
{
"grad_norm": 0.2831069529056549,
"learning_rate": 2.695739528930111e-05,
"loss": 0.0259,
"step": 4020
},
{
"grad_norm": 0.26757097244262695,
"learning_rate": 2.67131771199844e-05,
"loss": 0.0282,
"step": 4030
},
{
"grad_norm": 0.20798030495643616,
"learning_rate": 2.6469666341088677e-05,
"loss": 0.0255,
"step": 4040
},
{
"grad_norm": 0.20833858847618103,
"learning_rate": 2.6226870349809885e-05,
"loss": 0.0254,
"step": 4050
},
{
"grad_norm": 0.23676158487796783,
"learning_rate": 2.5984796521630737e-05,
"loss": 0.0339,
"step": 4060
},
{
"grad_norm": 0.2095239758491516,
"learning_rate": 2.574345221009653e-05,
"loss": 0.0275,
"step": 4070
},
{
"grad_norm": 0.27846553921699524,
"learning_rate": 2.5502844746591804e-05,
"loss": 0.0264,
"step": 4080
},
{
"grad_norm": 0.2405710518360138,
"learning_rate": 2.526298144011775e-05,
"loss": 0.0276,
"step": 4090
},
{
"grad_norm": 0.2077587991952896,
"learning_rate": 2.5023869577070013e-05,
"loss": 0.0312,
"step": 4100
},
{
"grad_norm": 0.2612724006175995,
"learning_rate": 2.478551642101743e-05,
"loss": 0.0301,
"step": 4110
},
{
"grad_norm": 0.32656130194664,
"learning_rate": 2.4547929212481435e-05,
"loss": 0.0276,
"step": 4120
},
{
"grad_norm": 0.25739291310310364,
"learning_rate": 2.4311115168716013e-05,
"loss": 0.0274,
"step": 4130
},
{
"grad_norm": 0.2871209979057312,
"learning_rate": 2.4075081483488494e-05,
"loss": 0.0334,
"step": 4140
},
{
"grad_norm": 0.2844501733779907,
"learning_rate": 2.3839835326861104e-05,
"loss": 0.0264,
"step": 4150
},
{
"grad_norm": 0.5137473940849304,
"learning_rate": 2.3605383844972966e-05,
"loss": 0.0235,
"step": 4160
},
{
"grad_norm": 0.26710981130599976,
"learning_rate": 2.3371734159823284e-05,
"loss": 0.0314,
"step": 4170
},
{
"grad_norm": 0.23191222548484802,
"learning_rate": 2.3138893369054766e-05,
"loss": 0.0292,
"step": 4180
},
{
"grad_norm": 0.25547853112220764,
"learning_rate": 2.2906868545738102e-05,
"loss": 0.0307,
"step": 4190
},
{
"grad_norm": 0.23825709521770477,
"learning_rate": 2.2675666738157186e-05,
"loss": 0.0269,
"step": 4200
},
{
"grad_norm": 0.2642747461795807,
"learning_rate": 2.2445294969594844e-05,
"loss": 0.022,
"step": 4210
},
{
"grad_norm": 0.4196590781211853,
"learning_rate": 2.22157602381196e-05,
"loss": 0.0257,
"step": 4220
},
{
"grad_norm": 0.11946013569831848,
"learning_rate": 2.1987069516373098e-05,
"loss": 0.0234,
"step": 4230
},
{
"grad_norm": 0.23782043159008026,
"learning_rate": 2.1759229751358217e-05,
"loss": 0.0304,
"step": 4240
},
{
"grad_norm": 0.2545646131038666,
"learning_rate": 2.1532247864228084e-05,
"loss": 0.0295,
"step": 4250
},
{
"grad_norm": 0.3075411915779114,
"learning_rate": 2.1306130750075865e-05,
"loss": 0.0395,
"step": 4260
},
{
"grad_norm": 0.178472101688385,
"learning_rate": 2.1080885277725236e-05,
"loss": 0.0262,
"step": 4270
},
{
"grad_norm": 0.23191122710704803,
"learning_rate": 2.085651828952175e-05,
"loss": 0.0239,
"step": 4280
},
{
"grad_norm": 0.23175159096717834,
"learning_rate": 2.063303660112506e-05,
"loss": 0.0253,
"step": 4290
},
{
"grad_norm": 0.24712395668029785,
"learning_rate": 2.0410447001301753e-05,
"loss": 0.0264,
"step": 4300
},
{
"grad_norm": 0.28210514783859253,
"learning_rate": 2.0188756251719203e-05,
"loss": 0.0238,
"step": 4310
},
{
"grad_norm": 0.1306670755147934,
"learning_rate": 1.9967971086740195e-05,
"loss": 0.026,
"step": 4320
},
{
"grad_norm": 0.1528860181570053,
"learning_rate": 1.974809821321827e-05,
"loss": 0.0235,
"step": 4330
},
{
"grad_norm": 0.23973195254802704,
"learning_rate": 1.9529144310294023e-05,
"loss": 0.0254,
"step": 4340
},
{
"grad_norm": 0.20007772743701935,
"learning_rate": 1.9311116029192278e-05,
"loss": 0.0274,
"step": 4350
},
{
"grad_norm": 0.4211285710334778,
"learning_rate": 1.909401999301993e-05,
"loss": 0.0228,
"step": 4360
},
{
"grad_norm": 0.25941383838653564,
"learning_rate": 1.887786279656482e-05,
"loss": 0.0267,
"step": 4370
},
{
"grad_norm": 0.1847459226846695,
"learning_rate": 1.8662651006095387e-05,
"loss": 0.0247,
"step": 4380
},
{
"grad_norm": 0.28263548016548157,
"learning_rate": 1.8448391159161204e-05,
"loss": 0.0301,
"step": 4390
},
{
"grad_norm": 0.2198208123445511,
"learning_rate": 1.8235089764394408e-05,
"loss": 0.0356,
"step": 4400
},
{
"grad_norm": 0.15997262299060822,
"learning_rate": 1.8022753301311935e-05,
"loss": 0.0283,
"step": 4410
},
{
"grad_norm": 0.2552344799041748,
"learning_rate": 1.7811388220118707e-05,
"loss": 0.0281,
"step": 4420
},
{
"grad_norm": 0.17792999744415283,
"learning_rate": 1.7601000941511757e-05,
"loss": 0.0269,
"step": 4430
},
{
"grad_norm": 0.3445883095264435,
"learning_rate": 1.7391597856485083e-05,
"loss": 0.0317,
"step": 4440
},
{
"grad_norm": 0.18865756690502167,
"learning_rate": 1.7183185326135543e-05,
"loss": 0.0221,
"step": 4450
},
{
"grad_norm": 0.25886648893356323,
"learning_rate": 1.6975769681469705e-05,
"loss": 0.0294,
"step": 4460
},
{
"grad_norm": 0.29419928789138794,
"learning_rate": 1.676935722321139e-05,
"loss": 0.0262,
"step": 4470
},
{
"grad_norm": 0.22673369944095612,
"learning_rate": 1.6563954221610355e-05,
"loss": 0.0258,
"step": 4480
},
{
"grad_norm": 0.18096527457237244,
"learning_rate": 1.6359566916251845e-05,
"loss": 0.0249,
"step": 4490
},
{
"grad_norm": 0.1566012054681778,
"learning_rate": 1.615620151586697e-05,
"loss": 0.0254,
"step": 4500
},
{
"grad_norm": 0.18056631088256836,
"learning_rate": 1.5953864198144135e-05,
"loss": 0.0317,
"step": 4510
},
{
"grad_norm": 0.24644680321216583,
"learning_rate": 1.5752561109541447e-05,
"loss": 0.0258,
"step": 4520
},
{
"grad_norm": 0.2510467767715454,
"learning_rate": 1.5552298365099882e-05,
"loss": 0.029,
"step": 4530
},
{
"grad_norm": 0.23443827033042908,
"learning_rate": 1.5353082048257596e-05,
"loss": 0.0264,
"step": 4540
},
{
"grad_norm": 0.1501440703868866,
"learning_rate": 1.5154918210665148e-05,
"loss": 0.0233,
"step": 4550
},
{
"grad_norm": 0.13923236727714539,
"learning_rate": 1.4957812872001614e-05,
"loss": 0.0269,
"step": 4560
},
{
"grad_norm": 0.22594907879829407,
"learning_rate": 1.4761772019791748e-05,
"loss": 0.03,
"step": 4570
},
{
"grad_norm": 0.15290258824825287,
"learning_rate": 1.4566801609224096e-05,
"loss": 0.0248,
"step": 4580
},
{
"grad_norm": 0.11978762596845627,
"learning_rate": 1.4372907562970079e-05,
"loss": 0.0273,
"step": 4590
},
{
"grad_norm": 0.22400222718715668,
"learning_rate": 1.4180095771004154e-05,
"loss": 0.028,
"step": 4600
},
{
"grad_norm": 0.20572899281978607,
"learning_rate": 1.3988372090424773e-05,
"loss": 0.0268,
"step": 4610
},
{
"grad_norm": 0.1658511757850647,
"learning_rate": 1.3797742345276521e-05,
"loss": 0.0327,
"step": 4620
},
{
"grad_norm": 0.19247505068778992,
"learning_rate": 1.3608212326373249e-05,
"loss": 0.0235,
"step": 4630
},
{
"grad_norm": 0.17291218042373657,
"learning_rate": 1.3419787791122062e-05,
"loss": 0.0296,
"step": 4640
},
{
"grad_norm": 0.28872936964035034,
"learning_rate": 1.323247446334847e-05,
"loss": 0.0257,
"step": 4650
},
{
"grad_norm": 0.25289255380630493,
"learning_rate": 1.3046278033122577e-05,
"loss": 0.034,
"step": 4660
},
{
"grad_norm": 0.16647827625274658,
"learning_rate": 1.286120415658611e-05,
"loss": 0.0275,
"step": 4670
},
{
"grad_norm": 0.17159391939640045,
"learning_rate": 1.2677258455780683e-05,
"loss": 0.0207,
"step": 4680
},
{
"grad_norm": 0.19143880903720856,
"learning_rate": 1.2494446518477022e-05,
"loss": 0.0298,
"step": 4690
},
{
"grad_norm": 0.20133113861083984,
"learning_rate": 1.2312773898005175e-05,
"loss": 0.0285,
"step": 4700
},
{
"grad_norm": 0.15403605997562408,
"learning_rate": 1.2132246113085822e-05,
"loss": 0.0234,
"step": 4710
},
{
"grad_norm": 0.16768252849578857,
"learning_rate": 1.1952868647662696e-05,
"loss": 0.0296,
"step": 4720
},
{
"grad_norm": 0.21265120804309845,
"learning_rate": 1.1774646950735913e-05,
"loss": 0.024,
"step": 4730
},
{
"grad_norm": 0.2079499512910843,
"learning_rate": 1.1597586436196473e-05,
"loss": 0.0251,
"step": 4740
},
{
"grad_norm": 0.26302212476730347,
"learning_rate": 1.1421692482661856e-05,
"loss": 0.0284,
"step": 4750
},
{
"grad_norm": 0.15336978435516357,
"learning_rate": 1.124697043331256e-05,
"loss": 0.0252,
"step": 4760
},
{
"grad_norm": 0.15700557827949524,
"learning_rate": 1.107342559572977e-05,
"loss": 0.0267,
"step": 4770
},
{
"grad_norm": 0.2047663778066635,
"learning_rate": 1.090106324173426e-05,
"loss": 0.0281,
"step": 4780
},
{
"grad_norm": 0.163861483335495,
"learning_rate": 1.0729888607226113e-05,
"loss": 0.0235,
"step": 4790
},
{
"grad_norm": 0.2103009968996048,
"learning_rate": 1.0559906892025745e-05,
"loss": 0.0264,
"step": 4800
},
{
"grad_norm": 0.24158141016960144,
"learning_rate": 1.0391123259715906e-05,
"loss": 0.0266,
"step": 4810
},
{
"grad_norm": 0.15435267984867096,
"learning_rate": 1.0223542837484839e-05,
"loss": 0.0233,
"step": 4820
},
{
"grad_norm": 0.21909579634666443,
"learning_rate": 1.0057170715970559e-05,
"loss": 0.0292,
"step": 4830
},
{
"grad_norm": 0.18775467574596405,
"learning_rate": 9.892011949106172e-06,
"loss": 0.032,
"step": 4840
},
{
"grad_norm": 0.10590514540672302,
"learning_rate": 9.728071553966339e-06,
"loss": 0.0264,
"step": 4850
},
{
"grad_norm": 0.19203981757164001,
"learning_rate": 9.56535451061496e-06,
"loss": 0.023,
"step": 4860
},
{
"grad_norm": 0.16413716971874237,
"learning_rate": 9.403865761953779e-06,
"loss": 0.0251,
"step": 4870
},
{
"grad_norm": 0.15964259207248688,
"learning_rate": 9.243610213572285e-06,
"loss": 0.0268,
"step": 4880
},
{
"grad_norm": 0.23685966432094574,
"learning_rate": 9.084592733598735e-06,
"loss": 0.0251,
"step": 4890
},
{
"grad_norm": 0.1267436146736145,
"learning_rate": 8.92681815255219e-06,
"loss": 0.0232,
"step": 4900
},
{
"grad_norm": 0.17809130251407623,
"learning_rate": 8.770291263195819e-06,
"loss": 0.0229,
"step": 4910
},
{
"grad_norm": 0.19903568923473358,
"learning_rate": 8.615016820391342e-06,
"loss": 0.0276,
"step": 4920
},
{
"grad_norm": 0.22169925272464752,
"learning_rate": 8.460999540954517e-06,
"loss": 0.0279,
"step": 4930
},
{
"grad_norm": 0.13805542886257172,
"learning_rate": 8.308244103511909e-06,
"loss": 0.0221,
"step": 4940
},
{
"grad_norm": 0.28144964575767517,
"learning_rate": 8.156755148358764e-06,
"loss": 0.0284,
"step": 4950
},
{
"grad_norm": 0.13647648692131042,
"learning_rate": 8.00653727731801e-06,
"loss": 0.0207,
"step": 4960
},
{
"grad_norm": 0.2985517382621765,
"learning_rate": 7.857595053600513e-06,
"loss": 0.0294,
"step": 4970
},
{
"grad_norm": 0.2604110538959503,
"learning_rate": 7.709933001666431e-06,
"loss": 0.028,
"step": 4980
},
{
"grad_norm": 0.179083913564682,
"learning_rate": 7.56355560708778e-06,
"loss": 0.0246,
"step": 4990
},
{
"grad_norm": 0.33425500988960266,
"learning_rate": 7.418467316412158e-06,
"loss": 0.027,
"step": 5000
}
],
"logging_steps": 10,
"max_steps": 6000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}