groot-pick-place-v5 / trainer_state.json
tolasing's picture
Upload folder using huggingface_hub
7e0b505 verified
Raw
History Blame Contribute Delete
126 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.3333333333333333,
"eval_steps": 500,
"global_step": 9000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"grad_norm": 2.1476383209228516,
"learning_rate": 3e-06,
"loss": 1.2658,
"step": 10
},
{
"grad_norm": 1.0018171072006226,
"learning_rate": 6.333333333333334e-06,
"loss": 1.1584,
"step": 20
},
{
"grad_norm": 0.6082165837287903,
"learning_rate": 9.666666666666667e-06,
"loss": 1.0354,
"step": 30
},
{
"grad_norm": 0.8369612693786621,
"learning_rate": 1.3000000000000001e-05,
"loss": 1.0311,
"step": 40
},
{
"grad_norm": 1.2067270278930664,
"learning_rate": 1.6333333333333335e-05,
"loss": 1.0284,
"step": 50
},
{
"grad_norm": 0.6714608073234558,
"learning_rate": 1.9666666666666666e-05,
"loss": 1.0189,
"step": 60
},
{
"grad_norm": 0.7627619504928589,
"learning_rate": 2.3000000000000003e-05,
"loss": 1.0143,
"step": 70
},
{
"grad_norm": 1.5823081731796265,
"learning_rate": 2.633333333333333e-05,
"loss": 0.8936,
"step": 80
},
{
"grad_norm": 3.6040964126586914,
"learning_rate": 2.9666666666666672e-05,
"loss": 0.7145,
"step": 90
},
{
"grad_norm": 4.273551940917969,
"learning_rate": 3.3e-05,
"loss": 0.467,
"step": 100
},
{
"grad_norm": 3.193614959716797,
"learning_rate": 3.633333333333333e-05,
"loss": 0.2897,
"step": 110
},
{
"grad_norm": 2.6662745475769043,
"learning_rate": 3.966666666666667e-05,
"loss": 0.2023,
"step": 120
},
{
"grad_norm": 3.4877500534057617,
"learning_rate": 4.3e-05,
"loss": 0.179,
"step": 130
},
{
"grad_norm": 3.9486441612243652,
"learning_rate": 4.633333333333333e-05,
"loss": 0.1781,
"step": 140
},
{
"grad_norm": 3.08587646484375,
"learning_rate": 4.966666666666667e-05,
"loss": 0.1523,
"step": 150
},
{
"grad_norm": 3.1493003368377686,
"learning_rate": 5.300000000000001e-05,
"loss": 0.1505,
"step": 160
},
{
"grad_norm": 2.5344488620758057,
"learning_rate": 5.633333333333334e-05,
"loss": 0.1294,
"step": 170
},
{
"grad_norm": 2.9247426986694336,
"learning_rate": 5.966666666666667e-05,
"loss": 0.1173,
"step": 180
},
{
"grad_norm": 1.8985925912857056,
"learning_rate": 6.3e-05,
"loss": 0.1138,
"step": 190
},
{
"grad_norm": 1.7166413068771362,
"learning_rate": 6.633333333333334e-05,
"loss": 0.1142,
"step": 200
},
{
"grad_norm": 1.4429770708084106,
"learning_rate": 6.966666666666668e-05,
"loss": 0.1212,
"step": 210
},
{
"grad_norm": 2.064023733139038,
"learning_rate": 7.3e-05,
"loss": 0.1127,
"step": 220
},
{
"grad_norm": 2.1307082176208496,
"learning_rate": 7.633333333333334e-05,
"loss": 0.1039,
"step": 230
},
{
"grad_norm": 1.7044156789779663,
"learning_rate": 7.966666666666666e-05,
"loss": 0.0969,
"step": 240
},
{
"grad_norm": 0.9484785199165344,
"learning_rate": 8.3e-05,
"loss": 0.0948,
"step": 250
},
{
"grad_norm": 1.0718821287155151,
"learning_rate": 8.633333333333334e-05,
"loss": 0.0936,
"step": 260
},
{
"grad_norm": 1.1008540391921997,
"learning_rate": 8.966666666666666e-05,
"loss": 0.1091,
"step": 270
},
{
"grad_norm": 1.158390760421753,
"learning_rate": 9.300000000000001e-05,
"loss": 0.0887,
"step": 280
},
{
"grad_norm": 1.1751158237457275,
"learning_rate": 9.633333333333335e-05,
"loss": 0.1126,
"step": 290
},
{
"grad_norm": 1.139798641204834,
"learning_rate": 9.966666666666667e-05,
"loss": 0.0857,
"step": 300
},
{
"grad_norm": 1.1303426027297974,
"learning_rate": 9.999938485971279e-05,
"loss": 0.0969,
"step": 310
},
{
"grad_norm": 1.1913203001022339,
"learning_rate": 9.999725846827562e-05,
"loss": 0.083,
"step": 320
},
{
"grad_norm": 1.4661716222763062,
"learning_rate": 9.999361329594254e-05,
"loss": 0.0823,
"step": 330
},
{
"grad_norm": 1.3807835578918457,
"learning_rate": 9.998844945344405e-05,
"loss": 0.0719,
"step": 340
},
{
"grad_norm": 0.8359068632125854,
"learning_rate": 9.99817670976436e-05,
"loss": 0.074,
"step": 350
},
{
"grad_norm": 1.560785174369812,
"learning_rate": 9.997356643153303e-05,
"loss": 0.0781,
"step": 360
},
{
"grad_norm": 1.1395645141601562,
"learning_rate": 9.996384770422629e-05,
"loss": 0.0911,
"step": 370
},
{
"grad_norm": 0.9609324336051941,
"learning_rate": 9.995261121095194e-05,
"loss": 0.0831,
"step": 380
},
{
"grad_norm": 1.0976556539535522,
"learning_rate": 9.993985729304408e-05,
"loss": 0.0812,
"step": 390
},
{
"grad_norm": 0.7917892336845398,
"learning_rate": 9.992558633793212e-05,
"loss": 0.0763,
"step": 400
},
{
"grad_norm": 1.223206639289856,
"learning_rate": 9.990979877912891e-05,
"loss": 0.0728,
"step": 410
},
{
"grad_norm": 0.6519767045974731,
"learning_rate": 9.989249509621759e-05,
"loss": 0.0644,
"step": 420
},
{
"grad_norm": 0.6069677472114563,
"learning_rate": 9.987367581483705e-05,
"loss": 0.0676,
"step": 430
},
{
"grad_norm": 0.7504929304122925,
"learning_rate": 9.985334150666592e-05,
"loss": 0.0781,
"step": 440
},
{
"grad_norm": 0.8308466076850891,
"learning_rate": 9.983149278940526e-05,
"loss": 0.0727,
"step": 450
},
{
"grad_norm": 0.6970736384391785,
"learning_rate": 9.980813032675974e-05,
"loss": 0.0802,
"step": 460
},
{
"grad_norm": 0.6909151077270508,
"learning_rate": 9.978325482841753e-05,
"loss": 0.0715,
"step": 470
},
{
"grad_norm": 0.6578675508499146,
"learning_rate": 9.975686705002867e-05,
"loss": 0.0787,
"step": 480
},
{
"grad_norm": 0.7847654819488525,
"learning_rate": 9.972896779318219e-05,
"loss": 0.0738,
"step": 490
},
{
"grad_norm": 0.7815350890159607,
"learning_rate": 9.969955790538175e-05,
"loss": 0.0682,
"step": 500
},
{
"grad_norm": 0.901800274848938,
"learning_rate": 9.966863828001982e-05,
"loss": 0.0706,
"step": 510
},
{
"grad_norm": 0.7130557894706726,
"learning_rate": 9.963620985635065e-05,
"loss": 0.0623,
"step": 520
},
{
"grad_norm": 0.6439282298088074,
"learning_rate": 9.960227361946164e-05,
"loss": 0.0676,
"step": 530
},
{
"grad_norm": 0.7865490913391113,
"learning_rate": 9.95668306002435e-05,
"loss": 0.0621,
"step": 540
},
{
"grad_norm": 0.3901299238204956,
"learning_rate": 9.952988187535886e-05,
"loss": 0.0631,
"step": 550
},
{
"grad_norm": 1.1427687406539917,
"learning_rate": 9.949142856720961e-05,
"loss": 0.0635,
"step": 560
},
{
"grad_norm": 0.8050365447998047,
"learning_rate": 9.945147184390278e-05,
"loss": 0.0549,
"step": 570
},
{
"grad_norm": 0.49396881461143494,
"learning_rate": 9.941001291921512e-05,
"loss": 0.0564,
"step": 580
},
{
"grad_norm": 0.7019354104995728,
"learning_rate": 9.936705305255612e-05,
"loss": 0.0553,
"step": 590
},
{
"grad_norm": 0.7188115119934082,
"learning_rate": 9.932259354892984e-05,
"loss": 0.0619,
"step": 600
},
{
"grad_norm": 1.1214160919189453,
"learning_rate": 9.927663575889521e-05,
"loss": 0.0614,
"step": 610
},
{
"grad_norm": 0.779168426990509,
"learning_rate": 9.922918107852504e-05,
"loss": 0.0502,
"step": 620
},
{
"grad_norm": 0.5766488909721375,
"learning_rate": 9.918023094936363e-05,
"loss": 0.0464,
"step": 630
},
{
"grad_norm": 0.4968787431716919,
"learning_rate": 9.912978685838294e-05,
"loss": 0.0523,
"step": 640
},
{
"grad_norm": 0.48410189151763916,
"learning_rate": 9.90778503379374e-05,
"loss": 0.0529,
"step": 650
},
{
"grad_norm": 0.7694746851921082,
"learning_rate": 9.902442296571743e-05,
"loss": 0.0585,
"step": 660
},
{
"grad_norm": 0.8091838955879211,
"learning_rate": 9.896950636470147e-05,
"loss": 0.0635,
"step": 670
},
{
"grad_norm": 0.6608409881591797,
"learning_rate": 9.891310220310666e-05,
"loss": 0.0591,
"step": 680
},
{
"grad_norm": 0.5572546124458313,
"learning_rate": 9.885521219433823e-05,
"loss": 0.0498,
"step": 690
},
{
"grad_norm": 0.7581210136413574,
"learning_rate": 9.879583809693738e-05,
"loss": 0.0572,
"step": 700
},
{
"grad_norm": 0.45685216784477234,
"learning_rate": 9.873498171452789e-05,
"loss": 0.0656,
"step": 710
},
{
"grad_norm": 0.8955629467964172,
"learning_rate": 9.867264489576135e-05,
"loss": 0.0506,
"step": 720
},
{
"grad_norm": 0.6478453874588013,
"learning_rate": 9.860882953426099e-05,
"loss": 0.0529,
"step": 730
},
{
"grad_norm": 0.3592980206012726,
"learning_rate": 9.854353756856412e-05,
"loss": 0.0382,
"step": 740
},
{
"grad_norm": 0.3558259904384613,
"learning_rate": 9.847677098206332e-05,
"loss": 0.0462,
"step": 750
},
{
"grad_norm": 0.3030086159706116,
"learning_rate": 9.840853180294608e-05,
"loss": 0.0388,
"step": 760
},
{
"grad_norm": 0.6828631162643433,
"learning_rate": 9.833882210413332e-05,
"loss": 0.0444,
"step": 770
},
{
"grad_norm": 0.6411892175674438,
"learning_rate": 9.826764400321633e-05,
"loss": 0.0746,
"step": 780
},
{
"grad_norm": 0.5807362794876099,
"learning_rate": 9.819499966239243e-05,
"loss": 0.0582,
"step": 790
},
{
"grad_norm": 0.46515199542045593,
"learning_rate": 9.812089128839938e-05,
"loss": 0.0567,
"step": 800
},
{
"grad_norm": 0.47056809067726135,
"learning_rate": 9.804532113244828e-05,
"loss": 0.0529,
"step": 810
},
{
"grad_norm": 0.34224045276641846,
"learning_rate": 9.796829149015517e-05,
"loss": 0.0445,
"step": 820
},
{
"grad_norm": 0.6688464879989624,
"learning_rate": 9.788980470147132e-05,
"loss": 0.0507,
"step": 830
},
{
"grad_norm": 0.8944596648216248,
"learning_rate": 9.780986315061218e-05,
"loss": 0.0509,
"step": 840
},
{
"grad_norm": 0.49517256021499634,
"learning_rate": 9.772846926598491e-05,
"loss": 0.0634,
"step": 850
},
{
"grad_norm": 0.6212198138237,
"learning_rate": 9.76456255201146e-05,
"loss": 0.0433,
"step": 860
},
{
"grad_norm": 0.4420333802700043,
"learning_rate": 9.756133442956923e-05,
"loss": 0.0476,
"step": 870
},
{
"grad_norm": 0.417572557926178,
"learning_rate": 9.747559855488313e-05,
"loss": 0.0468,
"step": 880
},
{
"grad_norm": 0.5029761791229248,
"learning_rate": 9.73884205004793e-05,
"loss": 0.0432,
"step": 890
},
{
"grad_norm": 0.8272815346717834,
"learning_rate": 9.729980291459019e-05,
"loss": 0.0504,
"step": 900
},
{
"grad_norm": 0.3368825912475586,
"learning_rate": 9.720974848917735e-05,
"loss": 0.0403,
"step": 910
},
{
"grad_norm": 0.5834298729896545,
"learning_rate": 9.711825995984957e-05,
"loss": 0.0465,
"step": 920
},
{
"grad_norm": 0.4559037387371063,
"learning_rate": 9.702534010577991e-05,
"loss": 0.064,
"step": 930
},
{
"grad_norm": 0.788367748260498,
"learning_rate": 9.693099174962103e-05,
"loss": 0.0446,
"step": 940
},
{
"grad_norm": 0.4215085208415985,
"learning_rate": 9.683521775741977e-05,
"loss": 0.0594,
"step": 950
},
{
"grad_norm": 0.29569482803344727,
"learning_rate": 9.673802103852979e-05,
"loss": 0.042,
"step": 960
},
{
"grad_norm": 0.3794245421886444,
"learning_rate": 9.663940454552342e-05,
"loss": 0.0435,
"step": 970
},
{
"grad_norm": 0.8238006830215454,
"learning_rate": 9.65393712741018e-05,
"loss": 0.0481,
"step": 980
},
{
"grad_norm": 0.4304267466068268,
"learning_rate": 9.6437924263004e-05,
"loss": 0.0515,
"step": 990
},
{
"grad_norm": 0.5858254432678223,
"learning_rate": 9.63350665939146e-05,
"loss": 0.0486,
"step": 1000
},
{
"grad_norm": 0.39110514521598816,
"learning_rate": 9.623080139137023e-05,
"loss": 0.0428,
"step": 1010
},
{
"grad_norm": 0.4019848108291626,
"learning_rate": 9.612513182266447e-05,
"loss": 0.0397,
"step": 1020
},
{
"grad_norm": 0.7421868443489075,
"learning_rate": 9.601806109775179e-05,
"loss": 0.0435,
"step": 1030
},
{
"grad_norm": 0.6537662744522095,
"learning_rate": 9.590959246914995e-05,
"loss": 0.0429,
"step": 1040
},
{
"grad_norm": 0.7396754622459412,
"learning_rate": 9.579972923184122e-05,
"loss": 0.0472,
"step": 1050
},
{
"grad_norm": 0.4307487905025482,
"learning_rate": 9.568847472317232e-05,
"loss": 0.0571,
"step": 1060
},
{
"grad_norm": 0.5497772097587585,
"learning_rate": 9.557583232275303e-05,
"loss": 0.0517,
"step": 1070
},
{
"grad_norm": 0.5181751251220703,
"learning_rate": 9.546180545235344e-05,
"loss": 0.0518,
"step": 1080
},
{
"grad_norm": 0.6198630928993225,
"learning_rate": 9.534639757580013e-05,
"loss": 0.0357,
"step": 1090
},
{
"grad_norm": 0.5379049777984619,
"learning_rate": 9.522961219887092e-05,
"loss": 0.0557,
"step": 1100
},
{
"grad_norm": 0.5486226081848145,
"learning_rate": 9.511145286918828e-05,
"loss": 0.0508,
"step": 1110
},
{
"grad_norm": 0.520230233669281,
"learning_rate": 9.499192317611167e-05,
"loss": 0.0479,
"step": 1120
},
{
"grad_norm": 0.4155549705028534,
"learning_rate": 9.487102675062851e-05,
"loss": 0.0524,
"step": 1130
},
{
"grad_norm": 0.6612143516540527,
"learning_rate": 9.474876726524374e-05,
"loss": 0.0662,
"step": 1140
},
{
"grad_norm": 1.0890560150146484,
"learning_rate": 9.462514843386845e-05,
"loss": 0.0589,
"step": 1150
},
{
"grad_norm": 0.6873958110809326,
"learning_rate": 9.450017401170689e-05,
"loss": 0.0533,
"step": 1160
},
{
"grad_norm": 0.5451594591140747,
"learning_rate": 9.437384779514256e-05,
"loss": 0.0432,
"step": 1170
},
{
"grad_norm": 0.3961072564125061,
"learning_rate": 9.424617362162271e-05,
"loss": 0.0448,
"step": 1180
},
{
"grad_norm": 0.7193255424499512,
"learning_rate": 9.411715536954196e-05,
"loss": 0.0424,
"step": 1190
},
{
"grad_norm": 0.6535919904708862,
"learning_rate": 9.39867969581243e-05,
"loss": 0.0484,
"step": 1200
},
{
"grad_norm": 0.7247876524925232,
"learning_rate": 9.385510234730415e-05,
"loss": 0.0423,
"step": 1210
},
{
"grad_norm": 0.4793953001499176,
"learning_rate": 9.372207553760603e-05,
"loss": 0.042,
"step": 1220
},
{
"grad_norm": 0.4487040042877197,
"learning_rate": 9.358772057002312e-05,
"loss": 0.048,
"step": 1230
},
{
"grad_norm": 0.40151041746139526,
"learning_rate": 9.345204152589428e-05,
"loss": 0.0428,
"step": 1240
},
{
"grad_norm": 0.9399805068969727,
"learning_rate": 9.331504252678037e-05,
"loss": 0.0586,
"step": 1250
},
{
"grad_norm": 0.350047767162323,
"learning_rate": 9.317672773433876e-05,
"loss": 0.0493,
"step": 1260
},
{
"grad_norm": 0.6832209229469299,
"learning_rate": 9.30371013501972e-05,
"loss": 0.0506,
"step": 1270
},
{
"grad_norm": 0.4315686821937561,
"learning_rate": 9.289616761582587e-05,
"loss": 0.0473,
"step": 1280
},
{
"grad_norm": 0.5023611187934875,
"learning_rate": 9.275393081240882e-05,
"loss": 0.0536,
"step": 1290
},
{
"grad_norm": 0.5154740214347839,
"learning_rate": 9.261039526071374e-05,
"loss": 0.0481,
"step": 1300
},
{
"grad_norm": 0.2950477600097656,
"learning_rate": 9.246556532096078e-05,
"loss": 0.0421,
"step": 1310
},
{
"grad_norm": 0.48403283953666687,
"learning_rate": 9.231944539269009e-05,
"loss": 0.0525,
"step": 1320
},
{
"grad_norm": 0.4625755846500397,
"learning_rate": 9.217203991462815e-05,
"loss": 0.0442,
"step": 1330
},
{
"grad_norm": 0.452635258436203,
"learning_rate": 9.202335336455296e-05,
"loss": 0.0492,
"step": 1340
},
{
"grad_norm": 0.4100465476512909,
"learning_rate": 9.187339025915802e-05,
"loss": 0.0387,
"step": 1350
},
{
"grad_norm": 0.44927647709846497,
"learning_rate": 9.17221551539151e-05,
"loss": 0.0453,
"step": 1360
},
{
"grad_norm": 0.3114309012889862,
"learning_rate": 9.156965264293586e-05,
"loss": 0.0339,
"step": 1370
},
{
"grad_norm": 0.30743640661239624,
"learning_rate": 9.141588735883232e-05,
"loss": 0.0354,
"step": 1380
},
{
"grad_norm": 0.44713470339775085,
"learning_rate": 9.126086397257612e-05,
"loss": 0.0379,
"step": 1390
},
{
"grad_norm": 0.6579342484474182,
"learning_rate": 9.110458719335659e-05,
"loss": 0.0493,
"step": 1400
},
{
"grad_norm": 0.6419163942337036,
"learning_rate": 9.094706176843777e-05,
"loss": 0.0381,
"step": 1410
},
{
"grad_norm": 0.5080713629722595,
"learning_rate": 9.078829248301417e-05,
"loss": 0.0484,
"step": 1420
},
{
"grad_norm": 0.5176462531089783,
"learning_rate": 9.062828416006539e-05,
"loss": 0.044,
"step": 1430
},
{
"grad_norm": 0.3638184666633606,
"learning_rate": 9.046704166020961e-05,
"loss": 0.0413,
"step": 1440
},
{
"grad_norm": 0.4040341377258301,
"learning_rate": 9.030456988155596e-05,
"loss": 0.044,
"step": 1450
},
{
"grad_norm": 0.2890731990337372,
"learning_rate": 9.014087375955573e-05,
"loss": 0.0395,
"step": 1460
},
{
"grad_norm": 0.38218948245048523,
"learning_rate": 8.997595826685243e-05,
"loss": 0.0371,
"step": 1470
},
{
"grad_norm": 0.49778327345848083,
"learning_rate": 8.980982841313074e-05,
"loss": 0.0457,
"step": 1480
},
{
"grad_norm": 0.4390401244163513,
"learning_rate": 8.964248924496435e-05,
"loss": 0.0369,
"step": 1490
},
{
"grad_norm": 0.3490327000617981,
"learning_rate": 8.947394584566258e-05,
"loss": 0.0483,
"step": 1500
},
{
"grad_norm": 0.3651290833950043,
"learning_rate": 8.930420333511606e-05,
"loss": 0.0369,
"step": 1510
},
{
"grad_norm": 0.5795438885688782,
"learning_rate": 8.913326686964117e-05,
"loss": 0.036,
"step": 1520
},
{
"grad_norm": 0.31590238213539124,
"learning_rate": 8.89611416418234e-05,
"loss": 0.0393,
"step": 1530
},
{
"grad_norm": 0.525740385055542,
"learning_rate": 8.878783288035957e-05,
"loss": 0.0461,
"step": 1540
},
{
"grad_norm": 0.469777375459671,
"learning_rate": 8.86133458498991e-05,
"loss": 0.044,
"step": 1550
},
{
"grad_norm": 0.607012927532196,
"learning_rate": 8.843768585088393e-05,
"loss": 0.0405,
"step": 1560
},
{
"grad_norm": 0.5046875476837158,
"learning_rate": 8.82608582193877e-05,
"loss": 0.0333,
"step": 1570
},
{
"grad_norm": 0.292531281709671,
"learning_rate": 8.80828683269535e-05,
"loss": 0.0358,
"step": 1580
},
{
"grad_norm": 0.2831183075904846,
"learning_rate": 8.790372158043074e-05,
"loss": 0.0386,
"step": 1590
},
{
"grad_norm": 0.42618659138679504,
"learning_rate": 8.772342342181095e-05,
"loss": 0.038,
"step": 1600
},
{
"grad_norm": 0.6016631126403809,
"learning_rate": 8.75419793280624e-05,
"loss": 0.0364,
"step": 1610
},
{
"grad_norm": 0.45318183302879333,
"learning_rate": 8.735939481096378e-05,
"loss": 0.0347,
"step": 1620
},
{
"grad_norm": 0.5996795296669006,
"learning_rate": 8.717567541693673e-05,
"loss": 0.0392,
"step": 1630
},
{
"grad_norm": 0.6204606890678406,
"learning_rate": 8.699082672687734e-05,
"loss": 0.0406,
"step": 1640
},
{
"grad_norm": 0.32462042570114136,
"learning_rate": 8.680485435598673e-05,
"loss": 0.0476,
"step": 1650
},
{
"grad_norm": 0.3754139840602875,
"learning_rate": 8.661776395360029e-05,
"loss": 0.0325,
"step": 1660
},
{
"grad_norm": 0.48614147305488586,
"learning_rate": 8.642956120301626e-05,
"loss": 0.0376,
"step": 1670
},
{
"grad_norm": 0.4615228474140167,
"learning_rate": 8.624025182132292e-05,
"loss": 0.0401,
"step": 1680
},
{
"grad_norm": 0.3228665888309479,
"learning_rate": 8.604984155922506e-05,
"loss": 0.0446,
"step": 1690
},
{
"grad_norm": 0.4279070198535919,
"learning_rate": 8.585833620086918e-05,
"loss": 0.0493,
"step": 1700
},
{
"grad_norm": 0.3909189999103546,
"learning_rate": 8.566574156366784e-05,
"loss": 0.0403,
"step": 1710
},
{
"grad_norm": 0.3734934628009796,
"learning_rate": 8.547206349812298e-05,
"loss": 0.0447,
"step": 1720
},
{
"grad_norm": 0.3394744098186493,
"learning_rate": 8.527730788764805e-05,
"loss": 0.0358,
"step": 1730
},
{
"grad_norm": 0.34820032119750977,
"learning_rate": 8.508148064838948e-05,
"loss": 0.0348,
"step": 1740
},
{
"grad_norm": 0.6352878212928772,
"learning_rate": 8.488458772904684e-05,
"loss": 0.0373,
"step": 1750
},
{
"grad_norm": 0.3595995306968689,
"learning_rate": 8.468663511069217e-05,
"loss": 0.0416,
"step": 1760
},
{
"grad_norm": 0.602392852306366,
"learning_rate": 8.448762880658825e-05,
"loss": 0.0365,
"step": 1770
},
{
"grad_norm": 0.4117213189601898,
"learning_rate": 8.428757486200603e-05,
"loss": 0.0379,
"step": 1780
},
{
"grad_norm": 0.35936439037323,
"learning_rate": 8.40864793540409e-05,
"loss": 0.0418,
"step": 1790
},
{
"grad_norm": 0.5282872319221497,
"learning_rate": 8.388434839142813e-05,
"loss": 0.0375,
"step": 1800
},
{
"grad_norm": 0.5019148588180542,
"learning_rate": 8.368118811435726e-05,
"loss": 0.0375,
"step": 1810
},
{
"grad_norm": 0.6031928658485413,
"learning_rate": 8.347700469428564e-05,
"loss": 0.0421,
"step": 1820
},
{
"grad_norm": 0.2310052067041397,
"learning_rate": 8.327180433375091e-05,
"loss": 0.0377,
"step": 1830
},
{
"grad_norm": 0.2899196147918701,
"learning_rate": 8.306559326618259e-05,
"loss": 0.0337,
"step": 1840
},
{
"grad_norm": 0.43216198682785034,
"learning_rate": 8.285837775571276e-05,
"loss": 0.0373,
"step": 1850
},
{
"grad_norm": 0.2763599157333374,
"learning_rate": 8.265016409698573e-05,
"loss": 0.0388,
"step": 1860
},
{
"grad_norm": 0.3530355989933014,
"learning_rate": 8.244095861496686e-05,
"loss": 0.0379,
"step": 1870
},
{
"grad_norm": 0.3489007353782654,
"learning_rate": 8.223076766475035e-05,
"loss": 0.0313,
"step": 1880
},
{
"grad_norm": 0.42826664447784424,
"learning_rate": 8.201959763136633e-05,
"loss": 0.033,
"step": 1890
},
{
"grad_norm": 0.3354351818561554,
"learning_rate": 8.180745492958674e-05,
"loss": 0.045,
"step": 1900
},
{
"grad_norm": 0.3736966550350189,
"learning_rate": 8.159434600373061e-05,
"loss": 0.0405,
"step": 1910
},
{
"grad_norm": 0.4795636236667633,
"learning_rate": 8.138027732746818e-05,
"loss": 0.0491,
"step": 1920
},
{
"grad_norm": 0.4792327880859375,
"learning_rate": 8.116525540362434e-05,
"loss": 0.0431,
"step": 1930
},
{
"grad_norm": 0.41160207986831665,
"learning_rate": 8.094928676398101e-05,
"loss": 0.0412,
"step": 1940
},
{
"grad_norm": 0.31209295988082886,
"learning_rate": 8.073237796907882e-05,
"loss": 0.0433,
"step": 1950
},
{
"grad_norm": 0.33164528012275696,
"learning_rate": 8.051453560801772e-05,
"loss": 0.0333,
"step": 1960
},
{
"grad_norm": 0.3001999258995056,
"learning_rate": 8.029576629825687e-05,
"loss": 0.0453,
"step": 1970
},
{
"grad_norm": 0.5621007680892944,
"learning_rate": 8.007607668541362e-05,
"loss": 0.0481,
"step": 1980
},
{
"grad_norm": 0.4241228997707367,
"learning_rate": 7.985547344306161e-05,
"loss": 0.0367,
"step": 1990
},
{
"grad_norm": 0.2405935376882553,
"learning_rate": 7.963396327252812e-05,
"loss": 0.0289,
"step": 2000
},
{
"grad_norm": 0.3709689974784851,
"learning_rate": 7.941155290269038e-05,
"loss": 0.0299,
"step": 2010
},
{
"grad_norm": 0.4156179428100586,
"learning_rate": 7.918824908977123e-05,
"loss": 0.0386,
"step": 2020
},
{
"grad_norm": 0.7034849524497986,
"learning_rate": 7.896405861713394e-05,
"loss": 0.0429,
"step": 2030
},
{
"grad_norm": 0.3684985637664795,
"learning_rate": 7.873898829507606e-05,
"loss": 0.0431,
"step": 2040
},
{
"grad_norm": 0.3061644732952118,
"learning_rate": 7.851304496062254e-05,
"loss": 0.0369,
"step": 2050
},
{
"grad_norm": 0.29207587242126465,
"learning_rate": 7.828623547731818e-05,
"loss": 0.0294,
"step": 2060
},
{
"grad_norm": 0.27505022287368774,
"learning_rate": 7.80585667350189e-05,
"loss": 0.0283,
"step": 2070
},
{
"grad_norm": 0.4933038353919983,
"learning_rate": 7.783004564968263e-05,
"loss": 0.0302,
"step": 2080
},
{
"grad_norm": 0.38244086503982544,
"learning_rate": 7.760067916315921e-05,
"loss": 0.0287,
"step": 2090
},
{
"grad_norm": 0.3276348412036896,
"learning_rate": 7.737047424297941e-05,
"loss": 0.0303,
"step": 2100
},
{
"grad_norm": 0.35118088126182556,
"learning_rate": 7.713943788214337e-05,
"loss": 0.0299,
"step": 2110
},
{
"grad_norm": 0.33147382736206055,
"learning_rate": 7.690757709890812e-05,
"loss": 0.0276,
"step": 2120
},
{
"grad_norm": 0.37490758299827576,
"learning_rate": 7.66748989365744e-05,
"loss": 0.0337,
"step": 2130
},
{
"grad_norm": 0.677511990070343,
"learning_rate": 7.644141046327271e-05,
"loss": 0.0436,
"step": 2140
},
{
"grad_norm": 0.330022394657135,
"learning_rate": 7.620711877174866e-05,
"loss": 0.0461,
"step": 2150
},
{
"grad_norm": 0.4340311586856842,
"learning_rate": 7.597203097914732e-05,
"loss": 0.0362,
"step": 2160
},
{
"grad_norm": 0.33995547890663147,
"learning_rate": 7.573615422679726e-05,
"loss": 0.0317,
"step": 2170
},
{
"grad_norm": 0.6235854625701904,
"learning_rate": 7.549949567999345e-05,
"loss": 0.0393,
"step": 2180
},
{
"grad_norm": 0.42266157269477844,
"learning_rate": 7.526206252777968e-05,
"loss": 0.0313,
"step": 2190
},
{
"grad_norm": 0.4183277189731598,
"learning_rate": 7.50238619827301e-05,
"loss": 0.0319,
"step": 2200
},
{
"grad_norm": 0.2806348502635956,
"learning_rate": 7.478490128073022e-05,
"loss": 0.0321,
"step": 2210
},
{
"grad_norm": 0.2528248131275177,
"learning_rate": 7.454518768075704e-05,
"loss": 0.0275,
"step": 2220
},
{
"grad_norm": 0.40887144207954407,
"learning_rate": 7.430472846465856e-05,
"loss": 0.036,
"step": 2230
},
{
"grad_norm": 0.22932802140712738,
"learning_rate": 7.406353093693253e-05,
"loss": 0.0307,
"step": 2240
},
{
"grad_norm": 0.4036305546760559,
"learning_rate": 7.382160242450469e-05,
"loss": 0.0359,
"step": 2250
},
{
"grad_norm": 0.5628826022148132,
"learning_rate": 7.357895027650598e-05,
"loss": 0.037,
"step": 2260
},
{
"grad_norm": 0.3429408371448517,
"learning_rate": 7.333558186404958e-05,
"loss": 0.026,
"step": 2270
},
{
"grad_norm": 0.359138548374176,
"learning_rate": 7.309150458000668e-05,
"loss": 0.0355,
"step": 2280
},
{
"grad_norm": 0.3372308909893036,
"learning_rate": 7.284672583878219e-05,
"loss": 0.0307,
"step": 2290
},
{
"grad_norm": 0.3284088373184204,
"learning_rate": 7.260125307608929e-05,
"loss": 0.0341,
"step": 2300
},
{
"grad_norm": 0.3641900420188904,
"learning_rate": 7.235509374872373e-05,
"loss": 0.0414,
"step": 2310
},
{
"grad_norm": 0.38477224111557007,
"learning_rate": 7.210825533433719e-05,
"loss": 0.0272,
"step": 2320
},
{
"grad_norm": 0.3984125852584839,
"learning_rate": 7.186074533121013e-05,
"loss": 0.0419,
"step": 2330
},
{
"grad_norm": 0.593463122844696,
"learning_rate": 7.161257125802413e-05,
"loss": 0.0342,
"step": 2340
},
{
"grad_norm": 0.5311180949211121,
"learning_rate": 7.136374065363334e-05,
"loss": 0.0394,
"step": 2350
},
{
"grad_norm": 0.5243760943412781,
"learning_rate": 7.11142610768356e-05,
"loss": 0.0354,
"step": 2360
},
{
"grad_norm": 0.23473842442035675,
"learning_rate": 7.086414010614276e-05,
"loss": 0.0326,
"step": 2370
},
{
"grad_norm": 0.3910692632198334,
"learning_rate": 7.061338533955043e-05,
"loss": 0.0274,
"step": 2380
},
{
"grad_norm": 0.37255534529685974,
"learning_rate": 7.036200439430725e-05,
"loss": 0.0388,
"step": 2390
},
{
"grad_norm": 0.3789617121219635,
"learning_rate": 7.01100049066835e-05,
"loss": 0.0366,
"step": 2400
},
{
"grad_norm": 0.5898721814155579,
"learning_rate": 6.985739453173903e-05,
"loss": 0.0289,
"step": 2410
},
{
"grad_norm": 0.3355694115161896,
"learning_rate": 6.960418094309085e-05,
"loss": 0.0308,
"step": 2420
},
{
"grad_norm": 0.3092407286167145,
"learning_rate": 6.93503718326799e-05,
"loss": 0.0304,
"step": 2430
},
{
"grad_norm": 0.43957892060279846,
"learning_rate": 6.909597491053751e-05,
"loss": 0.0342,
"step": 2440
},
{
"grad_norm": 0.2295580357313156,
"learning_rate": 6.884099790455113e-05,
"loss": 0.0275,
"step": 2450
},
{
"grad_norm": 0.4543008804321289,
"learning_rate": 6.858544856022952e-05,
"loss": 0.0369,
"step": 2460
},
{
"grad_norm": 0.3433627486228943,
"learning_rate": 6.83293346404676e-05,
"loss": 0.0291,
"step": 2470
},
{
"grad_norm": 0.29589101672172546,
"learning_rate": 6.80726639253105e-05,
"loss": 0.0276,
"step": 2480
},
{
"grad_norm": 0.4317607581615448,
"learning_rate": 6.781544421171732e-05,
"loss": 0.0257,
"step": 2490
},
{
"grad_norm": 0.3916417360305786,
"learning_rate": 6.755768331332424e-05,
"loss": 0.0359,
"step": 2500
},
{
"grad_norm": 0.3352242112159729,
"learning_rate": 6.729938906020713e-05,
"loss": 0.0337,
"step": 2510
},
{
"grad_norm": 0.31850340962409973,
"learning_rate": 6.704056929864376e-05,
"loss": 0.0329,
"step": 2520
},
{
"grad_norm": 0.46833932399749756,
"learning_rate": 6.67812318908754e-05,
"loss": 0.0306,
"step": 2530
},
{
"grad_norm": 0.35765767097473145,
"learning_rate": 6.6521384714868e-05,
"loss": 0.0362,
"step": 2540
},
{
"grad_norm": 0.4558477997779846,
"learning_rate": 6.626103566407295e-05,
"loss": 0.0412,
"step": 2550
},
{
"grad_norm": 0.3726062774658203,
"learning_rate": 6.600019264718713e-05,
"loss": 0.0326,
"step": 2560
},
{
"grad_norm": 0.2789278030395508,
"learning_rate": 6.573886358791285e-05,
"loss": 0.0292,
"step": 2570
},
{
"grad_norm": 0.3415791094303131,
"learning_rate": 6.547705642471703e-05,
"loss": 0.039,
"step": 2580
},
{
"grad_norm": 0.4107648432254791,
"learning_rate": 6.521477911059008e-05,
"loss": 0.0355,
"step": 2590
},
{
"grad_norm": 0.40447941422462463,
"learning_rate": 6.495203961280434e-05,
"loss": 0.0313,
"step": 2600
},
{
"grad_norm": 0.390096515417099,
"learning_rate": 6.468884591267204e-05,
"loss": 0.0291,
"step": 2610
},
{
"grad_norm": 0.3055766224861145,
"learning_rate": 6.44252060053028e-05,
"loss": 0.0338,
"step": 2620
},
{
"grad_norm": 0.4350510835647583,
"learning_rate": 6.416112789936086e-05,
"loss": 0.0326,
"step": 2630
},
{
"grad_norm": 0.3334941565990448,
"learning_rate": 6.389661961682173e-05,
"loss": 0.0273,
"step": 2640
},
{
"grad_norm": 0.28560584783554077,
"learning_rate": 6.363168919272846e-05,
"loss": 0.0312,
"step": 2650
},
{
"grad_norm": 0.19734404981136322,
"learning_rate": 6.336634467494768e-05,
"loss": 0.0469,
"step": 2660
},
{
"grad_norm": 0.48325496912002563,
"learning_rate": 6.310059412392505e-05,
"loss": 0.0372,
"step": 2670
},
{
"grad_norm": 0.33191633224487305,
"learning_rate": 6.283444561244042e-05,
"loss": 0.0369,
"step": 2680
},
{
"grad_norm": 0.21083933115005493,
"learning_rate": 6.256790722536251e-05,
"loss": 0.0316,
"step": 2690
},
{
"grad_norm": 0.32207825779914856,
"learning_rate": 6.230098705940354e-05,
"loss": 0.0321,
"step": 2700
},
{
"grad_norm": 0.4074423015117645,
"learning_rate": 6.203369322287306e-05,
"loss": 0.0321,
"step": 2710
},
{
"grad_norm": 0.4362260401248932,
"learning_rate": 6.17660338354317e-05,
"loss": 0.0312,
"step": 2720
},
{
"grad_norm": 0.29363614320755005,
"learning_rate": 6.149801702784456e-05,
"loss": 0.0308,
"step": 2730
},
{
"grad_norm": 0.45793551206588745,
"learning_rate": 6.122965094173424e-05,
"loss": 0.0305,
"step": 2740
},
{
"grad_norm": 0.4806693494319916,
"learning_rate": 6.0960943729333374e-05,
"loss": 0.0252,
"step": 2750
},
{
"grad_norm": 0.41287726163864136,
"learning_rate": 6.069190355323717e-05,
"loss": 0.0295,
"step": 2760
},
{
"grad_norm": 0.2462008148431778,
"learning_rate": 6.042253858615532e-05,
"loss": 0.0293,
"step": 2770
},
{
"grad_norm": 0.24040086567401886,
"learning_rate": 6.015285701066382e-05,
"loss": 0.0328,
"step": 2780
},
{
"grad_norm": 0.37454015016555786,
"learning_rate": 5.988286701895631e-05,
"loss": 0.0262,
"step": 2790
},
{
"grad_norm": 0.22840295732021332,
"learning_rate": 5.961257681259535e-05,
"loss": 0.0321,
"step": 2800
},
{
"grad_norm": 0.33319950103759766,
"learning_rate": 5.934199460226317e-05,
"loss": 0.0309,
"step": 2810
},
{
"grad_norm": 0.2775340676307678,
"learning_rate": 5.9071128607512285e-05,
"loss": 0.0284,
"step": 2820
},
{
"grad_norm": 0.315031498670578,
"learning_rate": 5.8799987056515804e-05,
"loss": 0.0307,
"step": 2830
},
{
"grad_norm": 0.39186322689056396,
"learning_rate": 5.8528578185817514e-05,
"loss": 0.0335,
"step": 2840
},
{
"grad_norm": 0.4090574383735657,
"learning_rate": 5.825691024008162e-05,
"loss": 0.0298,
"step": 2850
},
{
"grad_norm": 0.47943785786628723,
"learning_rate": 5.798499147184233e-05,
"loss": 0.0298,
"step": 2860
},
{
"grad_norm": 0.2602699398994446,
"learning_rate": 5.771283014125317e-05,
"loss": 0.0259,
"step": 2870
},
{
"grad_norm": 0.3160038888454437,
"learning_rate": 5.7440434515836064e-05,
"loss": 0.0268,
"step": 2880
},
{
"grad_norm": 0.24288243055343628,
"learning_rate": 5.7167812870230094e-05,
"loss": 0.0281,
"step": 2890
},
{
"grad_norm": 0.20718321204185486,
"learning_rate": 5.689497348594035e-05,
"loss": 0.0278,
"step": 2900
},
{
"grad_norm": 0.38176146149635315,
"learning_rate": 5.662192465108613e-05,
"loss": 0.0459,
"step": 2910
},
{
"grad_norm": 0.31435516476631165,
"learning_rate": 5.634867466014932e-05,
"loss": 0.0299,
"step": 2920
},
{
"grad_norm": 0.4356970489025116,
"learning_rate": 5.607523181372234e-05,
"loss": 0.0282,
"step": 2930
},
{
"grad_norm": 0.24084477126598358,
"learning_rate": 5.5801604418256117e-05,
"loss": 0.0283,
"step": 2940
},
{
"grad_norm": 0.28903084993362427,
"learning_rate": 5.552780078580756e-05,
"loss": 0.0274,
"step": 2950
},
{
"grad_norm": 0.24469222128391266,
"learning_rate": 5.525382923378728e-05,
"loss": 0.0335,
"step": 2960
},
{
"grad_norm": 0.23761104047298431,
"learning_rate": 5.49796980847068e-05,
"loss": 0.0307,
"step": 2970
},
{
"grad_norm": 0.23053614795207977,
"learning_rate": 5.470541566592573e-05,
"loss": 0.0241,
"step": 2980
},
{
"grad_norm": 0.33007940649986267,
"learning_rate": 5.443099030939887e-05,
"loss": 0.0263,
"step": 2990
},
{
"grad_norm": 0.3086564540863037,
"learning_rate": 5.415643035142309e-05,
"loss": 0.0327,
"step": 3000
},
{
"grad_norm": 0.328674852848053,
"learning_rate": 5.3881744132384104e-05,
"loss": 0.0291,
"step": 3010
},
{
"grad_norm": 0.3298027217388153,
"learning_rate": 5.360693999650303e-05,
"loss": 0.0391,
"step": 3020
},
{
"grad_norm": 0.2422970086336136,
"learning_rate": 5.3332026291583016e-05,
"loss": 0.0264,
"step": 3030
},
{
"grad_norm": 0.4864082634449005,
"learning_rate": 5.305701136875566e-05,
"loss": 0.0236,
"step": 3040
},
{
"grad_norm": 0.21486255526542664,
"learning_rate": 5.278190358222721e-05,
"loss": 0.0307,
"step": 3050
},
{
"grad_norm": 0.4227481484413147,
"learning_rate": 5.25067112890249e-05,
"loss": 0.0328,
"step": 3060
},
{
"grad_norm": 0.2608308792114258,
"learning_rate": 5.2231442848743064e-05,
"loss": 0.0317,
"step": 3070
},
{
"grad_norm": 0.3551100194454193,
"learning_rate": 5.1956106623289145e-05,
"loss": 0.0306,
"step": 3080
},
{
"grad_norm": 0.4440188407897949,
"learning_rate": 5.168071097662972e-05,
"loss": 0.0307,
"step": 3090
},
{
"grad_norm": 0.4040832817554474,
"learning_rate": 5.1405264274536445e-05,
"loss": 0.0287,
"step": 3100
},
{
"grad_norm": 0.39724764227867126,
"learning_rate": 5.112977488433188e-05,
"loss": 0.0349,
"step": 3110
},
{
"grad_norm": 0.3195992708206177,
"learning_rate": 5.085425117463533e-05,
"loss": 0.0402,
"step": 3120
},
{
"grad_norm": 0.5096355676651001,
"learning_rate": 5.057870151510864e-05,
"loss": 0.0392,
"step": 3130
},
{
"grad_norm": 0.4114478826522827,
"learning_rate": 5.030313427620197e-05,
"loss": 0.0358,
"step": 3140
},
{
"grad_norm": 0.26888328790664673,
"learning_rate": 5.0027557828899426e-05,
"loss": 0.0303,
"step": 3150
},
{
"grad_norm": 0.33315181732177734,
"learning_rate": 4.975198054446492e-05,
"loss": 0.0258,
"step": 3160
},
{
"grad_norm": 0.3892503082752228,
"learning_rate": 4.947641079418773e-05,
"loss": 0.0334,
"step": 3170
},
{
"grad_norm": 0.32576891779899597,
"learning_rate": 4.920085694912828e-05,
"loss": 0.0473,
"step": 3180
},
{
"grad_norm": 0.3438071310520172,
"learning_rate": 4.892532737986387e-05,
"loss": 0.0336,
"step": 3190
},
{
"grad_norm": 0.3293437361717224,
"learning_rate": 4.864983045623434e-05,
"loss": 0.0344,
"step": 3200
},
{
"grad_norm": 0.3349752128124237,
"learning_rate": 4.837437454708784e-05,
"loss": 0.0296,
"step": 3210
},
{
"grad_norm": 0.32625678181648254,
"learning_rate": 4.809896802002662e-05,
"loss": 0.0269,
"step": 3220
},
{
"grad_norm": 0.2607935070991516,
"learning_rate": 4.7823619241152854e-05,
"loss": 0.0288,
"step": 3230
},
{
"grad_norm": 0.39156603813171387,
"learning_rate": 4.754833657481445e-05,
"loss": 0.0257,
"step": 3240
},
{
"grad_norm": 0.37770044803619385,
"learning_rate": 4.7273128383351015e-05,
"loss": 0.0232,
"step": 3250
},
{
"grad_norm": 0.21835839748382568,
"learning_rate": 4.699800302683981e-05,
"loss": 0.0388,
"step": 3260
},
{
"grad_norm": 0.1701326221227646,
"learning_rate": 4.6722968862841806e-05,
"loss": 0.0303,
"step": 3270
},
{
"grad_norm": 0.2512141168117523,
"learning_rate": 4.6448034246147754e-05,
"loss": 0.0269,
"step": 3280
},
{
"grad_norm": 0.27361300587654114,
"learning_rate": 4.6173207528524476e-05,
"loss": 0.0275,
"step": 3290
},
{
"grad_norm": 0.41740429401397705,
"learning_rate": 4.58984970584611e-05,
"loss": 0.0225,
"step": 3300
},
{
"grad_norm": 0.21014976501464844,
"learning_rate": 4.562391118091544e-05,
"loss": 0.0221,
"step": 3310
},
{
"grad_norm": 0.1437355875968933,
"learning_rate": 4.534945823706056e-05,
"loss": 0.0215,
"step": 3320
},
{
"grad_norm": 0.4129693806171417,
"learning_rate": 4.507514656403137e-05,
"loss": 0.0283,
"step": 3330
},
{
"grad_norm": 0.30207327008247375,
"learning_rate": 4.480098449467132e-05,
"loss": 0.0252,
"step": 3340
},
{
"grad_norm": 0.29082217812538147,
"learning_rate": 4.452698035727929e-05,
"loss": 0.0363,
"step": 3350
},
{
"grad_norm": 0.40966904163360596,
"learning_rate": 4.425314247535668e-05,
"loss": 0.0227,
"step": 3360
},
{
"grad_norm": 0.26431578397750854,
"learning_rate": 4.3979479167354477e-05,
"loss": 0.0366,
"step": 3370
},
{
"grad_norm": 0.3912636637687683,
"learning_rate": 4.370599874642055e-05,
"loss": 0.0317,
"step": 3380
},
{
"grad_norm": 0.2687945067882538,
"learning_rate": 4.3432709520147205e-05,
"loss": 0.0289,
"step": 3390
},
{
"grad_norm": 0.32468682527542114,
"learning_rate": 4.315961979031875e-05,
"loss": 0.0325,
"step": 3400
},
{
"grad_norm": 0.4526863992214203,
"learning_rate": 4.2886737852659325e-05,
"loss": 0.0304,
"step": 3410
},
{
"grad_norm": 0.2836206257343292,
"learning_rate": 4.261407199658093e-05,
"loss": 0.03,
"step": 3420
},
{
"grad_norm": 0.26107683777809143,
"learning_rate": 4.234163050493158e-05,
"loss": 0.0247,
"step": 3430
},
{
"grad_norm": 0.362579882144928,
"learning_rate": 4.2069421653743706e-05,
"loss": 0.0305,
"step": 3440
},
{
"grad_norm": 0.27894529700279236,
"learning_rate": 4.179745371198276e-05,
"loss": 0.0218,
"step": 3450
},
{
"grad_norm": 0.3671315610408783,
"learning_rate": 4.1525734941296026e-05,
"loss": 0.0289,
"step": 3460
},
{
"grad_norm": 0.16687674820423126,
"learning_rate": 4.125427359576162e-05,
"loss": 0.0234,
"step": 3470
},
{
"grad_norm": 0.17282120883464813,
"learning_rate": 4.0983077921637815e-05,
"loss": 0.0307,
"step": 3480
},
{
"grad_norm": 0.2520134747028351,
"learning_rate": 4.07121561571125e-05,
"loss": 0.0274,
"step": 3490
},
{
"grad_norm": 0.21231453120708466,
"learning_rate": 4.044151653205292e-05,
"loss": 0.0355,
"step": 3500
},
{
"grad_norm": 0.27000704407691956,
"learning_rate": 4.0171167267755696e-05,
"loss": 0.0327,
"step": 3510
},
{
"grad_norm": 0.46360984444618225,
"learning_rate": 3.9901116576697083e-05,
"loss": 0.0284,
"step": 3520
},
{
"grad_norm": 0.26576513051986694,
"learning_rate": 3.963137266228349e-05,
"loss": 0.026,
"step": 3530
},
{
"grad_norm": 0.35086891055107117,
"learning_rate": 3.93619437186023e-05,
"loss": 0.0292,
"step": 3540
},
{
"grad_norm": 0.243887260556221,
"learning_rate": 3.9092837930172884e-05,
"loss": 0.0298,
"step": 3550
},
{
"grad_norm": 0.32888200879096985,
"learning_rate": 3.8824063471698105e-05,
"loss": 0.0278,
"step": 3560
},
{
"grad_norm": 0.168535977602005,
"learning_rate": 3.855562850781589e-05,
"loss": 0.0268,
"step": 3570
},
{
"grad_norm": 0.2635512053966522,
"learning_rate": 3.828754119285123e-05,
"loss": 0.0302,
"step": 3580
},
{
"grad_norm": 0.20049428939819336,
"learning_rate": 3.801980967056851e-05,
"loss": 0.0252,
"step": 3590
},
{
"grad_norm": 0.2980923652648926,
"learning_rate": 3.77524420739241e-05,
"loss": 0.031,
"step": 3600
},
{
"grad_norm": 0.3177404999732971,
"learning_rate": 3.748544652481927e-05,
"loss": 0.0221,
"step": 3610
},
{
"grad_norm": 0.2229740023612976,
"learning_rate": 3.721883113385353e-05,
"loss": 0.0249,
"step": 3620
},
{
"grad_norm": 0.38334640860557556,
"learning_rate": 3.695260400007819e-05,
"loss": 0.0294,
"step": 3630
},
{
"grad_norm": 0.34582415223121643,
"learning_rate": 3.6686773210750385e-05,
"loss": 0.0249,
"step": 3640
},
{
"grad_norm": 0.19283750653266907,
"learning_rate": 3.642134684108737e-05,
"loss": 0.0294,
"step": 3650
},
{
"grad_norm": 0.2977948486804962,
"learning_rate": 3.615633295402123e-05,
"loss": 0.0214,
"step": 3660
},
{
"grad_norm": 0.14953476190567017,
"learning_rate": 3.5891739599953945e-05,
"loss": 0.0237,
"step": 3670
},
{
"grad_norm": 0.15818743407726288,
"learning_rate": 3.5627574816512846e-05,
"loss": 0.0357,
"step": 3680
},
{
"grad_norm": 0.2496010810136795,
"learning_rate": 3.536384662830648e-05,
"loss": 0.0309,
"step": 3690
},
{
"grad_norm": 0.3513094186782837,
"learning_rate": 3.5100563046680764e-05,
"loss": 0.0236,
"step": 3700
},
{
"grad_norm": 0.20534180104732513,
"learning_rate": 3.483773206947572e-05,
"loss": 0.0302,
"step": 3710
},
{
"grad_norm": 0.23959755897521973,
"learning_rate": 3.457536168078247e-05,
"loss": 0.026,
"step": 3720
},
{
"grad_norm": 0.3402751386165619,
"learning_rate": 3.431345985070067e-05,
"loss": 0.0227,
"step": 3730
},
{
"grad_norm": 0.322955459356308,
"learning_rate": 3.40520345350965e-05,
"loss": 0.0312,
"step": 3740
},
{
"grad_norm": 0.39298954606056213,
"learning_rate": 3.379109367536089e-05,
"loss": 0.0305,
"step": 3750
},
{
"grad_norm": 0.3251369297504425,
"learning_rate": 3.3530645198168295e-05,
"loss": 0.0257,
"step": 3760
},
{
"grad_norm": 0.19268028438091278,
"learning_rate": 3.327069701523595e-05,
"loss": 0.0327,
"step": 3770
},
{
"grad_norm": 0.16670925915241241,
"learning_rate": 3.301125702308353e-05,
"loss": 0.0243,
"step": 3780
},
{
"grad_norm": 0.2705236077308655,
"learning_rate": 3.275233310279321e-05,
"loss": 0.0224,
"step": 3790
},
{
"grad_norm": 0.2979815602302551,
"learning_rate": 3.249393311977037e-05,
"loss": 0.028,
"step": 3800
},
{
"grad_norm": 0.32814258337020874,
"learning_rate": 3.223606492350451e-05,
"loss": 0.0225,
"step": 3810
},
{
"grad_norm": 0.32424116134643555,
"learning_rate": 3.197873634733096e-05,
"loss": 0.0278,
"step": 3820
},
{
"grad_norm": 0.21570391952991486,
"learning_rate": 3.172195520819285e-05,
"loss": 0.0294,
"step": 3830
},
{
"grad_norm": 0.24287858605384827,
"learning_rate": 3.146572930640362e-05,
"loss": 0.0256,
"step": 3840
},
{
"grad_norm": 0.2868352234363556,
"learning_rate": 3.121006642541014e-05,
"loss": 0.0275,
"step": 3850
},
{
"grad_norm": 0.3127637803554535,
"learning_rate": 3.095497433155626e-05,
"loss": 0.0269,
"step": 3860
},
{
"grad_norm": 0.42665526270866394,
"learning_rate": 3.070046077384682e-05,
"loss": 0.0314,
"step": 3870
},
{
"grad_norm": 0.20559321343898773,
"learning_rate": 3.0446533483712304e-05,
"loss": 0.0275,
"step": 3880
},
{
"grad_norm": 0.3054938018321991,
"learning_rate": 3.0193200174774038e-05,
"loss": 0.0295,
"step": 3890
},
{
"grad_norm": 0.2588619589805603,
"learning_rate": 2.994046854260974e-05,
"loss": 0.0266,
"step": 3900
},
{
"grad_norm": 0.2679649889469147,
"learning_rate": 2.9688346264519866e-05,
"loss": 0.0244,
"step": 3910
},
{
"grad_norm": 0.39090123772621155,
"learning_rate": 2.943684099929436e-05,
"loss": 0.0244,
"step": 3920
},
{
"grad_norm": 0.24833203852176666,
"learning_rate": 2.918596038697995e-05,
"loss": 0.0244,
"step": 3930
},
{
"grad_norm": 0.3929983675479889,
"learning_rate": 2.8935712048648112e-05,
"loss": 0.0303,
"step": 3940
},
{
"grad_norm": 0.28877899050712585,
"learning_rate": 2.8686103586163626e-05,
"loss": 0.0308,
"step": 3950
},
{
"grad_norm": 0.23107615113258362,
"learning_rate": 2.843714258195346e-05,
"loss": 0.0278,
"step": 3960
},
{
"grad_norm": 0.19419795274734497,
"learning_rate": 2.8188836598776662e-05,
"loss": 0.0304,
"step": 3970
},
{
"grad_norm": 0.2506042420864105,
"learning_rate": 2.7941193179494484e-05,
"loss": 0.0252,
"step": 3980
},
{
"grad_norm": 0.24775511026382446,
"learning_rate": 2.7694219846841262e-05,
"loss": 0.0247,
"step": 3990
},
{
"grad_norm": 0.18015913665294647,
"learning_rate": 2.7447924103195976e-05,
"loss": 0.0304,
"step": 4000
},
{
"grad_norm": 0.19404101371765137,
"learning_rate": 2.7202313430354253e-05,
"loss": 0.0257,
"step": 4010
},
{
"grad_norm": 0.26239460706710815,
"learning_rate": 2.695739528930111e-05,
"loss": 0.0245,
"step": 4020
},
{
"grad_norm": 0.345069020986557,
"learning_rate": 2.67131771199844e-05,
"loss": 0.0256,
"step": 4030
},
{
"grad_norm": 0.12073345482349396,
"learning_rate": 2.6469666341088677e-05,
"loss": 0.0243,
"step": 4040
},
{
"grad_norm": 0.14277902245521545,
"learning_rate": 2.6226870349809885e-05,
"loss": 0.0241,
"step": 4050
},
{
"grad_norm": 0.3298734724521637,
"learning_rate": 2.5984796521630737e-05,
"loss": 0.0325,
"step": 4060
},
{
"grad_norm": 0.17974033951759338,
"learning_rate": 2.574345221009653e-05,
"loss": 0.0287,
"step": 4070
},
{
"grad_norm": 0.30501413345336914,
"learning_rate": 2.5502844746591804e-05,
"loss": 0.0225,
"step": 4080
},
{
"grad_norm": 0.1621364802122116,
"learning_rate": 2.526298144011775e-05,
"loss": 0.0229,
"step": 4090
},
{
"grad_norm": 0.27389562129974365,
"learning_rate": 2.5023869577070013e-05,
"loss": 0.0297,
"step": 4100
},
{
"grad_norm": 0.27940860390663147,
"learning_rate": 2.478551642101743e-05,
"loss": 0.0268,
"step": 4110
},
{
"grad_norm": 0.24268212914466858,
"learning_rate": 2.4547929212481435e-05,
"loss": 0.0278,
"step": 4120
},
{
"grad_norm": 0.20361125469207764,
"learning_rate": 2.4311115168716013e-05,
"loss": 0.026,
"step": 4130
},
{
"grad_norm": 0.24622249603271484,
"learning_rate": 2.4075081483488494e-05,
"loss": 0.0291,
"step": 4140
},
{
"grad_norm": 0.24639320373535156,
"learning_rate": 2.3839835326861104e-05,
"loss": 0.023,
"step": 4150
},
{
"grad_norm": 0.24470673501491547,
"learning_rate": 2.3605383844972966e-05,
"loss": 0.0179,
"step": 4160
},
{
"grad_norm": 0.22621630132198334,
"learning_rate": 2.3371734159823284e-05,
"loss": 0.0287,
"step": 4170
},
{
"grad_norm": 0.2869281470775604,
"learning_rate": 2.3138893369054766e-05,
"loss": 0.0279,
"step": 4180
},
{
"grad_norm": 0.214973583817482,
"learning_rate": 2.2906868545738102e-05,
"loss": 0.0264,
"step": 4190
},
{
"grad_norm": 0.27452945709228516,
"learning_rate": 2.2675666738157186e-05,
"loss": 0.0272,
"step": 4200
},
{
"grad_norm": 0.1707894653081894,
"learning_rate": 2.2445294969594844e-05,
"loss": 0.0209,
"step": 4210
},
{
"grad_norm": 0.210554301738739,
"learning_rate": 2.22157602381196e-05,
"loss": 0.0211,
"step": 4220
},
{
"grad_norm": 0.16127753257751465,
"learning_rate": 2.1987069516373098e-05,
"loss": 0.0213,
"step": 4230
},
{
"grad_norm": 0.2200268805027008,
"learning_rate": 2.1759229751358217e-05,
"loss": 0.0272,
"step": 4240
},
{
"grad_norm": 0.20734037458896637,
"learning_rate": 2.1532247864228084e-05,
"loss": 0.0265,
"step": 4250
},
{
"grad_norm": 0.19393886625766754,
"learning_rate": 2.1306130750075865e-05,
"loss": 0.0285,
"step": 4260
},
{
"grad_norm": 0.21821701526641846,
"learning_rate": 2.1080885277725236e-05,
"loss": 0.0232,
"step": 4270
},
{
"grad_norm": 0.1954386681318283,
"learning_rate": 2.085651828952175e-05,
"loss": 0.0219,
"step": 4280
},
{
"grad_norm": 0.2786208689212799,
"learning_rate": 2.063303660112506e-05,
"loss": 0.0222,
"step": 4290
},
{
"grad_norm": 0.2469235360622406,
"learning_rate": 2.0410447001301753e-05,
"loss": 0.0289,
"step": 4300
},
{
"grad_norm": 0.29244598746299744,
"learning_rate": 2.0188756251719203e-05,
"loss": 0.0211,
"step": 4310
},
{
"grad_norm": 0.24272142350673676,
"learning_rate": 1.9967971086740195e-05,
"loss": 0.0244,
"step": 4320
},
{
"grad_norm": 0.15965189039707184,
"learning_rate": 1.974809821321827e-05,
"loss": 0.0217,
"step": 4330
},
{
"grad_norm": 0.3541429042816162,
"learning_rate": 1.9529144310294023e-05,
"loss": 0.0216,
"step": 4340
},
{
"grad_norm": 0.40942439436912537,
"learning_rate": 1.9311116029192278e-05,
"loss": 0.0228,
"step": 4350
},
{
"grad_norm": 0.28426092863082886,
"learning_rate": 1.909401999301993e-05,
"loss": 0.0209,
"step": 4360
},
{
"grad_norm": 0.21738092601299286,
"learning_rate": 1.887786279656482e-05,
"loss": 0.0223,
"step": 4370
},
{
"grad_norm": 0.20878979563713074,
"learning_rate": 1.8662651006095387e-05,
"loss": 0.0205,
"step": 4380
},
{
"grad_norm": 0.28445255756378174,
"learning_rate": 1.8448391159161204e-05,
"loss": 0.0254,
"step": 4390
},
{
"grad_norm": 0.3564216196537018,
"learning_rate": 1.8235089764394408e-05,
"loss": 0.0347,
"step": 4400
},
{
"grad_norm": 0.23867842555046082,
"learning_rate": 1.8022753301311935e-05,
"loss": 0.0253,
"step": 4410
},
{
"grad_norm": 0.23468661308288574,
"learning_rate": 1.7811388220118707e-05,
"loss": 0.0252,
"step": 4420
},
{
"grad_norm": 0.20441173017024994,
"learning_rate": 1.7601000941511757e-05,
"loss": 0.0243,
"step": 4430
},
{
"grad_norm": 0.24065618216991425,
"learning_rate": 1.7391597856485083e-05,
"loss": 0.028,
"step": 4440
},
{
"grad_norm": 0.4456522762775421,
"learning_rate": 1.7183185326135543e-05,
"loss": 0.0205,
"step": 4450
},
{
"grad_norm": 0.2482612282037735,
"learning_rate": 1.6975769681469705e-05,
"loss": 0.0273,
"step": 4460
},
{
"grad_norm": 0.2664894759654999,
"learning_rate": 1.676935722321139e-05,
"loss": 0.0211,
"step": 4470
},
{
"grad_norm": 0.18440550565719604,
"learning_rate": 1.6563954221610355e-05,
"loss": 0.0227,
"step": 4480
},
{
"grad_norm": 0.3048134446144104,
"learning_rate": 1.6359566916251845e-05,
"loss": 0.0243,
"step": 4490
},
{
"grad_norm": 0.23461848497390747,
"learning_rate": 1.615620151586697e-05,
"loss": 0.0223,
"step": 4500
},
{
"grad_norm": 0.27218717336654663,
"learning_rate": 1.5953864198144135e-05,
"loss": 0.0308,
"step": 4510
},
{
"grad_norm": 0.17611953616142273,
"learning_rate": 1.5752561109541447e-05,
"loss": 0.0224,
"step": 4520
},
{
"grad_norm": 0.14511722326278687,
"learning_rate": 1.5552298365099882e-05,
"loss": 0.0264,
"step": 4530
},
{
"grad_norm": 0.19246408343315125,
"learning_rate": 1.5353082048257596e-05,
"loss": 0.0207,
"step": 4540
},
{
"grad_norm": 0.2850362956523895,
"learning_rate": 1.5154918210665148e-05,
"loss": 0.0208,
"step": 4550
},
{
"grad_norm": 0.3816230297088623,
"learning_rate": 1.4957812872001614e-05,
"loss": 0.0231,
"step": 4560
},
{
"grad_norm": 0.30083927512168884,
"learning_rate": 1.4761772019791748e-05,
"loss": 0.0247,
"step": 4570
},
{
"grad_norm": 0.23763714730739594,
"learning_rate": 1.4566801609224096e-05,
"loss": 0.0228,
"step": 4580
},
{
"grad_norm": 0.18760724365711212,
"learning_rate": 1.4372907562970079e-05,
"loss": 0.0223,
"step": 4590
},
{
"grad_norm": 0.22761359810829163,
"learning_rate": 1.4180095771004154e-05,
"loss": 0.0227,
"step": 4600
},
{
"grad_norm": 0.27833399176597595,
"learning_rate": 1.3988372090424773e-05,
"loss": 0.0245,
"step": 4610
},
{
"grad_norm": 0.18956084549427032,
"learning_rate": 1.3797742345276521e-05,
"loss": 0.0278,
"step": 4620
},
{
"grad_norm": 0.2752882242202759,
"learning_rate": 1.3608212326373249e-05,
"loss": 0.024,
"step": 4630
},
{
"grad_norm": 0.2357792854309082,
"learning_rate": 1.3419787791122062e-05,
"loss": 0.029,
"step": 4640
},
{
"grad_norm": 0.2950059771537781,
"learning_rate": 1.323247446334847e-05,
"loss": 0.0211,
"step": 4650
},
{
"grad_norm": 0.25425857305526733,
"learning_rate": 1.3046278033122577e-05,
"loss": 0.0327,
"step": 4660
},
{
"grad_norm": 0.17522060871124268,
"learning_rate": 1.286120415658611e-05,
"loss": 0.0231,
"step": 4670
},
{
"grad_norm": 0.17624273896217346,
"learning_rate": 1.2677258455780683e-05,
"loss": 0.0162,
"step": 4680
},
{
"grad_norm": 0.19895870983600616,
"learning_rate": 1.2494446518477022e-05,
"loss": 0.0262,
"step": 4690
},
{
"grad_norm": 0.2301178127527237,
"learning_rate": 1.2312773898005175e-05,
"loss": 0.0274,
"step": 4700
},
{
"grad_norm": 0.2145126610994339,
"learning_rate": 1.2132246113085822e-05,
"loss": 0.0201,
"step": 4710
},
{
"grad_norm": 0.2308410108089447,
"learning_rate": 1.1952868647662696e-05,
"loss": 0.0239,
"step": 4720
},
{
"grad_norm": 0.24893225729465485,
"learning_rate": 1.1774646950735913e-05,
"loss": 0.0214,
"step": 4730
},
{
"grad_norm": 0.23895396292209625,
"learning_rate": 1.1597586436196473e-05,
"loss": 0.0188,
"step": 4740
},
{
"grad_norm": 0.29286670684814453,
"learning_rate": 1.1421692482661856e-05,
"loss": 0.026,
"step": 4750
},
{
"grad_norm": 0.32094335556030273,
"learning_rate": 1.124697043331256e-05,
"loss": 0.0234,
"step": 4760
},
{
"grad_norm": 0.22018274664878845,
"learning_rate": 1.107342559572977e-05,
"loss": 0.0248,
"step": 4770
},
{
"grad_norm": 0.15834014117717743,
"learning_rate": 1.090106324173426e-05,
"loss": 0.0259,
"step": 4780
},
{
"grad_norm": 0.25905129313468933,
"learning_rate": 1.0729888607226113e-05,
"loss": 0.0194,
"step": 4790
},
{
"grad_norm": 0.3370433747768402,
"learning_rate": 1.0559906892025745e-05,
"loss": 0.0223,
"step": 4800
},
{
"grad_norm": 0.25827088952064514,
"learning_rate": 1.0391123259715906e-05,
"loss": 0.0254,
"step": 4810
},
{
"grad_norm": 0.18139545619487762,
"learning_rate": 1.0223542837484839e-05,
"loss": 0.0203,
"step": 4820
},
{
"grad_norm": 0.3228249251842499,
"learning_rate": 1.0057170715970559e-05,
"loss": 0.0264,
"step": 4830
},
{
"grad_norm": 0.19505833089351654,
"learning_rate": 9.892011949106172e-06,
"loss": 0.0307,
"step": 4840
},
{
"grad_norm": 0.18679122626781464,
"learning_rate": 9.728071553966339e-06,
"loss": 0.0239,
"step": 4850
},
{
"grad_norm": 0.22244933247566223,
"learning_rate": 9.56535451061496e-06,
"loss": 0.0228,
"step": 4860
},
{
"grad_norm": 0.1451004296541214,
"learning_rate": 9.403865761953779e-06,
"loss": 0.021,
"step": 4870
},
{
"grad_norm": 0.17116180062294006,
"learning_rate": 9.243610213572285e-06,
"loss": 0.0243,
"step": 4880
},
{
"grad_norm": 0.27246227860450745,
"learning_rate": 9.084592733598735e-06,
"loss": 0.0219,
"step": 4890
},
{
"grad_norm": 0.19050748646259308,
"learning_rate": 8.92681815255219e-06,
"loss": 0.0206,
"step": 4900
},
{
"grad_norm": 0.16780933737754822,
"learning_rate": 8.770291263195819e-06,
"loss": 0.0186,
"step": 4910
},
{
"grad_norm": 0.19002662599086761,
"learning_rate": 8.615016820391342e-06,
"loss": 0.0276,
"step": 4920
},
{
"grad_norm": 0.24216674268245697,
"learning_rate": 8.460999540954517e-06,
"loss": 0.0246,
"step": 4930
},
{
"grad_norm": 0.30049100518226624,
"learning_rate": 8.308244103511909e-06,
"loss": 0.0202,
"step": 4940
},
{
"grad_norm": 0.19304461777210236,
"learning_rate": 8.156755148358764e-06,
"loss": 0.0256,
"step": 4950
},
{
"grad_norm": 0.21739085018634796,
"learning_rate": 8.00653727731801e-06,
"loss": 0.0202,
"step": 4960
},
{
"grad_norm": 0.2507297396659851,
"learning_rate": 7.857595053600513e-06,
"loss": 0.0295,
"step": 4970
},
{
"grad_norm": 0.20464932918548584,
"learning_rate": 7.709933001666431e-06,
"loss": 0.0283,
"step": 4980
},
{
"grad_norm": 0.2696041762828827,
"learning_rate": 7.56355560708778e-06,
"loss": 0.0211,
"step": 4990
},
{
"grad_norm": 0.41251495480537415,
"learning_rate": 7.418467316412158e-06,
"loss": 0.0252,
"step": 5000
},
{
"grad_norm": 0.19311721622943878,
"learning_rate": 7.2746725370277435e-06,
"loss": 0.0236,
"step": 5010
},
{
"grad_norm": 0.21397510170936584,
"learning_rate": 7.132175637029293e-06,
"loss": 0.0211,
"step": 5020
},
{
"grad_norm": 0.17316733300685883,
"learning_rate": 6.9909809450855345e-06,
"loss": 0.0215,
"step": 5030
},
{
"grad_norm": 0.24864064157009125,
"learning_rate": 6.851092750307686e-06,
"loss": 0.0218,
"step": 5040
},
{
"grad_norm": 0.21637900173664093,
"learning_rate": 6.712515302119077e-06,
"loss": 0.0223,
"step": 5050
},
{
"grad_norm": 0.29354986548423767,
"learning_rate": 6.575252810126143e-06,
"loss": 0.0238,
"step": 5060
},
{
"grad_norm": 0.17130261659622192,
"learning_rate": 6.439309443990532e-06,
"loss": 0.0159,
"step": 5070
},
{
"grad_norm": 0.243094339966774,
"learning_rate": 6.304689333302416e-06,
"loss": 0.0271,
"step": 5080
},
{
"grad_norm": 0.12733832001686096,
"learning_rate": 6.171396567455051e-06,
"loss": 0.0244,
"step": 5090
},
{
"grad_norm": 0.14348167181015015,
"learning_rate": 6.039435195520604e-06,
"loss": 0.0235,
"step": 5100
},
{
"grad_norm": 0.13294343650341034,
"learning_rate": 5.908809226127054e-06,
"loss": 0.0215,
"step": 5110
},
{
"grad_norm": 0.20097094774246216,
"learning_rate": 5.779522627336537e-06,
"loss": 0.0329,
"step": 5120
},
{
"grad_norm": 0.18011382222175598,
"learning_rate": 5.651579326524709e-06,
"loss": 0.0229,
"step": 5130
},
{
"grad_norm": 0.21075014770030975,
"learning_rate": 5.524983210261481e-06,
"loss": 0.0219,
"step": 5140
},
{
"grad_norm": 0.16863515973091125,
"learning_rate": 5.399738124192988e-06,
"loss": 0.0183,
"step": 5150
},
{
"grad_norm": 0.2895853817462921,
"learning_rate": 5.2758478729247164e-06,
"loss": 0.0222,
"step": 5160
},
{
"grad_norm": 0.10782874375581741,
"learning_rate": 5.153316219905946e-06,
"loss": 0.0143,
"step": 5170
},
{
"grad_norm": 0.15944276750087738,
"learning_rate": 5.032146887315448e-06,
"loss": 0.0216,
"step": 5180
},
{
"grad_norm": 0.2914569079875946,
"learning_rate": 4.91234355594839e-06,
"loss": 0.0244,
"step": 5190
},
{
"grad_norm": 0.3361850678920746,
"learning_rate": 4.7939098651045235e-06,
"loss": 0.0275,
"step": 5200
},
{
"grad_norm": 0.2176862359046936,
"learning_rate": 4.67684941247768e-06,
"loss": 0.022,
"step": 5210
},
{
"grad_norm": 0.14250448346138,
"learning_rate": 4.5611657540464036e-06,
"loss": 0.0247,
"step": 5220
},
{
"grad_norm": 0.23687288165092468,
"learning_rate": 4.446862403965984e-06,
"loss": 0.0277,
"step": 5230
},
{
"grad_norm": 0.14389674365520477,
"learning_rate": 4.333942834461702e-06,
"loss": 0.0194,
"step": 5240
},
{
"grad_norm": 0.21724650263786316,
"learning_rate": 4.222410475723326e-06,
"loss": 0.0172,
"step": 5250
},
{
"grad_norm": 0.18676047027111053,
"learning_rate": 4.112268715800943e-06,
"loss": 0.0268,
"step": 5260
},
{
"grad_norm": 0.16382524371147156,
"learning_rate": 4.003520900502028e-06,
"loss": 0.0232,
"step": 5270
},
{
"grad_norm": 0.18612629175186157,
"learning_rate": 3.8961703332898e-06,
"loss": 0.0249,
"step": 5280
},
{
"grad_norm": 0.14936773478984833,
"learning_rate": 3.790220275182854e-06,
"loss": 0.022,
"step": 5290
},
{
"grad_norm": 0.12002617865800858,
"learning_rate": 3.685673944656176e-06,
"loss": 0.0219,
"step": 5300
},
{
"grad_norm": 0.25049668550491333,
"learning_rate": 3.582534517543268e-06,
"loss": 0.02,
"step": 5310
},
{
"grad_norm": 0.1867896318435669,
"learning_rate": 3.4808051269397512e-06,
"loss": 0.0222,
"step": 5320
},
{
"grad_norm": 0.1993655264377594,
"learning_rate": 3.380488863108183e-06,
"loss": 0.0186,
"step": 5330
},
{
"grad_norm": 0.31545430421829224,
"learning_rate": 3.2815887733841365e-06,
"loss": 0.0244,
"step": 5340
},
{
"grad_norm": 0.2718031406402588,
"learning_rate": 3.1841078620836683e-06,
"loss": 0.0194,
"step": 5350
},
{
"grad_norm": 0.15710008144378662,
"learning_rate": 3.0880490904120874e-06,
"loss": 0.0237,
"step": 5360
},
{
"grad_norm": 0.22318512201309204,
"learning_rate": 2.9934153763739205e-06,
"loss": 0.0251,
"step": 5370
},
{
"grad_norm": 0.11480679363012314,
"learning_rate": 2.9002095946843277e-06,
"loss": 0.019,
"step": 5380
},
{
"grad_norm": 0.18399132788181305,
"learning_rate": 2.8084345766817676e-06,
"loss": 0.0216,
"step": 5390
},
{
"grad_norm": 0.1431991308927536,
"learning_rate": 2.718093110241976e-06,
"loss": 0.0211,
"step": 5400
},
{
"grad_norm": 0.18595671653747559,
"learning_rate": 2.6291879396933004e-06,
"loss": 0.0183,
"step": 5410
},
{
"grad_norm": 0.14346589148044586,
"learning_rate": 2.541721765733318e-06,
"loss": 0.0227,
"step": 5420
},
{
"grad_norm": 0.322017639875412,
"learning_rate": 2.455697245346783e-06,
"loss": 0.0207,
"step": 5430
},
{
"grad_norm": 0.2633446156978607,
"learning_rate": 2.371116991724953e-06,
"loss": 0.0206,
"step": 5440
},
{
"grad_norm": 0.47103169560432434,
"learning_rate": 2.2879835741861586e-06,
"loss": 0.0284,
"step": 5450
},
{
"grad_norm": 0.15965288877487183,
"learning_rate": 2.206299518097804e-06,
"loss": 0.0177,
"step": 5460
},
{
"grad_norm": 0.15988339483737946,
"learning_rate": 2.1260673047996227e-06,
"loss": 0.0219,
"step": 5470
},
{
"grad_norm": 0.22732670605182648,
"learning_rate": 2.047289371528299e-06,
"loss": 0.0211,
"step": 5480
},
{
"grad_norm": 0.23306666314601898,
"learning_rate": 1.96996811134344e-06,
"loss": 0.0169,
"step": 5490
},
{
"grad_norm": 0.14578266441822052,
"learning_rate": 1.8941058730549132e-06,
"loss": 0.0197,
"step": 5500
},
{
"grad_norm": 0.14337080717086792,
"learning_rate": 1.8197049611514194e-06,
"loss": 0.0223,
"step": 5510
},
{
"grad_norm": 0.12350290268659592,
"learning_rate": 1.7467676357305561e-06,
"loss": 0.0177,
"step": 5520
},
{
"grad_norm": 0.1414731740951538,
"learning_rate": 1.6752961124301415e-06,
"loss": 0.0262,
"step": 5530
},
{
"grad_norm": 0.12613792717456818,
"learning_rate": 1.6052925623609049e-06,
"loss": 0.0174,
"step": 5540
},
{
"grad_norm": 0.24162399768829346,
"learning_rate": 1.5367591120405256e-06,
"loss": 0.0179,
"step": 5550
},
{
"grad_norm": 0.1481933444738388,
"learning_rate": 1.4696978433290653e-06,
"loss": 0.0187,
"step": 5560
},
{
"grad_norm": 0.2521955072879791,
"learning_rate": 1.4041107933656928e-06,
"loss": 0.0159,
"step": 5570
},
{
"grad_norm": 0.1632024645805359,
"learning_rate": 1.339999954506821e-06,
"loss": 0.0213,
"step": 5580
},
{
"grad_norm": 0.20339828729629517,
"learning_rate": 1.2773672742655784e-06,
"loss": 0.0215,
"step": 5590
},
{
"grad_norm": 0.26193174719810486,
"learning_rate": 1.2162146552526399e-06,
"loss": 0.0247,
"step": 5600
},
{
"grad_norm": 0.1698426753282547,
"learning_rate": 1.1565439551184664e-06,
"loss": 0.0234,
"step": 5610
},
{
"grad_norm": 0.2011450082063675,
"learning_rate": 1.0983569864968346e-06,
"loss": 0.0232,
"step": 5620
},
{
"grad_norm": 0.1681407243013382,
"learning_rate": 1.0416555169497688e-06,
"loss": 0.0206,
"step": 5630
},
{
"grad_norm": 0.23886294662952423,
"learning_rate": 9.864412689139123e-07,
"loss": 0.027,
"step": 5640
},
{
"grad_norm": 0.12609030306339264,
"learning_rate": 9.327159196481138e-07,
"loss": 0.0129,
"step": 5650
},
{
"grad_norm": 0.1660696566104889,
"learning_rate": 8.804811011825398e-07,
"loss": 0.0218,
"step": 5660
},
{
"grad_norm": 0.15882883965969086,
"learning_rate": 8.297384002690866e-07,
"loss": 0.0214,
"step": 5670
},
{
"grad_norm": 0.2871585190296173,
"learning_rate": 7.804893583331696e-07,
"loss": 0.021,
"step": 5680
},
{
"grad_norm": 0.2237343043088913,
"learning_rate": 7.32735471426893e-07,
"loss": 0.0241,
"step": 5690
},
{
"grad_norm": 0.19407829642295837,
"learning_rate": 6.864781901836259e-07,
"loss": 0.0179,
"step": 5700
},
{
"grad_norm": 0.24691540002822876,
"learning_rate": 6.417189197739093e-07,
"loss": 0.028,
"step": 5710
},
{
"grad_norm": 0.21609312295913696,
"learning_rate": 5.984590198627849e-07,
"loss": 0.026,
"step": 5720
},
{
"grad_norm": 0.17725086212158203,
"learning_rate": 5.566998045685112e-07,
"loss": 0.0263,
"step": 5730
},
{
"grad_norm": 0.1782742589712143,
"learning_rate": 5.164425424226016e-07,
"loss": 0.0185,
"step": 5740
},
{
"grad_norm": 0.20926135778427124,
"learning_rate": 4.776884563313266e-07,
"loss": 0.0229,
"step": 5750
},
{
"grad_norm": 0.2255379855632782,
"learning_rate": 4.404387235385443e-07,
"loss": 0.0221,
"step": 5760
},
{
"grad_norm": 0.1183241754770279,
"learning_rate": 4.0469447558995065e-07,
"loss": 0.0225,
"step": 5770
},
{
"grad_norm": 0.18785372376441956,
"learning_rate": 3.7045679829870175e-07,
"loss": 0.0239,
"step": 5780
},
{
"grad_norm": 0.1176566556096077,
"learning_rate": 3.377267317124233e-07,
"loss": 0.0238,
"step": 5790
},
{
"grad_norm": 0.128651425242424,
"learning_rate": 3.0650527008162513e-07,
"loss": 0.0188,
"step": 5800
},
{
"grad_norm": 0.17648927867412567,
"learning_rate": 2.767933618295082e-07,
"loss": 0.0201,
"step": 5810
},
{
"grad_norm": 0.1903354972600937,
"learning_rate": 2.485919095231326e-07,
"loss": 0.0193,
"step": 5820
},
{
"grad_norm": 0.1696813404560089,
"learning_rate": 2.219017698460002e-07,
"loss": 0.0256,
"step": 5830
},
{
"grad_norm": 0.2743062973022461,
"learning_rate": 1.9672375357206452e-07,
"loss": 0.0255,
"step": 5840
},
{
"grad_norm": 0.23269422352313995,
"learning_rate": 1.73058625541056e-07,
"loss": 0.0283,
"step": 5850
},
{
"grad_norm": 0.17723648250102997,
"learning_rate": 1.5090710463527836e-07,
"loss": 0.0213,
"step": 5860
},
{
"grad_norm": 0.13669823110103607,
"learning_rate": 1.3026986375776485e-07,
"loss": 0.022,
"step": 5870
},
{
"grad_norm": 0.22522103786468506,
"learning_rate": 1.1114752981183917e-07,
"loss": 0.0342,
"step": 5880
},
{
"grad_norm": 0.19673998653888702,
"learning_rate": 9.354068368204739e-08,
"loss": 0.0184,
"step": 5890
},
{
"grad_norm": 0.1417853683233261,
"learning_rate": 7.744986021656076e-08,
"loss": 0.0212,
"step": 5900
},
{
"grad_norm": 0.2924249768257141,
"learning_rate": 6.287554821087783e-08,
"loss": 0.0207,
"step": 5910
},
{
"grad_norm": 0.14325696229934692,
"learning_rate": 4.981819039300284e-08,
"loss": 0.0236,
"step": 5920
},
{
"grad_norm": 0.22200936079025269,
"learning_rate": 3.827818341000655e-08,
"loss": 0.0184,
"step": 5930
},
{
"grad_norm": 0.14072245359420776,
"learning_rate": 2.8255877815946963e-08,
"loss": 0.0194,
"step": 5940
},
{
"grad_norm": 0.24255603551864624,
"learning_rate": 1.9751578061244504e-08,
"loss": 0.0205,
"step": 5950
},
{
"grad_norm": 0.15497086942195892,
"learning_rate": 1.2765542483417214e-08,
"loss": 0.0208,
"step": 5960
},
{
"grad_norm": 0.14815428853034973,
"learning_rate": 7.2979832992592365e-09,
"loss": 0.0186,
"step": 5970
},
{
"grad_norm": 0.2799966335296631,
"learning_rate": 3.349066598362649e-09,
"loss": 0.0236,
"step": 5980
},
{
"grad_norm": 0.15595677495002747,
"learning_rate": 9.189123380826114e-10,
"loss": 0.0171,
"step": 5990
},
{
"grad_norm": 0.21225546300411224,
"learning_rate": 7.594339912486703e-12,
"loss": 0.0227,
"step": 6000
},
{
"grad_norm": 0.29790252447128296,
"learning_rate": 2.727592425543899e-05,
"loss": 0.021,
"step": 6010
},
{
"grad_norm": 0.36792364716529846,
"learning_rate": 2.7112429307339438e-05,
"loss": 0.0255,
"step": 6020
},
{
"grad_norm": 0.5832757353782654,
"learning_rate": 2.6949243365107413e-05,
"loss": 0.0252,
"step": 6030
},
{
"grad_norm": 0.23331348598003387,
"learning_rate": 2.6786368631921836e-05,
"loss": 0.0302,
"step": 6040
},
{
"grad_norm": 0.24127094447612762,
"learning_rate": 2.662380730675996e-05,
"loss": 0.0205,
"step": 6050
},
{
"grad_norm": 0.34142670035362244,
"learning_rate": 2.6461561584367787e-05,
"loss": 0.0259,
"step": 6060
},
{
"grad_norm": 0.19131368398666382,
"learning_rate": 2.629963365523031e-05,
"loss": 0.0239,
"step": 6070
},
{
"grad_norm": 0.30079951882362366,
"learning_rate": 2.6138025705541973e-05,
"loss": 0.0301,
"step": 6080
},
{
"grad_norm": 0.4083310067653656,
"learning_rate": 2.5976739917177167e-05,
"loss": 0.0244,
"step": 6090
},
{
"grad_norm": 0.3078618347644806,
"learning_rate": 2.5815778467660823e-05,
"loss": 0.0199,
"step": 6100
},
{
"grad_norm": 0.12846031785011292,
"learning_rate": 2.565514353013891e-05,
"loss": 0.0176,
"step": 6110
},
{
"grad_norm": 0.3428780138492584,
"learning_rate": 2.5494837273349136e-05,
"loss": 0.032,
"step": 6120
},
{
"grad_norm": 0.2583243250846863,
"learning_rate": 2.5334861861591753e-05,
"loss": 0.0275,
"step": 6130
},
{
"grad_norm": 0.18178141117095947,
"learning_rate": 2.517521945470021e-05,
"loss": 0.0218,
"step": 6140
},
{
"grad_norm": 0.40674087405204773,
"learning_rate": 2.5015912208012005e-05,
"loss": 0.0215,
"step": 6150
},
{
"grad_norm": 0.3071826994419098,
"learning_rate": 2.485694227233971e-05,
"loss": 0.0331,
"step": 6160
},
{
"grad_norm": 0.23079386353492737,
"learning_rate": 2.469831179394182e-05,
"loss": 0.0308,
"step": 6170
},
{
"grad_norm": 0.5017572641372681,
"learning_rate": 2.4540022914493764e-05,
"loss": 0.0328,
"step": 6180
},
{
"grad_norm": 0.554499626159668,
"learning_rate": 2.438207777105911e-05,
"loss": 0.0307,
"step": 6190
},
{
"grad_norm": 0.2705341875553131,
"learning_rate": 2.422447849606057e-05,
"loss": 0.0249,
"step": 6200
},
{
"grad_norm": 0.28618431091308594,
"learning_rate": 2.4067227217251302e-05,
"loss": 0.0296,
"step": 6210
},
{
"grad_norm": 0.29002895951271057,
"learning_rate": 2.3910326057686127e-05,
"loss": 0.0258,
"step": 6220
},
{
"grad_norm": 0.3113899528980255,
"learning_rate": 2.3753777135692974e-05,
"loss": 0.0214,
"step": 6230
},
{
"grad_norm": 0.33459651470184326,
"learning_rate": 2.359758256484409e-05,
"loss": 0.0242,
"step": 6240
},
{
"grad_norm": 0.22960981726646423,
"learning_rate": 2.344174445392766e-05,
"loss": 0.0252,
"step": 6250
},
{
"grad_norm": 0.23525279760360718,
"learning_rate": 2.3286264906919348e-05,
"loss": 0.0262,
"step": 6260
},
{
"grad_norm": 0.21416305005550385,
"learning_rate": 2.3131146022953765e-05,
"loss": 0.026,
"step": 6270
},
{
"grad_norm": 0.31340327858924866,
"learning_rate": 2.2976389896296203e-05,
"loss": 0.0261,
"step": 6280
},
{
"grad_norm": 0.2814219892024994,
"learning_rate": 2.28219986163144e-05,
"loss": 0.0293,
"step": 6290
},
{
"grad_norm": 0.2639205753803253,
"learning_rate": 2.2667974267450253e-05,
"loss": 0.023,
"step": 6300
},
{
"grad_norm": 0.3291344940662384,
"learning_rate": 2.251431892919171e-05,
"loss": 0.0272,
"step": 6310
},
{
"grad_norm": 0.28297552466392517,
"learning_rate": 2.2361034676044677e-05,
"loss": 0.0234,
"step": 6320
},
{
"grad_norm": 0.3307504653930664,
"learning_rate": 2.2208123577505113e-05,
"loss": 0.0226,
"step": 6330
},
{
"grad_norm": 0.20939522981643677,
"learning_rate": 2.2055587698030877e-05,
"loss": 0.0227,
"step": 6340
},
{
"grad_norm": 0.36284714937210083,
"learning_rate": 2.1903429097014055e-05,
"loss": 0.02,
"step": 6350
},
{
"grad_norm": 0.17910777032375336,
"learning_rate": 2.1751649828753106e-05,
"loss": 0.0269,
"step": 6360
},
{
"grad_norm": 0.3476734161376953,
"learning_rate": 2.1600251942425066e-05,
"loss": 0.0275,
"step": 6370
},
{
"grad_norm": 0.19914166629314423,
"learning_rate": 2.1449237482057898e-05,
"loss": 0.0266,
"step": 6380
},
{
"grad_norm": 0.29321420192718506,
"learning_rate": 2.129860848650299e-05,
"loss": 0.0281,
"step": 6390
},
{
"grad_norm": 0.272433340549469,
"learning_rate": 2.1148366989407496e-05,
"loss": 0.0249,
"step": 6400
},
{
"grad_norm": 0.22041699290275574,
"learning_rate": 2.0998515019186967e-05,
"loss": 0.0285,
"step": 6410
},
{
"grad_norm": 0.2994774878025055,
"learning_rate": 2.0849054598997896e-05,
"loss": 0.0259,
"step": 6420
},
{
"grad_norm": 0.19432491064071655,
"learning_rate": 2.0699987746710554e-05,
"loss": 0.0195,
"step": 6430
},
{
"grad_norm": 0.35682758688926697,
"learning_rate": 2.0551316474881498e-05,
"loss": 0.0227,
"step": 6440
},
{
"grad_norm": 0.3229887783527374,
"learning_rate": 2.040304279072661e-05,
"loss": 0.0184,
"step": 6450
},
{
"grad_norm": 0.2759677469730377,
"learning_rate": 2.0255168696093968e-05,
"loss": 0.0201,
"step": 6460
},
{
"grad_norm": 0.27762091159820557,
"learning_rate": 2.010769618743671e-05,
"loss": 0.0269,
"step": 6470
},
{
"grad_norm": 0.3582492768764496,
"learning_rate": 1.9960627255786153e-05,
"loss": 0.021,
"step": 6480
},
{
"grad_norm": 0.2863319516181946,
"learning_rate": 1.981396388672496e-05,
"loss": 0.0219,
"step": 6490
},
{
"grad_norm": 0.3060336709022522,
"learning_rate": 1.9667708060360242e-05,
"loss": 0.0225,
"step": 6500
},
{
"grad_norm": 0.27475908398628235,
"learning_rate": 1.952186175129685e-05,
"loss": 0.025,
"step": 6510
},
{
"grad_norm": 0.2548336088657379,
"learning_rate": 1.937642692861076e-05,
"loss": 0.0294,
"step": 6520
},
{
"grad_norm": 0.21055921912193298,
"learning_rate": 1.9231405555822428e-05,
"loss": 0.0249,
"step": 6530
},
{
"grad_norm": 0.19206783175468445,
"learning_rate": 1.908679959087033e-05,
"loss": 0.0225,
"step": 6540
},
{
"grad_norm": 0.3079420030117035,
"learning_rate": 1.8942610986084486e-05,
"loss": 0.0237,
"step": 6550
},
{
"grad_norm": 0.18333545327186584,
"learning_rate": 1.8798841688160157e-05,
"loss": 0.0241,
"step": 6560
},
{
"grad_norm": 0.20045840740203857,
"learning_rate": 1.86554936381315e-05,
"loss": 0.0276,
"step": 6570
},
{
"grad_norm": 0.1402873545885086,
"learning_rate": 1.851256877134538e-05,
"loss": 0.0344,
"step": 6580
},
{
"grad_norm": 0.3743952512741089,
"learning_rate": 1.8370069017435286e-05,
"loss": 0.0247,
"step": 6590
},
{
"grad_norm": 0.20628072321414948,
"learning_rate": 1.8227996300295224e-05,
"loss": 0.0227,
"step": 6600
},
{
"grad_norm": 0.33474627137184143,
"learning_rate": 1.808635253805376e-05,
"loss": 0.0274,
"step": 6610
},
{
"grad_norm": 0.23404568433761597,
"learning_rate": 1.7945139643048113e-05,
"loss": 0.0233,
"step": 6620
},
{
"grad_norm": 0.23489069938659668,
"learning_rate": 1.780435952179838e-05,
"loss": 0.0303,
"step": 6630
},
{
"grad_norm": 0.272388756275177,
"learning_rate": 1.7664014074981742e-05,
"loss": 0.0265,
"step": 6640
},
{
"grad_norm": 0.17611105740070343,
"learning_rate": 1.7524105197406816e-05,
"loss": 0.0216,
"step": 6650
},
{
"grad_norm": 0.22890515625476837,
"learning_rate": 1.7384634777988133e-05,
"loss": 0.0205,
"step": 6660
},
{
"grad_norm": 0.2990550100803375,
"learning_rate": 1.7245604699720535e-05,
"loss": 0.0236,
"step": 6670
},
{
"grad_norm": 0.28004226088523865,
"learning_rate": 1.7107016839653804e-05,
"loss": 0.0255,
"step": 6680
},
{
"grad_norm": 0.3143727481365204,
"learning_rate": 1.696887306886737e-05,
"loss": 0.0235,
"step": 6690
},
{
"grad_norm": 0.27975836396217346,
"learning_rate": 1.6831175252444943e-05,
"loss": 0.0225,
"step": 6700
},
{
"grad_norm": 0.2756935656070709,
"learning_rate": 1.6693925249449412e-05,
"loss": 0.0213,
"step": 6710
},
{
"grad_norm": 0.34417200088500977,
"learning_rate": 1.6557124912897714e-05,
"loss": 0.0263,
"step": 6720
},
{
"grad_norm": 0.4200408458709717,
"learning_rate": 1.6420776089735827e-05,
"loss": 0.025,
"step": 6730
},
{
"grad_norm": 0.20470227301120758,
"learning_rate": 1.6284880620813848e-05,
"loss": 0.0285,
"step": 6740
},
{
"grad_norm": 0.18324626982212067,
"learning_rate": 1.6149440340861072e-05,
"loss": 0.0213,
"step": 6750
},
{
"grad_norm": 0.2063898742198944,
"learning_rate": 1.6014457078461353e-05,
"loss": 0.0264,
"step": 6760
},
{
"grad_norm": 0.2921772003173828,
"learning_rate": 1.587993265602827e-05,
"loss": 0.0178,
"step": 6770
},
{
"grad_norm": 0.19587600231170654,
"learning_rate": 1.574586888978059e-05,
"loss": 0.0276,
"step": 6780
},
{
"grad_norm": 0.2335367649793625,
"learning_rate": 1.5612267589717805e-05,
"loss": 0.0233,
"step": 6790
},
{
"grad_norm": 0.26285460591316223,
"learning_rate": 1.5479130559595574e-05,
"loss": 0.0306,
"step": 6800
},
{
"grad_norm": 0.19542640447616577,
"learning_rate": 1.5346459596901445e-05,
"loss": 0.0183,
"step": 6810
},
{
"grad_norm": 0.2286476492881775,
"learning_rate": 1.5214256492830598e-05,
"loss": 0.0192,
"step": 6820
},
{
"grad_norm": 0.19746887683868408,
"learning_rate": 1.5082523032261602e-05,
"loss": 0.019,
"step": 6830
},
{
"grad_norm": 0.29165586829185486,
"learning_rate": 1.4951260993732385e-05,
"loss": 0.0213,
"step": 6840
},
{
"grad_norm": 0.21441499888896942,
"learning_rate": 1.4820472149416154e-05,
"loss": 0.0167,
"step": 6850
},
{
"grad_norm": 0.2882816791534424,
"learning_rate": 1.4690158265097564e-05,
"loss": 0.0273,
"step": 6860
},
{
"grad_norm": 0.2803361713886261,
"learning_rate": 1.4560321100148772e-05,
"loss": 0.0274,
"step": 6870
},
{
"grad_norm": 0.3463282883167267,
"learning_rate": 1.443096240750571e-05,
"loss": 0.029,
"step": 6880
},
{
"grad_norm": 0.21605895459651947,
"learning_rate": 1.4302083933644523e-05,
"loss": 0.0248,
"step": 6890
},
{
"grad_norm": 0.257943332195282,
"learning_rate": 1.4173687418557818e-05,
"loss": 0.0226,
"step": 6900
},
{
"grad_norm": 0.33272460103034973,
"learning_rate": 1.4045774595731315e-05,
"loss": 0.0255,
"step": 6910
},
{
"grad_norm": 0.24251773953437805,
"learning_rate": 1.3918347192120351e-05,
"loss": 0.0294,
"step": 6920
},
{
"grad_norm": 0.1468239277601242,
"learning_rate": 1.3791406928126638e-05,
"loss": 0.0201,
"step": 6930
},
{
"grad_norm": 0.21084365248680115,
"learning_rate": 1.3664955517574968e-05,
"loss": 0.0174,
"step": 6940
},
{
"grad_norm": 0.20665758848190308,
"learning_rate": 1.3538994667690108e-05,
"loss": 0.0234,
"step": 6950
},
{
"grad_norm": 0.2259311079978943,
"learning_rate": 1.3413526079073796e-05,
"loss": 0.0257,
"step": 6960
},
{
"grad_norm": 0.2823341488838196,
"learning_rate": 1.328855144568168e-05,
"loss": 0.0212,
"step": 6970
},
{
"grad_norm": 0.16559848189353943,
"learning_rate": 1.3164072454800507e-05,
"loss": 0.0212,
"step": 6980
},
{
"grad_norm": 0.22839784622192383,
"learning_rate": 1.3040090787025378e-05,
"loss": 0.0278,
"step": 6990
},
{
"grad_norm": 0.20402885973453522,
"learning_rate": 1.2916608116236977e-05,
"loss": 0.0228,
"step": 7000
},
{
"grad_norm": 0.393628865480423,
"learning_rate": 1.2793626109579033e-05,
"loss": 0.0213,
"step": 7010
},
{
"grad_norm": 0.2810872495174408,
"learning_rate": 1.2671146427435775e-05,
"loss": 0.0165,
"step": 7020
},
{
"grad_norm": 0.3044673502445221,
"learning_rate": 1.2549170723409549e-05,
"loss": 0.0233,
"step": 7030
},
{
"grad_norm": 0.1715988963842392,
"learning_rate": 1.2427700644298462e-05,
"loss": 0.0261,
"step": 7040
},
{
"grad_norm": 0.22636812925338745,
"learning_rate": 1.2306737830074162e-05,
"loss": 0.0259,
"step": 7050
},
{
"grad_norm": 0.2515609562397003,
"learning_rate": 1.2186283913859726e-05,
"loss": 0.022,
"step": 7060
},
{
"grad_norm": 0.2545802593231201,
"learning_rate": 1.2066340521907549e-05,
"loss": 0.0248,
"step": 7070
},
{
"grad_norm": 0.20971673727035522,
"learning_rate": 1.1946909273577405e-05,
"loss": 0.0174,
"step": 7080
},
{
"grad_norm": 0.2228962481021881,
"learning_rate": 1.1827991781314667e-05,
"loss": 0.0175,
"step": 7090
},
{
"grad_norm": 0.176900252699852,
"learning_rate": 1.1709589650628411e-05,
"loss": 0.0207,
"step": 7100
},
{
"grad_norm": 0.21749795973300934,
"learning_rate": 1.159170448006981e-05,
"loss": 0.0206,
"step": 7110
},
{
"grad_norm": 0.34848442673683167,
"learning_rate": 1.1474337861210543e-05,
"loss": 0.0238,
"step": 7120
},
{
"grad_norm": 0.16689461469650269,
"learning_rate": 1.1357491378621321e-05,
"loss": 0.0214,
"step": 7130
},
{
"grad_norm": 0.2874811589717865,
"learning_rate": 1.1241166609850456e-05,
"loss": 0.0233,
"step": 7140
},
{
"grad_norm": 0.17507481575012207,
"learning_rate": 1.1125365125402582e-05,
"loss": 0.0212,
"step": 7150
},
{
"grad_norm": 0.20651166141033173,
"learning_rate": 1.101008848871749e-05,
"loss": 0.026,
"step": 7160
},
{
"grad_norm": 0.23986105620861053,
"learning_rate": 1.0895338256148946e-05,
"loss": 0.0218,
"step": 7170
},
{
"grad_norm": 0.347660630941391,
"learning_rate": 1.0781115976943717e-05,
"loss": 0.025,
"step": 7180
},
{
"grad_norm": 0.3137768507003784,
"learning_rate": 1.0667423193220689e-05,
"loss": 0.0327,
"step": 7190
},
{
"grad_norm": 0.23858413100242615,
"learning_rate": 1.0554261439949975e-05,
"loss": 0.0213,
"step": 7200
},
{
"grad_norm": 0.2236446738243103,
"learning_rate": 1.0441632244932237e-05,
"loss": 0.0152,
"step": 7210
},
{
"grad_norm": 0.27083760499954224,
"learning_rate": 1.032953712877806e-05,
"loss": 0.0281,
"step": 7220
},
{
"grad_norm": 0.14618238806724548,
"learning_rate": 1.0217977604887408e-05,
"loss": 0.0189,
"step": 7230
},
{
"grad_norm": 0.16847121715545654,
"learning_rate": 1.0106955179429183e-05,
"loss": 0.0228,
"step": 7240
},
{
"grad_norm": 0.1947457641363144,
"learning_rate": 9.996471351320907e-06,
"loss": 0.0202,
"step": 7250
},
{
"grad_norm": 0.26655104756355286,
"learning_rate": 9.886527612208512e-06,
"loss": 0.0205,
"step": 7260
},
{
"grad_norm": 0.31789079308509827,
"learning_rate": 9.777125446446133e-06,
"loss": 0.0216,
"step": 7270
},
{
"grad_norm": 0.3424108624458313,
"learning_rate": 9.668266331076103e-06,
"loss": 0.0188,
"step": 7280
},
{
"grad_norm": 0.15899164974689484,
"learning_rate": 9.559951735809059e-06,
"loss": 0.0192,
"step": 7290
},
{
"grad_norm": 0.14476493000984192,
"learning_rate": 9.452183123004e-06,
"loss": 0.0235,
"step": 7300
},
{
"grad_norm": 0.14348042011260986,
"learning_rate": 9.344961947648623e-06,
"loss": 0.0225,
"step": 7310
},
{
"grad_norm": 0.3830648362636566,
"learning_rate": 9.238289657339649e-06,
"loss": 0.0223,
"step": 7320
},
{
"grad_norm": 0.20327335596084595,
"learning_rate": 9.132167692263289e-06,
"loss": 0.0241,
"step": 7330
},
{
"grad_norm": 0.1945178508758545,
"learning_rate": 9.02659748517578e-06,
"loss": 0.0234,
"step": 7340
},
{
"grad_norm": 0.1814965158700943,
"learning_rate": 8.921580461384066e-06,
"loss": 0.0171,
"step": 7350
},
{
"grad_norm": 0.18689940869808197,
"learning_rate": 8.817118038726558e-06,
"loss": 0.0209,
"step": 7360
},
{
"grad_norm": 0.2710418105125427,
"learning_rate": 8.713211627553958e-06,
"loss": 0.0196,
"step": 7370
},
{
"grad_norm": 0.1881471872329712,
"learning_rate": 8.609862630710231e-06,
"loss": 0.022,
"step": 7380
},
{
"grad_norm": 0.14768840372562408,
"learning_rate": 8.507072443513702e-06,
"loss": 0.0227,
"step": 7390
},
{
"grad_norm": 0.2851825952529907,
"learning_rate": 8.404842453738172e-06,
"loss": 0.0211,
"step": 7400
},
{
"grad_norm": 0.341522216796875,
"learning_rate": 8.303174041594192e-06,
"loss": 0.0198,
"step": 7410
},
{
"grad_norm": 0.16477511823177338,
"learning_rate": 8.202068579710431e-06,
"loss": 0.0229,
"step": 7420
},
{
"grad_norm": 0.17231206595897675,
"learning_rate": 8.101527433115169e-06,
"loss": 0.0202,
"step": 7430
},
{
"grad_norm": 0.13956135511398315,
"learning_rate": 8.001551959217829e-06,
"loss": 0.0215,
"step": 7440
},
{
"grad_norm": 0.14954547584056854,
"learning_rate": 7.902143507790661e-06,
"loss": 0.0189,
"step": 7450
},
{
"grad_norm": 0.16716261208057404,
"learning_rate": 7.803303420950554e-06,
"loss": 0.0227,
"step": 7460
},
{
"grad_norm": 0.195988267660141,
"learning_rate": 7.705033033140857e-06,
"loss": 0.0213,
"step": 7470
},
{
"grad_norm": 0.2251002937555313,
"learning_rate": 7.607333671113409e-06,
"loss": 0.0204,
"step": 7480
},
{
"grad_norm": 0.2585914134979248,
"learning_rate": 7.510206653910612e-06,
"loss": 0.0241,
"step": 7490
},
{
"grad_norm": 0.18494385480880737,
"learning_rate": 7.413653292847617e-06,
"loss": 0.0186,
"step": 7500
},
{
"grad_norm": 0.2994157671928406,
"learning_rate": 7.317674891494625e-06,
"loss": 0.019,
"step": 7510
},
{
"grad_norm": 0.17607255280017853,
"learning_rate": 7.222272745659275e-06,
"loss": 0.0226,
"step": 7520
},
{
"grad_norm": 0.23287427425384521,
"learning_rate": 7.1274481433691755e-06,
"loss": 0.0243,
"step": 7530
},
{
"grad_norm": 0.09976483136415482,
"learning_rate": 7.0332023648544965e-06,
"loss": 0.0173,
"step": 7540
},
{
"grad_norm": 0.2103106826543808,
"learning_rate": 6.939536682530678e-06,
"loss": 0.0159,
"step": 7550
},
{
"grad_norm": 0.3099997043609619,
"learning_rate": 6.846452360981292e-06,
"loss": 0.0199,
"step": 7560
},
{
"grad_norm": 0.2214270383119583,
"learning_rate": 6.753950656940905e-06,
"loss": 0.0212,
"step": 7570
},
{
"grad_norm": 0.20791272819042206,
"learning_rate": 6.662032819278147e-06,
"loss": 0.025,
"step": 7580
},
{
"grad_norm": 0.20438748598098755,
"learning_rate": 6.570700088978876e-06,
"loss": 0.0225,
"step": 7590
},
{
"grad_norm": 0.23018181324005127,
"learning_rate": 6.479953699129382e-06,
"loss": 0.026,
"step": 7600
},
{
"grad_norm": 0.17940333485603333,
"learning_rate": 6.3897948748997345e-06,
"loss": 0.0183,
"step": 7610
},
{
"grad_norm": 0.1848093569278717,
"learning_rate": 6.300224833527285e-06,
"loss": 0.0187,
"step": 7620
},
{
"grad_norm": 0.18162129819393158,
"learning_rate": 6.211244784300197e-06,
"loss": 0.0204,
"step": 7630
},
{
"grad_norm": 0.20585574209690094,
"learning_rate": 6.1228559285411216e-06,
"loss": 0.0243,
"step": 7640
},
{
"grad_norm": 0.15938639640808105,
"learning_rate": 6.0350594595909914e-06,
"loss": 0.0239,
"step": 7650
},
{
"grad_norm": 0.16458004713058472,
"learning_rate": 5.947856562792925e-06,
"loss": 0.0183,
"step": 7660
},
{
"grad_norm": 0.1435135304927826,
"learning_rate": 5.861248415476184e-06,
"loss": 0.025,
"step": 7670
},
{
"grad_norm": 0.18994653224945068,
"learning_rate": 5.775236186940286e-06,
"loss": 0.0181,
"step": 7680
},
{
"grad_norm": 0.14257881045341492,
"learning_rate": 5.689821038439263e-06,
"loss": 0.0179,
"step": 7690
},
{
"grad_norm": 0.22278554737567902,
"learning_rate": 5.605004123165919e-06,
"loss": 0.0203,
"step": 7700
},
{
"grad_norm": 0.20522241294384003,
"learning_rate": 5.5207865862363035e-06,
"loss": 0.0222,
"step": 7710
},
{
"grad_norm": 0.2745155990123749,
"learning_rate": 5.437169564674233e-06,
"loss": 0.0179,
"step": 7720
},
{
"grad_norm": 0.10246184468269348,
"learning_rate": 5.354154187395954e-06,
"loss": 0.0276,
"step": 7730
},
{
"grad_norm": 0.16371352970600128,
"learning_rate": 5.271741575194883e-06,
"loss": 0.0169,
"step": 7740
},
{
"grad_norm": 0.1587575376033783,
"learning_rate": 5.1899328407264855e-06,
"loss": 0.0208,
"step": 7750
},
{
"grad_norm": 0.16685296595096588,
"learning_rate": 5.108729088493269e-06,
"loss": 0.0204,
"step": 7760
},
{
"grad_norm": 0.22429557144641876,
"learning_rate": 5.028131414829845e-06,
"loss": 0.0144,
"step": 7770
},
{
"grad_norm": 0.20360174775123596,
"learning_rate": 4.948140907888121e-06,
"loss": 0.0233,
"step": 7780
},
{
"grad_norm": 0.24377571046352386,
"learning_rate": 4.868758647622657e-06,
"loss": 0.021,
"step": 7790
},
{
"grad_norm": 0.19975610077381134,
"learning_rate": 4.78998570577604e-06,
"loss": 0.02,
"step": 7800
},
{
"grad_norm": 0.24063648283481598,
"learning_rate": 4.711823145864419e-06,
"loss": 0.0233,
"step": 7810
},
{
"grad_norm": 0.245465949177742,
"learning_rate": 4.634272023163166e-06,
"loss": 0.0249,
"step": 7820
},
{
"grad_norm": 0.10845431685447693,
"learning_rate": 4.557333384692619e-06,
"loss": 0.0281,
"step": 7830
},
{
"grad_norm": 0.5066738724708557,
"learning_rate": 4.48100826920394e-06,
"loss": 0.0235,
"step": 7840
},
{
"grad_norm": 0.1477183997631073,
"learning_rate": 4.40529770716509e-06,
"loss": 0.0182,
"step": 7850
},
{
"grad_norm": 0.20134125649929047,
"learning_rate": 4.330202720746956e-06,
"loss": 0.0168,
"step": 7860
},
{
"grad_norm": 0.32067152857780457,
"learning_rate": 4.255724323809479e-06,
"loss": 0.0204,
"step": 7870
},
{
"grad_norm": 0.16830016672611237,
"learning_rate": 4.181863521888019e-06,
"loss": 0.0217,
"step": 7880
},
{
"grad_norm": 0.14914914965629578,
"learning_rate": 4.108621312179772e-06,
"loss": 0.0177,
"step": 7890
},
{
"grad_norm": 0.3066643178462982,
"learning_rate": 4.03599868353029e-06,
"loss": 0.0158,
"step": 7900
},
{
"grad_norm": 0.21244697272777557,
"learning_rate": 3.963996616420146e-06,
"loss": 0.0274,
"step": 7910
},
{
"grad_norm": 0.11710469424724579,
"learning_rate": 3.892616082951689e-06,
"loss": 0.0213,
"step": 7920
},
{
"grad_norm": 0.4161364734172821,
"learning_rate": 3.821858046835913e-06,
"loss": 0.0193,
"step": 7930
},
{
"grad_norm": 0.2539246082305908,
"learning_rate": 3.751723463379464e-06,
"loss": 0.0197,
"step": 7940
},
{
"grad_norm": 0.14983807504177094,
"learning_rate": 3.6822132794717246e-06,
"loss": 0.0235,
"step": 7950
},
{
"grad_norm": 0.29121822118759155,
"learning_rate": 3.6133284335720605e-06,
"loss": 0.028,
"step": 7960
},
{
"grad_norm": 0.24540068209171295,
"learning_rate": 3.5450698556970973e-06,
"loss": 0.0248,
"step": 7970
},
{
"grad_norm": 0.2647874355316162,
"learning_rate": 3.4774384674082007e-06,
"loss": 0.0254,
"step": 7980
},
{
"grad_norm": 0.12534739077091217,
"learning_rate": 3.41043518179906e-06,
"loss": 0.0227,
"step": 7990
},
{
"grad_norm": 0.10827510803937912,
"learning_rate": 3.3440609034832815e-06,
"loss": 0.0236,
"step": 8000
},
{
"grad_norm": 0.15810130536556244,
"learning_rate": 3.278316528582259e-06,
"loss": 0.0243,
"step": 8010
},
{
"grad_norm": 0.13566693663597107,
"learning_rate": 3.213202944713023e-06,
"loss": 0.0202,
"step": 8020
},
{
"grad_norm": 0.10108888894319534,
"learning_rate": 3.1487210309762706e-06,
"loss": 0.0204,
"step": 8030
},
{
"grad_norm": 0.15295878052711487,
"learning_rate": 3.0848716579445102e-06,
"loss": 0.0165,
"step": 8040
},
{
"grad_norm": 0.15363246202468872,
"learning_rate": 3.021655687650282e-06,
"loss": 0.0177,
"step": 8050
},
{
"grad_norm": 0.17233219742774963,
"learning_rate": 2.9590739735745553e-06,
"loss": 0.0244,
"step": 8060
},
{
"grad_norm": 0.202011838555336,
"learning_rate": 2.8971273606351658e-06,
"loss": 0.0201,
"step": 8070
},
{
"grad_norm": 0.18432720005512238,
"learning_rate": 2.8358166851754297e-06,
"loss": 0.0221,
"step": 8080
},
{
"grad_norm": 0.1623450666666031,
"learning_rate": 2.7751427749528658e-06,
"loss": 0.0168,
"step": 8090
},
{
"grad_norm": 0.1833294779062271,
"learning_rate": 2.7151064491279855e-06,
"loss": 0.0153,
"step": 8100
},
{
"grad_norm": 0.1472567915916443,
"learning_rate": 2.6557085182532582e-06,
"loss": 0.0208,
"step": 8110
},
{
"grad_norm": 0.08988528698682785,
"learning_rate": 2.5969497842621595e-06,
"loss": 0.0228,
"step": 8120
},
{
"grad_norm": 0.11376199126243591,
"learning_rate": 2.5388310404583504e-06,
"loss": 0.0145,
"step": 8130
},
{
"grad_norm": 0.19667993485927582,
"learning_rate": 2.481353071504966e-06,
"loss": 0.0233,
"step": 8140
},
{
"grad_norm": 0.21620571613311768,
"learning_rate": 2.424516653413994e-06,
"loss": 0.0217,
"step": 8150
},
{
"grad_norm": 0.13917125761508942,
"learning_rate": 2.368322553535862e-06,
"loss": 0.0219,
"step": 8160
},
{
"grad_norm": 0.18975970149040222,
"learning_rate": 2.3127715305490073e-06,
"loss": 0.02,
"step": 8170
},
{
"grad_norm": 0.13851851224899292,
"learning_rate": 2.257864334449672e-06,
"loss": 0.0223,
"step": 8180
},
{
"grad_norm": 0.15608343482017517,
"learning_rate": 2.20360170654178e-06,
"loss": 0.0224,
"step": 8190
},
{
"grad_norm": 0.1645740568637848,
"learning_rate": 2.149984379426906e-06,
"loss": 0.0209,
"step": 8200
},
{
"grad_norm": 0.26837170124053955,
"learning_rate": 2.0970130769943917e-06,
"loss": 0.0183,
"step": 8210
},
{
"grad_norm": 0.1264772266149521,
"learning_rate": 2.0446885144115912e-06,
"loss": 0.0246,
"step": 8220
},
{
"grad_norm": 0.29881325364112854,
"learning_rate": 1.9930113981142028e-06,
"loss": 0.0257,
"step": 8230
},
{
"grad_norm": 0.13800419867038727,
"learning_rate": 1.9419824257967233e-06,
"loss": 0.0166,
"step": 8240
},
{
"grad_norm": 0.19800323247909546,
"learning_rate": 1.891602286403027e-06,
"loss": 0.017,
"step": 8250
},
{
"grad_norm": 0.1734076291322708,
"learning_rate": 1.841871660117095e-06,
"loss": 0.0204,
"step": 8260
},
{
"grad_norm": 0.2505193054676056,
"learning_rate": 1.79279121835379e-06,
"loss": 0.0303,
"step": 8270
},
{
"grad_norm": 0.1082741841673851,
"learning_rate": 1.7443616237498183e-06,
"loss": 0.0182,
"step": 8280
},
{
"grad_norm": 0.20857660472393036,
"learning_rate": 1.696583530154794e-06,
"loss": 0.0211,
"step": 8290
},
{
"grad_norm": 0.15800972282886505,
"learning_rate": 1.6494575826223656e-06,
"loss": 0.0154,
"step": 8300
},
{
"grad_norm": 0.2536732852458954,
"learning_rate": 1.6029844174015373e-06,
"loss": 0.0243,
"step": 8310
},
{
"grad_norm": 0.20237207412719727,
"learning_rate": 1.5571646619281066e-06,
"loss": 0.025,
"step": 8320
},
{
"grad_norm": 0.2454453557729721,
"learning_rate": 1.5119989348161334e-06,
"loss": 0.0213,
"step": 8330
},
{
"grad_norm": 0.11848924309015274,
"learning_rate": 1.4674878458496354e-06,
"loss": 0.018,
"step": 8340
},
{
"grad_norm": 0.11526887118816376,
"learning_rate": 1.4236319959743227e-06,
"loss": 0.0225,
"step": 8350
},
{
"grad_norm": 0.15515992045402527,
"learning_rate": 1.3804319772895258e-06,
"loss": 0.0182,
"step": 8360
},
{
"grad_norm": 0.14200261235237122,
"learning_rate": 1.337888373040158e-06,
"loss": 0.0134,
"step": 8370
},
{
"grad_norm": 0.10732973366975784,
"learning_rate": 1.2960017576088446e-06,
"loss": 0.0196,
"step": 8380
},
{
"grad_norm": 0.2091865837574005,
"learning_rate": 1.2547726965082218e-06,
"loss": 0.0217,
"step": 8390
},
{
"grad_norm": 0.16020621359348297,
"learning_rate": 1.214201746373239e-06,
"loss": 0.0153,
"step": 8400
},
{
"grad_norm": 0.2096724808216095,
"learning_rate": 1.1742894549536477e-06,
"loss": 0.0168,
"step": 8410
},
{
"grad_norm": 0.11459730565547943,
"learning_rate": 1.1350363611066673e-06,
"loss": 0.0215,
"step": 8420
},
{
"grad_norm": 0.15194177627563477,
"learning_rate": 1.0964429947896425e-06,
"loss": 0.0167,
"step": 8430
},
{
"grad_norm": 0.18774618208408356,
"learning_rate": 1.0585098770529157e-06,
"loss": 0.0182,
"step": 8440
},
{
"grad_norm": 0.14489156007766724,
"learning_rate": 1.0212375200327973e-06,
"loss": 0.0171,
"step": 8450
},
{
"grad_norm": 0.15997833013534546,
"learning_rate": 9.846264269446459e-07,
"loss": 0.0168,
"step": 8460
},
{
"grad_norm": 0.22002941370010376,
"learning_rate": 9.486770920760668e-07,
"loss": 0.0158,
"step": 8470
},
{
"grad_norm": 0.19405952095985413,
"learning_rate": 9.133900007802398e-07,
"loss": 0.0242,
"step": 8480
},
{
"grad_norm": 0.12186640501022339,
"learning_rate": 8.787656294694025e-07,
"loss": 0.0234,
"step": 8490
},
{
"grad_norm": 0.16742081940174103,
"learning_rate": 8.448044456083493e-07,
"loss": 0.0154,
"step": 8500
},
{
"grad_norm": 0.21662619709968567,
"learning_rate": 8.115069077081649e-07,
"loss": 0.0203,
"step": 8510
},
{
"grad_norm": 0.1955857127904892,
"learning_rate": 7.788734653200569e-07,
"loss": 0.0209,
"step": 8520
},
{
"grad_norm": 0.11155475676059723,
"learning_rate": 7.469045590292323e-07,
"loss": 0.0157,
"step": 8530
},
{
"grad_norm": 0.29821738600730896,
"learning_rate": 7.156006204489696e-07,
"loss": 0.0157,
"step": 8540
},
{
"grad_norm": 0.11047469824552536,
"learning_rate": 6.849620722148065e-07,
"loss": 0.0214,
"step": 8550
},
{
"grad_norm": 0.13593296706676483,
"learning_rate": 6.549893279788277e-07,
"loss": 0.0163,
"step": 8560
},
{
"grad_norm": 0.16028732061386108,
"learning_rate": 6.256827924040642e-07,
"loss": 0.0249,
"step": 8570
},
{
"grad_norm": 0.22085900604724884,
"learning_rate": 5.970428611590473e-07,
"loss": 0.0171,
"step": 8580
},
{
"grad_norm": 0.19247432053089142,
"learning_rate": 5.690699209124573e-07,
"loss": 0.0177,
"step": 8590
},
{
"grad_norm": 0.2937515377998352,
"learning_rate": 5.417643493279112e-07,
"loss": 0.0285,
"step": 8600
},
{
"grad_norm": 0.18159297108650208,
"learning_rate": 5.151265150588503e-07,
"loss": 0.0218,
"step": 8610
},
{
"grad_norm": 0.18706293404102325,
"learning_rate": 4.891567777435879e-07,
"loss": 0.023,
"step": 8620
},
{
"grad_norm": 0.17344366014003754,
"learning_rate": 4.638554880004198e-07,
"loss": 0.0167,
"step": 8630
},
{
"grad_norm": 0.12474612146615982,
"learning_rate": 4.392229874229159e-07,
"loss": 0.016,
"step": 8640
},
{
"grad_norm": 0.18510790169239044,
"learning_rate": 4.1525960857530243e-07,
"loss": 0.0258,
"step": 8650
},
{
"grad_norm": 0.22793252766132355,
"learning_rate": 3.9196567498797077e-07,
"loss": 0.0215,
"step": 8660
},
{
"grad_norm": 0.2246151566505432,
"learning_rate": 3.693415011531087e-07,
"loss": 0.0194,
"step": 8670
},
{
"grad_norm": 0.2726196348667145,
"learning_rate": 3.4738739252045405e-07,
"loss": 0.025,
"step": 8680
},
{
"grad_norm": 0.12718743085861206,
"learning_rate": 3.261036454931643e-07,
"loss": 0.0173,
"step": 8690
},
{
"grad_norm": 0.1293998807668686,
"learning_rate": 3.054905474238423e-07,
"loss": 0.0191,
"step": 8700
},
{
"grad_norm": 0.12940746545791626,
"learning_rate": 2.85548376610606e-07,
"loss": 0.0158,
"step": 8710
},
{
"grad_norm": 0.11670862883329391,
"learning_rate": 2.6627740229339673e-07,
"loss": 0.0173,
"step": 8720
},
{
"grad_norm": 0.13993482291698456,
"learning_rate": 2.476778846502825e-07,
"loss": 0.0205,
"step": 8730
},
{
"grad_norm": 0.24328497052192688,
"learning_rate": 2.2975007479397738e-07,
"loss": 0.0204,
"step": 8740
},
{
"grad_norm": 0.1302046924829483,
"learning_rate": 2.1249421476846054e-07,
"loss": 0.0201,
"step": 8750
},
{
"grad_norm": 0.22330981492996216,
"learning_rate": 1.9591053754567935e-07,
"loss": 0.0227,
"step": 8760
},
{
"grad_norm": 0.1316378116607666,
"learning_rate": 1.799992670224182e-07,
"loss": 0.0192,
"step": 8770
},
{
"grad_norm": 0.18792611360549927,
"learning_rate": 1.6476061801727894e-07,
"loss": 0.021,
"step": 8780
},
{
"grad_norm": 0.11966916173696518,
"learning_rate": 1.5019479626778298e-07,
"loss": 0.0257,
"step": 8790
},
{
"grad_norm": 0.123959019780159,
"learning_rate": 1.3630199842758484e-07,
"loss": 0.0186,
"step": 8800
},
{
"grad_norm": 0.1272306889295578,
"learning_rate": 1.2308241206380744e-07,
"loss": 0.0206,
"step": 8810
},
{
"grad_norm": 0.3675824701786041,
"learning_rate": 1.1053621565454974e-07,
"loss": 0.0226,
"step": 8820
},
{
"grad_norm": 0.17444156110286713,
"learning_rate": 9.866357858642205e-08,
"loss": 0.0191,
"step": 8830
},
{
"grad_norm": 0.11050330102443695,
"learning_rate": 8.74646611523089e-08,
"loss": 0.0168,
"step": 8840
},
{
"grad_norm": 0.2335355579853058,
"learning_rate": 7.693961454916521e-08,
"loss": 0.0195,
"step": 8850
},
{
"grad_norm": 0.2811698615550995,
"learning_rate": 6.708858087601244e-08,
"loss": 0.0217,
"step": 8860
},
{
"grad_norm": 0.09265216439962387,
"learning_rate": 5.791169313197897e-08,
"loss": 0.0179,
"step": 8870
},
{
"grad_norm": 0.11499057710170746,
"learning_rate": 4.9409075214545965e-08,
"loss": 0.0208,
"step": 8880
},
{
"grad_norm": 0.1252926141023636,
"learning_rate": 4.158084191783762e-08,
"loss": 0.0179,
"step": 8890
},
{
"grad_norm": 0.26581233739852905,
"learning_rate": 3.442709893110574e-08,
"loss": 0.0216,
"step": 8900
},
{
"grad_norm": 0.2547733783721924,
"learning_rate": 2.794794283727531e-08,
"loss": 0.0155,
"step": 8910
},
{
"grad_norm": 0.22705094516277313,
"learning_rate": 2.214346111164556e-08,
"loss": 0.0172,
"step": 8920
},
{
"grad_norm": 0.12886379659175873,
"learning_rate": 1.7013732120724214e-08,
"loss": 0.0149,
"step": 8930
},
{
"grad_norm": 0.26090526580810547,
"learning_rate": 1.2558825121161688e-08,
"loss": 0.016,
"step": 8940
},
{
"grad_norm": 0.10405340790748596,
"learning_rate": 8.778800258801844e-09,
"loss": 0.0218,
"step": 8950
},
{
"grad_norm": 0.18438711762428284,
"learning_rate": 5.673708567888181e-09,
"loss": 0.0186,
"step": 8960
},
{
"grad_norm": 0.15594296157360077,
"learning_rate": 3.2435919703754923e-09,
"loss": 0.0148,
"step": 8970
},
{
"grad_norm": 0.1541050672531128,
"learning_rate": 1.4884832753414569e-09,
"loss": 0.0211,
"step": 8980
},
{
"grad_norm": 0.12811198830604553,
"learning_rate": 4.084061785758486e-10,
"loss": 0.0194,
"step": 8990
},
{
"grad_norm": 0.12370117753744125,
"learning_rate": 3.3752622308202976e-12,
"loss": 0.0205,
"step": 9000
}
],
"logging_steps": 10,
"max_steps": 9000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 2000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}