upper_body_60-task_cls_4obj_100 / trainer_state.json
gubam's picture
Upload folder using huggingface_hub
c47482d verified
Raw
History Blame Contribute Delete
420 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 30000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"grad_norm": 1.497167944908142,
"learning_rate": 6.000000000000001e-07,
"loss": 1.3862,
"step": 10
},
{
"grad_norm": 1.3765490055084229,
"learning_rate": 1.2666666666666667e-06,
"loss": 1.3712,
"step": 20
},
{
"grad_norm": 1.2162531614303589,
"learning_rate": 1.9333333333333336e-06,
"loss": 1.3714,
"step": 30
},
{
"grad_norm": 0.8798138499259949,
"learning_rate": 2.6e-06,
"loss": 1.3439,
"step": 40
},
{
"grad_norm": 0.74563068151474,
"learning_rate": 3.2666666666666666e-06,
"loss": 1.314,
"step": 50
},
{
"grad_norm": 0.9039244651794434,
"learning_rate": 3.9333333333333335e-06,
"loss": 1.274,
"step": 60
},
{
"grad_norm": 0.991463840007782,
"learning_rate": 4.6e-06,
"loss": 1.2117,
"step": 70
},
{
"grad_norm": 0.8518133163452148,
"learning_rate": 5.266666666666667e-06,
"loss": 1.1661,
"step": 80
},
{
"grad_norm": 0.723469614982605,
"learning_rate": 5.933333333333334e-06,
"loss": 1.1264,
"step": 90
},
{
"grad_norm": 0.6089727282524109,
"learning_rate": 6.6e-06,
"loss": 1.1035,
"step": 100
},
{
"grad_norm": 0.6829776763916016,
"learning_rate": 7.266666666666668e-06,
"loss": 1.0938,
"step": 110
},
{
"grad_norm": 0.5374065637588501,
"learning_rate": 7.933333333333334e-06,
"loss": 1.0935,
"step": 120
},
{
"grad_norm": 0.7095082998275757,
"learning_rate": 8.599999999999999e-06,
"loss": 1.0789,
"step": 130
},
{
"grad_norm": 0.4934196472167969,
"learning_rate": 9.266666666666667e-06,
"loss": 1.0773,
"step": 140
},
{
"grad_norm": 0.87713623046875,
"learning_rate": 9.933333333333334e-06,
"loss": 1.0759,
"step": 150
},
{
"grad_norm": 0.5606873631477356,
"learning_rate": 1.06e-05,
"loss": 1.0713,
"step": 160
},
{
"grad_norm": 0.6145502328872681,
"learning_rate": 1.1266666666666667e-05,
"loss": 1.068,
"step": 170
},
{
"grad_norm": 0.77350252866745,
"learning_rate": 1.1933333333333333e-05,
"loss": 1.0732,
"step": 180
},
{
"grad_norm": 0.7174309492111206,
"learning_rate": 1.2600000000000001e-05,
"loss": 1.0733,
"step": 190
},
{
"grad_norm": 0.46515318751335144,
"learning_rate": 1.3266666666666666e-05,
"loss": 1.0611,
"step": 200
},
{
"grad_norm": 0.66241055727005,
"learning_rate": 1.3933333333333334e-05,
"loss": 1.0705,
"step": 210
},
{
"grad_norm": 0.5770873427391052,
"learning_rate": 1.4599999999999999e-05,
"loss": 1.0709,
"step": 220
},
{
"grad_norm": 0.5283482074737549,
"learning_rate": 1.5266666666666667e-05,
"loss": 1.0578,
"step": 230
},
{
"grad_norm": 0.5669373869895935,
"learning_rate": 1.5933333333333332e-05,
"loss": 1.0548,
"step": 240
},
{
"grad_norm": 0.6653897166252136,
"learning_rate": 1.66e-05,
"loss": 1.0562,
"step": 250
},
{
"grad_norm": 0.6418729424476624,
"learning_rate": 1.726666666666667e-05,
"loss": 1.0548,
"step": 260
},
{
"grad_norm": 0.642163872718811,
"learning_rate": 1.7933333333333337e-05,
"loss": 1.0527,
"step": 270
},
{
"grad_norm": 0.6322233080863953,
"learning_rate": 1.86e-05,
"loss": 1.0533,
"step": 280
},
{
"grad_norm": 0.5504233241081238,
"learning_rate": 1.926666666666667e-05,
"loss": 1.0418,
"step": 290
},
{
"grad_norm": 0.5050532817840576,
"learning_rate": 1.9933333333333334e-05,
"loss": 1.0322,
"step": 300
},
{
"grad_norm": 0.727378249168396,
"learning_rate": 2.06e-05,
"loss": 1.0304,
"step": 310
},
{
"grad_norm": 0.589513897895813,
"learning_rate": 2.1266666666666667e-05,
"loss": 1.0368,
"step": 320
},
{
"grad_norm": 0.6995396614074707,
"learning_rate": 2.1933333333333332e-05,
"loss": 1.0283,
"step": 330
},
{
"grad_norm": 0.773554801940918,
"learning_rate": 2.26e-05,
"loss": 1.0288,
"step": 340
},
{
"grad_norm": 0.7324805855751038,
"learning_rate": 2.326666666666667e-05,
"loss": 1.019,
"step": 350
},
{
"grad_norm": 0.556996762752533,
"learning_rate": 2.3933333333333337e-05,
"loss": 1.0126,
"step": 360
},
{
"grad_norm": 0.7883577346801758,
"learning_rate": 2.46e-05,
"loss": 1.0099,
"step": 370
},
{
"grad_norm": 0.5862056016921997,
"learning_rate": 2.5266666666666666e-05,
"loss": 1.0048,
"step": 380
},
{
"grad_norm": 0.6650963425636292,
"learning_rate": 2.5933333333333338e-05,
"loss": 0.9887,
"step": 390
},
{
"grad_norm": 0.7218946814537048,
"learning_rate": 2.6600000000000003e-05,
"loss": 0.9745,
"step": 400
},
{
"grad_norm": 0.8010955452919006,
"learning_rate": 2.7266666666666668e-05,
"loss": 0.9748,
"step": 410
},
{
"grad_norm": 0.7480322122573853,
"learning_rate": 2.7933333333333332e-05,
"loss": 0.9664,
"step": 420
},
{
"grad_norm": 0.7678745985031128,
"learning_rate": 2.86e-05,
"loss": 0.959,
"step": 430
},
{
"grad_norm": 0.7401871681213379,
"learning_rate": 2.926666666666667e-05,
"loss": 0.9445,
"step": 440
},
{
"grad_norm": 0.8685915470123291,
"learning_rate": 2.9933333333333337e-05,
"loss": 0.9351,
"step": 450
},
{
"grad_norm": 0.7339940071105957,
"learning_rate": 3.06e-05,
"loss": 0.9219,
"step": 460
},
{
"grad_norm": 0.8734851479530334,
"learning_rate": 3.126666666666666e-05,
"loss": 0.9021,
"step": 470
},
{
"grad_norm": 0.7715612053871155,
"learning_rate": 3.1933333333333335e-05,
"loss": 0.8883,
"step": 480
},
{
"grad_norm": 1.0519468784332275,
"learning_rate": 3.26e-05,
"loss": 0.8771,
"step": 490
},
{
"grad_norm": 0.9699525833129883,
"learning_rate": 3.326666666666667e-05,
"loss": 0.8607,
"step": 500
},
{
"grad_norm": 1.0209293365478516,
"learning_rate": 3.3933333333333336e-05,
"loss": 0.8641,
"step": 510
},
{
"grad_norm": 0.8366290330886841,
"learning_rate": 3.46e-05,
"loss": 0.8452,
"step": 520
},
{
"grad_norm": 0.9080895185470581,
"learning_rate": 3.526666666666667e-05,
"loss": 0.8281,
"step": 530
},
{
"grad_norm": 0.8733110427856445,
"learning_rate": 3.593333333333334e-05,
"loss": 0.7995,
"step": 540
},
{
"grad_norm": 0.9767991304397583,
"learning_rate": 3.66e-05,
"loss": 0.8148,
"step": 550
},
{
"grad_norm": 1.2055487632751465,
"learning_rate": 3.726666666666667e-05,
"loss": 0.7953,
"step": 560
},
{
"grad_norm": 0.9383611679077148,
"learning_rate": 3.793333333333334e-05,
"loss": 0.7654,
"step": 570
},
{
"grad_norm": 1.064436435699463,
"learning_rate": 3.86e-05,
"loss": 0.7548,
"step": 580
},
{
"grad_norm": 1.0268619060516357,
"learning_rate": 3.926666666666667e-05,
"loss": 0.7569,
"step": 590
},
{
"grad_norm": 1.1436439752578735,
"learning_rate": 3.993333333333333e-05,
"loss": 0.7439,
"step": 600
},
{
"grad_norm": 1.0441279411315918,
"learning_rate": 4.0600000000000004e-05,
"loss": 0.7237,
"step": 610
},
{
"grad_norm": 1.0106209516525269,
"learning_rate": 4.126666666666667e-05,
"loss": 0.724,
"step": 620
},
{
"grad_norm": 0.9051573276519775,
"learning_rate": 4.1933333333333334e-05,
"loss": 0.7183,
"step": 630
},
{
"grad_norm": 1.2043668031692505,
"learning_rate": 4.26e-05,
"loss": 0.7111,
"step": 640
},
{
"grad_norm": 1.1984174251556396,
"learning_rate": 4.3266666666666664e-05,
"loss": 0.6893,
"step": 650
},
{
"grad_norm": 1.0618480443954468,
"learning_rate": 4.3933333333333335e-05,
"loss": 0.6911,
"step": 660
},
{
"grad_norm": 1.0270366668701172,
"learning_rate": 4.46e-05,
"loss": 0.671,
"step": 670
},
{
"grad_norm": 0.9954440593719482,
"learning_rate": 4.526666666666667e-05,
"loss": 0.6683,
"step": 680
},
{
"grad_norm": 1.089008092880249,
"learning_rate": 4.5933333333333336e-05,
"loss": 0.6616,
"step": 690
},
{
"grad_norm": 1.2877507209777832,
"learning_rate": 4.660000000000001e-05,
"loss": 0.6632,
"step": 700
},
{
"grad_norm": 0.9327141642570496,
"learning_rate": 4.726666666666667e-05,
"loss": 0.6458,
"step": 710
},
{
"grad_norm": 1.3218812942504883,
"learning_rate": 4.793333333333334e-05,
"loss": 0.6373,
"step": 720
},
{
"grad_norm": 0.8969250321388245,
"learning_rate": 4.86e-05,
"loss": 0.6259,
"step": 730
},
{
"grad_norm": 0.8581910133361816,
"learning_rate": 4.926666666666667e-05,
"loss": 0.6083,
"step": 740
},
{
"grad_norm": 1.2754545211791992,
"learning_rate": 4.993333333333334e-05,
"loss": 0.621,
"step": 750
},
{
"grad_norm": 0.9609273672103882,
"learning_rate": 5.0600000000000003e-05,
"loss": 0.6125,
"step": 760
},
{
"grad_norm": 0.8969170451164246,
"learning_rate": 5.1266666666666675e-05,
"loss": 0.6055,
"step": 770
},
{
"grad_norm": 0.9942559599876404,
"learning_rate": 5.193333333333333e-05,
"loss": 0.6279,
"step": 780
},
{
"grad_norm": 0.8410463929176331,
"learning_rate": 5.2600000000000005e-05,
"loss": 0.6012,
"step": 790
},
{
"grad_norm": 0.9585829973220825,
"learning_rate": 5.326666666666666e-05,
"loss": 0.5983,
"step": 800
},
{
"grad_norm": 0.9032193422317505,
"learning_rate": 5.3933333333333334e-05,
"loss": 0.5916,
"step": 810
},
{
"grad_norm": 1.078798532485962,
"learning_rate": 5.4600000000000006e-05,
"loss": 0.5978,
"step": 820
},
{
"grad_norm": 1.0203973054885864,
"learning_rate": 5.5266666666666664e-05,
"loss": 0.5996,
"step": 830
},
{
"grad_norm": 1.1048483848571777,
"learning_rate": 5.5933333333333335e-05,
"loss": 0.5924,
"step": 840
},
{
"grad_norm": 1.0069968700408936,
"learning_rate": 5.66e-05,
"loss": 0.5853,
"step": 850
},
{
"grad_norm": 1.1509100198745728,
"learning_rate": 5.726666666666667e-05,
"loss": 0.566,
"step": 860
},
{
"grad_norm": 0.9662450551986694,
"learning_rate": 5.7933333333333337e-05,
"loss": 0.5791,
"step": 870
},
{
"grad_norm": 1.0324444770812988,
"learning_rate": 5.86e-05,
"loss": 0.5627,
"step": 880
},
{
"grad_norm": 1.153710126876831,
"learning_rate": 5.926666666666667e-05,
"loss": 0.568,
"step": 890
},
{
"grad_norm": 1.1811585426330566,
"learning_rate": 5.9933333333333345e-05,
"loss": 0.5628,
"step": 900
},
{
"grad_norm": 0.9662961363792419,
"learning_rate": 6.06e-05,
"loss": 0.5588,
"step": 910
},
{
"grad_norm": 0.9198316931724548,
"learning_rate": 6.126666666666667e-05,
"loss": 0.544,
"step": 920
},
{
"grad_norm": 1.1507964134216309,
"learning_rate": 6.193333333333333e-05,
"loss": 0.5426,
"step": 930
},
{
"grad_norm": 0.9815706014633179,
"learning_rate": 6.26e-05,
"loss": 0.5457,
"step": 940
},
{
"grad_norm": 0.9707251787185669,
"learning_rate": 6.326666666666667e-05,
"loss": 0.5374,
"step": 950
},
{
"grad_norm": 1.0989611148834229,
"learning_rate": 6.393333333333333e-05,
"loss": 0.5404,
"step": 960
},
{
"grad_norm": 1.0847727060317993,
"learning_rate": 6.460000000000001e-05,
"loss": 0.5583,
"step": 970
},
{
"grad_norm": 1.0409971475601196,
"learning_rate": 6.526666666666666e-05,
"loss": 0.5429,
"step": 980
},
{
"grad_norm": 0.9299330711364746,
"learning_rate": 6.593333333333334e-05,
"loss": 0.541,
"step": 990
},
{
"grad_norm": 1.0520474910736084,
"learning_rate": 6.66e-05,
"loss": 0.5447,
"step": 1000
},
{
"grad_norm": 0.889153003692627,
"learning_rate": 6.726666666666667e-05,
"loss": 0.5346,
"step": 1010
},
{
"grad_norm": 0.9422370195388794,
"learning_rate": 6.793333333333334e-05,
"loss": 0.5562,
"step": 1020
},
{
"grad_norm": 0.8990776538848877,
"learning_rate": 6.860000000000001e-05,
"loss": 0.5303,
"step": 1030
},
{
"grad_norm": 0.8331694602966309,
"learning_rate": 6.926666666666667e-05,
"loss": 0.5329,
"step": 1040
},
{
"grad_norm": 1.011156439781189,
"learning_rate": 6.993333333333334e-05,
"loss": 0.5459,
"step": 1050
},
{
"grad_norm": 0.9624161124229431,
"learning_rate": 7.06e-05,
"loss": 0.5488,
"step": 1060
},
{
"grad_norm": 0.850904643535614,
"learning_rate": 7.126666666666667e-05,
"loss": 0.5386,
"step": 1070
},
{
"grad_norm": 0.9043158292770386,
"learning_rate": 7.193333333333334e-05,
"loss": 0.5413,
"step": 1080
},
{
"grad_norm": 0.7124777436256409,
"learning_rate": 7.26e-05,
"loss": 0.5301,
"step": 1090
},
{
"grad_norm": 0.9083084464073181,
"learning_rate": 7.326666666666667e-05,
"loss": 0.5106,
"step": 1100
},
{
"grad_norm": 0.9425422549247742,
"learning_rate": 7.393333333333333e-05,
"loss": 0.5404,
"step": 1110
},
{
"grad_norm": 1.0150341987609863,
"learning_rate": 7.46e-05,
"loss": 0.5165,
"step": 1120
},
{
"grad_norm": 0.9683831334114075,
"learning_rate": 7.526666666666668e-05,
"loss": 0.5314,
"step": 1130
},
{
"grad_norm": 0.94903963804245,
"learning_rate": 7.593333333333334e-05,
"loss": 0.5483,
"step": 1140
},
{
"grad_norm": 0.9994460344314575,
"learning_rate": 7.66e-05,
"loss": 0.5411,
"step": 1150
},
{
"grad_norm": 0.8009998202323914,
"learning_rate": 7.726666666666667e-05,
"loss": 0.5255,
"step": 1160
},
{
"grad_norm": 0.7843228578567505,
"learning_rate": 7.793333333333333e-05,
"loss": 0.525,
"step": 1170
},
{
"grad_norm": 0.924232542514801,
"learning_rate": 7.860000000000001e-05,
"loss": 0.5139,
"step": 1180
},
{
"grad_norm": 0.9923452138900757,
"learning_rate": 7.926666666666666e-05,
"loss": 0.5137,
"step": 1190
},
{
"grad_norm": 0.9879342317581177,
"learning_rate": 7.993333333333334e-05,
"loss": 0.5157,
"step": 1200
},
{
"grad_norm": 1.165381908416748,
"learning_rate": 8.060000000000001e-05,
"loss": 0.5244,
"step": 1210
},
{
"grad_norm": 1.091857671737671,
"learning_rate": 8.126666666666667e-05,
"loss": 0.5063,
"step": 1220
},
{
"grad_norm": 1.0989230871200562,
"learning_rate": 8.193333333333334e-05,
"loss": 0.5123,
"step": 1230
},
{
"grad_norm": 0.8473907709121704,
"learning_rate": 8.26e-05,
"loss": 0.5086,
"step": 1240
},
{
"grad_norm": 1.0479320287704468,
"learning_rate": 8.326666666666667e-05,
"loss": 0.4947,
"step": 1250
},
{
"grad_norm": 0.8895961046218872,
"learning_rate": 8.393333333333335e-05,
"loss": 0.4908,
"step": 1260
},
{
"grad_norm": 0.9734536409378052,
"learning_rate": 8.46e-05,
"loss": 0.4824,
"step": 1270
},
{
"grad_norm": 0.9393786787986755,
"learning_rate": 8.526666666666667e-05,
"loss": 0.4945,
"step": 1280
},
{
"grad_norm": 0.8548934459686279,
"learning_rate": 8.593333333333333e-05,
"loss": 0.4918,
"step": 1290
},
{
"grad_norm": 0.833220899105072,
"learning_rate": 8.66e-05,
"loss": 0.4902,
"step": 1300
},
{
"grad_norm": 0.849828839302063,
"learning_rate": 8.726666666666667e-05,
"loss": 0.4775,
"step": 1310
},
{
"grad_norm": 0.8431593775749207,
"learning_rate": 8.793333333333333e-05,
"loss": 0.4811,
"step": 1320
},
{
"grad_norm": 0.9460287690162659,
"learning_rate": 8.86e-05,
"loss": 0.4694,
"step": 1330
},
{
"grad_norm": 0.7675583362579346,
"learning_rate": 8.926666666666668e-05,
"loss": 0.4631,
"step": 1340
},
{
"grad_norm": 1.0189882516860962,
"learning_rate": 8.993333333333334e-05,
"loss": 0.4856,
"step": 1350
},
{
"grad_norm": 0.9207183718681335,
"learning_rate": 9.06e-05,
"loss": 0.4744,
"step": 1360
},
{
"grad_norm": 1.1016457080841064,
"learning_rate": 9.126666666666667e-05,
"loss": 0.4672,
"step": 1370
},
{
"grad_norm": 0.7799882888793945,
"learning_rate": 9.193333333333334e-05,
"loss": 0.4636,
"step": 1380
},
{
"grad_norm": 0.9405816793441772,
"learning_rate": 9.260000000000001e-05,
"loss": 0.4692,
"step": 1390
},
{
"grad_norm": 0.8900464773178101,
"learning_rate": 9.326666666666667e-05,
"loss": 0.4641,
"step": 1400
},
{
"grad_norm": 0.8389847874641418,
"learning_rate": 9.393333333333334e-05,
"loss": 0.4766,
"step": 1410
},
{
"grad_norm": 0.8662654161453247,
"learning_rate": 9.46e-05,
"loss": 0.4716,
"step": 1420
},
{
"grad_norm": 0.9146279692649841,
"learning_rate": 9.526666666666667e-05,
"loss": 0.4578,
"step": 1430
},
{
"grad_norm": 0.8330411911010742,
"learning_rate": 9.593333333333334e-05,
"loss": 0.4729,
"step": 1440
},
{
"grad_norm": 0.7738226652145386,
"learning_rate": 9.66e-05,
"loss": 0.4795,
"step": 1450
},
{
"grad_norm": 0.7489462494850159,
"learning_rate": 9.726666666666667e-05,
"loss": 0.4688,
"step": 1460
},
{
"grad_norm": 0.8304651379585266,
"learning_rate": 9.793333333333333e-05,
"loss": 0.4729,
"step": 1470
},
{
"grad_norm": 0.9070173501968384,
"learning_rate": 9.86e-05,
"loss": 0.4658,
"step": 1480
},
{
"grad_norm": 0.8511922359466553,
"learning_rate": 9.926666666666668e-05,
"loss": 0.4622,
"step": 1490
},
{
"grad_norm": 0.7142724990844727,
"learning_rate": 9.993333333333334e-05,
"loss": 0.4733,
"step": 1500
},
{
"grad_norm": 0.8550265431404114,
"learning_rate": 9.999997539434007e-05,
"loss": 0.4661,
"step": 1510
},
{
"grad_norm": 0.8205634355545044,
"learning_rate": 9.999989033776898e-05,
"loss": 0.4752,
"step": 1520
},
{
"grad_norm": 0.8192468285560608,
"learning_rate": 9.999974452661641e-05,
"loss": 0.4754,
"step": 1530
},
{
"grad_norm": 0.7811523079872131,
"learning_rate": 9.999953796105959e-05,
"loss": 0.4665,
"step": 1540
},
{
"grad_norm": 0.8306159377098083,
"learning_rate": 9.999927064134949e-05,
"loss": 0.4572,
"step": 1550
},
{
"grad_norm": 0.8542218208312988,
"learning_rate": 9.999894256781095e-05,
"loss": 0.4738,
"step": 1560
},
{
"grad_norm": 0.7783997654914856,
"learning_rate": 9.99985537408426e-05,
"loss": 0.4547,
"step": 1570
},
{
"grad_norm": 0.8400068283081055,
"learning_rate": 9.999810416091688e-05,
"loss": 0.4493,
"step": 1580
},
{
"grad_norm": 0.8952717185020447,
"learning_rate": 9.99975938285801e-05,
"loss": 0.4717,
"step": 1590
},
{
"grad_norm": 0.9472647905349731,
"learning_rate": 9.999702274445236e-05,
"loss": 0.4563,
"step": 1600
},
{
"grad_norm": 0.7749526500701904,
"learning_rate": 9.999639090922756e-05,
"loss": 0.4384,
"step": 1610
},
{
"grad_norm": 0.7800078392028809,
"learning_rate": 9.999569832367346e-05,
"loss": 0.4657,
"step": 1620
},
{
"grad_norm": 0.7521607875823975,
"learning_rate": 9.999494498863162e-05,
"loss": 0.4713,
"step": 1630
},
{
"grad_norm": 0.8178069591522217,
"learning_rate": 9.99941309050174e-05,
"loss": 0.4553,
"step": 1640
},
{
"grad_norm": 0.7313920855522156,
"learning_rate": 9.999325607381999e-05,
"loss": 0.45,
"step": 1650
},
{
"grad_norm": 0.8572726845741272,
"learning_rate": 9.999232049610238e-05,
"loss": 0.4517,
"step": 1660
},
{
"grad_norm": 0.8027247190475464,
"learning_rate": 9.999132417300142e-05,
"loss": 0.4345,
"step": 1670
},
{
"grad_norm": 0.8615437150001526,
"learning_rate": 9.99902671057277e-05,
"loss": 0.4379,
"step": 1680
},
{
"grad_norm": 0.6692933440208435,
"learning_rate": 9.998914929556569e-05,
"loss": 0.4298,
"step": 1690
},
{
"grad_norm": 0.6929596066474915,
"learning_rate": 9.998797074387361e-05,
"loss": 0.439,
"step": 1700
},
{
"grad_norm": 0.824317991733551,
"learning_rate": 9.99867314520835e-05,
"loss": 0.4482,
"step": 1710
},
{
"grad_norm": 0.738298773765564,
"learning_rate": 9.998543142170126e-05,
"loss": 0.4345,
"step": 1720
},
{
"grad_norm": 0.7953527569770813,
"learning_rate": 9.99840706543065e-05,
"loss": 0.4279,
"step": 1730
},
{
"grad_norm": 0.7345385551452637,
"learning_rate": 9.998264915155274e-05,
"loss": 0.4285,
"step": 1740
},
{
"grad_norm": 0.9032840728759766,
"learning_rate": 9.998116691516718e-05,
"loss": 0.4352,
"step": 1750
},
{
"grad_norm": 0.6412612795829773,
"learning_rate": 9.997962394695091e-05,
"loss": 0.4353,
"step": 1760
},
{
"grad_norm": 0.7473551630973816,
"learning_rate": 9.997802024877875e-05,
"loss": 0.4224,
"step": 1770
},
{
"grad_norm": 0.785769522190094,
"learning_rate": 9.99763558225994e-05,
"loss": 0.4407,
"step": 1780
},
{
"grad_norm": 0.6304603815078735,
"learning_rate": 9.997463067043526e-05,
"loss": 0.4201,
"step": 1790
},
{
"grad_norm": 0.7144474983215332,
"learning_rate": 9.997284479438253e-05,
"loss": 0.4263,
"step": 1800
},
{
"grad_norm": 0.659083902835846,
"learning_rate": 9.997099819661127e-05,
"loss": 0.4196,
"step": 1810
},
{
"grad_norm": 0.7157149910926819,
"learning_rate": 9.996909087936524e-05,
"loss": 0.4284,
"step": 1820
},
{
"grad_norm": 0.8297467231750488,
"learning_rate": 9.996712284496201e-05,
"loss": 0.4253,
"step": 1830
},
{
"grad_norm": 0.7346173524856567,
"learning_rate": 9.996509409579293e-05,
"loss": 0.4365,
"step": 1840
},
{
"grad_norm": 0.6862539052963257,
"learning_rate": 9.996300463432312e-05,
"loss": 0.4456,
"step": 1850
},
{
"grad_norm": 0.7219457626342773,
"learning_rate": 9.996085446309148e-05,
"loss": 0.4372,
"step": 1860
},
{
"grad_norm": 0.7898635268211365,
"learning_rate": 9.995864358471066e-05,
"loss": 0.4328,
"step": 1870
},
{
"grad_norm": 0.6751769781112671,
"learning_rate": 9.99563720018671e-05,
"loss": 0.4142,
"step": 1880
},
{
"grad_norm": 0.7996441721916199,
"learning_rate": 9.995403971732098e-05,
"loss": 0.4227,
"step": 1890
},
{
"grad_norm": 0.6541454792022705,
"learning_rate": 9.995164673390625e-05,
"loss": 0.4201,
"step": 1900
},
{
"grad_norm": 0.7181878685951233,
"learning_rate": 9.994919305453059e-05,
"loss": 0.4391,
"step": 1910
},
{
"grad_norm": 0.8541218638420105,
"learning_rate": 9.994667868217548e-05,
"loss": 0.4296,
"step": 1920
},
{
"grad_norm": 0.6868025064468384,
"learning_rate": 9.99441036198961e-05,
"loss": 0.421,
"step": 1930
},
{
"grad_norm": 0.7327513098716736,
"learning_rate": 9.99414678708214e-05,
"loss": 0.4209,
"step": 1940
},
{
"grad_norm": 0.9473391175270081,
"learning_rate": 9.993877143815407e-05,
"loss": 0.4247,
"step": 1950
},
{
"grad_norm": 0.8402295708656311,
"learning_rate": 9.993601432517053e-05,
"loss": 0.4333,
"step": 1960
},
{
"grad_norm": 0.7014586329460144,
"learning_rate": 9.993319653522091e-05,
"loss": 0.42,
"step": 1970
},
{
"grad_norm": 0.5906397700309753,
"learning_rate": 9.993031807172911e-05,
"loss": 0.4331,
"step": 1980
},
{
"grad_norm": 0.7896249294281006,
"learning_rate": 9.992737893819273e-05,
"loss": 0.4166,
"step": 1990
},
{
"grad_norm": 0.7049199938774109,
"learning_rate": 9.992437913818312e-05,
"loss": 0.4244,
"step": 2000
},
{
"grad_norm": 0.7133922576904297,
"learning_rate": 9.992131867534526e-05,
"loss": 0.4378,
"step": 2010
},
{
"grad_norm": 0.793612539768219,
"learning_rate": 9.991819755339796e-05,
"loss": 0.4287,
"step": 2020
},
{
"grad_norm": 0.7581238746643066,
"learning_rate": 9.991501577613365e-05,
"loss": 0.4138,
"step": 2030
},
{
"grad_norm": 0.7746891975402832,
"learning_rate": 9.99117733474185e-05,
"loss": 0.4303,
"step": 2040
},
{
"grad_norm": 0.6476732492446899,
"learning_rate": 9.990847027119234e-05,
"loss": 0.4272,
"step": 2050
},
{
"grad_norm": 0.7977179288864136,
"learning_rate": 9.990510655146877e-05,
"loss": 0.4243,
"step": 2060
},
{
"grad_norm": 0.6552172899246216,
"learning_rate": 9.990168219233496e-05,
"loss": 0.4159,
"step": 2070
},
{
"grad_norm": 0.7039655447006226,
"learning_rate": 9.989819719795188e-05,
"loss": 0.4175,
"step": 2080
},
{
"grad_norm": 0.763336718082428,
"learning_rate": 9.989465157255412e-05,
"loss": 0.4169,
"step": 2090
},
{
"grad_norm": 0.8805884718894958,
"learning_rate": 9.989104532044994e-05,
"loss": 0.4321,
"step": 2100
},
{
"grad_norm": 0.8546279668807983,
"learning_rate": 9.988737844602128e-05,
"loss": 0.4448,
"step": 2110
},
{
"grad_norm": 0.7598029971122742,
"learning_rate": 9.988365095372372e-05,
"loss": 0.4235,
"step": 2120
},
{
"grad_norm": 0.8650438189506531,
"learning_rate": 9.987986284808654e-05,
"loss": 0.4219,
"step": 2130
},
{
"grad_norm": 0.9486260414123535,
"learning_rate": 9.987601413371264e-05,
"loss": 0.4092,
"step": 2140
},
{
"grad_norm": 0.7379596829414368,
"learning_rate": 9.987210481527855e-05,
"loss": 0.428,
"step": 2150
},
{
"grad_norm": 0.5985510349273682,
"learning_rate": 9.98681348975345e-05,
"loss": 0.416,
"step": 2160
},
{
"grad_norm": 0.7316847443580627,
"learning_rate": 9.986410438530427e-05,
"loss": 0.4097,
"step": 2170
},
{
"grad_norm": 0.7186805605888367,
"learning_rate": 9.986001328348534e-05,
"loss": 0.4234,
"step": 2180
},
{
"grad_norm": 0.7606436610221863,
"learning_rate": 9.985586159704878e-05,
"loss": 0.4102,
"step": 2190
},
{
"grad_norm": 0.6335312128067017,
"learning_rate": 9.985164933103929e-05,
"loss": 0.4036,
"step": 2200
},
{
"grad_norm": 0.8029035925865173,
"learning_rate": 9.984737649057513e-05,
"loss": 0.4152,
"step": 2210
},
{
"grad_norm": 0.8585079908370972,
"learning_rate": 9.984304308084827e-05,
"loss": 0.4015,
"step": 2220
},
{
"grad_norm": 0.8344599604606628,
"learning_rate": 9.983864910712416e-05,
"loss": 0.4136,
"step": 2230
},
{
"grad_norm": 0.7427005171775818,
"learning_rate": 9.98341945747419e-05,
"loss": 0.4046,
"step": 2240
},
{
"grad_norm": 0.6998052000999451,
"learning_rate": 9.98296794891142e-05,
"loss": 0.4148,
"step": 2250
},
{
"grad_norm": 0.8317843675613403,
"learning_rate": 9.982510385572725e-05,
"loss": 0.4001,
"step": 2260
},
{
"grad_norm": 0.775233805179596,
"learning_rate": 9.982046768014094e-05,
"loss": 0.3878,
"step": 2270
},
{
"grad_norm": 0.6719950437545776,
"learning_rate": 9.981577096798863e-05,
"loss": 0.3963,
"step": 2280
},
{
"grad_norm": 0.8318219184875488,
"learning_rate": 9.981101372497727e-05,
"loss": 0.4185,
"step": 2290
},
{
"grad_norm": 0.7903131246566772,
"learning_rate": 9.980619595688737e-05,
"loss": 0.3859,
"step": 2300
},
{
"grad_norm": 0.7791595458984375,
"learning_rate": 9.980131766957295e-05,
"loss": 0.3909,
"step": 2310
},
{
"grad_norm": 0.6508023738861084,
"learning_rate": 9.979637886896163e-05,
"loss": 0.3849,
"step": 2320
},
{
"grad_norm": 0.7626515626907349,
"learning_rate": 9.979137956105447e-05,
"loss": 0.3987,
"step": 2330
},
{
"grad_norm": 0.7466723322868347,
"learning_rate": 9.978631975192613e-05,
"loss": 0.3889,
"step": 2340
},
{
"grad_norm": 0.6682965159416199,
"learning_rate": 9.978119944772475e-05,
"loss": 0.3886,
"step": 2350
},
{
"grad_norm": 0.6208071708679199,
"learning_rate": 9.977601865467197e-05,
"loss": 0.388,
"step": 2360
},
{
"grad_norm": 0.7388435006141663,
"learning_rate": 9.977077737906297e-05,
"loss": 0.386,
"step": 2370
},
{
"grad_norm": 0.7926000952720642,
"learning_rate": 9.976547562726636e-05,
"loss": 0.3886,
"step": 2380
},
{
"grad_norm": 0.8842079639434814,
"learning_rate": 9.976011340572429e-05,
"loss": 0.3964,
"step": 2390
},
{
"grad_norm": 0.6615763902664185,
"learning_rate": 9.975469072095237e-05,
"loss": 0.3874,
"step": 2400
},
{
"grad_norm": 0.588610053062439,
"learning_rate": 9.974920757953965e-05,
"loss": 0.3804,
"step": 2410
},
{
"grad_norm": 0.7170594930648804,
"learning_rate": 9.97436639881487e-05,
"loss": 0.4035,
"step": 2420
},
{
"grad_norm": 0.7512857913970947,
"learning_rate": 9.973805995351545e-05,
"loss": 0.3805,
"step": 2430
},
{
"grad_norm": 0.8106023073196411,
"learning_rate": 9.973239548244939e-05,
"loss": 0.383,
"step": 2440
},
{
"grad_norm": 0.7844937443733215,
"learning_rate": 9.972667058183333e-05,
"loss": 0.3785,
"step": 2450
},
{
"grad_norm": 0.6774455308914185,
"learning_rate": 9.972088525862362e-05,
"loss": 0.3682,
"step": 2460
},
{
"grad_norm": 0.710124671459198,
"learning_rate": 9.971503951984995e-05,
"loss": 0.3771,
"step": 2470
},
{
"grad_norm": 0.726272463798523,
"learning_rate": 9.970913337261543e-05,
"loss": 0.3934,
"step": 2480
},
{
"grad_norm": 0.7088817954063416,
"learning_rate": 9.97031668240966e-05,
"loss": 0.3718,
"step": 2490
},
{
"grad_norm": 0.6792616844177246,
"learning_rate": 9.969713988154339e-05,
"loss": 0.3664,
"step": 2500
},
{
"grad_norm": 0.8668208122253418,
"learning_rate": 9.969105255227906e-05,
"loss": 0.3671,
"step": 2510
},
{
"grad_norm": 0.7345057725906372,
"learning_rate": 9.968490484370035e-05,
"loss": 0.3747,
"step": 2520
},
{
"grad_norm": 0.5861186981201172,
"learning_rate": 9.967869676327726e-05,
"loss": 0.3643,
"step": 2530
},
{
"grad_norm": 0.6670198440551758,
"learning_rate": 9.96724283185532e-05,
"loss": 0.3698,
"step": 2540
},
{
"grad_norm": 0.7445732951164246,
"learning_rate": 9.966609951714494e-05,
"loss": 0.3802,
"step": 2550
},
{
"grad_norm": 0.7027205228805542,
"learning_rate": 9.965971036674255e-05,
"loss": 0.3742,
"step": 2560
},
{
"grad_norm": 0.6619306802749634,
"learning_rate": 9.965326087510947e-05,
"loss": 0.3695,
"step": 2570
},
{
"grad_norm": 0.6221873760223389,
"learning_rate": 9.964675105008244e-05,
"loss": 0.3708,
"step": 2580
},
{
"grad_norm": 0.708595335483551,
"learning_rate": 9.964018089957147e-05,
"loss": 0.3731,
"step": 2590
},
{
"grad_norm": 0.6284708976745605,
"learning_rate": 9.963355043155996e-05,
"loss": 0.374,
"step": 2600
},
{
"grad_norm": 0.8080525994300842,
"learning_rate": 9.962685965410455e-05,
"loss": 0.3803,
"step": 2610
},
{
"grad_norm": 0.6886402368545532,
"learning_rate": 9.962010857533514e-05,
"loss": 0.3603,
"step": 2620
},
{
"grad_norm": 0.6571764945983887,
"learning_rate": 9.961329720345493e-05,
"loss": 0.3646,
"step": 2630
},
{
"grad_norm": 0.7243236899375916,
"learning_rate": 9.96064255467404e-05,
"loss": 0.3762,
"step": 2640
},
{
"grad_norm": 0.7539463043212891,
"learning_rate": 9.959949361354126e-05,
"loss": 0.3743,
"step": 2650
},
{
"grad_norm": 0.6598141193389893,
"learning_rate": 9.959250141228045e-05,
"loss": 0.3734,
"step": 2660
},
{
"grad_norm": 0.7238931059837341,
"learning_rate": 9.958544895145414e-05,
"loss": 0.3812,
"step": 2670
},
{
"grad_norm": 0.764260470867157,
"learning_rate": 9.957833623963177e-05,
"loss": 0.3877,
"step": 2680
},
{
"grad_norm": 0.744884192943573,
"learning_rate": 9.957116328545593e-05,
"loss": 0.3603,
"step": 2690
},
{
"grad_norm": 0.5897803902626038,
"learning_rate": 9.956393009764244e-05,
"loss": 0.3524,
"step": 2700
},
{
"grad_norm": 0.7628750205039978,
"learning_rate": 9.955663668498032e-05,
"loss": 0.3617,
"step": 2710
},
{
"grad_norm": 0.6826313138008118,
"learning_rate": 9.954928305633173e-05,
"loss": 0.3674,
"step": 2720
},
{
"grad_norm": 0.6923078298568726,
"learning_rate": 9.954186922063204e-05,
"loss": 0.355,
"step": 2730
},
{
"grad_norm": 0.6269809007644653,
"learning_rate": 9.953439518688974e-05,
"loss": 0.3526,
"step": 2740
},
{
"grad_norm": 0.6172541379928589,
"learning_rate": 9.952686096418652e-05,
"loss": 0.3624,
"step": 2750
},
{
"grad_norm": 0.6197295188903809,
"learning_rate": 9.951926656167714e-05,
"loss": 0.3501,
"step": 2760
},
{
"grad_norm": 0.7210254669189453,
"learning_rate": 9.951161198858953e-05,
"loss": 0.3543,
"step": 2770
},
{
"grad_norm": 0.6547584533691406,
"learning_rate": 9.950389725422471e-05,
"loss": 0.3577,
"step": 2780
},
{
"grad_norm": 0.6652739644050598,
"learning_rate": 9.949612236795682e-05,
"loss": 0.3583,
"step": 2790
},
{
"grad_norm": 0.7988615036010742,
"learning_rate": 9.948828733923305e-05,
"loss": 0.347,
"step": 2800
},
{
"grad_norm": 0.851736307144165,
"learning_rate": 9.948039217757374e-05,
"loss": 0.3782,
"step": 2810
},
{
"grad_norm": 0.7384556531906128,
"learning_rate": 9.947243689257225e-05,
"loss": 0.3517,
"step": 2820
},
{
"grad_norm": 0.8288888931274414,
"learning_rate": 9.946442149389497e-05,
"loss": 0.3396,
"step": 2830
},
{
"grad_norm": 0.7147480249404907,
"learning_rate": 9.945634599128139e-05,
"loss": 0.3572,
"step": 2840
},
{
"grad_norm": 0.7184939384460449,
"learning_rate": 9.944821039454402e-05,
"loss": 0.3587,
"step": 2850
},
{
"grad_norm": 0.7544175386428833,
"learning_rate": 9.944001471356835e-05,
"loss": 0.3436,
"step": 2860
},
{
"grad_norm": 0.7547817230224609,
"learning_rate": 9.94317589583129e-05,
"loss": 0.3358,
"step": 2870
},
{
"grad_norm": 0.5905265808105469,
"learning_rate": 9.942344313880922e-05,
"loss": 0.3348,
"step": 2880
},
{
"grad_norm": 0.6228170394897461,
"learning_rate": 9.941506726516179e-05,
"loss": 0.3428,
"step": 2890
},
{
"grad_norm": 0.759371817111969,
"learning_rate": 9.94066313475481e-05,
"loss": 0.3461,
"step": 2900
},
{
"grad_norm": 0.7357356548309326,
"learning_rate": 9.939813539621857e-05,
"loss": 0.3313,
"step": 2910
},
{
"grad_norm": 0.590205192565918,
"learning_rate": 9.93895794214966e-05,
"loss": 0.3396,
"step": 2920
},
{
"grad_norm": 0.5977882742881775,
"learning_rate": 9.938096343377852e-05,
"loss": 0.3248,
"step": 2930
},
{
"grad_norm": 0.6735866665840149,
"learning_rate": 9.937228744353353e-05,
"loss": 0.3313,
"step": 2940
},
{
"grad_norm": 0.7506408095359802,
"learning_rate": 9.936355146130379e-05,
"loss": 0.3313,
"step": 2950
},
{
"grad_norm": 0.5893487930297852,
"learning_rate": 9.935475549770435e-05,
"loss": 0.3358,
"step": 2960
},
{
"grad_norm": 0.835304856300354,
"learning_rate": 9.934589956342315e-05,
"loss": 0.3317,
"step": 2970
},
{
"grad_norm": 0.6585904955863953,
"learning_rate": 9.933698366922093e-05,
"loss": 0.3356,
"step": 2980
},
{
"grad_norm": 0.7953837513923645,
"learning_rate": 9.93280078259314e-05,
"loss": 0.3429,
"step": 2990
},
{
"grad_norm": 0.6197541952133179,
"learning_rate": 9.931897204446104e-05,
"loss": 0.3384,
"step": 3000
},
{
"grad_norm": 0.7005690336227417,
"learning_rate": 9.930987633578915e-05,
"loss": 0.3306,
"step": 3010
},
{
"grad_norm": 0.6685798764228821,
"learning_rate": 9.93007207109679e-05,
"loss": 0.3305,
"step": 3020
},
{
"grad_norm": 0.6820731163024902,
"learning_rate": 9.929150518112224e-05,
"loss": 0.3309,
"step": 3030
},
{
"grad_norm": 0.6705907583236694,
"learning_rate": 9.928222975744991e-05,
"loss": 0.3346,
"step": 3040
},
{
"grad_norm": 0.5994101166725159,
"learning_rate": 9.92728944512214e-05,
"loss": 0.3338,
"step": 3050
},
{
"grad_norm": 0.7535775899887085,
"learning_rate": 9.926349927378001e-05,
"loss": 0.3406,
"step": 3060
},
{
"grad_norm": 0.6269504427909851,
"learning_rate": 9.925404423654174e-05,
"loss": 0.3187,
"step": 3070
},
{
"grad_norm": 0.6764205098152161,
"learning_rate": 9.924452935099537e-05,
"loss": 0.3415,
"step": 3080
},
{
"grad_norm": 0.7323486804962158,
"learning_rate": 9.92349546287024e-05,
"loss": 0.3569,
"step": 3090
},
{
"grad_norm": 0.6715821623802185,
"learning_rate": 9.9225320081297e-05,
"loss": 0.3389,
"step": 3100
},
{
"grad_norm": 0.6851534247398376,
"learning_rate": 9.921562572048606e-05,
"loss": 0.3301,
"step": 3110
},
{
"grad_norm": 0.6987500786781311,
"learning_rate": 9.920587155804913e-05,
"loss": 0.3318,
"step": 3120
},
{
"grad_norm": 0.6074950098991394,
"learning_rate": 9.919605760583845e-05,
"loss": 0.3298,
"step": 3130
},
{
"grad_norm": 0.6436213254928589,
"learning_rate": 9.91861838757789e-05,
"loss": 0.3388,
"step": 3140
},
{
"grad_norm": 0.6318737268447876,
"learning_rate": 9.917625037986798e-05,
"loss": 0.3311,
"step": 3150
},
{
"grad_norm": 0.6454184055328369,
"learning_rate": 9.916625713017583e-05,
"loss": 0.3425,
"step": 3160
},
{
"grad_norm": 0.6960732936859131,
"learning_rate": 9.915620413884519e-05,
"loss": 0.3262,
"step": 3170
},
{
"grad_norm": 0.7481048703193665,
"learning_rate": 9.914609141809139e-05,
"loss": 0.3445,
"step": 3180
},
{
"grad_norm": 0.7493470311164856,
"learning_rate": 9.913591898020235e-05,
"loss": 0.3369,
"step": 3190
},
{
"grad_norm": 0.6892242431640625,
"learning_rate": 9.912568683753853e-05,
"loss": 0.3382,
"step": 3200
},
{
"grad_norm": 0.5738986134529114,
"learning_rate": 9.911539500253295e-05,
"loss": 0.326,
"step": 3210
},
{
"grad_norm": 0.7782413959503174,
"learning_rate": 9.910504348769118e-05,
"loss": 0.3535,
"step": 3220
},
{
"grad_norm": 0.7134031653404236,
"learning_rate": 9.909463230559127e-05,
"loss": 0.3402,
"step": 3230
},
{
"grad_norm": 0.7463231086730957,
"learning_rate": 9.908416146888376e-05,
"loss": 0.3268,
"step": 3240
},
{
"grad_norm": 0.7446751594543457,
"learning_rate": 9.907363099029175e-05,
"loss": 0.3315,
"step": 3250
},
{
"grad_norm": 0.7117018103599548,
"learning_rate": 9.906304088261073e-05,
"loss": 0.3264,
"step": 3260
},
{
"grad_norm": 0.7543739080429077,
"learning_rate": 9.905239115870872e-05,
"loss": 0.326,
"step": 3270
},
{
"grad_norm": 0.7214993834495544,
"learning_rate": 9.90416818315261e-05,
"loss": 0.3138,
"step": 3280
},
{
"grad_norm": 0.6942511796951294,
"learning_rate": 9.903091291407573e-05,
"loss": 0.3163,
"step": 3290
},
{
"grad_norm": 0.5961774587631226,
"learning_rate": 9.902008441944286e-05,
"loss": 0.3077,
"step": 3300
},
{
"grad_norm": 0.7515019774436951,
"learning_rate": 9.900919636078512e-05,
"loss": 0.3132,
"step": 3310
},
{
"grad_norm": 0.6976336240768433,
"learning_rate": 9.899824875133255e-05,
"loss": 0.3254,
"step": 3320
},
{
"grad_norm": 0.8105019927024841,
"learning_rate": 9.898724160438749e-05,
"loss": 0.3064,
"step": 3330
},
{
"grad_norm": 0.6638350486755371,
"learning_rate": 9.89761749333247e-05,
"loss": 0.3076,
"step": 3340
},
{
"grad_norm": 0.8114433288574219,
"learning_rate": 9.896504875159122e-05,
"loss": 0.3044,
"step": 3350
},
{
"grad_norm": 0.7312133312225342,
"learning_rate": 9.89538630727064e-05,
"loss": 0.3005,
"step": 3360
},
{
"grad_norm": 0.6905005574226379,
"learning_rate": 9.894261791026189e-05,
"loss": 0.3072,
"step": 3370
},
{
"grad_norm": 0.7884241938591003,
"learning_rate": 9.893131327792165e-05,
"loss": 0.3316,
"step": 3380
},
{
"grad_norm": 0.8003658652305603,
"learning_rate": 9.891994918942182e-05,
"loss": 0.2936,
"step": 3390
},
{
"grad_norm": 0.7533016800880432,
"learning_rate": 9.890852565857092e-05,
"loss": 0.3012,
"step": 3400
},
{
"grad_norm": 0.7855885028839111,
"learning_rate": 9.889704269924954e-05,
"loss": 0.2976,
"step": 3410
},
{
"grad_norm": 0.6067432165145874,
"learning_rate": 9.888550032541059e-05,
"loss": 0.2838,
"step": 3420
},
{
"grad_norm": 0.7256248593330383,
"learning_rate": 9.887389855107916e-05,
"loss": 0.2872,
"step": 3430
},
{
"grad_norm": 0.8641679883003235,
"learning_rate": 9.886223739035248e-05,
"loss": 0.2961,
"step": 3440
},
{
"grad_norm": 0.7299965023994446,
"learning_rate": 9.885051685739997e-05,
"loss": 0.2789,
"step": 3450
},
{
"grad_norm": 0.9069922566413879,
"learning_rate": 9.883873696646316e-05,
"loss": 0.274,
"step": 3460
},
{
"grad_norm": 0.7453370094299316,
"learning_rate": 9.882689773185575e-05,
"loss": 0.2914,
"step": 3470
},
{
"grad_norm": 0.6731999516487122,
"learning_rate": 9.881499916796353e-05,
"loss": 0.2901,
"step": 3480
},
{
"grad_norm": 0.7556237578392029,
"learning_rate": 9.880304128924434e-05,
"loss": 0.2786,
"step": 3490
},
{
"grad_norm": 0.6947938203811646,
"learning_rate": 9.879102411022817e-05,
"loss": 0.2673,
"step": 3500
},
{
"grad_norm": 0.7781992554664612,
"learning_rate": 9.877894764551703e-05,
"loss": 0.2745,
"step": 3510
},
{
"grad_norm": 0.7071310877799988,
"learning_rate": 9.876681190978494e-05,
"loss": 0.2717,
"step": 3520
},
{
"grad_norm": 0.7303004860877991,
"learning_rate": 9.875461691777797e-05,
"loss": 0.2744,
"step": 3530
},
{
"grad_norm": 0.7480106949806213,
"learning_rate": 9.874236268431417e-05,
"loss": 0.2757,
"step": 3540
},
{
"grad_norm": 0.7309539914131165,
"learning_rate": 9.873004922428361e-05,
"loss": 0.2727,
"step": 3550
},
{
"grad_norm": 0.7038681507110596,
"learning_rate": 9.871767655264829e-05,
"loss": 0.265,
"step": 3560
},
{
"grad_norm": 0.873786985874176,
"learning_rate": 9.87052446844422e-05,
"loss": 0.2763,
"step": 3570
},
{
"grad_norm": 0.7259548902511597,
"learning_rate": 9.869275363477122e-05,
"loss": 0.2838,
"step": 3580
},
{
"grad_norm": 0.9347019791603088,
"learning_rate": 9.868020341881312e-05,
"loss": 0.2859,
"step": 3590
},
{
"grad_norm": 0.843194305896759,
"learning_rate": 9.866759405181765e-05,
"loss": 0.2794,
"step": 3600
},
{
"grad_norm": 0.6475412249565125,
"learning_rate": 9.865492554910633e-05,
"loss": 0.2794,
"step": 3610
},
{
"grad_norm": 0.7971134185791016,
"learning_rate": 9.864219792607262e-05,
"loss": 0.2724,
"step": 3620
},
{
"grad_norm": 0.6854262948036194,
"learning_rate": 9.862941119818177e-05,
"loss": 0.2603,
"step": 3630
},
{
"grad_norm": 0.7395080924034119,
"learning_rate": 9.861656538097086e-05,
"loss": 0.2649,
"step": 3640
},
{
"grad_norm": 0.7772731184959412,
"learning_rate": 9.860366049004877e-05,
"loss": 0.2729,
"step": 3650
},
{
"grad_norm": 0.6858628392219543,
"learning_rate": 9.859069654109615e-05,
"loss": 0.2735,
"step": 3660
},
{
"grad_norm": 0.7978370785713196,
"learning_rate": 9.857767354986545e-05,
"loss": 0.2731,
"step": 3670
},
{
"grad_norm": 0.7753234505653381,
"learning_rate": 9.856459153218078e-05,
"loss": 0.2678,
"step": 3680
},
{
"grad_norm": 0.9582557678222656,
"learning_rate": 9.855145050393808e-05,
"loss": 0.2628,
"step": 3690
},
{
"grad_norm": 0.6748051643371582,
"learning_rate": 9.85382504811049e-05,
"loss": 0.2602,
"step": 3700
},
{
"grad_norm": 0.9243506193161011,
"learning_rate": 9.852499147972054e-05,
"loss": 0.2565,
"step": 3710
},
{
"grad_norm": 0.7661805152893066,
"learning_rate": 9.851167351589592e-05,
"loss": 0.2602,
"step": 3720
},
{
"grad_norm": 0.7572988867759705,
"learning_rate": 9.849829660581363e-05,
"loss": 0.246,
"step": 3730
},
{
"grad_norm": 0.7904701232910156,
"learning_rate": 9.848486076572787e-05,
"loss": 0.2466,
"step": 3740
},
{
"grad_norm": 0.8415496945381165,
"learning_rate": 9.847136601196446e-05,
"loss": 0.2559,
"step": 3750
},
{
"grad_norm": 0.6188539266586304,
"learning_rate": 9.845781236092078e-05,
"loss": 0.2454,
"step": 3760
},
{
"grad_norm": 0.9642737507820129,
"learning_rate": 9.844419982906583e-05,
"loss": 0.2586,
"step": 3770
},
{
"grad_norm": 0.6799070239067078,
"learning_rate": 9.843052843294008e-05,
"loss": 0.2465,
"step": 3780
},
{
"grad_norm": 0.6631637811660767,
"learning_rate": 9.841679818915559e-05,
"loss": 0.2444,
"step": 3790
},
{
"grad_norm": 0.7599271535873413,
"learning_rate": 9.840300911439591e-05,
"loss": 0.262,
"step": 3800
},
{
"grad_norm": 0.7756125926971436,
"learning_rate": 9.838916122541603e-05,
"loss": 0.2307,
"step": 3810
},
{
"grad_norm": 0.8162533044815063,
"learning_rate": 9.837525453904246e-05,
"loss": 0.2429,
"step": 3820
},
{
"grad_norm": 0.6748949885368347,
"learning_rate": 9.836128907217314e-05,
"loss": 0.2357,
"step": 3830
},
{
"grad_norm": 0.7403388023376465,
"learning_rate": 9.834726484177743e-05,
"loss": 0.2289,
"step": 3840
},
{
"grad_norm": 0.811571478843689,
"learning_rate": 9.833318186489609e-05,
"loss": 0.2486,
"step": 3850
},
{
"grad_norm": 0.787054717540741,
"learning_rate": 9.831904015864126e-05,
"loss": 0.232,
"step": 3860
},
{
"grad_norm": 0.9414629340171814,
"learning_rate": 9.830483974019645e-05,
"loss": 0.2305,
"step": 3870
},
{
"grad_norm": 0.6645330786705017,
"learning_rate": 9.82905806268165e-05,
"loss": 0.2207,
"step": 3880
},
{
"grad_norm": 0.6679629683494568,
"learning_rate": 9.82762628358276e-05,
"loss": 0.2353,
"step": 3890
},
{
"grad_norm": 0.7740006446838379,
"learning_rate": 9.826188638462718e-05,
"loss": 0.2347,
"step": 3900
},
{
"grad_norm": 0.716414213180542,
"learning_rate": 9.824745129068402e-05,
"loss": 0.2251,
"step": 3910
},
{
"grad_norm": 0.7028761506080627,
"learning_rate": 9.82329575715381e-05,
"loss": 0.2415,
"step": 3920
},
{
"grad_norm": 0.8053025603294373,
"learning_rate": 9.821840524480066e-05,
"loss": 0.2237,
"step": 3930
},
{
"grad_norm": 0.909868597984314,
"learning_rate": 9.820379432815414e-05,
"loss": 0.2385,
"step": 3940
},
{
"grad_norm": 0.9696028232574463,
"learning_rate": 9.81891248393522e-05,
"loss": 0.2289,
"step": 3950
},
{
"grad_norm": 0.7611873149871826,
"learning_rate": 9.817439679621963e-05,
"loss": 0.2335,
"step": 3960
},
{
"grad_norm": 0.820307731628418,
"learning_rate": 9.815961021665243e-05,
"loss": 0.2251,
"step": 3970
},
{
"grad_norm": 0.7758439779281616,
"learning_rate": 9.814476511861763e-05,
"loss": 0.2256,
"step": 3980
},
{
"grad_norm": 0.8019824028015137,
"learning_rate": 9.812986152015348e-05,
"loss": 0.2319,
"step": 3990
},
{
"grad_norm": 0.8140775561332703,
"learning_rate": 9.811489943936922e-05,
"loss": 0.2286,
"step": 4000
},
{
"grad_norm": 0.7253543734550476,
"learning_rate": 9.809987889444522e-05,
"loss": 0.2179,
"step": 4010
},
{
"grad_norm": 0.7287288904190063,
"learning_rate": 9.808479990363282e-05,
"loss": 0.2214,
"step": 4020
},
{
"grad_norm": 0.8032832741737366,
"learning_rate": 9.806966248525445e-05,
"loss": 0.225,
"step": 4030
},
{
"grad_norm": 0.6941774487495422,
"learning_rate": 9.805446665770348e-05,
"loss": 0.1991,
"step": 4040
},
{
"grad_norm": 0.7072444558143616,
"learning_rate": 9.803921243944429e-05,
"loss": 0.2116,
"step": 4050
},
{
"grad_norm": 0.731121301651001,
"learning_rate": 9.802389984901218e-05,
"loss": 0.2158,
"step": 4060
},
{
"grad_norm": 0.6477376222610474,
"learning_rate": 9.80085289050134e-05,
"loss": 0.2036,
"step": 4070
},
{
"grad_norm": 0.7232305407524109,
"learning_rate": 9.799309962612508e-05,
"loss": 0.2151,
"step": 4080
},
{
"grad_norm": 0.7788245677947998,
"learning_rate": 9.797761203109527e-05,
"loss": 0.2083,
"step": 4090
},
{
"grad_norm": 0.7749330401420593,
"learning_rate": 9.796206613874283e-05,
"loss": 0.2027,
"step": 4100
},
{
"grad_norm": 0.7190510034561157,
"learning_rate": 9.794646196795754e-05,
"loss": 0.2134,
"step": 4110
},
{
"grad_norm": 0.7808344960212708,
"learning_rate": 9.793079953769987e-05,
"loss": 0.2132,
"step": 4120
},
{
"grad_norm": 0.7859423160552979,
"learning_rate": 9.79150788670012e-05,
"loss": 0.2091,
"step": 4130
},
{
"grad_norm": 0.736219048500061,
"learning_rate": 9.78992999749636e-05,
"loss": 0.2036,
"step": 4140
},
{
"grad_norm": 0.7088515758514404,
"learning_rate": 9.788346288075994e-05,
"loss": 0.2138,
"step": 4150
},
{
"grad_norm": 0.849895715713501,
"learning_rate": 9.786756760363373e-05,
"loss": 0.2025,
"step": 4160
},
{
"grad_norm": 0.8397395014762878,
"learning_rate": 9.78516141628993e-05,
"loss": 0.2031,
"step": 4170
},
{
"grad_norm": 0.748094379901886,
"learning_rate": 9.783560257794154e-05,
"loss": 0.1984,
"step": 4180
},
{
"grad_norm": 0.7007679343223572,
"learning_rate": 9.781953286821603e-05,
"loss": 0.2024,
"step": 4190
},
{
"grad_norm": 0.8917890191078186,
"learning_rate": 9.780340505324901e-05,
"loss": 0.2038,
"step": 4200
},
{
"grad_norm": 0.6705322861671448,
"learning_rate": 9.778721915263727e-05,
"loss": 0.2,
"step": 4210
},
{
"grad_norm": 0.8025335669517517,
"learning_rate": 9.777097518604824e-05,
"loss": 0.2161,
"step": 4220
},
{
"grad_norm": 0.8322722911834717,
"learning_rate": 9.775467317321984e-05,
"loss": 0.2203,
"step": 4230
},
{
"grad_norm": 0.6298625469207764,
"learning_rate": 9.773831313396055e-05,
"loss": 0.2164,
"step": 4240
},
{
"grad_norm": 0.7074111104011536,
"learning_rate": 9.77218950881494e-05,
"loss": 0.2071,
"step": 4250
},
{
"grad_norm": 0.666023850440979,
"learning_rate": 9.770541905573583e-05,
"loss": 0.2137,
"step": 4260
},
{
"grad_norm": 0.7591691613197327,
"learning_rate": 9.768888505673976e-05,
"loss": 0.2071,
"step": 4270
},
{
"grad_norm": 0.7407532930374146,
"learning_rate": 9.767229311125162e-05,
"loss": 0.1982,
"step": 4280
},
{
"grad_norm": 0.7608945965766907,
"learning_rate": 9.765564323943211e-05,
"loss": 0.2003,
"step": 4290
},
{
"grad_norm": 0.7440188527107239,
"learning_rate": 9.763893546151244e-05,
"loss": 0.1941,
"step": 4300
},
{
"grad_norm": 0.7300034761428833,
"learning_rate": 9.762216979779412e-05,
"loss": 0.1951,
"step": 4310
},
{
"grad_norm": 0.834522008895874,
"learning_rate": 9.760534626864902e-05,
"loss": 0.2216,
"step": 4320
},
{
"grad_norm": 1.0703771114349365,
"learning_rate": 9.758846489451931e-05,
"loss": 0.1892,
"step": 4330
},
{
"grad_norm": 0.7994539141654968,
"learning_rate": 9.757152569591748e-05,
"loss": 0.1866,
"step": 4340
},
{
"grad_norm": 0.7248221635818481,
"learning_rate": 9.75545286934262e-05,
"loss": 0.1776,
"step": 4350
},
{
"grad_norm": 0.6876189708709717,
"learning_rate": 9.753747390769847e-05,
"loss": 0.1852,
"step": 4360
},
{
"grad_norm": 0.7878427505493164,
"learning_rate": 9.752036135945744e-05,
"loss": 0.1896,
"step": 4370
},
{
"grad_norm": 1.0356807708740234,
"learning_rate": 9.750319106949649e-05,
"loss": 0.1889,
"step": 4380
},
{
"grad_norm": 0.7777647376060486,
"learning_rate": 9.748596305867913e-05,
"loss": 0.1689,
"step": 4390
},
{
"grad_norm": 0.6953495740890503,
"learning_rate": 9.746867734793903e-05,
"loss": 0.1843,
"step": 4400
},
{
"grad_norm": 0.6768542528152466,
"learning_rate": 9.745133395827993e-05,
"loss": 0.1764,
"step": 4410
},
{
"grad_norm": 0.6815414428710938,
"learning_rate": 9.743393291077572e-05,
"loss": 0.1922,
"step": 4420
},
{
"grad_norm": 0.7767486572265625,
"learning_rate": 9.741647422657028e-05,
"loss": 0.1766,
"step": 4430
},
{
"grad_norm": 0.9260280728340149,
"learning_rate": 9.739895792687758e-05,
"loss": 0.1726,
"step": 4440
},
{
"grad_norm": 0.8812766671180725,
"learning_rate": 9.738138403298157e-05,
"loss": 0.1848,
"step": 4450
},
{
"grad_norm": 0.8485357165336609,
"learning_rate": 9.736375256623619e-05,
"loss": 0.1761,
"step": 4460
},
{
"grad_norm": 0.8344379663467407,
"learning_rate": 9.734606354806533e-05,
"loss": 0.1728,
"step": 4470
},
{
"grad_norm": 0.759541392326355,
"learning_rate": 9.73283169999628e-05,
"loss": 0.1789,
"step": 4480
},
{
"grad_norm": 0.7452002763748169,
"learning_rate": 9.731051294349238e-05,
"loss": 0.1752,
"step": 4490
},
{
"grad_norm": 0.7703397870063782,
"learning_rate": 9.729265140028762e-05,
"loss": 0.164,
"step": 4500
},
{
"grad_norm": 0.7867411971092224,
"learning_rate": 9.727473239205201e-05,
"loss": 0.1931,
"step": 4510
},
{
"grad_norm": 0.765819787979126,
"learning_rate": 9.725675594055883e-05,
"loss": 0.1688,
"step": 4520
},
{
"grad_norm": 0.849748969078064,
"learning_rate": 9.723872206765116e-05,
"loss": 0.1626,
"step": 4530
},
{
"grad_norm": 0.7978370189666748,
"learning_rate": 9.722063079524185e-05,
"loss": 0.1783,
"step": 4540
},
{
"grad_norm": 0.8076903223991394,
"learning_rate": 9.720248214531351e-05,
"loss": 0.1679,
"step": 4550
},
{
"grad_norm": 0.8124512434005737,
"learning_rate": 9.718427613991848e-05,
"loss": 0.1535,
"step": 4560
},
{
"grad_norm": 0.8864796161651611,
"learning_rate": 9.716601280117873e-05,
"loss": 0.1688,
"step": 4570
},
{
"grad_norm": 0.947771430015564,
"learning_rate": 9.714769215128596e-05,
"loss": 0.1591,
"step": 4580
},
{
"grad_norm": 0.6856046319007874,
"learning_rate": 9.712931421250152e-05,
"loss": 0.1491,
"step": 4590
},
{
"grad_norm": 1.0799272060394287,
"learning_rate": 9.711087900715627e-05,
"loss": 0.1508,
"step": 4600
},
{
"grad_norm": 0.9266803860664368,
"learning_rate": 9.709238655765078e-05,
"loss": 0.1497,
"step": 4610
},
{
"grad_norm": 0.891480028629303,
"learning_rate": 9.707383688645511e-05,
"loss": 0.1411,
"step": 4620
},
{
"grad_norm": 0.866604745388031,
"learning_rate": 9.705523001610883e-05,
"loss": 0.1513,
"step": 4630
},
{
"grad_norm": 0.9172372221946716,
"learning_rate": 9.703656596922107e-05,
"loss": 0.1519,
"step": 4640
},
{
"grad_norm": 0.7589128017425537,
"learning_rate": 9.70178447684704e-05,
"loss": 0.1361,
"step": 4650
},
{
"grad_norm": 0.7801403403282166,
"learning_rate": 9.699906643660483e-05,
"loss": 0.144,
"step": 4660
},
{
"grad_norm": 0.8865439891815186,
"learning_rate": 9.698023099644185e-05,
"loss": 0.1438,
"step": 4670
},
{
"grad_norm": 0.8262112736701965,
"learning_rate": 9.696133847086823e-05,
"loss": 0.1401,
"step": 4680
},
{
"grad_norm": 0.9018675088882446,
"learning_rate": 9.694238888284022e-05,
"loss": 0.1606,
"step": 4690
},
{
"grad_norm": 0.6830954551696777,
"learning_rate": 9.692338225538333e-05,
"loss": 0.1384,
"step": 4700
},
{
"grad_norm": 0.7323166131973267,
"learning_rate": 9.690431861159241e-05,
"loss": 0.1341,
"step": 4710
},
{
"grad_norm": 0.8567700386047363,
"learning_rate": 9.688519797463161e-05,
"loss": 0.1255,
"step": 4720
},
{
"grad_norm": 0.7289927005767822,
"learning_rate": 9.686602036773426e-05,
"loss": 0.1479,
"step": 4730
},
{
"grad_norm": 0.8099994659423828,
"learning_rate": 9.684678581420302e-05,
"loss": 0.1332,
"step": 4740
},
{
"grad_norm": 0.7734137773513794,
"learning_rate": 9.682749433740962e-05,
"loss": 0.1181,
"step": 4750
},
{
"grad_norm": 0.7820311784744263,
"learning_rate": 9.680814596079507e-05,
"loss": 0.1287,
"step": 4760
},
{
"grad_norm": 0.6994800567626953,
"learning_rate": 9.678874070786945e-05,
"loss": 0.1357,
"step": 4770
},
{
"grad_norm": 0.9670836925506592,
"learning_rate": 9.676927860221199e-05,
"loss": 0.1494,
"step": 4780
},
{
"grad_norm": 0.760356068611145,
"learning_rate": 9.674975966747097e-05,
"loss": 0.118,
"step": 4790
},
{
"grad_norm": 1.0088319778442383,
"learning_rate": 9.673018392736374e-05,
"loss": 0.1218,
"step": 4800
},
{
"grad_norm": 0.7568646669387817,
"learning_rate": 9.671055140567667e-05,
"loss": 0.1218,
"step": 4810
},
{
"grad_norm": 0.8963031768798828,
"learning_rate": 9.669086212626511e-05,
"loss": 0.1243,
"step": 4820
},
{
"grad_norm": 0.896043062210083,
"learning_rate": 9.667111611305341e-05,
"loss": 0.1158,
"step": 4830
},
{
"grad_norm": 0.9382137060165405,
"learning_rate": 9.665131339003486e-05,
"loss": 0.1189,
"step": 4840
},
{
"grad_norm": 0.795920729637146,
"learning_rate": 9.663145398127158e-05,
"loss": 0.1106,
"step": 4850
},
{
"grad_norm": 1.0919344425201416,
"learning_rate": 9.661153791089467e-05,
"loss": 0.1038,
"step": 4860
},
{
"grad_norm": 0.8370903134346008,
"learning_rate": 9.659156520310402e-05,
"loss": 0.1045,
"step": 4870
},
{
"grad_norm": 0.8273882865905762,
"learning_rate": 9.657153588216835e-05,
"loss": 0.1046,
"step": 4880
},
{
"grad_norm": 0.862650454044342,
"learning_rate": 9.655144997242516e-05,
"loss": 0.1191,
"step": 4890
},
{
"grad_norm": 0.8968775868415833,
"learning_rate": 9.653130749828075e-05,
"loss": 0.0906,
"step": 4900
},
{
"grad_norm": 0.8365146517753601,
"learning_rate": 9.65111084842101e-05,
"loss": 0.096,
"step": 4910
},
{
"grad_norm": 1.0482263565063477,
"learning_rate": 9.649085295475695e-05,
"loss": 0.1002,
"step": 4920
},
{
"grad_norm": 0.9175655841827393,
"learning_rate": 9.647054093453365e-05,
"loss": 0.1005,
"step": 4930
},
{
"grad_norm": 0.8757518529891968,
"learning_rate": 9.645017244822123e-05,
"loss": 0.085,
"step": 4940
},
{
"grad_norm": 0.8668944835662842,
"learning_rate": 9.642974752056931e-05,
"loss": 0.0945,
"step": 4950
},
{
"grad_norm": 0.7647026777267456,
"learning_rate": 9.640926617639613e-05,
"loss": 0.0874,
"step": 4960
},
{
"grad_norm": 0.8276016116142273,
"learning_rate": 9.638872844058843e-05,
"loss": 0.1022,
"step": 4970
},
{
"grad_norm": 0.7343841195106506,
"learning_rate": 9.63681343381015e-05,
"loss": 0.0958,
"step": 4980
},
{
"grad_norm": 0.8836742043495178,
"learning_rate": 9.634748389395914e-05,
"loss": 0.091,
"step": 4990
},
{
"grad_norm": 0.9837483763694763,
"learning_rate": 9.632677713325353e-05,
"loss": 0.0951,
"step": 5000
},
{
"grad_norm": 0.8699268102645874,
"learning_rate": 9.63060140811454e-05,
"loss": 0.0917,
"step": 5010
},
{
"grad_norm": 0.822909414768219,
"learning_rate": 9.628519476286379e-05,
"loss": 0.0839,
"step": 5020
},
{
"grad_norm": 0.8628169894218445,
"learning_rate": 9.626431920370612e-05,
"loss": 0.0709,
"step": 5030
},
{
"grad_norm": 1.0187662839889526,
"learning_rate": 9.624338742903819e-05,
"loss": 0.0761,
"step": 5040
},
{
"grad_norm": 0.8035425543785095,
"learning_rate": 9.622239946429406e-05,
"loss": 0.0905,
"step": 5050
},
{
"grad_norm": 1.024082899093628,
"learning_rate": 9.620135533497609e-05,
"loss": 0.0613,
"step": 5060
},
{
"grad_norm": 0.8901798725128174,
"learning_rate": 9.61802550666549e-05,
"loss": 0.102,
"step": 5070
},
{
"grad_norm": 0.917160153388977,
"learning_rate": 9.615909868496928e-05,
"loss": 0.0861,
"step": 5080
},
{
"grad_norm": 0.7845200300216675,
"learning_rate": 9.613788621562622e-05,
"loss": 0.0712,
"step": 5090
},
{
"grad_norm": 0.7655529975891113,
"learning_rate": 9.611661768440093e-05,
"loss": 0.0702,
"step": 5100
},
{
"grad_norm": 0.8100429773330688,
"learning_rate": 9.609529311713661e-05,
"loss": 0.0787,
"step": 5110
},
{
"grad_norm": 0.8901064991950989,
"learning_rate": 9.607391253974466e-05,
"loss": 0.0925,
"step": 5120
},
{
"grad_norm": 0.8169835209846497,
"learning_rate": 9.605247597820448e-05,
"loss": 0.0915,
"step": 5130
},
{
"grad_norm": 0.814324676990509,
"learning_rate": 9.603098345856354e-05,
"loss": 0.0789,
"step": 5140
},
{
"grad_norm": 0.8646579384803772,
"learning_rate": 9.600943500693725e-05,
"loss": 0.0634,
"step": 5150
},
{
"grad_norm": 0.737809419631958,
"learning_rate": 9.598783064950902e-05,
"loss": 0.0892,
"step": 5160
},
{
"grad_norm": 0.8735009431838989,
"learning_rate": 9.596617041253018e-05,
"loss": 0.079,
"step": 5170
},
{
"grad_norm": 0.7089709043502808,
"learning_rate": 9.594445432231996e-05,
"loss": 0.0693,
"step": 5180
},
{
"grad_norm": 0.9700685739517212,
"learning_rate": 9.592268240526547e-05,
"loss": 0.0969,
"step": 5190
},
{
"grad_norm": 0.8081687688827515,
"learning_rate": 9.590085468782162e-05,
"loss": 0.0625,
"step": 5200
},
{
"grad_norm": 0.9224503040313721,
"learning_rate": 9.587897119651116e-05,
"loss": 0.08,
"step": 5210
},
{
"grad_norm": 0.9574503898620605,
"learning_rate": 9.585703195792459e-05,
"loss": 0.0736,
"step": 5220
},
{
"grad_norm": 0.8965017199516296,
"learning_rate": 9.583503699872016e-05,
"loss": 0.068,
"step": 5230
},
{
"grad_norm": 0.8844852447509766,
"learning_rate": 9.581298634562381e-05,
"loss": 0.0703,
"step": 5240
},
{
"grad_norm": 0.8397418856620789,
"learning_rate": 9.579088002542917e-05,
"loss": 0.0947,
"step": 5250
},
{
"grad_norm": 0.9070932865142822,
"learning_rate": 9.57687180649975e-05,
"loss": 0.07,
"step": 5260
},
{
"grad_norm": 0.8467302918434143,
"learning_rate": 9.574650049125768e-05,
"loss": 0.0747,
"step": 5270
},
{
"grad_norm": 0.9094961881637573,
"learning_rate": 9.572422733120614e-05,
"loss": 0.0779,
"step": 5280
},
{
"grad_norm": 0.851626992225647,
"learning_rate": 9.570189861190689e-05,
"loss": 0.0808,
"step": 5290
},
{
"grad_norm": 0.7709887623786926,
"learning_rate": 9.56795143604914e-05,
"loss": 0.0645,
"step": 5300
},
{
"grad_norm": 0.7786981463432312,
"learning_rate": 9.565707460415869e-05,
"loss": 0.0715,
"step": 5310
},
{
"grad_norm": 0.6938153505325317,
"learning_rate": 9.563457937017515e-05,
"loss": 0.0792,
"step": 5320
},
{
"grad_norm": 0.9235035181045532,
"learning_rate": 9.56120286858746e-05,
"loss": 0.0855,
"step": 5330
},
{
"grad_norm": 0.8447647094726562,
"learning_rate": 9.558942257865829e-05,
"loss": 0.0709,
"step": 5340
},
{
"grad_norm": 0.8521619439125061,
"learning_rate": 9.556676107599472e-05,
"loss": 0.0589,
"step": 5350
},
{
"grad_norm": 0.663080632686615,
"learning_rate": 9.554404420541978e-05,
"loss": 0.0668,
"step": 5360
},
{
"grad_norm": 0.8164395689964294,
"learning_rate": 9.55212719945366e-05,
"loss": 0.0654,
"step": 5370
},
{
"grad_norm": 0.7483910918235779,
"learning_rate": 9.549844447101559e-05,
"loss": 0.0643,
"step": 5380
},
{
"grad_norm": 0.8366538882255554,
"learning_rate": 9.547556166259433e-05,
"loss": 0.0659,
"step": 5390
},
{
"grad_norm": 0.8382837772369385,
"learning_rate": 9.545262359707756e-05,
"loss": 0.0655,
"step": 5400
},
{
"grad_norm": 0.7271720767021179,
"learning_rate": 9.542963030233724e-05,
"loss": 0.0604,
"step": 5410
},
{
"grad_norm": 0.8077654242515564,
"learning_rate": 9.540658180631237e-05,
"loss": 0.0595,
"step": 5420
},
{
"grad_norm": 0.6384038329124451,
"learning_rate": 9.538347813700904e-05,
"loss": 0.0565,
"step": 5430
},
{
"grad_norm": 0.8646925687789917,
"learning_rate": 9.536031932250036e-05,
"loss": 0.0825,
"step": 5440
},
{
"grad_norm": 0.7276682257652283,
"learning_rate": 9.533710539092653e-05,
"loss": 0.0559,
"step": 5450
},
{
"grad_norm": 0.7854685187339783,
"learning_rate": 9.531383637049464e-05,
"loss": 0.0567,
"step": 5460
},
{
"grad_norm": 0.8306655287742615,
"learning_rate": 9.529051228947875e-05,
"loss": 0.0527,
"step": 5470
},
{
"grad_norm": 0.9192001223564148,
"learning_rate": 9.52671331762198e-05,
"loss": 0.0627,
"step": 5480
},
{
"grad_norm": 0.9559565186500549,
"learning_rate": 9.524369905912565e-05,
"loss": 0.0605,
"step": 5490
},
{
"grad_norm": 0.8344584703445435,
"learning_rate": 9.522020996667092e-05,
"loss": 0.0672,
"step": 5500
},
{
"grad_norm": 0.6726586818695068,
"learning_rate": 9.519666592739709e-05,
"loss": 0.0639,
"step": 5510
},
{
"grad_norm": 0.9621972441673279,
"learning_rate": 9.517306696991241e-05,
"loss": 0.0567,
"step": 5520
},
{
"grad_norm": 0.7269927263259888,
"learning_rate": 9.51494131228918e-05,
"loss": 0.0596,
"step": 5530
},
{
"grad_norm": 0.870598316192627,
"learning_rate": 9.512570441507695e-05,
"loss": 0.0564,
"step": 5540
},
{
"grad_norm": 0.8606002330780029,
"learning_rate": 9.510194087527615e-05,
"loss": 0.0581,
"step": 5550
},
{
"grad_norm": 0.7414759397506714,
"learning_rate": 9.507812253236435e-05,
"loss": 0.0569,
"step": 5560
},
{
"grad_norm": 0.8925358057022095,
"learning_rate": 9.505424941528309e-05,
"loss": 0.0607,
"step": 5570
},
{
"grad_norm": 0.751079797744751,
"learning_rate": 9.503032155304046e-05,
"loss": 0.075,
"step": 5580
},
{
"grad_norm": 0.8519660830497742,
"learning_rate": 9.500633897471106e-05,
"loss": 0.0611,
"step": 5590
},
{
"grad_norm": 0.7957033514976501,
"learning_rate": 9.498230170943596e-05,
"loss": 0.0745,
"step": 5600
},
{
"grad_norm": 0.8079347610473633,
"learning_rate": 9.495820978642275e-05,
"loss": 0.0635,
"step": 5610
},
{
"grad_norm": 0.795558512210846,
"learning_rate": 9.493406323494535e-05,
"loss": 0.0544,
"step": 5620
},
{
"grad_norm": 0.8301962614059448,
"learning_rate": 9.490986208434413e-05,
"loss": 0.0515,
"step": 5630
},
{
"grad_norm": 0.913617730140686,
"learning_rate": 9.488560636402577e-05,
"loss": 0.0596,
"step": 5640
},
{
"grad_norm": 0.7896845936775208,
"learning_rate": 9.486129610346321e-05,
"loss": 0.0699,
"step": 5650
},
{
"grad_norm": 0.777924120426178,
"learning_rate": 9.483693133219576e-05,
"loss": 0.0581,
"step": 5660
},
{
"grad_norm": 0.9253566265106201,
"learning_rate": 9.481251207982888e-05,
"loss": 0.0589,
"step": 5670
},
{
"grad_norm": 0.7859442234039307,
"learning_rate": 9.47880383760343e-05,
"loss": 0.0658,
"step": 5680
},
{
"grad_norm": 0.8194724917411804,
"learning_rate": 9.476351025054983e-05,
"loss": 0.0595,
"step": 5690
},
{
"grad_norm": 0.7006014585494995,
"learning_rate": 9.473892773317952e-05,
"loss": 0.0638,
"step": 5700
},
{
"grad_norm": 0.8078575730323792,
"learning_rate": 9.471429085379338e-05,
"loss": 0.0792,
"step": 5710
},
{
"grad_norm": 0.7397284507751465,
"learning_rate": 9.468959964232757e-05,
"loss": 0.0583,
"step": 5720
},
{
"grad_norm": 0.7037245631217957,
"learning_rate": 9.466485412878425e-05,
"loss": 0.0606,
"step": 5730
},
{
"grad_norm": 0.7415469288825989,
"learning_rate": 9.464005434323154e-05,
"loss": 0.0763,
"step": 5740
},
{
"grad_norm": 0.6046724319458008,
"learning_rate": 9.461520031580352e-05,
"loss": 0.0572,
"step": 5750
},
{
"grad_norm": 0.6844933032989502,
"learning_rate": 9.459029207670019e-05,
"loss": 0.062,
"step": 5760
},
{
"grad_norm": 0.7399989366531372,
"learning_rate": 9.456532965618737e-05,
"loss": 0.0611,
"step": 5770
},
{
"grad_norm": 0.7355150580406189,
"learning_rate": 9.454031308459681e-05,
"loss": 0.0751,
"step": 5780
},
{
"grad_norm": 0.8606380820274353,
"learning_rate": 9.451524239232595e-05,
"loss": 0.0597,
"step": 5790
},
{
"grad_norm": 0.8436996936798096,
"learning_rate": 9.449011760983809e-05,
"loss": 0.0705,
"step": 5800
},
{
"grad_norm": 0.6689172387123108,
"learning_rate": 9.446493876766218e-05,
"loss": 0.0626,
"step": 5810
},
{
"grad_norm": 0.7019214630126953,
"learning_rate": 9.44397058963929e-05,
"loss": 0.0646,
"step": 5820
},
{
"grad_norm": 0.7588643431663513,
"learning_rate": 9.441441902669056e-05,
"loss": 0.0661,
"step": 5830
},
{
"grad_norm": 0.7251266241073608,
"learning_rate": 9.43890781892811e-05,
"loss": 0.0605,
"step": 5840
},
{
"grad_norm": 0.841923177242279,
"learning_rate": 9.436368341495603e-05,
"loss": 0.0528,
"step": 5850
},
{
"grad_norm": 0.8010355830192566,
"learning_rate": 9.43382347345724e-05,
"loss": 0.062,
"step": 5860
},
{
"grad_norm": 0.6700798869132996,
"learning_rate": 9.431273217905273e-05,
"loss": 0.0633,
"step": 5870
},
{
"grad_norm": 0.8948337435722351,
"learning_rate": 9.428717577938504e-05,
"loss": 0.073,
"step": 5880
},
{
"grad_norm": 0.7211799025535583,
"learning_rate": 9.426156556662276e-05,
"loss": 0.0647,
"step": 5890
},
{
"grad_norm": 0.7889621257781982,
"learning_rate": 9.423590157188474e-05,
"loss": 0.0826,
"step": 5900
},
{
"grad_norm": 0.6979097723960876,
"learning_rate": 9.421018382635513e-05,
"loss": 0.0744,
"step": 5910
},
{
"grad_norm": 0.7292050123214722,
"learning_rate": 9.418441236128343e-05,
"loss": 0.0565,
"step": 5920
},
{
"grad_norm": 0.7622458338737488,
"learning_rate": 9.41585872079844e-05,
"loss": 0.0526,
"step": 5930
},
{
"grad_norm": 0.7774178385734558,
"learning_rate": 9.413270839783802e-05,
"loss": 0.059,
"step": 5940
},
{
"grad_norm": 0.9345079064369202,
"learning_rate": 9.41067759622895e-05,
"loss": 0.0591,
"step": 5950
},
{
"grad_norm": 0.6906028985977173,
"learning_rate": 9.408078993284917e-05,
"loss": 0.0504,
"step": 5960
},
{
"grad_norm": 0.7807912826538086,
"learning_rate": 9.405475034109254e-05,
"loss": 0.0642,
"step": 5970
},
{
"grad_norm": 0.7987840175628662,
"learning_rate": 9.402865721866015e-05,
"loss": 0.0596,
"step": 5980
},
{
"grad_norm": 0.7883781790733337,
"learning_rate": 9.400251059725762e-05,
"loss": 0.0812,
"step": 5990
},
{
"grad_norm": 0.7730071544647217,
"learning_rate": 9.397631050865554e-05,
"loss": 0.0828,
"step": 6000
},
{
"grad_norm": 0.8008111119270325,
"learning_rate": 9.395005698468949e-05,
"loss": 0.0612,
"step": 6010
},
{
"grad_norm": 0.7160155773162842,
"learning_rate": 9.392375005726e-05,
"loss": 0.0601,
"step": 6020
},
{
"grad_norm": 0.8973404765129089,
"learning_rate": 9.389738975833242e-05,
"loss": 0.0639,
"step": 6030
},
{
"grad_norm": 0.7334786653518677,
"learning_rate": 9.387097611993707e-05,
"loss": 0.0579,
"step": 6040
},
{
"grad_norm": 0.7777971625328064,
"learning_rate": 9.384450917416894e-05,
"loss": 0.0642,
"step": 6050
},
{
"grad_norm": 0.6371760368347168,
"learning_rate": 9.381798895318792e-05,
"loss": 0.0502,
"step": 6060
},
{
"grad_norm": 0.6775743365287781,
"learning_rate": 9.379141548921853e-05,
"loss": 0.0596,
"step": 6070
},
{
"grad_norm": 0.7282953858375549,
"learning_rate": 9.376478881455009e-05,
"loss": 0.0692,
"step": 6080
},
{
"grad_norm": 0.773247241973877,
"learning_rate": 9.373810896153647e-05,
"loss": 0.0639,
"step": 6090
},
{
"grad_norm": 0.7960075736045837,
"learning_rate": 9.371137596259623e-05,
"loss": 0.0733,
"step": 6100
},
{
"grad_norm": 0.8895021080970764,
"learning_rate": 9.368458985021248e-05,
"loss": 0.0487,
"step": 6110
},
{
"grad_norm": 0.7914479970932007,
"learning_rate": 9.365775065693287e-05,
"loss": 0.0565,
"step": 6120
},
{
"grad_norm": 0.7280433177947998,
"learning_rate": 9.363085841536957e-05,
"loss": 0.0498,
"step": 6130
},
{
"grad_norm": 0.7270240783691406,
"learning_rate": 9.360391315819917e-05,
"loss": 0.0527,
"step": 6140
},
{
"grad_norm": 0.7264328002929688,
"learning_rate": 9.357691491816269e-05,
"loss": 0.0737,
"step": 6150
},
{
"grad_norm": 0.7558386921882629,
"learning_rate": 9.354986372806557e-05,
"loss": 0.0788,
"step": 6160
},
{
"grad_norm": 0.7712228298187256,
"learning_rate": 9.352275962077752e-05,
"loss": 0.0662,
"step": 6170
},
{
"grad_norm": 0.8861271142959595,
"learning_rate": 9.349560262923262e-05,
"loss": 0.0535,
"step": 6180
},
{
"grad_norm": 0.7875964045524597,
"learning_rate": 9.346839278642913e-05,
"loss": 0.0722,
"step": 6190
},
{
"grad_norm": 0.6420438289642334,
"learning_rate": 9.344113012542963e-05,
"loss": 0.0521,
"step": 6200
},
{
"grad_norm": 0.8272718191146851,
"learning_rate": 9.341381467936079e-05,
"loss": 0.0632,
"step": 6210
},
{
"grad_norm": 0.9079589247703552,
"learning_rate": 9.338644648141346e-05,
"loss": 0.0505,
"step": 6220
},
{
"grad_norm": 0.7319368720054626,
"learning_rate": 9.335902556484257e-05,
"loss": 0.0608,
"step": 6230
},
{
"grad_norm": 0.7756045460700989,
"learning_rate": 9.333155196296712e-05,
"loss": 0.0661,
"step": 6240
},
{
"grad_norm": 0.794601559638977,
"learning_rate": 9.330402570917016e-05,
"loss": 0.0685,
"step": 6250
},
{
"grad_norm": 0.6460059881210327,
"learning_rate": 9.327644683689865e-05,
"loss": 0.0484,
"step": 6260
},
{
"grad_norm": 0.7627430558204651,
"learning_rate": 9.324881537966354e-05,
"loss": 0.0558,
"step": 6270
},
{
"grad_norm": 0.706190824508667,
"learning_rate": 9.322113137103964e-05,
"loss": 0.0581,
"step": 6280
},
{
"grad_norm": 0.7619665861129761,
"learning_rate": 9.319339484466564e-05,
"loss": 0.0591,
"step": 6290
},
{
"grad_norm": 0.7956165671348572,
"learning_rate": 9.316560583424403e-05,
"loss": 0.0588,
"step": 6300
},
{
"grad_norm": 0.7355234026908875,
"learning_rate": 9.313776437354108e-05,
"loss": 0.0517,
"step": 6310
},
{
"grad_norm": 0.7918558716773987,
"learning_rate": 9.31098704963868e-05,
"loss": 0.05,
"step": 6320
},
{
"grad_norm": 0.6687741875648499,
"learning_rate": 9.308192423667485e-05,
"loss": 0.0511,
"step": 6330
},
{
"grad_norm": 0.5716345906257629,
"learning_rate": 9.305392562836262e-05,
"loss": 0.0656,
"step": 6340
},
{
"grad_norm": 0.7398003339767456,
"learning_rate": 9.3025874705471e-05,
"loss": 0.0539,
"step": 6350
},
{
"grad_norm": 0.8083683252334595,
"learning_rate": 9.299777150208457e-05,
"loss": 0.0513,
"step": 6360
},
{
"grad_norm": 0.744621992111206,
"learning_rate": 9.296961605235133e-05,
"loss": 0.0533,
"step": 6370
},
{
"grad_norm": 0.7401552796363831,
"learning_rate": 9.29414083904828e-05,
"loss": 0.0502,
"step": 6380
},
{
"grad_norm": 0.7192158102989197,
"learning_rate": 9.291314855075397e-05,
"loss": 0.062,
"step": 6390
},
{
"grad_norm": 0.7841793298721313,
"learning_rate": 9.288483656750322e-05,
"loss": 0.0588,
"step": 6400
},
{
"grad_norm": 0.8342483043670654,
"learning_rate": 9.285647247513225e-05,
"loss": 0.0456,
"step": 6410
},
{
"grad_norm": 0.7881906628608704,
"learning_rate": 9.282805630810614e-05,
"loss": 0.0583,
"step": 6420
},
{
"grad_norm": 0.9873470664024353,
"learning_rate": 9.279958810095317e-05,
"loss": 0.0519,
"step": 6430
},
{
"grad_norm": 0.6749179363250732,
"learning_rate": 9.277106788826494e-05,
"loss": 0.053,
"step": 6440
},
{
"grad_norm": 0.6616987586021423,
"learning_rate": 9.274249570469618e-05,
"loss": 0.0499,
"step": 6450
},
{
"grad_norm": 0.809191882610321,
"learning_rate": 9.271387158496476e-05,
"loss": 0.0499,
"step": 6460
},
{
"grad_norm": 0.6982690095901489,
"learning_rate": 9.268519556385173e-05,
"loss": 0.0496,
"step": 6470
},
{
"grad_norm": 0.8391900062561035,
"learning_rate": 9.265646767620113e-05,
"loss": 0.0489,
"step": 6480
},
{
"grad_norm": 0.7313620448112488,
"learning_rate": 9.262768795692006e-05,
"loss": 0.06,
"step": 6490
},
{
"grad_norm": 0.7192261219024658,
"learning_rate": 9.259885644097861e-05,
"loss": 0.0432,
"step": 6500
},
{
"grad_norm": 0.7688872814178467,
"learning_rate": 9.256997316340976e-05,
"loss": 0.0432,
"step": 6510
},
{
"grad_norm": 0.8148354887962341,
"learning_rate": 9.254103815930943e-05,
"loss": 0.0537,
"step": 6520
},
{
"grad_norm": 0.7512215971946716,
"learning_rate": 9.251205146383636e-05,
"loss": 0.0487,
"step": 6530
},
{
"grad_norm": 0.7448429465293884,
"learning_rate": 9.248301311221216e-05,
"loss": 0.0572,
"step": 6540
},
{
"grad_norm": 0.6850093007087708,
"learning_rate": 9.245392313972115e-05,
"loss": 0.0459,
"step": 6550
},
{
"grad_norm": 0.8869002461433411,
"learning_rate": 9.242478158171037e-05,
"loss": 0.0533,
"step": 6560
},
{
"grad_norm": 0.8239165544509888,
"learning_rate": 9.239558847358957e-05,
"loss": 0.0572,
"step": 6570
},
{
"grad_norm": 0.6669193506240845,
"learning_rate": 9.236634385083114e-05,
"loss": 0.0412,
"step": 6580
},
{
"grad_norm": 0.8102614879608154,
"learning_rate": 9.233704774897005e-05,
"loss": 0.0618,
"step": 6590
},
{
"grad_norm": 0.7019132971763611,
"learning_rate": 9.230770020360382e-05,
"loss": 0.0582,
"step": 6600
},
{
"grad_norm": 0.6381471157073975,
"learning_rate": 9.227830125039247e-05,
"loss": 0.0441,
"step": 6610
},
{
"grad_norm": 0.7567533850669861,
"learning_rate": 9.224885092505853e-05,
"loss": 0.0428,
"step": 6620
},
{
"grad_norm": 0.7329815030097961,
"learning_rate": 9.221934926338691e-05,
"loss": 0.0444,
"step": 6630
},
{
"grad_norm": 0.9483146071434021,
"learning_rate": 9.21897963012249e-05,
"loss": 0.0441,
"step": 6640
},
{
"grad_norm": 0.7093248963356018,
"learning_rate": 9.216019207448217e-05,
"loss": 0.0615,
"step": 6650
},
{
"grad_norm": 0.6419187784194946,
"learning_rate": 9.21305366191306e-05,
"loss": 0.0409,
"step": 6660
},
{
"grad_norm": 0.5810886025428772,
"learning_rate": 9.21008299712044e-05,
"loss": 0.0388,
"step": 6670
},
{
"grad_norm": 0.7291533350944519,
"learning_rate": 9.207107216679993e-05,
"loss": 0.058,
"step": 6680
},
{
"grad_norm": 0.8118759989738464,
"learning_rate": 9.204126324207574e-05,
"loss": 0.054,
"step": 6690
},
{
"grad_norm": 0.6670665144920349,
"learning_rate": 9.201140323325248e-05,
"loss": 0.0552,
"step": 6700
},
{
"grad_norm": 0.6652207374572754,
"learning_rate": 9.198149217661287e-05,
"loss": 0.0578,
"step": 6710
},
{
"grad_norm": 0.7636727094650269,
"learning_rate": 9.195153010850167e-05,
"loss": 0.0604,
"step": 6720
},
{
"grad_norm": 0.7503575086593628,
"learning_rate": 9.192151706532562e-05,
"loss": 0.0538,
"step": 6730
},
{
"grad_norm": 0.6445121169090271,
"learning_rate": 9.189145308355337e-05,
"loss": 0.0794,
"step": 6740
},
{
"grad_norm": 0.6689440011978149,
"learning_rate": 9.186133819971555e-05,
"loss": 0.0504,
"step": 6750
},
{
"grad_norm": 0.7566612958908081,
"learning_rate": 9.183117245040455e-05,
"loss": 0.0487,
"step": 6760
},
{
"grad_norm": 0.7625004053115845,
"learning_rate": 9.180095587227458e-05,
"loss": 0.0454,
"step": 6770
},
{
"grad_norm": 0.7175439596176147,
"learning_rate": 9.177068850204167e-05,
"loss": 0.057,
"step": 6780
},
{
"grad_norm": 0.7054648399353027,
"learning_rate": 9.174037037648351e-05,
"loss": 0.0459,
"step": 6790
},
{
"grad_norm": 0.5674930214881897,
"learning_rate": 9.171000153243947e-05,
"loss": 0.055,
"step": 6800
},
{
"grad_norm": 0.7952650785446167,
"learning_rate": 9.167958200681057e-05,
"loss": 0.0407,
"step": 6810
},
{
"grad_norm": 0.8852372169494629,
"learning_rate": 9.164911183655943e-05,
"loss": 0.0584,
"step": 6820
},
{
"grad_norm": 0.7131014466285706,
"learning_rate": 9.161859105871013e-05,
"loss": 0.0545,
"step": 6830
},
{
"grad_norm": 0.6972997784614563,
"learning_rate": 9.158801971034832e-05,
"loss": 0.0563,
"step": 6840
},
{
"grad_norm": 0.762372612953186,
"learning_rate": 9.155739782862107e-05,
"loss": 0.0483,
"step": 6850
},
{
"grad_norm": 0.5963253974914551,
"learning_rate": 9.152672545073687e-05,
"loss": 0.0432,
"step": 6860
},
{
"grad_norm": 0.7276193499565125,
"learning_rate": 9.149600261396551e-05,
"loss": 0.0649,
"step": 6870
},
{
"grad_norm": 0.6192886233329773,
"learning_rate": 9.146522935563817e-05,
"loss": 0.0578,
"step": 6880
},
{
"grad_norm": 0.7328228950500488,
"learning_rate": 9.143440571314723e-05,
"loss": 0.0505,
"step": 6890
},
{
"grad_norm": 0.7885982394218445,
"learning_rate": 9.140353172394636e-05,
"loss": 0.0561,
"step": 6900
},
{
"grad_norm": 0.819013774394989,
"learning_rate": 9.137260742555032e-05,
"loss": 0.045,
"step": 6910
},
{
"grad_norm": 0.8304609060287476,
"learning_rate": 9.13416328555351e-05,
"loss": 0.0575,
"step": 6920
},
{
"grad_norm": 0.7522127628326416,
"learning_rate": 9.131060805153769e-05,
"loss": 0.057,
"step": 6930
},
{
"grad_norm": 0.7928993105888367,
"learning_rate": 9.127953305125618e-05,
"loss": 0.043,
"step": 6940
},
{
"grad_norm": 0.8068320751190186,
"learning_rate": 9.124840789244958e-05,
"loss": 0.0473,
"step": 6950
},
{
"grad_norm": 0.6784055829048157,
"learning_rate": 9.121723261293794e-05,
"loss": 0.0443,
"step": 6960
},
{
"grad_norm": 0.8421730995178223,
"learning_rate": 9.118600725060214e-05,
"loss": 0.0518,
"step": 6970
},
{
"grad_norm": 0.7832958698272705,
"learning_rate": 9.115473184338392e-05,
"loss": 0.0572,
"step": 6980
},
{
"grad_norm": 0.8210423588752747,
"learning_rate": 9.11234064292859e-05,
"loss": 0.0399,
"step": 6990
},
{
"grad_norm": 0.7533814311027527,
"learning_rate": 9.109203104637137e-05,
"loss": 0.0505,
"step": 7000
},
{
"grad_norm": 0.7397049069404602,
"learning_rate": 9.10606057327644e-05,
"loss": 0.058,
"step": 7010
},
{
"grad_norm": 0.5646662712097168,
"learning_rate": 9.10291305266497e-05,
"loss": 0.05,
"step": 7020
},
{
"grad_norm": 0.9847374558448792,
"learning_rate": 9.099760546627261e-05,
"loss": 0.0722,
"step": 7030
},
{
"grad_norm": 0.7427952289581299,
"learning_rate": 9.096603058993907e-05,
"loss": 0.0507,
"step": 7040
},
{
"grad_norm": 0.7624571323394775,
"learning_rate": 9.093440593601552e-05,
"loss": 0.0464,
"step": 7050
},
{
"grad_norm": 0.6838902831077576,
"learning_rate": 9.090273154292889e-05,
"loss": 0.045,
"step": 7060
},
{
"grad_norm": 0.6658808588981628,
"learning_rate": 9.087100744916656e-05,
"loss": 0.0509,
"step": 7070
},
{
"grad_norm": 0.9025909900665283,
"learning_rate": 9.08392336932763e-05,
"loss": 0.0662,
"step": 7080
},
{
"grad_norm": 0.948261559009552,
"learning_rate": 9.080741031386619e-05,
"loss": 0.0513,
"step": 7090
},
{
"grad_norm": 0.8182303309440613,
"learning_rate": 9.077553734960468e-05,
"loss": 0.0443,
"step": 7100
},
{
"grad_norm": 0.716875433921814,
"learning_rate": 9.074361483922041e-05,
"loss": 0.0608,
"step": 7110
},
{
"grad_norm": 0.8959944248199463,
"learning_rate": 9.071164282150224e-05,
"loss": 0.0459,
"step": 7120
},
{
"grad_norm": 0.6726084351539612,
"learning_rate": 9.067962133529919e-05,
"loss": 0.0414,
"step": 7130
},
{
"grad_norm": 0.8475841879844666,
"learning_rate": 9.064755041952036e-05,
"loss": 0.0641,
"step": 7140
},
{
"grad_norm": 0.6371098160743713,
"learning_rate": 9.061543011313497e-05,
"loss": 0.0617,
"step": 7150
},
{
"grad_norm": 0.8916845917701721,
"learning_rate": 9.05832604551722e-05,
"loss": 0.06,
"step": 7160
},
{
"grad_norm": 0.7170267701148987,
"learning_rate": 9.055104148472124e-05,
"loss": 0.0567,
"step": 7170
},
{
"grad_norm": 0.8654351830482483,
"learning_rate": 9.051877324093114e-05,
"loss": 0.0486,
"step": 7180
},
{
"grad_norm": 0.6471521258354187,
"learning_rate": 9.04864557630109e-05,
"loss": 0.0571,
"step": 7190
},
{
"grad_norm": 0.70100998878479,
"learning_rate": 9.045408909022928e-05,
"loss": 0.0459,
"step": 7200
},
{
"grad_norm": 0.5567395091056824,
"learning_rate": 9.042167326191484e-05,
"loss": 0.0491,
"step": 7210
},
{
"grad_norm": 0.6932230591773987,
"learning_rate": 9.038920831745587e-05,
"loss": 0.0484,
"step": 7220
},
{
"grad_norm": 0.7523159384727478,
"learning_rate": 9.035669429630036e-05,
"loss": 0.0493,
"step": 7230
},
{
"grad_norm": 0.6686680316925049,
"learning_rate": 9.032413123795588e-05,
"loss": 0.0529,
"step": 7240
},
{
"grad_norm": 0.6976425051689148,
"learning_rate": 9.029151918198962e-05,
"loss": 0.0504,
"step": 7250
},
{
"grad_norm": 0.6891392469406128,
"learning_rate": 9.025885816802833e-05,
"loss": 0.0404,
"step": 7260
},
{
"grad_norm": 0.6735789775848389,
"learning_rate": 9.02261482357582e-05,
"loss": 0.0452,
"step": 7270
},
{
"grad_norm": 0.7384276390075684,
"learning_rate": 9.019338942492485e-05,
"loss": 0.0484,
"step": 7280
},
{
"grad_norm": 0.6800125241279602,
"learning_rate": 9.01605817753334e-05,
"loss": 0.0486,
"step": 7290
},
{
"grad_norm": 0.7764517664909363,
"learning_rate": 9.012772532684818e-05,
"loss": 0.0445,
"step": 7300
},
{
"grad_norm": 0.5980468392372131,
"learning_rate": 9.00948201193929e-05,
"loss": 0.0477,
"step": 7310
},
{
"grad_norm": 0.6909130811691284,
"learning_rate": 9.006186619295047e-05,
"loss": 0.0375,
"step": 7320
},
{
"grad_norm": 0.723732590675354,
"learning_rate": 9.002886358756305e-05,
"loss": 0.044,
"step": 7330
},
{
"grad_norm": 0.9257233142852783,
"learning_rate": 8.999581234333188e-05,
"loss": 0.0533,
"step": 7340
},
{
"grad_norm": 0.746911346912384,
"learning_rate": 8.996271250041734e-05,
"loss": 0.0441,
"step": 7350
},
{
"grad_norm": 0.7061902284622192,
"learning_rate": 8.99295640990389e-05,
"loss": 0.0465,
"step": 7360
},
{
"grad_norm": 0.6909142732620239,
"learning_rate": 8.989636717947494e-05,
"loss": 0.0485,
"step": 7370
},
{
"grad_norm": 0.594556987285614,
"learning_rate": 8.986312178206291e-05,
"loss": 0.0431,
"step": 7380
},
{
"grad_norm": 0.6893497109413147,
"learning_rate": 8.982982794719904e-05,
"loss": 0.0439,
"step": 7390
},
{
"grad_norm": 0.6575574278831482,
"learning_rate": 8.979648571533852e-05,
"loss": 0.0406,
"step": 7400
},
{
"grad_norm": 0.6159623861312866,
"learning_rate": 8.976309512699528e-05,
"loss": 0.047,
"step": 7410
},
{
"grad_norm": 0.7083145380020142,
"learning_rate": 8.972965622274205e-05,
"loss": 0.0569,
"step": 7420
},
{
"grad_norm": 0.8037659525871277,
"learning_rate": 8.969616904321026e-05,
"loss": 0.0522,
"step": 7430
},
{
"grad_norm": 0.6126889586448669,
"learning_rate": 8.966263362908998e-05,
"loss": 0.0456,
"step": 7440
},
{
"grad_norm": 0.7288218140602112,
"learning_rate": 8.962905002112989e-05,
"loss": 0.0434,
"step": 7450
},
{
"grad_norm": 0.610046923160553,
"learning_rate": 8.959541826013724e-05,
"loss": 0.0436,
"step": 7460
},
{
"grad_norm": 0.7315624356269836,
"learning_rate": 8.956173838697781e-05,
"loss": 0.0546,
"step": 7470
},
{
"grad_norm": 0.8750680088996887,
"learning_rate": 8.952801044257581e-05,
"loss": 0.0589,
"step": 7480
},
{
"grad_norm": 0.706566333770752,
"learning_rate": 8.949423446791387e-05,
"loss": 0.0491,
"step": 7490
},
{
"grad_norm": 0.87822026014328,
"learning_rate": 8.9460410504033e-05,
"loss": 0.0438,
"step": 7500
},
{
"grad_norm": 0.7777622938156128,
"learning_rate": 8.942653859203248e-05,
"loss": 0.0427,
"step": 7510
},
{
"grad_norm": 0.7306796908378601,
"learning_rate": 8.93926187730699e-05,
"loss": 0.04,
"step": 7520
},
{
"grad_norm": 0.624380886554718,
"learning_rate": 8.935865108836102e-05,
"loss": 0.0352,
"step": 7530
},
{
"grad_norm": 0.6734282970428467,
"learning_rate": 8.932463557917981e-05,
"loss": 0.0451,
"step": 7540
},
{
"grad_norm": 0.7374112010002136,
"learning_rate": 8.929057228685829e-05,
"loss": 0.0427,
"step": 7550
},
{
"grad_norm": 0.7810717821121216,
"learning_rate": 8.925646125278658e-05,
"loss": 0.0428,
"step": 7560
},
{
"grad_norm": 0.7465488910675049,
"learning_rate": 8.92223025184128e-05,
"loss": 0.0545,
"step": 7570
},
{
"grad_norm": 0.7289469242095947,
"learning_rate": 8.918809612524303e-05,
"loss": 0.0566,
"step": 7580
},
{
"grad_norm": 0.6863954067230225,
"learning_rate": 8.91538421148413e-05,
"loss": 0.0532,
"step": 7590
},
{
"grad_norm": 0.6845741868019104,
"learning_rate": 8.91195405288294e-05,
"loss": 0.0653,
"step": 7600
},
{
"grad_norm": 0.6781623363494873,
"learning_rate": 8.908519140888705e-05,
"loss": 0.0561,
"step": 7610
},
{
"grad_norm": 0.7280553579330444,
"learning_rate": 8.905079479675164e-05,
"loss": 0.052,
"step": 7620
},
{
"grad_norm": 0.5901615023612976,
"learning_rate": 8.901635073421831e-05,
"loss": 0.0533,
"step": 7630
},
{
"grad_norm": 0.6767765879631042,
"learning_rate": 8.898185926313983e-05,
"loss": 0.0482,
"step": 7640
},
{
"grad_norm": 0.5965691804885864,
"learning_rate": 8.894732042542659e-05,
"loss": 0.0418,
"step": 7650
},
{
"grad_norm": 0.6716552376747131,
"learning_rate": 8.891273426304654e-05,
"loss": 0.0403,
"step": 7660
},
{
"grad_norm": 0.8393347859382629,
"learning_rate": 8.887810081802515e-05,
"loss": 0.0531,
"step": 7670
},
{
"grad_norm": 0.6679224371910095,
"learning_rate": 8.884342013244529e-05,
"loss": 0.0515,
"step": 7680
},
{
"grad_norm": 0.5963760614395142,
"learning_rate": 8.880869224844725e-05,
"loss": 0.0474,
"step": 7690
},
{
"grad_norm": 0.6980289816856384,
"learning_rate": 8.877391720822875e-05,
"loss": 0.0367,
"step": 7700
},
{
"grad_norm": 0.8115483522415161,
"learning_rate": 8.873909505404468e-05,
"loss": 0.0539,
"step": 7710
},
{
"grad_norm": 0.6974862813949585,
"learning_rate": 8.870422582820726e-05,
"loss": 0.053,
"step": 7720
},
{
"grad_norm": 0.6965011954307556,
"learning_rate": 8.866930957308588e-05,
"loss": 0.0499,
"step": 7730
},
{
"grad_norm": 0.6952334046363831,
"learning_rate": 8.86343463311071e-05,
"loss": 0.0484,
"step": 7740
},
{
"grad_norm": 0.7690582871437073,
"learning_rate": 8.859933614475452e-05,
"loss": 0.0397,
"step": 7750
},
{
"grad_norm": 0.6073489785194397,
"learning_rate": 8.85642790565689e-05,
"loss": 0.045,
"step": 7760
},
{
"grad_norm": 0.8936102986335754,
"learning_rate": 8.852917510914782e-05,
"loss": 0.0502,
"step": 7770
},
{
"grad_norm": 0.8381749391555786,
"learning_rate": 8.84940243451459e-05,
"loss": 0.048,
"step": 7780
},
{
"grad_norm": 0.7113655805587769,
"learning_rate": 8.845882680727469e-05,
"loss": 0.0482,
"step": 7790
},
{
"grad_norm": 0.5558151602745056,
"learning_rate": 8.842358253830246e-05,
"loss": 0.0432,
"step": 7800
},
{
"grad_norm": 0.6490152478218079,
"learning_rate": 8.838829158105434e-05,
"loss": 0.0421,
"step": 7810
},
{
"grad_norm": 0.5585162043571472,
"learning_rate": 8.835295397841217e-05,
"loss": 0.0332,
"step": 7820
},
{
"grad_norm": 0.6778039336204529,
"learning_rate": 8.831756977331447e-05,
"loss": 0.0404,
"step": 7830
},
{
"grad_norm": 0.759754478931427,
"learning_rate": 8.828213900875638e-05,
"loss": 0.0364,
"step": 7840
},
{
"grad_norm": 0.7292488813400269,
"learning_rate": 8.824666172778964e-05,
"loss": 0.0425,
"step": 7850
},
{
"grad_norm": 0.6859617829322815,
"learning_rate": 8.821113797352245e-05,
"loss": 0.0469,
"step": 7860
},
{
"grad_norm": 0.6054825186729431,
"learning_rate": 8.817556778911956e-05,
"loss": 0.0506,
"step": 7870
},
{
"grad_norm": 0.7315418124198914,
"learning_rate": 8.81399512178021e-05,
"loss": 0.0459,
"step": 7880
},
{
"grad_norm": 0.621973991394043,
"learning_rate": 8.810428830284752e-05,
"loss": 0.0411,
"step": 7890
},
{
"grad_norm": 0.6084474325180054,
"learning_rate": 8.806857908758967e-05,
"loss": 0.0572,
"step": 7900
},
{
"grad_norm": 0.6649889945983887,
"learning_rate": 8.80328236154186e-05,
"loss": 0.051,
"step": 7910
},
{
"grad_norm": 0.7248353958129883,
"learning_rate": 8.799702192978056e-05,
"loss": 0.0699,
"step": 7920
},
{
"grad_norm": 0.6285856366157532,
"learning_rate": 8.7961174074178e-05,
"loss": 0.04,
"step": 7930
},
{
"grad_norm": 0.6644786596298218,
"learning_rate": 8.792528009216942e-05,
"loss": 0.0453,
"step": 7940
},
{
"grad_norm": 0.6715810894966125,
"learning_rate": 8.788934002736943e-05,
"loss": 0.0416,
"step": 7950
},
{
"grad_norm": 0.69179767370224,
"learning_rate": 8.785335392344857e-05,
"loss": 0.0403,
"step": 7960
},
{
"grad_norm": 0.665648877620697,
"learning_rate": 8.781732182413336e-05,
"loss": 0.039,
"step": 7970
},
{
"grad_norm": 0.7149850726127625,
"learning_rate": 8.778124377320618e-05,
"loss": 0.0429,
"step": 7980
},
{
"grad_norm": 0.6621801853179932,
"learning_rate": 8.774511981450529e-05,
"loss": 0.0378,
"step": 7990
},
{
"grad_norm": 0.7296872138977051,
"learning_rate": 8.770894999192467e-05,
"loss": 0.0384,
"step": 8000
},
{
"grad_norm": 0.5579106211662292,
"learning_rate": 8.767273434941412e-05,
"loss": 0.0371,
"step": 8010
},
{
"grad_norm": 0.7171684503555298,
"learning_rate": 8.7636472930979e-05,
"loss": 0.0379,
"step": 8020
},
{
"grad_norm": 0.5831372737884521,
"learning_rate": 8.76001657806804e-05,
"loss": 0.0376,
"step": 8030
},
{
"grad_norm": 0.6490166783332825,
"learning_rate": 8.756381294263491e-05,
"loss": 0.0385,
"step": 8040
},
{
"grad_norm": 0.6851412653923035,
"learning_rate": 8.752741446101464e-05,
"loss": 0.0469,
"step": 8050
},
{
"grad_norm": 0.7253557443618774,
"learning_rate": 8.749097038004722e-05,
"loss": 0.0381,
"step": 8060
},
{
"grad_norm": 0.8101983666419983,
"learning_rate": 8.745448074401561e-05,
"loss": 0.0385,
"step": 8070
},
{
"grad_norm": 0.5129411220550537,
"learning_rate": 8.741794559725818e-05,
"loss": 0.0565,
"step": 8080
},
{
"grad_norm": 0.8387269973754883,
"learning_rate": 8.738136498416856e-05,
"loss": 0.0496,
"step": 8090
},
{
"grad_norm": 0.8249434232711792,
"learning_rate": 8.734473894919564e-05,
"loss": 0.0507,
"step": 8100
},
{
"grad_norm": 0.7114320397377014,
"learning_rate": 8.730806753684353e-05,
"loss": 0.0543,
"step": 8110
},
{
"grad_norm": 0.7107694745063782,
"learning_rate": 8.727135079167144e-05,
"loss": 0.0416,
"step": 8120
},
{
"grad_norm": 0.6892553567886353,
"learning_rate": 8.723458875829368e-05,
"loss": 0.053,
"step": 8130
},
{
"grad_norm": 0.6339153051376343,
"learning_rate": 8.719778148137958e-05,
"loss": 0.0455,
"step": 8140
},
{
"grad_norm": 0.7327772378921509,
"learning_rate": 8.716092900565346e-05,
"loss": 0.0435,
"step": 8150
},
{
"grad_norm": 0.6891136765480042,
"learning_rate": 8.712403137589456e-05,
"loss": 0.0506,
"step": 8160
},
{
"grad_norm": 0.7031595706939697,
"learning_rate": 8.708708863693697e-05,
"loss": 0.0464,
"step": 8170
},
{
"grad_norm": 0.4746207296848297,
"learning_rate": 8.70501008336696e-05,
"loss": 0.0444,
"step": 8180
},
{
"grad_norm": 0.6223254799842834,
"learning_rate": 8.701306801103611e-05,
"loss": 0.0388,
"step": 8190
},
{
"grad_norm": 0.727079451084137,
"learning_rate": 8.697599021403489e-05,
"loss": 0.0421,
"step": 8200
},
{
"grad_norm": 0.6117185950279236,
"learning_rate": 8.693886748771896e-05,
"loss": 0.0326,
"step": 8210
},
{
"grad_norm": 0.6069605946540833,
"learning_rate": 8.690169987719593e-05,
"loss": 0.0387,
"step": 8220
},
{
"grad_norm": 0.5669341683387756,
"learning_rate": 8.686448742762791e-05,
"loss": 0.0356,
"step": 8230
},
{
"grad_norm": 0.6107902526855469,
"learning_rate": 8.682723018423161e-05,
"loss": 0.0409,
"step": 8240
},
{
"grad_norm": 0.7450816035270691,
"learning_rate": 8.678992819227804e-05,
"loss": 0.0412,
"step": 8250
},
{
"grad_norm": 0.6763024926185608,
"learning_rate": 8.675258149709265e-05,
"loss": 0.0334,
"step": 8260
},
{
"grad_norm": 0.6744876503944397,
"learning_rate": 8.67151901440552e-05,
"loss": 0.0352,
"step": 8270
},
{
"grad_norm": 0.5323679447174072,
"learning_rate": 8.667775417859971e-05,
"loss": 0.0476,
"step": 8280
},
{
"grad_norm": 0.6425528526306152,
"learning_rate": 8.664027364621441e-05,
"loss": 0.0324,
"step": 8290
},
{
"grad_norm": 0.6528747081756592,
"learning_rate": 8.660274859244168e-05,
"loss": 0.0455,
"step": 8300
},
{
"grad_norm": 0.7584214806556702,
"learning_rate": 8.656517906287799e-05,
"loss": 0.0405,
"step": 8310
},
{
"grad_norm": 0.6350160837173462,
"learning_rate": 8.652756510317387e-05,
"loss": 0.0358,
"step": 8320
},
{
"grad_norm": 0.5808953642845154,
"learning_rate": 8.648990675903382e-05,
"loss": 0.0407,
"step": 8330
},
{
"grad_norm": 0.5569411516189575,
"learning_rate": 8.64522040762163e-05,
"loss": 0.0355,
"step": 8340
},
{
"grad_norm": 0.6639098525047302,
"learning_rate": 8.64144571005336e-05,
"loss": 0.0472,
"step": 8350
},
{
"grad_norm": 0.7112576961517334,
"learning_rate": 8.637666587785184e-05,
"loss": 0.0384,
"step": 8360
},
{
"grad_norm": 0.7281445264816284,
"learning_rate": 8.633883045409096e-05,
"loss": 0.0388,
"step": 8370
},
{
"grad_norm": 0.5406919121742249,
"learning_rate": 8.630095087522457e-05,
"loss": 0.0405,
"step": 8380
},
{
"grad_norm": 0.571739137172699,
"learning_rate": 8.62630271872799e-05,
"loss": 0.041,
"step": 8390
},
{
"grad_norm": 0.7128417491912842,
"learning_rate": 8.62250594363378e-05,
"loss": 0.0351,
"step": 8400
},
{
"grad_norm": 0.6335889101028442,
"learning_rate": 8.61870476685327e-05,
"loss": 0.0386,
"step": 8410
},
{
"grad_norm": 0.7824411392211914,
"learning_rate": 8.614899193005248e-05,
"loss": 0.0357,
"step": 8420
},
{
"grad_norm": 0.8190487027168274,
"learning_rate": 8.611089226713842e-05,
"loss": 0.0426,
"step": 8430
},
{
"grad_norm": 0.6597971320152283,
"learning_rate": 8.607274872608521e-05,
"loss": 0.0382,
"step": 8440
},
{
"grad_norm": 0.6724768280982971,
"learning_rate": 8.603456135324088e-05,
"loss": 0.0401,
"step": 8450
},
{
"grad_norm": 0.6130894422531128,
"learning_rate": 8.599633019500665e-05,
"loss": 0.0385,
"step": 8460
},
{
"grad_norm": 0.6459059715270996,
"learning_rate": 8.595805529783702e-05,
"loss": 0.034,
"step": 8470
},
{
"grad_norm": 0.759242594242096,
"learning_rate": 8.591973670823958e-05,
"loss": 0.033,
"step": 8480
},
{
"grad_norm": 0.7036397457122803,
"learning_rate": 8.588137447277503e-05,
"loss": 0.0453,
"step": 8490
},
{
"grad_norm": 0.6174358129501343,
"learning_rate": 8.584296863805709e-05,
"loss": 0.0487,
"step": 8500
},
{
"grad_norm": 0.6935808062553406,
"learning_rate": 8.580451925075248e-05,
"loss": 0.0368,
"step": 8510
},
{
"grad_norm": 0.7373861074447632,
"learning_rate": 8.576602635758085e-05,
"loss": 0.0339,
"step": 8520
},
{
"grad_norm": 0.7253849506378174,
"learning_rate": 8.572749000531469e-05,
"loss": 0.0418,
"step": 8530
},
{
"grad_norm": 0.6872004866600037,
"learning_rate": 8.568891024077926e-05,
"loss": 0.041,
"step": 8540
},
{
"grad_norm": 0.6780731081962585,
"learning_rate": 8.565028711085265e-05,
"loss": 0.0474,
"step": 8550
},
{
"grad_norm": 0.7518962621688843,
"learning_rate": 8.561162066246561e-05,
"loss": 0.0435,
"step": 8560
},
{
"grad_norm": 0.7724444270133972,
"learning_rate": 8.557291094260152e-05,
"loss": 0.0716,
"step": 8570
},
{
"grad_norm": 0.9539865851402283,
"learning_rate": 8.553415799829629e-05,
"loss": 0.0473,
"step": 8580
},
{
"grad_norm": 0.5941001176834106,
"learning_rate": 8.549536187663847e-05,
"loss": 0.0418,
"step": 8590
},
{
"grad_norm": 0.5229378342628479,
"learning_rate": 8.545652262476897e-05,
"loss": 0.0349,
"step": 8600
},
{
"grad_norm": 0.6420212388038635,
"learning_rate": 8.541764028988115e-05,
"loss": 0.0345,
"step": 8610
},
{
"grad_norm": 0.6257814168930054,
"learning_rate": 8.537871491922071e-05,
"loss": 0.0434,
"step": 8620
},
{
"grad_norm": 0.6820962429046631,
"learning_rate": 8.533974656008566e-05,
"loss": 0.0342,
"step": 8630
},
{
"grad_norm": 0.8080209493637085,
"learning_rate": 8.530073525982621e-05,
"loss": 0.0412,
"step": 8640
},
{
"grad_norm": 0.6080604791641235,
"learning_rate": 8.526168106584476e-05,
"loss": 0.0475,
"step": 8650
},
{
"grad_norm": 0.5313056707382202,
"learning_rate": 8.522258402559586e-05,
"loss": 0.0387,
"step": 8660
},
{
"grad_norm": 0.6141229271888733,
"learning_rate": 8.518344418658613e-05,
"loss": 0.0362,
"step": 8670
},
{
"grad_norm": 0.7168601751327515,
"learning_rate": 8.51442615963741e-05,
"loss": 0.0326,
"step": 8680
},
{
"grad_norm": 0.6128477454185486,
"learning_rate": 8.510503630257034e-05,
"loss": 0.0372,
"step": 8690
},
{
"grad_norm": 0.755458652973175,
"learning_rate": 8.50657683528373e-05,
"loss": 0.0357,
"step": 8700
},
{
"grad_norm": 0.6050610542297363,
"learning_rate": 8.502645779488922e-05,
"loss": 0.0427,
"step": 8710
},
{
"grad_norm": 0.6760846972465515,
"learning_rate": 8.498710467649213e-05,
"loss": 0.06,
"step": 8720
},
{
"grad_norm": 0.6286876201629639,
"learning_rate": 8.494770904546381e-05,
"loss": 0.042,
"step": 8730
},
{
"grad_norm": 0.8376249074935913,
"learning_rate": 8.490827094967363e-05,
"loss": 0.0374,
"step": 8740
},
{
"grad_norm": 0.7126126885414124,
"learning_rate": 8.486879043704263e-05,
"loss": 0.0476,
"step": 8750
},
{
"grad_norm": 0.6995195746421814,
"learning_rate": 8.482926755554332e-05,
"loss": 0.0381,
"step": 8760
},
{
"grad_norm": 0.6092513799667358,
"learning_rate": 8.478970235319975e-05,
"loss": 0.042,
"step": 8770
},
{
"grad_norm": 0.7193765044212341,
"learning_rate": 8.475009487808737e-05,
"loss": 0.0499,
"step": 8780
},
{
"grad_norm": 0.7832525372505188,
"learning_rate": 8.471044517833298e-05,
"loss": 0.0416,
"step": 8790
},
{
"grad_norm": 0.5836358666419983,
"learning_rate": 8.467075330211474e-05,
"loss": 0.0322,
"step": 8800
},
{
"grad_norm": 0.6432408690452576,
"learning_rate": 8.463101929766198e-05,
"loss": 0.0362,
"step": 8810
},
{
"grad_norm": 0.6151525378227234,
"learning_rate": 8.459124321325528e-05,
"loss": 0.0363,
"step": 8820
},
{
"grad_norm": 0.6140194535255432,
"learning_rate": 8.455142509722634e-05,
"loss": 0.0414,
"step": 8830
},
{
"grad_norm": 0.6025208234786987,
"learning_rate": 8.451156499795792e-05,
"loss": 0.0359,
"step": 8840
},
{
"grad_norm": 0.5990309119224548,
"learning_rate": 8.447166296388381e-05,
"loss": 0.0365,
"step": 8850
},
{
"grad_norm": 0.634741485118866,
"learning_rate": 8.443171904348872e-05,
"loss": 0.0355,
"step": 8860
},
{
"grad_norm": 0.807852566242218,
"learning_rate": 8.43917332853083e-05,
"loss": 0.0405,
"step": 8870
},
{
"grad_norm": 0.617822527885437,
"learning_rate": 8.435170573792901e-05,
"loss": 0.0379,
"step": 8880
},
{
"grad_norm": 0.5730050802230835,
"learning_rate": 8.431163644998808e-05,
"loss": 0.0418,
"step": 8890
},
{
"grad_norm": 0.6506041884422302,
"learning_rate": 8.42715254701735e-05,
"loss": 0.0377,
"step": 8900
},
{
"grad_norm": 0.6470239758491516,
"learning_rate": 8.42313728472239e-05,
"loss": 0.0341,
"step": 8910
},
{
"grad_norm": 0.8019155859947205,
"learning_rate": 8.419117862992844e-05,
"loss": 0.0522,
"step": 8920
},
{
"grad_norm": 0.6734329462051392,
"learning_rate": 8.415094286712694e-05,
"loss": 0.039,
"step": 8930
},
{
"grad_norm": 0.5875380039215088,
"learning_rate": 8.411066560770965e-05,
"loss": 0.0402,
"step": 8940
},
{
"grad_norm": 0.5588849782943726,
"learning_rate": 8.407034690061721e-05,
"loss": 0.0346,
"step": 8950
},
{
"grad_norm": 0.540400505065918,
"learning_rate": 8.402998679484067e-05,
"loss": 0.0307,
"step": 8960
},
{
"grad_norm": 0.7245332598686218,
"learning_rate": 8.398958533942135e-05,
"loss": 0.0296,
"step": 8970
},
{
"grad_norm": 0.600703239440918,
"learning_rate": 8.394914258345083e-05,
"loss": 0.0396,
"step": 8980
},
{
"grad_norm": 0.6987263560295105,
"learning_rate": 8.390865857607089e-05,
"loss": 0.0625,
"step": 8990
},
{
"grad_norm": 0.5313953757286072,
"learning_rate": 8.386813336647339e-05,
"loss": 0.0401,
"step": 9000
},
{
"grad_norm": 0.5456262826919556,
"learning_rate": 8.38275670039003e-05,
"loss": 0.0306,
"step": 9010
},
{
"grad_norm": 0.5934560298919678,
"learning_rate": 8.378695953764356e-05,
"loss": 0.0375,
"step": 9020
},
{
"grad_norm": 0.7319691181182861,
"learning_rate": 8.374631101704509e-05,
"loss": 0.0476,
"step": 9030
},
{
"grad_norm": 0.5224413275718689,
"learning_rate": 8.370562149149665e-05,
"loss": 0.0449,
"step": 9040
},
{
"grad_norm": 0.6204515695571899,
"learning_rate": 8.366489101043989e-05,
"loss": 0.0355,
"step": 9050
},
{
"grad_norm": 0.5916332006454468,
"learning_rate": 8.362411962336614e-05,
"loss": 0.043,
"step": 9060
},
{
"grad_norm": 0.558459997177124,
"learning_rate": 8.358330737981651e-05,
"loss": 0.0301,
"step": 9070
},
{
"grad_norm": 0.7058679461479187,
"learning_rate": 8.35424543293817e-05,
"loss": 0.0377,
"step": 9080
},
{
"grad_norm": 0.7708312273025513,
"learning_rate": 8.350156052170206e-05,
"loss": 0.0463,
"step": 9090
},
{
"grad_norm": 0.6086065769195557,
"learning_rate": 8.346062600646738e-05,
"loss": 0.0483,
"step": 9100
},
{
"grad_norm": 0.6130119562149048,
"learning_rate": 8.341965083341697e-05,
"loss": 0.0401,
"step": 9110
},
{
"grad_norm": 0.5543129444122314,
"learning_rate": 8.337863505233953e-05,
"loss": 0.0398,
"step": 9120
},
{
"grad_norm": 0.551835298538208,
"learning_rate": 8.33375787130731e-05,
"loss": 0.0409,
"step": 9130
},
{
"grad_norm": 0.6582505702972412,
"learning_rate": 8.3296481865505e-05,
"loss": 0.0317,
"step": 9140
},
{
"grad_norm": 0.6819815039634705,
"learning_rate": 8.325534455957181e-05,
"loss": 0.045,
"step": 9150
},
{
"grad_norm": 0.8029713034629822,
"learning_rate": 8.321416684525917e-05,
"loss": 0.0307,
"step": 9160
},
{
"grad_norm": 0.5987986326217651,
"learning_rate": 8.317294877260193e-05,
"loss": 0.0378,
"step": 9170
},
{
"grad_norm": 0.5609573721885681,
"learning_rate": 8.313169039168394e-05,
"loss": 0.0359,
"step": 9180
},
{
"grad_norm": 0.613232433795929,
"learning_rate": 8.3090391752638e-05,
"loss": 0.0283,
"step": 9190
},
{
"grad_norm": 0.6214397549629211,
"learning_rate": 8.304905290564586e-05,
"loss": 0.0389,
"step": 9200
},
{
"grad_norm": 0.6382613182067871,
"learning_rate": 8.300767390093813e-05,
"loss": 0.0443,
"step": 9210
},
{
"grad_norm": 0.6291276812553406,
"learning_rate": 8.296625478879417e-05,
"loss": 0.0438,
"step": 9220
},
{
"grad_norm": 0.6240870952606201,
"learning_rate": 8.292479561954214e-05,
"loss": 0.0445,
"step": 9230
},
{
"grad_norm": 0.6337258815765381,
"learning_rate": 8.288329644355883e-05,
"loss": 0.0453,
"step": 9240
},
{
"grad_norm": 0.7190439105033875,
"learning_rate": 8.284175731126965e-05,
"loss": 0.035,
"step": 9250
},
{
"grad_norm": 0.6192926168441772,
"learning_rate": 8.280017827314854e-05,
"loss": 0.0346,
"step": 9260
},
{
"grad_norm": 0.8758948445320129,
"learning_rate": 8.2758559379718e-05,
"loss": 0.0428,
"step": 9270
},
{
"grad_norm": 0.8559668660163879,
"learning_rate": 8.271690068154887e-05,
"loss": 0.0436,
"step": 9280
},
{
"grad_norm": 0.7683869004249573,
"learning_rate": 8.26752022292604e-05,
"loss": 0.0509,
"step": 9290
},
{
"grad_norm": 0.7226149439811707,
"learning_rate": 8.263346407352018e-05,
"loss": 0.0494,
"step": 9300
},
{
"grad_norm": 0.770819365978241,
"learning_rate": 8.259168626504395e-05,
"loss": 0.041,
"step": 9310
},
{
"grad_norm": 0.6311588883399963,
"learning_rate": 8.254986885459572e-05,
"loss": 0.0563,
"step": 9320
},
{
"grad_norm": 0.7456294298171997,
"learning_rate": 8.250801189298757e-05,
"loss": 0.0397,
"step": 9330
},
{
"grad_norm": 0.6444935202598572,
"learning_rate": 8.246611543107967e-05,
"loss": 0.045,
"step": 9340
},
{
"grad_norm": 0.6856061220169067,
"learning_rate": 8.242417951978018e-05,
"loss": 0.0362,
"step": 9350
},
{
"grad_norm": 0.7016450762748718,
"learning_rate": 8.238220421004517e-05,
"loss": 0.0336,
"step": 9360
},
{
"grad_norm": 0.5704193711280823,
"learning_rate": 8.234018955287859e-05,
"loss": 0.0523,
"step": 9370
},
{
"grad_norm": 0.6714090704917908,
"learning_rate": 8.229813559933224e-05,
"loss": 0.039,
"step": 9380
},
{
"grad_norm": 0.7125976085662842,
"learning_rate": 8.22560424005056e-05,
"loss": 0.0376,
"step": 9390
},
{
"grad_norm": 0.7148918509483337,
"learning_rate": 8.22139100075459e-05,
"loss": 0.0511,
"step": 9400
},
{
"grad_norm": 0.6266578435897827,
"learning_rate": 8.217173847164799e-05,
"loss": 0.0446,
"step": 9410
},
{
"grad_norm": 0.6550078392028809,
"learning_rate": 8.212952784405422e-05,
"loss": 0.0385,
"step": 9420
},
{
"grad_norm": 0.740433394908905,
"learning_rate": 8.208727817605451e-05,
"loss": 0.0329,
"step": 9430
},
{
"grad_norm": 0.6755595207214355,
"learning_rate": 8.204498951898618e-05,
"loss": 0.0385,
"step": 9440
},
{
"grad_norm": 0.8082408905029297,
"learning_rate": 8.200266192423395e-05,
"loss": 0.0347,
"step": 9450
},
{
"grad_norm": 0.7583034038543701,
"learning_rate": 8.196029544322981e-05,
"loss": 0.033,
"step": 9460
},
{
"grad_norm": 0.6213312149047852,
"learning_rate": 8.191789012745305e-05,
"loss": 0.0604,
"step": 9470
},
{
"grad_norm": 0.6822872161865234,
"learning_rate": 8.187544602843014e-05,
"loss": 0.0357,
"step": 9480
},
{
"grad_norm": 0.6919769048690796,
"learning_rate": 8.183296319773466e-05,
"loss": 0.0356,
"step": 9490
},
{
"grad_norm": 0.7437395453453064,
"learning_rate": 8.179044168698721e-05,
"loss": 0.0355,
"step": 9500
},
{
"grad_norm": 0.7025963068008423,
"learning_rate": 8.174788154785548e-05,
"loss": 0.0406,
"step": 9510
},
{
"grad_norm": 0.719218373298645,
"learning_rate": 8.170528283205403e-05,
"loss": 0.0443,
"step": 9520
},
{
"grad_norm": 0.5219499468803406,
"learning_rate": 8.166264559134435e-05,
"loss": 0.0396,
"step": 9530
},
{
"grad_norm": 0.6852062940597534,
"learning_rate": 8.161996987753467e-05,
"loss": 0.0497,
"step": 9540
},
{
"grad_norm": 0.67015141248703,
"learning_rate": 8.157725574248e-05,
"loss": 0.0401,
"step": 9550
},
{
"grad_norm": 0.6463527083396912,
"learning_rate": 8.153450323808209e-05,
"loss": 0.0439,
"step": 9560
},
{
"grad_norm": 0.6911786198616028,
"learning_rate": 8.149171241628923e-05,
"loss": 0.055,
"step": 9570
},
{
"grad_norm": 0.7392476201057434,
"learning_rate": 8.144888332909631e-05,
"loss": 0.0451,
"step": 9580
},
{
"grad_norm": 0.7043988108634949,
"learning_rate": 8.14060160285447e-05,
"loss": 0.0411,
"step": 9590
},
{
"grad_norm": 0.6849877834320068,
"learning_rate": 8.136311056672223e-05,
"loss": 0.0357,
"step": 9600
},
{
"grad_norm": 0.6231916546821594,
"learning_rate": 8.132016699576308e-05,
"loss": 0.0366,
"step": 9610
},
{
"grad_norm": 0.6831573247909546,
"learning_rate": 8.127718536784771e-05,
"loss": 0.0425,
"step": 9620
},
{
"grad_norm": 0.49121764302253723,
"learning_rate": 8.12341657352029e-05,
"loss": 0.0306,
"step": 9630
},
{
"grad_norm": 0.6878386735916138,
"learning_rate": 8.119110815010151e-05,
"loss": 0.0361,
"step": 9640
},
{
"grad_norm": 0.7074189186096191,
"learning_rate": 8.114801266486261e-05,
"loss": 0.0407,
"step": 9650
},
{
"grad_norm": 0.7856928706169128,
"learning_rate": 8.110487933185122e-05,
"loss": 0.0404,
"step": 9660
},
{
"grad_norm": 0.6212881803512573,
"learning_rate": 8.106170820347847e-05,
"loss": 0.0413,
"step": 9670
},
{
"grad_norm": 0.6857605576515198,
"learning_rate": 8.101849933220132e-05,
"loss": 0.0369,
"step": 9680
},
{
"grad_norm": 0.6190844774246216,
"learning_rate": 8.097525277052264e-05,
"loss": 0.0371,
"step": 9690
},
{
"grad_norm": 0.6972376108169556,
"learning_rate": 8.093196857099106e-05,
"loss": 0.0359,
"step": 9700
},
{
"grad_norm": 0.7533993721008301,
"learning_rate": 8.088864678620096e-05,
"loss": 0.0377,
"step": 9710
},
{
"grad_norm": 0.6172062158584595,
"learning_rate": 8.084528746879243e-05,
"loss": 0.0395,
"step": 9720
},
{
"grad_norm": 0.6637015342712402,
"learning_rate": 8.080189067145108e-05,
"loss": 0.0455,
"step": 9730
},
{
"grad_norm": 0.7848032712936401,
"learning_rate": 8.075845644690813e-05,
"loss": 0.0377,
"step": 9740
},
{
"grad_norm": 0.728754460811615,
"learning_rate": 8.07149848479403e-05,
"loss": 0.0381,
"step": 9750
},
{
"grad_norm": 0.6640408635139465,
"learning_rate": 8.067147592736962e-05,
"loss": 0.0438,
"step": 9760
},
{
"grad_norm": 0.6077513694763184,
"learning_rate": 8.062792973806357e-05,
"loss": 0.0365,
"step": 9770
},
{
"grad_norm": 0.6503734588623047,
"learning_rate": 8.058434633293485e-05,
"loss": 0.0367,
"step": 9780
},
{
"grad_norm": 0.5871071815490723,
"learning_rate": 8.054072576494143e-05,
"loss": 0.033,
"step": 9790
},
{
"grad_norm": 0.7790606617927551,
"learning_rate": 8.04970680870864e-05,
"loss": 0.0366,
"step": 9800
},
{
"grad_norm": 0.7164578437805176,
"learning_rate": 8.045337335241793e-05,
"loss": 0.0392,
"step": 9810
},
{
"grad_norm": 0.8188837766647339,
"learning_rate": 8.040964161402932e-05,
"loss": 0.0386,
"step": 9820
},
{
"grad_norm": 0.5662012696266174,
"learning_rate": 8.036587292505869e-05,
"loss": 0.037,
"step": 9830
},
{
"grad_norm": 0.7119251489639282,
"learning_rate": 8.032206733868912e-05,
"loss": 0.0502,
"step": 9840
},
{
"grad_norm": 0.7550401091575623,
"learning_rate": 8.027822490814858e-05,
"loss": 0.0457,
"step": 9850
},
{
"grad_norm": 0.5830123424530029,
"learning_rate": 8.023434568670972e-05,
"loss": 0.0339,
"step": 9860
},
{
"grad_norm": 0.6764442920684814,
"learning_rate": 8.019042972768992e-05,
"loss": 0.0357,
"step": 9870
},
{
"grad_norm": 0.6409294605255127,
"learning_rate": 8.014647708445124e-05,
"loss": 0.0407,
"step": 9880
},
{
"grad_norm": 0.7694797515869141,
"learning_rate": 8.010248781040027e-05,
"loss": 0.0342,
"step": 9890
},
{
"grad_norm": 0.5252751111984253,
"learning_rate": 8.005846195898814e-05,
"loss": 0.0378,
"step": 9900
},
{
"grad_norm": 0.5659962296485901,
"learning_rate": 8.00143995837104e-05,
"loss": 0.0418,
"step": 9910
},
{
"grad_norm": 0.7089165449142456,
"learning_rate": 7.997030073810698e-05,
"loss": 0.0455,
"step": 9920
},
{
"grad_norm": 0.7163835167884827,
"learning_rate": 7.992616547576217e-05,
"loss": 0.0356,
"step": 9930
},
{
"grad_norm": 0.630772590637207,
"learning_rate": 7.988199385030445e-05,
"loss": 0.0371,
"step": 9940
},
{
"grad_norm": 0.5773047208786011,
"learning_rate": 7.98377859154065e-05,
"loss": 0.0301,
"step": 9950
},
{
"grad_norm": 0.5738563537597656,
"learning_rate": 7.979354172478516e-05,
"loss": 0.0313,
"step": 9960
},
{
"grad_norm": 0.6947473287582397,
"learning_rate": 7.974926133220127e-05,
"loss": 0.0398,
"step": 9970
},
{
"grad_norm": 0.612501859664917,
"learning_rate": 7.970494479145967e-05,
"loss": 0.0356,
"step": 9980
},
{
"grad_norm": 0.7582767605781555,
"learning_rate": 7.966059215640919e-05,
"loss": 0.0411,
"step": 9990
},
{
"grad_norm": 0.7247899770736694,
"learning_rate": 7.961620348094241e-05,
"loss": 0.0432,
"step": 10000
},
{
"grad_norm": 0.7437941431999207,
"learning_rate": 7.957177881899577e-05,
"loss": 0.0416,
"step": 10010
},
{
"grad_norm": 0.5598741769790649,
"learning_rate": 7.952731822454944e-05,
"loss": 0.0408,
"step": 10020
},
{
"grad_norm": 0.7919538617134094,
"learning_rate": 7.948282175162722e-05,
"loss": 0.0329,
"step": 10030
},
{
"grad_norm": 0.684561014175415,
"learning_rate": 7.943828945429652e-05,
"loss": 0.0413,
"step": 10040
},
{
"grad_norm": 0.8000953793525696,
"learning_rate": 7.939372138666827e-05,
"loss": 0.0377,
"step": 10050
},
{
"grad_norm": 0.6715127229690552,
"learning_rate": 7.934911760289692e-05,
"loss": 0.0428,
"step": 10060
},
{
"grad_norm": 0.5884119272232056,
"learning_rate": 7.930447815718022e-05,
"loss": 0.0274,
"step": 10070
},
{
"grad_norm": 0.7886438369750977,
"learning_rate": 7.925980310375934e-05,
"loss": 0.0386,
"step": 10080
},
{
"grad_norm": 0.6328913569450378,
"learning_rate": 7.921509249691865e-05,
"loss": 0.0372,
"step": 10090
},
{
"grad_norm": 0.6320500373840332,
"learning_rate": 7.917034639098579e-05,
"loss": 0.0384,
"step": 10100
},
{
"grad_norm": 0.613904595375061,
"learning_rate": 7.912556484033146e-05,
"loss": 0.0299,
"step": 10110
},
{
"grad_norm": 0.5378363728523254,
"learning_rate": 7.908074789936952e-05,
"loss": 0.0295,
"step": 10120
},
{
"grad_norm": 0.6810264587402344,
"learning_rate": 7.903589562255673e-05,
"loss": 0.0367,
"step": 10130
},
{
"grad_norm": 0.6502209305763245,
"learning_rate": 7.899100806439285e-05,
"loss": 0.0338,
"step": 10140
},
{
"grad_norm": 0.5764700174331665,
"learning_rate": 7.894608527942049e-05,
"loss": 0.0445,
"step": 10150
},
{
"grad_norm": 0.7452075481414795,
"learning_rate": 7.89011273222251e-05,
"loss": 0.0406,
"step": 10160
},
{
"grad_norm": 0.6723094582557678,
"learning_rate": 7.885613424743482e-05,
"loss": 0.0351,
"step": 10170
},
{
"grad_norm": 0.7193927764892578,
"learning_rate": 7.881110610972044e-05,
"loss": 0.0502,
"step": 10180
},
{
"grad_norm": 0.6798580884933472,
"learning_rate": 7.876604296379544e-05,
"loss": 0.0514,
"step": 10190
},
{
"grad_norm": 0.7610640525817871,
"learning_rate": 7.872094486441579e-05,
"loss": 0.0465,
"step": 10200
},
{
"grad_norm": 0.7539733648300171,
"learning_rate": 7.867581186637991e-05,
"loss": 0.0393,
"step": 10210
},
{
"grad_norm": 0.7211567759513855,
"learning_rate": 7.863064402452866e-05,
"loss": 0.0337,
"step": 10220
},
{
"grad_norm": 0.6181485652923584,
"learning_rate": 7.858544139374524e-05,
"loss": 0.0471,
"step": 10230
},
{
"grad_norm": 0.5199313163757324,
"learning_rate": 7.854020402895508e-05,
"loss": 0.0382,
"step": 10240
},
{
"grad_norm": 0.6421964764595032,
"learning_rate": 7.849493198512587e-05,
"loss": 0.0335,
"step": 10250
},
{
"grad_norm": 0.5942662358283997,
"learning_rate": 7.84496253172674e-05,
"loss": 0.0324,
"step": 10260
},
{
"grad_norm": 0.6493527293205261,
"learning_rate": 7.840428408043155e-05,
"loss": 0.0389,
"step": 10270
},
{
"grad_norm": 0.6417429447174072,
"learning_rate": 7.835890832971218e-05,
"loss": 0.0393,
"step": 10280
},
{
"grad_norm": 0.6768302321434021,
"learning_rate": 7.831349812024512e-05,
"loss": 0.0468,
"step": 10290
},
{
"grad_norm": 0.6727313995361328,
"learning_rate": 7.826805350720807e-05,
"loss": 0.0363,
"step": 10300
},
{
"grad_norm": 0.6983158588409424,
"learning_rate": 7.822257454582049e-05,
"loss": 0.0322,
"step": 10310
},
{
"grad_norm": 0.6224091053009033,
"learning_rate": 7.817706129134363e-05,
"loss": 0.0424,
"step": 10320
},
{
"grad_norm": 0.5504922270774841,
"learning_rate": 7.813151379908036e-05,
"loss": 0.0323,
"step": 10330
},
{
"grad_norm": 0.853366494178772,
"learning_rate": 7.808593212437523e-05,
"loss": 0.0383,
"step": 10340
},
{
"grad_norm": 0.822418749332428,
"learning_rate": 7.804031632261421e-05,
"loss": 0.0478,
"step": 10350
},
{
"grad_norm": 0.6383507251739502,
"learning_rate": 7.799466644922484e-05,
"loss": 0.0318,
"step": 10360
},
{
"grad_norm": 0.7397134900093079,
"learning_rate": 7.794898255967602e-05,
"loss": 0.036,
"step": 10370
},
{
"grad_norm": 0.6186095476150513,
"learning_rate": 7.790326470947795e-05,
"loss": 0.0354,
"step": 10380
},
{
"grad_norm": 0.6787047982215881,
"learning_rate": 7.785751295418217e-05,
"loss": 0.0401,
"step": 10390
},
{
"grad_norm": 0.673518180847168,
"learning_rate": 7.781172734938136e-05,
"loss": 0.0415,
"step": 10400
},
{
"grad_norm": 0.6681343913078308,
"learning_rate": 7.776590795070933e-05,
"loss": 0.0374,
"step": 10410
},
{
"grad_norm": 0.5259663462638855,
"learning_rate": 7.772005481384099e-05,
"loss": 0.042,
"step": 10420
},
{
"grad_norm": 0.8224720358848572,
"learning_rate": 7.767416799449223e-05,
"loss": 0.0303,
"step": 10430
},
{
"grad_norm": 0.6709105968475342,
"learning_rate": 7.762824754841984e-05,
"loss": 0.0339,
"step": 10440
},
{
"grad_norm": 0.6275613307952881,
"learning_rate": 7.758229353142152e-05,
"loss": 0.0421,
"step": 10450
},
{
"grad_norm": 0.6834690570831299,
"learning_rate": 7.753630599933572e-05,
"loss": 0.0342,
"step": 10460
},
{
"grad_norm": 0.7321269512176514,
"learning_rate": 7.749028500804161e-05,
"loss": 0.0357,
"step": 10470
},
{
"grad_norm": 0.6390467286109924,
"learning_rate": 7.744423061345906e-05,
"loss": 0.0331,
"step": 10480
},
{
"grad_norm": 0.6164873242378235,
"learning_rate": 7.73981428715485e-05,
"loss": 0.0298,
"step": 10490
},
{
"grad_norm": 0.6307353377342224,
"learning_rate": 7.735202183831085e-05,
"loss": 0.0282,
"step": 10500
},
{
"grad_norm": 0.8414798974990845,
"learning_rate": 7.730586756978757e-05,
"loss": 0.045,
"step": 10510
},
{
"grad_norm": 0.5747160911560059,
"learning_rate": 7.72596801220604e-05,
"loss": 0.028,
"step": 10520
},
{
"grad_norm": 0.5569151043891907,
"learning_rate": 7.721345955125147e-05,
"loss": 0.0371,
"step": 10530
},
{
"grad_norm": 0.6454663872718811,
"learning_rate": 7.71672059135231e-05,
"loss": 0.0304,
"step": 10540
},
{
"grad_norm": 0.5511540174484253,
"learning_rate": 7.712091926507787e-05,
"loss": 0.0269,
"step": 10550
},
{
"grad_norm": 0.7155776023864746,
"learning_rate": 7.70745996621584e-05,
"loss": 0.034,
"step": 10560
},
{
"grad_norm": 0.7590230107307434,
"learning_rate": 7.702824716104735e-05,
"loss": 0.0314,
"step": 10570
},
{
"grad_norm": 0.6551960110664368,
"learning_rate": 7.698186181806743e-05,
"loss": 0.0374,
"step": 10580
},
{
"grad_norm": 0.6793715357780457,
"learning_rate": 7.693544368958116e-05,
"loss": 0.0368,
"step": 10590
},
{
"grad_norm": 0.6801042556762695,
"learning_rate": 7.688899283199096e-05,
"loss": 0.0458,
"step": 10600
},
{
"grad_norm": 0.7142727971076965,
"learning_rate": 7.684250930173901e-05,
"loss": 0.0521,
"step": 10610
},
{
"grad_norm": 0.6396678686141968,
"learning_rate": 7.679599315530718e-05,
"loss": 0.0494,
"step": 10620
},
{
"grad_norm": 0.6586829423904419,
"learning_rate": 7.674944444921695e-05,
"loss": 0.031,
"step": 10630
},
{
"grad_norm": 0.6611000895500183,
"learning_rate": 7.670286324002944e-05,
"loss": 0.0401,
"step": 10640
},
{
"grad_norm": 0.6456710696220398,
"learning_rate": 7.665624958434514e-05,
"loss": 0.0396,
"step": 10650
},
{
"grad_norm": 0.6632084250450134,
"learning_rate": 7.66096035388041e-05,
"loss": 0.0406,
"step": 10660
},
{
"grad_norm": 0.5977802276611328,
"learning_rate": 7.656292516008563e-05,
"loss": 0.039,
"step": 10670
},
{
"grad_norm": 0.5965768694877625,
"learning_rate": 7.651621450490837e-05,
"loss": 0.0288,
"step": 10680
},
{
"grad_norm": 0.6352524757385254,
"learning_rate": 7.646947163003017e-05,
"loss": 0.028,
"step": 10690
},
{
"grad_norm": 0.5458565354347229,
"learning_rate": 7.642269659224803e-05,
"loss": 0.0546,
"step": 10700
},
{
"grad_norm": 0.6172992587089539,
"learning_rate": 7.637588944839803e-05,
"loss": 0.0378,
"step": 10710
},
{
"grad_norm": 0.7162474989891052,
"learning_rate": 7.632905025535528e-05,
"loss": 0.0375,
"step": 10720
},
{
"grad_norm": 0.5355373620986938,
"learning_rate": 7.628217907003378e-05,
"loss": 0.0312,
"step": 10730
},
{
"grad_norm": 0.49585744738578796,
"learning_rate": 7.623527594938649e-05,
"loss": 0.0291,
"step": 10740
},
{
"grad_norm": 0.6132726073265076,
"learning_rate": 7.618834095040509e-05,
"loss": 0.0366,
"step": 10750
},
{
"grad_norm": 0.7078750729560852,
"learning_rate": 7.614137413012e-05,
"loss": 0.0402,
"step": 10760
},
{
"grad_norm": 0.6871320009231567,
"learning_rate": 7.609437554560042e-05,
"loss": 0.0404,
"step": 10770
},
{
"grad_norm": 0.6165637969970703,
"learning_rate": 7.604734525395398e-05,
"loss": 0.0381,
"step": 10780
},
{
"grad_norm": 0.6198022961616516,
"learning_rate": 7.600028331232699e-05,
"loss": 0.0318,
"step": 10790
},
{
"grad_norm": 0.6509577035903931,
"learning_rate": 7.595318977790408e-05,
"loss": 0.044,
"step": 10800
},
{
"grad_norm": 0.7306013703346252,
"learning_rate": 7.590606470790836e-05,
"loss": 0.033,
"step": 10810
},
{
"grad_norm": 0.707929790019989,
"learning_rate": 7.585890815960125e-05,
"loss": 0.0396,
"step": 10820
},
{
"grad_norm": 0.7060425281524658,
"learning_rate": 7.581172019028238e-05,
"loss": 0.0378,
"step": 10830
},
{
"grad_norm": 0.6853958964347839,
"learning_rate": 7.576450085728959e-05,
"loss": 0.0435,
"step": 10840
},
{
"grad_norm": 0.7887881398200989,
"learning_rate": 7.571725021799885e-05,
"loss": 0.0374,
"step": 10850
},
{
"grad_norm": 0.6388114094734192,
"learning_rate": 7.566996832982409e-05,
"loss": 0.031,
"step": 10860
},
{
"grad_norm": 0.537063479423523,
"learning_rate": 7.56226552502173e-05,
"loss": 0.0345,
"step": 10870
},
{
"grad_norm": 0.6638509631156921,
"learning_rate": 7.557531103666833e-05,
"loss": 0.0393,
"step": 10880
},
{
"grad_norm": 0.6527905464172363,
"learning_rate": 7.552793574670485e-05,
"loss": 0.0368,
"step": 10890
},
{
"grad_norm": 0.6603227853775024,
"learning_rate": 7.54805294378923e-05,
"loss": 0.0422,
"step": 10900
},
{
"grad_norm": 0.6587973833084106,
"learning_rate": 7.543309216783384e-05,
"loss": 0.0378,
"step": 10910
},
{
"grad_norm": 0.7895269393920898,
"learning_rate": 7.53856239941702e-05,
"loss": 0.0331,
"step": 10920
},
{
"grad_norm": 0.5479774475097656,
"learning_rate": 7.533812497457972e-05,
"loss": 0.046,
"step": 10930
},
{
"grad_norm": 0.5610347390174866,
"learning_rate": 7.529059516677814e-05,
"loss": 0.0417,
"step": 10940
},
{
"grad_norm": 0.7604182362556458,
"learning_rate": 7.524303462851872e-05,
"loss": 0.0336,
"step": 10950
},
{
"grad_norm": 0.6482762098312378,
"learning_rate": 7.519544341759192e-05,
"loss": 0.0318,
"step": 10960
},
{
"grad_norm": 0.6553546786308289,
"learning_rate": 7.514782159182562e-05,
"loss": 0.0314,
"step": 10970
},
{
"grad_norm": 0.5254109501838684,
"learning_rate": 7.510016920908481e-05,
"loss": 0.0343,
"step": 10980
},
{
"grad_norm": 0.5273259282112122,
"learning_rate": 7.505248632727158e-05,
"loss": 0.0341,
"step": 10990
},
{
"grad_norm": 0.639883279800415,
"learning_rate": 7.50047730043252e-05,
"loss": 0.0328,
"step": 11000
},
{
"grad_norm": 0.6890296339988708,
"learning_rate": 7.495702929822183e-05,
"loss": 0.0335,
"step": 11010
},
{
"grad_norm": 0.6245216727256775,
"learning_rate": 7.490925526697455e-05,
"loss": 0.0355,
"step": 11020
},
{
"grad_norm": 0.7273987531661987,
"learning_rate": 7.486145096863334e-05,
"loss": 0.0363,
"step": 11030
},
{
"grad_norm": 0.6850669384002686,
"learning_rate": 7.481361646128491e-05,
"loss": 0.0359,
"step": 11040
},
{
"grad_norm": 0.7155725359916687,
"learning_rate": 7.476575180305271e-05,
"loss": 0.0324,
"step": 11050
},
{
"grad_norm": 0.7474520802497864,
"learning_rate": 7.471785705209682e-05,
"loss": 0.0413,
"step": 11060
},
{
"grad_norm": 0.4675997197628021,
"learning_rate": 7.466993226661387e-05,
"loss": 0.0299,
"step": 11070
},
{
"grad_norm": 0.5993349552154541,
"learning_rate": 7.462197750483699e-05,
"loss": 0.0373,
"step": 11080
},
{
"grad_norm": 0.513828694820404,
"learning_rate": 7.457399282503574e-05,
"loss": 0.0297,
"step": 11090
},
{
"grad_norm": 0.5940067768096924,
"learning_rate": 7.452597828551604e-05,
"loss": 0.0384,
"step": 11100
},
{
"grad_norm": 0.6551719307899475,
"learning_rate": 7.447793394462006e-05,
"loss": 0.0371,
"step": 11110
},
{
"grad_norm": 0.6518399715423584,
"learning_rate": 7.442985986072624e-05,
"loss": 0.0346,
"step": 11120
},
{
"grad_norm": 0.6839174628257751,
"learning_rate": 7.438175609224908e-05,
"loss": 0.0406,
"step": 11130
},
{
"grad_norm": 0.5911973118782043,
"learning_rate": 7.433362269763924e-05,
"loss": 0.0305,
"step": 11140
},
{
"grad_norm": 0.6020933389663696,
"learning_rate": 7.428545973538329e-05,
"loss": 0.0227,
"step": 11150
},
{
"grad_norm": 0.6545886993408203,
"learning_rate": 7.42372672640038e-05,
"loss": 0.0328,
"step": 11160
},
{
"grad_norm": 0.5474277138710022,
"learning_rate": 7.418904534205917e-05,
"loss": 0.0296,
"step": 11170
},
{
"grad_norm": 0.532030463218689,
"learning_rate": 7.414079402814356e-05,
"loss": 0.0319,
"step": 11180
},
{
"grad_norm": 0.5599130988121033,
"learning_rate": 7.409251338088687e-05,
"loss": 0.0348,
"step": 11190
},
{
"grad_norm": 0.6050534844398499,
"learning_rate": 7.404420345895467e-05,
"loss": 0.0346,
"step": 11200
},
{
"grad_norm": 0.6966410875320435,
"learning_rate": 7.399586432104804e-05,
"loss": 0.0348,
"step": 11210
},
{
"grad_norm": 0.48129600286483765,
"learning_rate": 7.394749602590358e-05,
"loss": 0.0317,
"step": 11220
},
{
"grad_norm": 0.605392336845398,
"learning_rate": 7.389909863229336e-05,
"loss": 0.0433,
"step": 11230
},
{
"grad_norm": 0.6171544194221497,
"learning_rate": 7.385067219902477e-05,
"loss": 0.0306,
"step": 11240
},
{
"grad_norm": 0.7285153269767761,
"learning_rate": 7.380221678494049e-05,
"loss": 0.0269,
"step": 11250
},
{
"grad_norm": 0.5658971071243286,
"learning_rate": 7.37537324489184e-05,
"loss": 0.0257,
"step": 11260
},
{
"grad_norm": 0.5757483839988708,
"learning_rate": 7.370521924987155e-05,
"loss": 0.027,
"step": 11270
},
{
"grad_norm": 0.5941859483718872,
"learning_rate": 7.365667724674806e-05,
"loss": 0.0425,
"step": 11280
},
{
"grad_norm": 0.7511237263679504,
"learning_rate": 7.360810649853105e-05,
"loss": 0.0359,
"step": 11290
},
{
"grad_norm": 0.650376558303833,
"learning_rate": 7.355950706423853e-05,
"loss": 0.0328,
"step": 11300
},
{
"grad_norm": 0.7988902926445007,
"learning_rate": 7.351087900292342e-05,
"loss": 0.033,
"step": 11310
},
{
"grad_norm": 0.6878522634506226,
"learning_rate": 7.346222237367339e-05,
"loss": 0.0328,
"step": 11320
},
{
"grad_norm": 0.618222713470459,
"learning_rate": 7.341353723561081e-05,
"loss": 0.0319,
"step": 11330
},
{
"grad_norm": 0.5047088265419006,
"learning_rate": 7.336482364789277e-05,
"loss": 0.0242,
"step": 11340
},
{
"grad_norm": 0.537241518497467,
"learning_rate": 7.331608166971082e-05,
"loss": 0.0467,
"step": 11350
},
{
"grad_norm": 0.7081823348999023,
"learning_rate": 7.326731136029108e-05,
"loss": 0.032,
"step": 11360
},
{
"grad_norm": 0.6761635541915894,
"learning_rate": 7.321851277889408e-05,
"loss": 0.0266,
"step": 11370
},
{
"grad_norm": 0.49546825885772705,
"learning_rate": 7.316968598481469e-05,
"loss": 0.0319,
"step": 11380
},
{
"grad_norm": 0.6182970404624939,
"learning_rate": 7.312083103738206e-05,
"loss": 0.0399,
"step": 11390
},
{
"grad_norm": 0.594307005405426,
"learning_rate": 7.307194799595958e-05,
"loss": 0.0378,
"step": 11400
},
{
"grad_norm": 0.7331461906433105,
"learning_rate": 7.302303691994473e-05,
"loss": 0.0289,
"step": 11410
},
{
"grad_norm": 0.5764034986495972,
"learning_rate": 7.29740978687691e-05,
"loss": 0.0459,
"step": 11420
},
{
"grad_norm": 0.7816040515899658,
"learning_rate": 7.292513090189825e-05,
"loss": 0.0331,
"step": 11430
},
{
"grad_norm": 0.7596520781517029,
"learning_rate": 7.287613607883163e-05,
"loss": 0.0329,
"step": 11440
},
{
"grad_norm": 0.571803092956543,
"learning_rate": 7.282711345910263e-05,
"loss": 0.033,
"step": 11450
},
{
"grad_norm": 0.6962886452674866,
"learning_rate": 7.27780631022783e-05,
"loss": 0.0393,
"step": 11460
},
{
"grad_norm": 0.6777133941650391,
"learning_rate": 7.272898506795948e-05,
"loss": 0.0304,
"step": 11470
},
{
"grad_norm": 0.6623858213424683,
"learning_rate": 7.267987941578058e-05,
"loss": 0.0294,
"step": 11480
},
{
"grad_norm": 0.7173178791999817,
"learning_rate": 7.263074620540962e-05,
"loss": 0.0357,
"step": 11490
},
{
"grad_norm": 0.5767369866371155,
"learning_rate": 7.25815854965481e-05,
"loss": 0.0312,
"step": 11500
},
{
"grad_norm": 0.5995810627937317,
"learning_rate": 7.253239734893089e-05,
"loss": 0.0261,
"step": 11510
},
{
"grad_norm": 0.5689314007759094,
"learning_rate": 7.248318182232623e-05,
"loss": 0.0292,
"step": 11520
},
{
"grad_norm": 0.7716387510299683,
"learning_rate": 7.243393897653565e-05,
"loss": 0.0269,
"step": 11530
},
{
"grad_norm": 0.4963926076889038,
"learning_rate": 7.23846688713938e-05,
"loss": 0.0325,
"step": 11540
},
{
"grad_norm": 0.6989874839782715,
"learning_rate": 7.233537156676854e-05,
"loss": 0.0363,
"step": 11550
},
{
"grad_norm": 0.6602086424827576,
"learning_rate": 7.228604712256076e-05,
"loss": 0.0325,
"step": 11560
},
{
"grad_norm": 0.5852044820785522,
"learning_rate": 7.223669559870426e-05,
"loss": 0.0283,
"step": 11570
},
{
"grad_norm": 0.6695244908332825,
"learning_rate": 7.218731705516585e-05,
"loss": 0.0363,
"step": 11580
},
{
"grad_norm": 0.593602180480957,
"learning_rate": 7.21379115519451e-05,
"loss": 0.0394,
"step": 11590
},
{
"grad_norm": 0.6178359389305115,
"learning_rate": 7.20884791490743e-05,
"loss": 0.0311,
"step": 11600
},
{
"grad_norm": 0.6107975244522095,
"learning_rate": 7.203901990661857e-05,
"loss": 0.04,
"step": 11610
},
{
"grad_norm": 0.7388432621955872,
"learning_rate": 7.198953388467549e-05,
"loss": 0.0496,
"step": 11620
},
{
"grad_norm": 0.6027284860610962,
"learning_rate": 7.194002114337526e-05,
"loss": 0.0519,
"step": 11630
},
{
"grad_norm": 0.716697096824646,
"learning_rate": 7.189048174288054e-05,
"loss": 0.0334,
"step": 11640
},
{
"grad_norm": 0.5934044122695923,
"learning_rate": 7.184091574338636e-05,
"loss": 0.0334,
"step": 11650
},
{
"grad_norm": 0.5458714365959167,
"learning_rate": 7.179132320512009e-05,
"loss": 0.0321,
"step": 11660
},
{
"grad_norm": 0.5490051507949829,
"learning_rate": 7.174170418834134e-05,
"loss": 0.0237,
"step": 11670
},
{
"grad_norm": 0.6042584776878357,
"learning_rate": 7.169205875334189e-05,
"loss": 0.0344,
"step": 11680
},
{
"grad_norm": 0.6334197521209717,
"learning_rate": 7.164238696044562e-05,
"loss": 0.0323,
"step": 11690
},
{
"grad_norm": 0.5060211420059204,
"learning_rate": 7.159268887000846e-05,
"loss": 0.0349,
"step": 11700
},
{
"grad_norm": 0.6472730040550232,
"learning_rate": 7.154296454241827e-05,
"loss": 0.0399,
"step": 11710
},
{
"grad_norm": 0.5644455552101135,
"learning_rate": 7.149321403809479e-05,
"loss": 0.036,
"step": 11720
},
{
"grad_norm": 0.5377004146575928,
"learning_rate": 7.144343741748956e-05,
"loss": 0.0294,
"step": 11730
},
{
"grad_norm": 0.649018406867981,
"learning_rate": 7.139363474108589e-05,
"loss": 0.0301,
"step": 11740
},
{
"grad_norm": 0.6406453251838684,
"learning_rate": 7.134380606939874e-05,
"loss": 0.0384,
"step": 11750
},
{
"grad_norm": 0.6631308197975159,
"learning_rate": 7.12939514629746e-05,
"loss": 0.0302,
"step": 11760
},
{
"grad_norm": 0.5578881502151489,
"learning_rate": 7.124407098239155e-05,
"loss": 0.0301,
"step": 11770
},
{
"grad_norm": 0.7225531339645386,
"learning_rate": 7.119416468825908e-05,
"loss": 0.0276,
"step": 11780
},
{
"grad_norm": 0.6244858503341675,
"learning_rate": 7.114423264121804e-05,
"loss": 0.0333,
"step": 11790
},
{
"grad_norm": 0.6503976583480835,
"learning_rate": 7.109427490194056e-05,
"loss": 0.0395,
"step": 11800
},
{
"grad_norm": 0.7467981576919556,
"learning_rate": 7.104429153113001e-05,
"loss": 0.0476,
"step": 11810
},
{
"grad_norm": 0.7426225543022156,
"learning_rate": 7.099428258952092e-05,
"loss": 0.0382,
"step": 11820
},
{
"grad_norm": 0.7167433500289917,
"learning_rate": 7.094424813787883e-05,
"loss": 0.05,
"step": 11830
},
{
"grad_norm": 0.6713168025016785,
"learning_rate": 7.089418823700034e-05,
"loss": 0.0332,
"step": 11840
},
{
"grad_norm": 0.5983794331550598,
"learning_rate": 7.084410294771298e-05,
"loss": 0.0388,
"step": 11850
},
{
"grad_norm": 0.6211906671524048,
"learning_rate": 7.079399233087504e-05,
"loss": 0.0398,
"step": 11860
},
{
"grad_norm": 0.5773683786392212,
"learning_rate": 7.074385644737568e-05,
"loss": 0.0334,
"step": 11870
},
{
"grad_norm": 0.6725842356681824,
"learning_rate": 7.069369535813473e-05,
"loss": 0.0491,
"step": 11880
},
{
"grad_norm": 0.6285985708236694,
"learning_rate": 7.06435091241026e-05,
"loss": 0.0318,
"step": 11890
},
{
"grad_norm": 0.7551082372665405,
"learning_rate": 7.059329780626034e-05,
"loss": 0.0355,
"step": 11900
},
{
"grad_norm": 0.6254936456680298,
"learning_rate": 7.054306146561944e-05,
"loss": 0.0387,
"step": 11910
},
{
"grad_norm": 0.6552140116691589,
"learning_rate": 7.049280016322178e-05,
"loss": 0.0357,
"step": 11920
},
{
"grad_norm": 0.5803700089454651,
"learning_rate": 7.044251396013957e-05,
"loss": 0.0378,
"step": 11930
},
{
"grad_norm": 0.6197409629821777,
"learning_rate": 7.039220291747528e-05,
"loss": 0.033,
"step": 11940
},
{
"grad_norm": 0.6032425165176392,
"learning_rate": 7.034186709636159e-05,
"loss": 0.0277,
"step": 11950
},
{
"grad_norm": 0.7099371552467346,
"learning_rate": 7.029150655796129e-05,
"loss": 0.0319,
"step": 11960
},
{
"grad_norm": 0.6506530046463013,
"learning_rate": 7.024112136346712e-05,
"loss": 0.0336,
"step": 11970
},
{
"grad_norm": 0.64686119556427,
"learning_rate": 7.01907115741019e-05,
"loss": 0.0366,
"step": 11980
},
{
"grad_norm": 0.6405677199363708,
"learning_rate": 7.014027725111825e-05,
"loss": 0.037,
"step": 11990
},
{
"grad_norm": 0.6229185461997986,
"learning_rate": 7.008981845579864e-05,
"loss": 0.0442,
"step": 12000
},
{
"grad_norm": 0.6924329996109009,
"learning_rate": 7.003933524945528e-05,
"loss": 0.0378,
"step": 12010
},
{
"grad_norm": 0.5977965593338013,
"learning_rate": 6.998882769342998e-05,
"loss": 0.0311,
"step": 12020
},
{
"grad_norm": 0.616434633731842,
"learning_rate": 6.993829584909423e-05,
"loss": 0.0316,
"step": 12030
},
{
"grad_norm": 0.6260470747947693,
"learning_rate": 6.988773977784895e-05,
"loss": 0.0337,
"step": 12040
},
{
"grad_norm": 0.78681480884552,
"learning_rate": 6.983715954112454e-05,
"loss": 0.0392,
"step": 12050
},
{
"grad_norm": 0.6005724668502808,
"learning_rate": 6.978655520038079e-05,
"loss": 0.0316,
"step": 12060
},
{
"grad_norm": 0.6713159084320068,
"learning_rate": 6.97359268171067e-05,
"loss": 0.0522,
"step": 12070
},
{
"grad_norm": 0.7689294219017029,
"learning_rate": 6.968527445282056e-05,
"loss": 0.0347,
"step": 12080
},
{
"grad_norm": 0.6034610271453857,
"learning_rate": 6.963459816906977e-05,
"loss": 0.0272,
"step": 12090
},
{
"grad_norm": 0.5005972981452942,
"learning_rate": 6.958389802743077e-05,
"loss": 0.0399,
"step": 12100
},
{
"grad_norm": 0.7032498121261597,
"learning_rate": 6.953317408950903e-05,
"loss": 0.0316,
"step": 12110
},
{
"grad_norm": 0.727918267250061,
"learning_rate": 6.94824264169389e-05,
"loss": 0.0304,
"step": 12120
},
{
"grad_norm": 0.6104573011398315,
"learning_rate": 6.94316550713836e-05,
"loss": 0.0354,
"step": 12130
},
{
"grad_norm": 0.7801100015640259,
"learning_rate": 6.938086011453513e-05,
"loss": 0.0288,
"step": 12140
},
{
"grad_norm": 0.5958548188209534,
"learning_rate": 6.93300416081141e-05,
"loss": 0.0302,
"step": 12150
},
{
"grad_norm": 0.608515739440918,
"learning_rate": 6.927919961386984e-05,
"loss": 0.0296,
"step": 12160
},
{
"grad_norm": 0.5884709358215332,
"learning_rate": 6.922833419358013e-05,
"loss": 0.0299,
"step": 12170
},
{
"grad_norm": 0.6651334166526794,
"learning_rate": 6.917744540905124e-05,
"loss": 0.0412,
"step": 12180
},
{
"grad_norm": 0.6754817962646484,
"learning_rate": 6.912653332211787e-05,
"loss": 0.0303,
"step": 12190
},
{
"grad_norm": 0.5638582706451416,
"learning_rate": 6.9075597994643e-05,
"loss": 0.0425,
"step": 12200
},
{
"grad_norm": 0.5522139668464661,
"learning_rate": 6.902463948851787e-05,
"loss": 0.0492,
"step": 12210
},
{
"grad_norm": 0.612118124961853,
"learning_rate": 6.897365786566184e-05,
"loss": 0.0311,
"step": 12220
},
{
"grad_norm": 0.6508922576904297,
"learning_rate": 6.892265318802242e-05,
"loss": 0.0318,
"step": 12230
},
{
"grad_norm": 0.7115322947502136,
"learning_rate": 6.887162551757507e-05,
"loss": 0.0282,
"step": 12240
},
{
"grad_norm": 0.6574018597602844,
"learning_rate": 6.882057491632326e-05,
"loss": 0.0424,
"step": 12250
},
{
"grad_norm": 0.6815955638885498,
"learning_rate": 6.876950144629824e-05,
"loss": 0.0428,
"step": 12260
},
{
"grad_norm": 0.5280714631080627,
"learning_rate": 6.87184051695591e-05,
"loss": 0.0358,
"step": 12270
},
{
"grad_norm": 0.5930230617523193,
"learning_rate": 6.866728614819268e-05,
"loss": 0.0329,
"step": 12280
},
{
"grad_norm": 0.5141021609306335,
"learning_rate": 6.861614444431337e-05,
"loss": 0.0388,
"step": 12290
},
{
"grad_norm": 0.5407117605209351,
"learning_rate": 6.856498012006318e-05,
"loss": 0.0263,
"step": 12300
},
{
"grad_norm": 0.6878035068511963,
"learning_rate": 6.851379323761157e-05,
"loss": 0.0342,
"step": 12310
},
{
"grad_norm": 0.8023784756660461,
"learning_rate": 6.846258385915545e-05,
"loss": 0.0369,
"step": 12320
},
{
"grad_norm": 0.5518360137939453,
"learning_rate": 6.841135204691902e-05,
"loss": 0.0271,
"step": 12330
},
{
"grad_norm": 0.752828061580658,
"learning_rate": 6.836009786315377e-05,
"loss": 0.0356,
"step": 12340
},
{
"grad_norm": 0.54446941614151,
"learning_rate": 6.830882137013839e-05,
"loss": 0.0333,
"step": 12350
},
{
"grad_norm": 0.635636031627655,
"learning_rate": 6.825752263017863e-05,
"loss": 0.0362,
"step": 12360
},
{
"grad_norm": 0.6407579183578491,
"learning_rate": 6.820620170560731e-05,
"loss": 0.0301,
"step": 12370
},
{
"grad_norm": 0.661153256893158,
"learning_rate": 6.815485865878418e-05,
"loss": 0.0331,
"step": 12380
},
{
"grad_norm": 0.677038848400116,
"learning_rate": 6.810349355209588e-05,
"loss": 0.035,
"step": 12390
},
{
"grad_norm": 0.6581941843032837,
"learning_rate": 6.805210644795588e-05,
"loss": 0.0392,
"step": 12400
},
{
"grad_norm": 0.6731207966804504,
"learning_rate": 6.800069740880433e-05,
"loss": 0.0427,
"step": 12410
},
{
"grad_norm": 0.6513192057609558,
"learning_rate": 6.794926649710807e-05,
"loss": 0.0291,
"step": 12420
},
{
"grad_norm": 0.6166074872016907,
"learning_rate": 6.789781377536052e-05,
"loss": 0.0443,
"step": 12430
},
{
"grad_norm": 0.6841890215873718,
"learning_rate": 6.784633930608158e-05,
"loss": 0.0417,
"step": 12440
},
{
"grad_norm": 0.5105737447738647,
"learning_rate": 6.779484315181759e-05,
"loss": 0.0325,
"step": 12450
},
{
"grad_norm": 0.5187857747077942,
"learning_rate": 6.774332537514122e-05,
"loss": 0.0386,
"step": 12460
},
{
"grad_norm": 0.5860793590545654,
"learning_rate": 6.769178603865143e-05,
"loss": 0.0305,
"step": 12470
},
{
"grad_norm": 0.5588794946670532,
"learning_rate": 6.764022520497337e-05,
"loss": 0.0265,
"step": 12480
},
{
"grad_norm": 0.6861588358879089,
"learning_rate": 6.758864293675833e-05,
"loss": 0.0291,
"step": 12490
},
{
"grad_norm": 0.6288846135139465,
"learning_rate": 6.753703929668362e-05,
"loss": 0.0313,
"step": 12500
},
{
"grad_norm": 0.6623268723487854,
"learning_rate": 6.748541434745253e-05,
"loss": 0.0442,
"step": 12510
},
{
"grad_norm": 0.6872732639312744,
"learning_rate": 6.743376815179424e-05,
"loss": 0.0343,
"step": 12520
},
{
"grad_norm": 0.5934109687805176,
"learning_rate": 6.738210077246376e-05,
"loss": 0.0353,
"step": 12530
},
{
"grad_norm": 0.6127296686172485,
"learning_rate": 6.733041227224181e-05,
"loss": 0.0432,
"step": 12540
},
{
"grad_norm": 0.7586196064949036,
"learning_rate": 6.72787027139348e-05,
"loss": 0.0392,
"step": 12550
},
{
"grad_norm": 0.6353082060813904,
"learning_rate": 6.72269721603747e-05,
"loss": 0.0304,
"step": 12560
},
{
"grad_norm": 0.5351864695549011,
"learning_rate": 6.717522067441904e-05,
"loss": 0.0311,
"step": 12570
},
{
"grad_norm": 0.58354252576828,
"learning_rate": 6.712344831895074e-05,
"loss": 0.0326,
"step": 12580
},
{
"grad_norm": 0.5963569283485413,
"learning_rate": 6.707165515687811e-05,
"loss": 0.0385,
"step": 12590
},
{
"grad_norm": 0.5581371784210205,
"learning_rate": 6.70198412511347e-05,
"loss": 0.0343,
"step": 12600
},
{
"grad_norm": 0.6134654879570007,
"learning_rate": 6.69680066646793e-05,
"loss": 0.0271,
"step": 12610
},
{
"grad_norm": 0.5863450765609741,
"learning_rate": 6.691615146049583e-05,
"loss": 0.0298,
"step": 12620
},
{
"grad_norm": 0.5359828472137451,
"learning_rate": 6.686427570159324e-05,
"loss": 0.0262,
"step": 12630
},
{
"grad_norm": 0.6880347728729248,
"learning_rate": 6.681237945100548e-05,
"loss": 0.0398,
"step": 12640
},
{
"grad_norm": 0.5505436062812805,
"learning_rate": 6.676046277179139e-05,
"loss": 0.0289,
"step": 12650
},
{
"grad_norm": 0.5953719019889832,
"learning_rate": 6.670852572703462e-05,
"loss": 0.0411,
"step": 12660
},
{
"grad_norm": 0.5068235397338867,
"learning_rate": 6.665656837984359e-05,
"loss": 0.0306,
"step": 12670
},
{
"grad_norm": 0.5572917461395264,
"learning_rate": 6.660459079335135e-05,
"loss": 0.0375,
"step": 12680
},
{
"grad_norm": 0.6635385155677795,
"learning_rate": 6.655259303071558e-05,
"loss": 0.0316,
"step": 12690
},
{
"grad_norm": 0.6144265532493591,
"learning_rate": 6.650057515511848e-05,
"loss": 0.0295,
"step": 12700
},
{
"grad_norm": 0.49584537744522095,
"learning_rate": 6.644853722976667e-05,
"loss": 0.0277,
"step": 12710
},
{
"grad_norm": 0.5529701113700867,
"learning_rate": 6.639647931789113e-05,
"loss": 0.0282,
"step": 12720
},
{
"grad_norm": 0.6346619129180908,
"learning_rate": 6.634440148274713e-05,
"loss": 0.0287,
"step": 12730
},
{
"grad_norm": 0.753279983997345,
"learning_rate": 6.629230378761415e-05,
"loss": 0.034,
"step": 12740
},
{
"grad_norm": 0.6205716133117676,
"learning_rate": 6.624018629579581e-05,
"loss": 0.0284,
"step": 12750
},
{
"grad_norm": 0.7444762587547302,
"learning_rate": 6.618804907061976e-05,
"loss": 0.0376,
"step": 12760
},
{
"grad_norm": 0.5614160895347595,
"learning_rate": 6.613589217543766e-05,
"loss": 0.0409,
"step": 12770
},
{
"grad_norm": 0.6681153178215027,
"learning_rate": 6.608371567362504e-05,
"loss": 0.035,
"step": 12780
},
{
"grad_norm": 0.8152887225151062,
"learning_rate": 6.60315196285813e-05,
"loss": 0.0375,
"step": 12790
},
{
"grad_norm": 0.5622035264968872,
"learning_rate": 6.597930410372954e-05,
"loss": 0.0301,
"step": 12800
},
{
"grad_norm": 0.6008073091506958,
"learning_rate": 6.592706916251653e-05,
"loss": 0.0334,
"step": 12810
},
{
"grad_norm": 0.5802395343780518,
"learning_rate": 6.587481486841267e-05,
"loss": 0.032,
"step": 12820
},
{
"grad_norm": 0.5756995677947998,
"learning_rate": 6.582254128491184e-05,
"loss": 0.0338,
"step": 12830
},
{
"grad_norm": 0.6084258556365967,
"learning_rate": 6.577024847553139e-05,
"loss": 0.0296,
"step": 12840
},
{
"grad_norm": 0.6573593020439148,
"learning_rate": 6.571793650381201e-05,
"loss": 0.0276,
"step": 12850
},
{
"grad_norm": 0.679627537727356,
"learning_rate": 6.566560543331766e-05,
"loss": 0.0327,
"step": 12860
},
{
"grad_norm": 0.5940977931022644,
"learning_rate": 6.561325532763554e-05,
"loss": 0.0272,
"step": 12870
},
{
"grad_norm": 0.6621523499488831,
"learning_rate": 6.556088625037597e-05,
"loss": 0.0255,
"step": 12880
},
{
"grad_norm": 0.5121192932128906,
"learning_rate": 6.55084982651723e-05,
"loss": 0.0276,
"step": 12890
},
{
"grad_norm": 0.5393102765083313,
"learning_rate": 6.54560914356809e-05,
"loss": 0.0387,
"step": 12900
},
{
"grad_norm": 0.4858904778957367,
"learning_rate": 6.540366582558097e-05,
"loss": 0.024,
"step": 12910
},
{
"grad_norm": 0.7111104130744934,
"learning_rate": 6.53512214985746e-05,
"loss": 0.0232,
"step": 12920
},
{
"grad_norm": 0.5576184391975403,
"learning_rate": 6.529875851838658e-05,
"loss": 0.0274,
"step": 12930
},
{
"grad_norm": 0.5276832580566406,
"learning_rate": 6.52462769487644e-05,
"loss": 0.0279,
"step": 12940
},
{
"grad_norm": 0.5785664916038513,
"learning_rate": 6.519377685347807e-05,
"loss": 0.0222,
"step": 12950
},
{
"grad_norm": 0.7197355031967163,
"learning_rate": 6.514125829632021e-05,
"loss": 0.0252,
"step": 12960
},
{
"grad_norm": 0.6110815405845642,
"learning_rate": 6.508872134110577e-05,
"loss": 0.0264,
"step": 12970
},
{
"grad_norm": 0.553406834602356,
"learning_rate": 6.503616605167213e-05,
"loss": 0.0253,
"step": 12980
},
{
"grad_norm": 0.5752418637275696,
"learning_rate": 6.498359249187893e-05,
"loss": 0.0265,
"step": 12990
},
{
"grad_norm": 0.6047320365905762,
"learning_rate": 6.493100072560799e-05,
"loss": 0.0301,
"step": 13000
},
{
"grad_norm": 0.526272177696228,
"learning_rate": 6.487839081676327e-05,
"loss": 0.0457,
"step": 13010
},
{
"grad_norm": 0.6036477088928223,
"learning_rate": 6.482576282927076e-05,
"loss": 0.0395,
"step": 13020
},
{
"grad_norm": 0.6669952869415283,
"learning_rate": 6.477311682707845e-05,
"loss": 0.037,
"step": 13030
},
{
"grad_norm": 0.5997741222381592,
"learning_rate": 6.472045287415616e-05,
"loss": 0.0267,
"step": 13040
},
{
"grad_norm": 0.5792614817619324,
"learning_rate": 6.466777103449558e-05,
"loss": 0.0233,
"step": 13050
},
{
"grad_norm": 0.5351223349571228,
"learning_rate": 6.461507137211012e-05,
"loss": 0.0273,
"step": 13060
},
{
"grad_norm": 0.6230636835098267,
"learning_rate": 6.456235395103483e-05,
"loss": 0.0383,
"step": 13070
},
{
"grad_norm": 0.6492744088172913,
"learning_rate": 6.450961883532634e-05,
"loss": 0.0266,
"step": 13080
},
{
"grad_norm": 0.633439302444458,
"learning_rate": 6.445686608906283e-05,
"loss": 0.0282,
"step": 13090
},
{
"grad_norm": 0.5529469847679138,
"learning_rate": 6.44040957763438e-05,
"loss": 0.0296,
"step": 13100
},
{
"grad_norm": 0.556348443031311,
"learning_rate": 6.435130796129018e-05,
"loss": 0.0237,
"step": 13110
},
{
"grad_norm": 0.6205959916114807,
"learning_rate": 6.429850270804416e-05,
"loss": 0.0231,
"step": 13120
},
{
"grad_norm": 0.6284688115119934,
"learning_rate": 6.424568008076909e-05,
"loss": 0.0336,
"step": 13130
},
{
"grad_norm": 0.5601750016212463,
"learning_rate": 6.419284014364943e-05,
"loss": 0.0339,
"step": 13140
},
{
"grad_norm": 0.579443097114563,
"learning_rate": 6.41399829608907e-05,
"loss": 0.0311,
"step": 13150
},
{
"grad_norm": 0.48059356212615967,
"learning_rate": 6.408710859671938e-05,
"loss": 0.0228,
"step": 13160
},
{
"grad_norm": 0.6346940994262695,
"learning_rate": 6.403421711538278e-05,
"loss": 0.0277,
"step": 13170
},
{
"grad_norm": 0.5606958866119385,
"learning_rate": 6.398130858114903e-05,
"loss": 0.0273,
"step": 13180
},
{
"grad_norm": 0.5713655352592468,
"learning_rate": 6.392838305830701e-05,
"loss": 0.0229,
"step": 13190
},
{
"grad_norm": 0.6509492993354797,
"learning_rate": 6.387544061116621e-05,
"loss": 0.0312,
"step": 13200
},
{
"grad_norm": 0.4779699146747589,
"learning_rate": 6.382248130405671e-05,
"loss": 0.0197,
"step": 13210
},
{
"grad_norm": 0.5044391751289368,
"learning_rate": 6.376950520132905e-05,
"loss": 0.0251,
"step": 13220
},
{
"grad_norm": 0.583042562007904,
"learning_rate": 6.371651236735418e-05,
"loss": 0.0282,
"step": 13230
},
{
"grad_norm": 0.7058117389678955,
"learning_rate": 6.366350286652341e-05,
"loss": 0.0412,
"step": 13240
},
{
"grad_norm": 0.6108366250991821,
"learning_rate": 6.361047676324827e-05,
"loss": 0.0259,
"step": 13250
},
{
"grad_norm": 0.6422246098518372,
"learning_rate": 6.355743412196047e-05,
"loss": 0.0288,
"step": 13260
},
{
"grad_norm": 0.5584660768508911,
"learning_rate": 6.350437500711184e-05,
"loss": 0.0273,
"step": 13270
},
{
"grad_norm": 0.6778274178504944,
"learning_rate": 6.345129948317418e-05,
"loss": 0.0273,
"step": 13280
},
{
"grad_norm": 0.6109534502029419,
"learning_rate": 6.33982076146393e-05,
"loss": 0.0271,
"step": 13290
},
{
"grad_norm": 0.6794594526290894,
"learning_rate": 6.334509946601879e-05,
"loss": 0.0237,
"step": 13300
},
{
"grad_norm": 0.6256342530250549,
"learning_rate": 6.329197510184406e-05,
"loss": 0.033,
"step": 13310
},
{
"grad_norm": 0.6385566592216492,
"learning_rate": 6.323883458666624e-05,
"loss": 0.0386,
"step": 13320
},
{
"grad_norm": 0.5630882978439331,
"learning_rate": 6.318567798505605e-05,
"loss": 0.0282,
"step": 13330
},
{
"grad_norm": 0.5479835867881775,
"learning_rate": 6.313250536160378e-05,
"loss": 0.0303,
"step": 13340
},
{
"grad_norm": 0.5947245359420776,
"learning_rate": 6.307931678091917e-05,
"loss": 0.0234,
"step": 13350
},
{
"grad_norm": 0.705656111240387,
"learning_rate": 6.302611230763138e-05,
"loss": 0.0289,
"step": 13360
},
{
"grad_norm": 0.6083478331565857,
"learning_rate": 6.297289200638887e-05,
"loss": 0.0351,
"step": 13370
},
{
"grad_norm": 0.5655539631843567,
"learning_rate": 6.29196559418593e-05,
"loss": 0.0324,
"step": 13380
},
{
"grad_norm": 0.6297789812088013,
"learning_rate": 6.286640417872951e-05,
"loss": 0.034,
"step": 13390
},
{
"grad_norm": 0.5360618233680725,
"learning_rate": 6.281313678170542e-05,
"loss": 0.039,
"step": 13400
},
{
"grad_norm": 0.6041401624679565,
"learning_rate": 6.275985381551194e-05,
"loss": 0.0276,
"step": 13410
},
{
"grad_norm": 0.47424963116645813,
"learning_rate": 6.270655534489292e-05,
"loss": 0.0308,
"step": 13420
},
{
"grad_norm": 0.6943671107292175,
"learning_rate": 6.265324143461098e-05,
"loss": 0.0275,
"step": 13430
},
{
"grad_norm": 0.6331978440284729,
"learning_rate": 6.259991214944758e-05,
"loss": 0.0293,
"step": 13440
},
{
"grad_norm": 0.6188458800315857,
"learning_rate": 6.254656755420283e-05,
"loss": 0.0395,
"step": 13450
},
{
"grad_norm": 0.5554435849189758,
"learning_rate": 6.249320771369545e-05,
"loss": 0.0375,
"step": 13460
},
{
"grad_norm": 0.6950687170028687,
"learning_rate": 6.243983269276263e-05,
"loss": 0.0447,
"step": 13470
},
{
"grad_norm": 0.5953019261360168,
"learning_rate": 6.238644255626012e-05,
"loss": 0.0331,
"step": 13480
},
{
"grad_norm": 0.6013931632041931,
"learning_rate": 6.233303736906193e-05,
"loss": 0.0319,
"step": 13490
},
{
"grad_norm": 0.6358153223991394,
"learning_rate": 6.227961719606039e-05,
"loss": 0.0203,
"step": 13500
},
{
"grad_norm": 0.5066450834274292,
"learning_rate": 6.222618210216608e-05,
"loss": 0.0336,
"step": 13510
},
{
"grad_norm": 0.5191952586174011,
"learning_rate": 6.217273215230767e-05,
"loss": 0.0377,
"step": 13520
},
{
"grad_norm": 0.5887081027030945,
"learning_rate": 6.211926741143188e-05,
"loss": 0.0347,
"step": 13530
},
{
"grad_norm": 0.5157421827316284,
"learning_rate": 6.20657879445034e-05,
"loss": 0.0325,
"step": 13540
},
{
"grad_norm": 0.5061367154121399,
"learning_rate": 6.201229381650485e-05,
"loss": 0.0246,
"step": 13550
},
{
"grad_norm": 0.5677783489227295,
"learning_rate": 6.195878509243661e-05,
"loss": 0.0312,
"step": 13560
},
{
"grad_norm": 0.5754074454307556,
"learning_rate": 6.190526183731685e-05,
"loss": 0.0252,
"step": 13570
},
{
"grad_norm": 0.6771786212921143,
"learning_rate": 6.185172411618138e-05,
"loss": 0.0296,
"step": 13580
},
{
"grad_norm": 0.5526571273803711,
"learning_rate": 6.179817199408355e-05,
"loss": 0.0367,
"step": 13590
},
{
"grad_norm": 0.6066913604736328,
"learning_rate": 6.174460553609426e-05,
"loss": 0.0432,
"step": 13600
},
{
"grad_norm": 0.5527705550193787,
"learning_rate": 6.16910248073018e-05,
"loss": 0.0234,
"step": 13610
},
{
"grad_norm": 0.5730995535850525,
"learning_rate": 6.16374298728118e-05,
"loss": 0.0395,
"step": 13620
},
{
"grad_norm": 0.6747135519981384,
"learning_rate": 6.158382079774716e-05,
"loss": 0.0277,
"step": 13630
},
{
"grad_norm": 0.6275179982185364,
"learning_rate": 6.153019764724799e-05,
"loss": 0.028,
"step": 13640
},
{
"grad_norm": 0.5738186240196228,
"learning_rate": 6.147656048647144e-05,
"loss": 0.031,
"step": 13650
},
{
"grad_norm": 0.6589876413345337,
"learning_rate": 6.142290938059173e-05,
"loss": 0.0324,
"step": 13660
},
{
"grad_norm": 0.5665801763534546,
"learning_rate": 6.136924439480001e-05,
"loss": 0.0252,
"step": 13670
},
{
"grad_norm": 0.5784894824028015,
"learning_rate": 6.13155655943043e-05,
"loss": 0.0279,
"step": 13680
},
{
"grad_norm": 0.6009385585784912,
"learning_rate": 6.126187304432941e-05,
"loss": 0.0257,
"step": 13690
},
{
"grad_norm": 0.5812541842460632,
"learning_rate": 6.120816681011684e-05,
"loss": 0.0296,
"step": 13700
},
{
"grad_norm": 0.5980089902877808,
"learning_rate": 6.115444695692474e-05,
"loss": 0.0268,
"step": 13710
},
{
"grad_norm": 0.5743216872215271,
"learning_rate": 6.110071355002779e-05,
"loss": 0.023,
"step": 13720
},
{
"grad_norm": 0.4781140387058258,
"learning_rate": 6.104696665471714e-05,
"loss": 0.0231,
"step": 13730
},
{
"grad_norm": 0.5402002334594727,
"learning_rate": 6.099320633630036e-05,
"loss": 0.0225,
"step": 13740
},
{
"grad_norm": 0.5916728973388672,
"learning_rate": 6.093943266010128e-05,
"loss": 0.0265,
"step": 13750
},
{
"grad_norm": 0.49914225935935974,
"learning_rate": 6.088564569146e-05,
"loss": 0.0427,
"step": 13760
},
{
"grad_norm": 0.6978329420089722,
"learning_rate": 6.083184549573274e-05,
"loss": 0.0285,
"step": 13770
},
{
"grad_norm": 0.7084382772445679,
"learning_rate": 6.077803213829184e-05,
"loss": 0.0293,
"step": 13780
},
{
"grad_norm": 0.6758463382720947,
"learning_rate": 6.0724205684525594e-05,
"loss": 0.0283,
"step": 13790
},
{
"grad_norm": 0.6244635581970215,
"learning_rate": 6.0670366199838215e-05,
"loss": 0.0257,
"step": 13800
},
{
"grad_norm": 0.627186119556427,
"learning_rate": 6.061651374964974e-05,
"loss": 0.0379,
"step": 13810
},
{
"grad_norm": 0.6106190085411072,
"learning_rate": 6.0562648399396004e-05,
"loss": 0.0323,
"step": 13820
},
{
"grad_norm": 0.5287126898765564,
"learning_rate": 6.050877021452845e-05,
"loss": 0.0331,
"step": 13830
},
{
"grad_norm": 0.449434369802475,
"learning_rate": 6.04548792605142e-05,
"loss": 0.0351,
"step": 13840
},
{
"grad_norm": 0.7566120624542236,
"learning_rate": 6.04009756028358e-05,
"loss": 0.0262,
"step": 13850
},
{
"grad_norm": 0.4557020366191864,
"learning_rate": 6.0347059306991294e-05,
"loss": 0.0308,
"step": 13860
},
{
"grad_norm": 0.4765125811100006,
"learning_rate": 6.029313043849407e-05,
"loss": 0.0334,
"step": 13870
},
{
"grad_norm": 0.5916638970375061,
"learning_rate": 6.023918906287279e-05,
"loss": 0.0233,
"step": 13880
},
{
"grad_norm": 0.8654065728187561,
"learning_rate": 6.01852352456713e-05,
"loss": 0.0406,
"step": 13890
},
{
"grad_norm": 0.6977891325950623,
"learning_rate": 6.013126905244858e-05,
"loss": 0.0243,
"step": 13900
},
{
"grad_norm": 0.596157968044281,
"learning_rate": 6.007729054877864e-05,
"loss": 0.0307,
"step": 13910
},
{
"grad_norm": 0.5382691621780396,
"learning_rate": 6.0023299800250466e-05,
"loss": 0.0349,
"step": 13920
},
{
"grad_norm": 0.7043864727020264,
"learning_rate": 5.9969296872467894e-05,
"loss": 0.0311,
"step": 13930
},
{
"grad_norm": 0.5829411745071411,
"learning_rate": 5.9915281831049585e-05,
"loss": 0.0293,
"step": 13940
},
{
"grad_norm": 0.6979572176933289,
"learning_rate": 5.98612547416289e-05,
"loss": 0.0317,
"step": 13950
},
{
"grad_norm": 0.5189060568809509,
"learning_rate": 5.9807215669853855e-05,
"loss": 0.0311,
"step": 13960
},
{
"grad_norm": 0.7373045682907104,
"learning_rate": 5.9753164681387e-05,
"loss": 0.0536,
"step": 13970
},
{
"grad_norm": 0.6852480173110962,
"learning_rate": 5.969910184190539e-05,
"loss": 0.0382,
"step": 13980
},
{
"grad_norm": 0.688338577747345,
"learning_rate": 5.9645027217100477e-05,
"loss": 0.0318,
"step": 13990
},
{
"grad_norm": 0.6396709084510803,
"learning_rate": 5.959094087267804e-05,
"loss": 0.0325,
"step": 14000
},
{
"grad_norm": 0.563651978969574,
"learning_rate": 5.9536842874358065e-05,
"loss": 0.0289,
"step": 14010
},
{
"grad_norm": 0.6133570075035095,
"learning_rate": 5.9482733287874734e-05,
"loss": 0.0388,
"step": 14020
},
{
"grad_norm": 0.7192299962043762,
"learning_rate": 5.942861217897631e-05,
"loss": 0.0329,
"step": 14030
},
{
"grad_norm": 0.5441362261772156,
"learning_rate": 5.9374479613425e-05,
"loss": 0.0231,
"step": 14040
},
{
"grad_norm": 0.615674614906311,
"learning_rate": 5.932033565699704e-05,
"loss": 0.0238,
"step": 14050
},
{
"grad_norm": 0.5434221625328064,
"learning_rate": 5.926618037548237e-05,
"loss": 0.0291,
"step": 14060
},
{
"grad_norm": 0.6541652679443359,
"learning_rate": 5.9212013834684824e-05,
"loss": 0.0319,
"step": 14070
},
{
"grad_norm": 0.49798721075057983,
"learning_rate": 5.9157836100421824e-05,
"loss": 0.0275,
"step": 14080
},
{
"grad_norm": 0.6375170946121216,
"learning_rate": 5.910364723852444e-05,
"loss": 0.0276,
"step": 14090
},
{
"grad_norm": 0.4796459376811981,
"learning_rate": 5.904944731483723e-05,
"loss": 0.0314,
"step": 14100
},
{
"grad_norm": 0.5173145532608032,
"learning_rate": 5.899523639521825e-05,
"loss": 0.031,
"step": 14110
},
{
"grad_norm": 0.496674507856369,
"learning_rate": 5.894101454553883e-05,
"loss": 0.0251,
"step": 14120
},
{
"grad_norm": 0.5368170738220215,
"learning_rate": 5.888678183168368e-05,
"loss": 0.0376,
"step": 14130
},
{
"grad_norm": 0.5114181637763977,
"learning_rate": 5.883253831955061e-05,
"loss": 0.0283,
"step": 14140
},
{
"grad_norm": 0.7223478555679321,
"learning_rate": 5.8778284075050625e-05,
"loss": 0.0261,
"step": 14150
},
{
"grad_norm": 0.6599705815315247,
"learning_rate": 5.872401916410777e-05,
"loss": 0.0474,
"step": 14160
},
{
"grad_norm": 0.6466548442840576,
"learning_rate": 5.8669743652659014e-05,
"loss": 0.0311,
"step": 14170
},
{
"grad_norm": 0.5883294343948364,
"learning_rate": 5.86154576066542e-05,
"loss": 0.0286,
"step": 14180
},
{
"grad_norm": 0.597936749458313,
"learning_rate": 5.856116109205602e-05,
"loss": 0.0359,
"step": 14190
},
{
"grad_norm": 0.5859272480010986,
"learning_rate": 5.850685417483983e-05,
"loss": 0.0304,
"step": 14200
},
{
"grad_norm": 0.7173089385032654,
"learning_rate": 5.8452536920993684e-05,
"loss": 0.0341,
"step": 14210
},
{
"grad_norm": 0.7008039951324463,
"learning_rate": 5.839820939651817e-05,
"loss": 0.0357,
"step": 14220
},
{
"grad_norm": 0.5297600626945496,
"learning_rate": 5.834387166742632e-05,
"loss": 0.0304,
"step": 14230
},
{
"grad_norm": 0.694340169429779,
"learning_rate": 5.828952379974364e-05,
"loss": 0.034,
"step": 14240
},
{
"grad_norm": 0.5987368822097778,
"learning_rate": 5.8235165859507864e-05,
"loss": 0.037,
"step": 14250
},
{
"grad_norm": 0.6396300792694092,
"learning_rate": 5.818079791276907e-05,
"loss": 0.0294,
"step": 14260
},
{
"grad_norm": 0.7714502811431885,
"learning_rate": 5.8126420025589415e-05,
"loss": 0.0402,
"step": 14270
},
{
"grad_norm": 0.5708024501800537,
"learning_rate": 5.807203226404313e-05,
"loss": 0.0292,
"step": 14280
},
{
"grad_norm": 0.6370897889137268,
"learning_rate": 5.801763469421652e-05,
"loss": 0.0252,
"step": 14290
},
{
"grad_norm": 0.562615692615509,
"learning_rate": 5.796322738220774e-05,
"loss": 0.0357,
"step": 14300
},
{
"grad_norm": 0.739847719669342,
"learning_rate": 5.79088103941268e-05,
"loss": 0.0298,
"step": 14310
},
{
"grad_norm": 0.6278020143508911,
"learning_rate": 5.785438379609549e-05,
"loss": 0.0411,
"step": 14320
},
{
"grad_norm": 0.7169325947761536,
"learning_rate": 5.779994765424724e-05,
"loss": 0.027,
"step": 14330
},
{
"grad_norm": 0.5885417461395264,
"learning_rate": 5.7745502034727114e-05,
"loss": 0.0338,
"step": 14340
},
{
"grad_norm": 0.8366875648498535,
"learning_rate": 5.769104700369164e-05,
"loss": 0.033,
"step": 14350
},
{
"grad_norm": 0.5692504048347473,
"learning_rate": 5.7636582627308854e-05,
"loss": 0.025,
"step": 14360
},
{
"grad_norm": 0.5334132313728333,
"learning_rate": 5.7582108971758095e-05,
"loss": 0.0285,
"step": 14370
},
{
"grad_norm": 0.4853151738643646,
"learning_rate": 5.752762610322999e-05,
"loss": 0.0294,
"step": 14380
},
{
"grad_norm": 0.5374132394790649,
"learning_rate": 5.747313408792636e-05,
"loss": 0.0393,
"step": 14390
},
{
"grad_norm": 0.5966346859931946,
"learning_rate": 5.741863299206014e-05,
"loss": 0.0292,
"step": 14400
},
{
"grad_norm": 0.6312227845191956,
"learning_rate": 5.73641228818553e-05,
"loss": 0.0286,
"step": 14410
},
{
"grad_norm": 0.5129473805427551,
"learning_rate": 5.730960382354676e-05,
"loss": 0.0273,
"step": 14420
},
{
"grad_norm": 0.6088635921478271,
"learning_rate": 5.725507588338035e-05,
"loss": 0.0308,
"step": 14430
},
{
"grad_norm": 0.5573912262916565,
"learning_rate": 5.7200539127612604e-05,
"loss": 0.0264,
"step": 14440
},
{
"grad_norm": 0.6307456493377686,
"learning_rate": 5.714599362251087e-05,
"loss": 0.0319,
"step": 14450
},
{
"grad_norm": 0.6305881142616272,
"learning_rate": 5.709143943435307e-05,
"loss": 0.03,
"step": 14460
},
{
"grad_norm": 0.6245845556259155,
"learning_rate": 5.7036876629427646e-05,
"loss": 0.0348,
"step": 14470
},
{
"grad_norm": 0.6379913687705994,
"learning_rate": 5.698230527403361e-05,
"loss": 0.0281,
"step": 14480
},
{
"grad_norm": 0.5301382541656494,
"learning_rate": 5.6927725434480274e-05,
"loss": 0.0298,
"step": 14490
},
{
"grad_norm": 0.4236861765384674,
"learning_rate": 5.687313717708728e-05,
"loss": 0.0415,
"step": 14500
},
{
"grad_norm": 0.6002101302146912,
"learning_rate": 5.681854056818453e-05,
"loss": 0.0336,
"step": 14510
},
{
"grad_norm": 0.6859078407287598,
"learning_rate": 5.676393567411205e-05,
"loss": 0.0266,
"step": 14520
},
{
"grad_norm": 0.5183367729187012,
"learning_rate": 5.670932256121991e-05,
"loss": 0.049,
"step": 14530
},
{
"grad_norm": 0.5718507766723633,
"learning_rate": 5.6654701295868216e-05,
"loss": 0.03,
"step": 14540
},
{
"grad_norm": 0.5576866865158081,
"learning_rate": 5.660007194442697e-05,
"loss": 0.03,
"step": 14550
},
{
"grad_norm": 0.6183775663375854,
"learning_rate": 5.6545434573275946e-05,
"loss": 0.0347,
"step": 14560
},
{
"grad_norm": 0.5022698640823364,
"learning_rate": 5.649078924880472e-05,
"loss": 0.0239,
"step": 14570
},
{
"grad_norm": 0.6661533713340759,
"learning_rate": 5.643613603741252e-05,
"loss": 0.0351,
"step": 14580
},
{
"grad_norm": 0.5069316625595093,
"learning_rate": 5.6381475005508154e-05,
"loss": 0.0398,
"step": 14590
},
{
"grad_norm": 0.5089695453643799,
"learning_rate": 5.632680621950992e-05,
"loss": 0.0365,
"step": 14600
},
{
"grad_norm": 0.5341720581054688,
"learning_rate": 5.627212974584555e-05,
"loss": 0.0283,
"step": 14610
},
{
"grad_norm": 0.6401216983795166,
"learning_rate": 5.62174456509521e-05,
"loss": 0.0275,
"step": 14620
},
{
"grad_norm": 0.5458536148071289,
"learning_rate": 5.616275400127594e-05,
"loss": 0.0297,
"step": 14630
},
{
"grad_norm": 0.5753058195114136,
"learning_rate": 5.610805486327254e-05,
"loss": 0.026,
"step": 14640
},
{
"grad_norm": 0.5529103875160217,
"learning_rate": 5.6053348303406536e-05,
"loss": 0.0267,
"step": 14650
},
{
"grad_norm": 0.6850414276123047,
"learning_rate": 5.599863438815156e-05,
"loss": 0.0313,
"step": 14660
},
{
"grad_norm": 0.6585615277290344,
"learning_rate": 5.594391318399017e-05,
"loss": 0.0357,
"step": 14670
},
{
"grad_norm": 0.7079808115959167,
"learning_rate": 5.588918475741377e-05,
"loss": 0.0298,
"step": 14680
},
{
"grad_norm": 0.5729900002479553,
"learning_rate": 5.5834449174922586e-05,
"loss": 0.03,
"step": 14690
},
{
"grad_norm": 0.5670976638793945,
"learning_rate": 5.57797065030255e-05,
"loss": 0.0339,
"step": 14700
},
{
"grad_norm": 0.5758898258209229,
"learning_rate": 5.5724956808240016e-05,
"loss": 0.0341,
"step": 14710
},
{
"grad_norm": 0.7799325585365295,
"learning_rate": 5.567020015709219e-05,
"loss": 0.0336,
"step": 14720
},
{
"grad_norm": 0.6034257411956787,
"learning_rate": 5.561543661611649e-05,
"loss": 0.0291,
"step": 14730
},
{
"grad_norm": 0.5102670192718506,
"learning_rate": 5.556066625185583e-05,
"loss": 0.0355,
"step": 14740
},
{
"grad_norm": 0.6423075795173645,
"learning_rate": 5.550588913086131e-05,
"loss": 0.0324,
"step": 14750
},
{
"grad_norm": 0.6715309023857117,
"learning_rate": 5.545110531969234e-05,
"loss": 0.0302,
"step": 14760
},
{
"grad_norm": 0.668735921382904,
"learning_rate": 5.539631488491641e-05,
"loss": 0.0291,
"step": 14770
},
{
"grad_norm": 0.6001492142677307,
"learning_rate": 5.534151789310904e-05,
"loss": 0.0273,
"step": 14780
},
{
"grad_norm": 0.6185621023178101,
"learning_rate": 5.528671441085376e-05,
"loss": 0.0246,
"step": 14790
},
{
"grad_norm": 0.7435315251350403,
"learning_rate": 5.523190450474197e-05,
"loss": 0.03,
"step": 14800
},
{
"grad_norm": 0.4512796401977539,
"learning_rate": 5.5177088241372865e-05,
"loss": 0.0241,
"step": 14810
},
{
"grad_norm": 0.6752473711967468,
"learning_rate": 5.512226568735338e-05,
"loss": 0.0287,
"step": 14820
},
{
"grad_norm": 0.6391697525978088,
"learning_rate": 5.506743690929809e-05,
"loss": 0.0383,
"step": 14830
},
{
"grad_norm": 0.5191895365715027,
"learning_rate": 5.501260197382912e-05,
"loss": 0.0412,
"step": 14840
},
{
"grad_norm": 0.6540213823318481,
"learning_rate": 5.49577609475761e-05,
"loss": 0.0495,
"step": 14850
},
{
"grad_norm": 0.5827611088752747,
"learning_rate": 5.490291389717603e-05,
"loss": 0.0249,
"step": 14860
},
{
"grad_norm": 0.6659421920776367,
"learning_rate": 5.484806088927329e-05,
"loss": 0.0262,
"step": 14870
},
{
"grad_norm": 0.7835679054260254,
"learning_rate": 5.479320199051942e-05,
"loss": 0.0302,
"step": 14880
},
{
"grad_norm": 0.6480026841163635,
"learning_rate": 5.473833726757314e-05,
"loss": 0.0265,
"step": 14890
},
{
"grad_norm": 0.5586690306663513,
"learning_rate": 5.4683466787100304e-05,
"loss": 0.0272,
"step": 14900
},
{
"grad_norm": 0.4808818995952606,
"learning_rate": 5.462859061577369e-05,
"loss": 0.0285,
"step": 14910
},
{
"grad_norm": 0.4624554216861725,
"learning_rate": 5.4573708820273026e-05,
"loss": 0.027,
"step": 14920
},
{
"grad_norm": 0.7141984105110168,
"learning_rate": 5.451882146728488e-05,
"loss": 0.0272,
"step": 14930
},
{
"grad_norm": 0.5207654237747192,
"learning_rate": 5.446392862350255e-05,
"loss": 0.0342,
"step": 14940
},
{
"grad_norm": 0.574707567691803,
"learning_rate": 5.440903035562603e-05,
"loss": 0.0445,
"step": 14950
},
{
"grad_norm": 0.5535931587219238,
"learning_rate": 5.435412673036188e-05,
"loss": 0.0353,
"step": 14960
},
{
"grad_norm": 0.6730747818946838,
"learning_rate": 5.429921781442318e-05,
"loss": 0.0315,
"step": 14970
},
{
"grad_norm": 0.5502081513404846,
"learning_rate": 5.424430367452945e-05,
"loss": 0.0286,
"step": 14980
},
{
"grad_norm": 0.4824492633342743,
"learning_rate": 5.418938437740655e-05,
"loss": 0.0299,
"step": 14990
},
{
"grad_norm": 0.6315788626670837,
"learning_rate": 5.413445998978658e-05,
"loss": 0.0371,
"step": 15000
},
{
"grad_norm": 0.4424601197242737,
"learning_rate": 5.407953057840789e-05,
"loss": 0.023,
"step": 15010
},
{
"grad_norm": 0.5388809442520142,
"learning_rate": 5.4024596210014853e-05,
"loss": 0.0259,
"step": 15020
},
{
"grad_norm": 0.6487328410148621,
"learning_rate": 5.3969656951357935e-05,
"loss": 0.0243,
"step": 15030
},
{
"grad_norm": 0.5870911478996277,
"learning_rate": 5.3914712869193504e-05,
"loss": 0.0378,
"step": 15040
},
{
"grad_norm": 0.6013197898864746,
"learning_rate": 5.385976403028381e-05,
"loss": 0.0387,
"step": 15050
},
{
"grad_norm": 0.6203992366790771,
"learning_rate": 5.3804810501396875e-05,
"loss": 0.0384,
"step": 15060
},
{
"grad_norm": 0.4616369605064392,
"learning_rate": 5.374985234930639e-05,
"loss": 0.0285,
"step": 15070
},
{
"grad_norm": 0.5302886366844177,
"learning_rate": 5.3694889640791724e-05,
"loss": 0.0285,
"step": 15080
},
{
"grad_norm": 0.6262102127075195,
"learning_rate": 5.3639922442637736e-05,
"loss": 0.0296,
"step": 15090
},
{
"grad_norm": 0.5695949792861938,
"learning_rate": 5.358495082163476e-05,
"loss": 0.0311,
"step": 15100
},
{
"grad_norm": 0.5180025100708008,
"learning_rate": 5.35299748445785e-05,
"loss": 0.0272,
"step": 15110
},
{
"grad_norm": 0.5484063625335693,
"learning_rate": 5.3474994578269944e-05,
"loss": 0.0268,
"step": 15120
},
{
"grad_norm": 0.4623980224132538,
"learning_rate": 5.34200100895153e-05,
"loss": 0.0248,
"step": 15130
},
{
"grad_norm": 0.5041223764419556,
"learning_rate": 5.3365021445125916e-05,
"loss": 0.0199,
"step": 15140
},
{
"grad_norm": 0.5685341358184814,
"learning_rate": 5.331002871191817e-05,
"loss": 0.0296,
"step": 15150
},
{
"grad_norm": 0.6017773747444153,
"learning_rate": 5.3255031956713443e-05,
"loss": 0.0283,
"step": 15160
},
{
"grad_norm": 0.5817410945892334,
"learning_rate": 5.320003124633795e-05,
"loss": 0.035,
"step": 15170
},
{
"grad_norm": 0.5924304723739624,
"learning_rate": 5.314502664762274e-05,
"loss": 0.0248,
"step": 15180
},
{
"grad_norm": 0.6906862258911133,
"learning_rate": 5.3090018227403604e-05,
"loss": 0.0276,
"step": 15190
},
{
"grad_norm": 0.6142193675041199,
"learning_rate": 5.3035006052520955e-05,
"loss": 0.0329,
"step": 15200
},
{
"grad_norm": 0.5247135758399963,
"learning_rate": 5.297999018981976e-05,
"loss": 0.0253,
"step": 15210
},
{
"grad_norm": 0.7159953117370605,
"learning_rate": 5.2924970706149505e-05,
"loss": 0.0263,
"step": 15220
},
{
"grad_norm": 0.47811299562454224,
"learning_rate": 5.2869947668364014e-05,
"loss": 0.0231,
"step": 15230
},
{
"grad_norm": 0.57129967212677,
"learning_rate": 5.2814921143321504e-05,
"loss": 0.023,
"step": 15240
},
{
"grad_norm": 0.6653215289115906,
"learning_rate": 5.275989119788436e-05,
"loss": 0.039,
"step": 15250
},
{
"grad_norm": 0.6285099387168884,
"learning_rate": 5.270485789891919e-05,
"loss": 0.0304,
"step": 15260
},
{
"grad_norm": 0.504889726638794,
"learning_rate": 5.264982131329661e-05,
"loss": 0.0278,
"step": 15270
},
{
"grad_norm": 0.5001536011695862,
"learning_rate": 5.259478150789128e-05,
"loss": 0.0248,
"step": 15280
},
{
"grad_norm": 0.6013836860656738,
"learning_rate": 5.253973854958173e-05,
"loss": 0.0324,
"step": 15290
},
{
"grad_norm": 0.462444007396698,
"learning_rate": 5.2484692505250376e-05,
"loss": 0.0277,
"step": 15300
},
{
"grad_norm": 0.4197489619255066,
"learning_rate": 5.242964344178333e-05,
"loss": 0.0328,
"step": 15310
},
{
"grad_norm": 0.5642027854919434,
"learning_rate": 5.237459142607041e-05,
"loss": 0.0231,
"step": 15320
},
{
"grad_norm": 0.5360603928565979,
"learning_rate": 5.231953652500498e-05,
"loss": 0.0245,
"step": 15330
},
{
"grad_norm": 0.651279091835022,
"learning_rate": 5.2264478805483976e-05,
"loss": 0.0253,
"step": 15340
},
{
"grad_norm": 0.6155446767807007,
"learning_rate": 5.220941833440768e-05,
"loss": 0.0321,
"step": 15350
},
{
"grad_norm": 0.5636153817176819,
"learning_rate": 5.215435517867977e-05,
"loss": 0.0258,
"step": 15360
},
{
"grad_norm": 0.46555501222610474,
"learning_rate": 5.209928940520719e-05,
"loss": 0.028,
"step": 15370
},
{
"grad_norm": 0.647098958492279,
"learning_rate": 5.204422108090004e-05,
"loss": 0.0272,
"step": 15380
},
{
"grad_norm": 0.5118846297264099,
"learning_rate": 5.19891502726715e-05,
"loss": 0.0267,
"step": 15390
},
{
"grad_norm": 0.5053704977035522,
"learning_rate": 5.193407704743782e-05,
"loss": 0.0257,
"step": 15400
},
{
"grad_norm": 0.5166581869125366,
"learning_rate": 5.187900147211815e-05,
"loss": 0.0235,
"step": 15410
},
{
"grad_norm": 0.5306247472763062,
"learning_rate": 5.182392361363453e-05,
"loss": 0.0288,
"step": 15420
},
{
"grad_norm": 0.5358672142028809,
"learning_rate": 5.176884353891172e-05,
"loss": 0.0434,
"step": 15430
},
{
"grad_norm": 0.5390897989273071,
"learning_rate": 5.171376131487722e-05,
"loss": 0.0243,
"step": 15440
},
{
"grad_norm": 0.6452733874320984,
"learning_rate": 5.165867700846113e-05,
"loss": 0.0332,
"step": 15450
},
{
"grad_norm": 0.6073983907699585,
"learning_rate": 5.160359068659606e-05,
"loss": 0.0252,
"step": 15460
},
{
"grad_norm": 0.48896294832229614,
"learning_rate": 5.154850241621712e-05,
"loss": 0.0224,
"step": 15470
},
{
"grad_norm": 0.524262011051178,
"learning_rate": 5.149341226426172e-05,
"loss": 0.0307,
"step": 15480
},
{
"grad_norm": 0.5278933048248291,
"learning_rate": 5.14383202976696e-05,
"loss": 0.0296,
"step": 15490
},
{
"grad_norm": 0.5647032260894775,
"learning_rate": 5.1383226583382684e-05,
"loss": 0.0281,
"step": 15500
},
{
"grad_norm": 0.5894825458526611,
"learning_rate": 5.132813118834504e-05,
"loss": 0.0293,
"step": 15510
},
{
"grad_norm": 0.47027528285980225,
"learning_rate": 5.1273034179502777e-05,
"loss": 0.0204,
"step": 15520
},
{
"grad_norm": 0.5253623127937317,
"learning_rate": 5.1217935623803945e-05,
"loss": 0.0212,
"step": 15530
},
{
"grad_norm": 0.5147504806518555,
"learning_rate": 5.1162835588198476e-05,
"loss": 0.0236,
"step": 15540
},
{
"grad_norm": 0.46429577469825745,
"learning_rate": 5.110773413963813e-05,
"loss": 0.0218,
"step": 15550
},
{
"grad_norm": 0.5662097334861755,
"learning_rate": 5.105263134507636e-05,
"loss": 0.0295,
"step": 15560
},
{
"grad_norm": 0.5501077175140381,
"learning_rate": 5.099752727146824e-05,
"loss": 0.0289,
"step": 15570
},
{
"grad_norm": 0.5169740915298462,
"learning_rate": 5.094242198577042e-05,
"loss": 0.0349,
"step": 15580
},
{
"grad_norm": 0.4682682454586029,
"learning_rate": 5.088731555494102e-05,
"loss": 0.0226,
"step": 15590
},
{
"grad_norm": 0.5549875497817993,
"learning_rate": 5.0832208045939556e-05,
"loss": 0.0223,
"step": 15600
},
{
"grad_norm": 0.39353492856025696,
"learning_rate": 5.0777099525726844e-05,
"loss": 0.0291,
"step": 15610
},
{
"grad_norm": 0.5529068112373352,
"learning_rate": 5.072199006126494e-05,
"loss": 0.0225,
"step": 15620
},
{
"grad_norm": 0.5920098423957825,
"learning_rate": 5.0666879719517026e-05,
"loss": 0.0266,
"step": 15630
},
{
"grad_norm": 0.6010950207710266,
"learning_rate": 5.061176856744737e-05,
"loss": 0.0234,
"step": 15640
},
{
"grad_norm": 0.6867124438285828,
"learning_rate": 5.0556656672021205e-05,
"loss": 0.0277,
"step": 15650
},
{
"grad_norm": 0.5782086849212646,
"learning_rate": 5.050154410020472e-05,
"loss": 0.0288,
"step": 15660
},
{
"grad_norm": 0.5920884609222412,
"learning_rate": 5.044643091896485e-05,
"loss": 0.0265,
"step": 15670
},
{
"grad_norm": 0.5702138543128967,
"learning_rate": 5.039131719526932e-05,
"loss": 0.0221,
"step": 15680
},
{
"grad_norm": 0.5448335409164429,
"learning_rate": 5.03362029960865e-05,
"loss": 0.0253,
"step": 15690
},
{
"grad_norm": 0.49450764060020447,
"learning_rate": 5.028108838838533e-05,
"loss": 0.0273,
"step": 15700
},
{
"grad_norm": 0.5342084765434265,
"learning_rate": 5.0225973439135276e-05,
"loss": 0.0326,
"step": 15710
},
{
"grad_norm": 0.6638140082359314,
"learning_rate": 5.017085821530617e-05,
"loss": 0.0245,
"step": 15720
},
{
"grad_norm": 0.7417370080947876,
"learning_rate": 5.011574278386822e-05,
"loss": 0.0343,
"step": 15730
},
{
"grad_norm": 0.5454708933830261,
"learning_rate": 5.006062721179189e-05,
"loss": 0.0241,
"step": 15740
},
{
"grad_norm": 0.7073826193809509,
"learning_rate": 5.000551156604777e-05,
"loss": 0.0336,
"step": 15750
},
{
"grad_norm": 0.5999709963798523,
"learning_rate": 4.9950395913606594e-05,
"loss": 0.0259,
"step": 15760
},
{
"grad_norm": 0.6195284128189087,
"learning_rate": 4.989528032143903e-05,
"loss": 0.0305,
"step": 15770
},
{
"grad_norm": 0.48954397439956665,
"learning_rate": 4.984016485651578e-05,
"loss": 0.0248,
"step": 15780
},
{
"grad_norm": 0.6889064908027649,
"learning_rate": 4.9785049585807274e-05,
"loss": 0.0327,
"step": 15790
},
{
"grad_norm": 0.6040080189704895,
"learning_rate": 4.972993457628381e-05,
"loss": 0.0276,
"step": 15800
},
{
"grad_norm": 0.5439534187316895,
"learning_rate": 4.9674819894915306e-05,
"loss": 0.0263,
"step": 15810
},
{
"grad_norm": 0.5559313893318176,
"learning_rate": 4.9619705608671265e-05,
"loss": 0.0242,
"step": 15820
},
{
"grad_norm": 0.5409719347953796,
"learning_rate": 4.956459178452079e-05,
"loss": 0.0266,
"step": 15830
},
{
"grad_norm": 0.5941179990768433,
"learning_rate": 4.950947848943235e-05,
"loss": 0.0242,
"step": 15840
},
{
"grad_norm": 0.552200436592102,
"learning_rate": 4.9454365790373805e-05,
"loss": 0.027,
"step": 15850
},
{
"grad_norm": 0.48444920778274536,
"learning_rate": 4.939925375431226e-05,
"loss": 0.0261,
"step": 15860
},
{
"grad_norm": 0.6010574102401733,
"learning_rate": 4.9344142448214046e-05,
"loss": 0.0323,
"step": 15870
},
{
"grad_norm": 0.5929127335548401,
"learning_rate": 4.928903193904461e-05,
"loss": 0.0316,
"step": 15880
},
{
"grad_norm": 0.5855652689933777,
"learning_rate": 4.923392229376842e-05,
"loss": 0.0277,
"step": 15890
},
{
"grad_norm": 0.6001634001731873,
"learning_rate": 4.9178813579348865e-05,
"loss": 0.0215,
"step": 15900
},
{
"grad_norm": 0.5325583815574646,
"learning_rate": 4.9123705862748253e-05,
"loss": 0.0236,
"step": 15910
},
{
"grad_norm": 0.584514319896698,
"learning_rate": 4.9068599210927627e-05,
"loss": 0.0298,
"step": 15920
},
{
"grad_norm": 0.4741508662700653,
"learning_rate": 4.901349369084681e-05,
"loss": 0.0308,
"step": 15930
},
{
"grad_norm": 0.504326581954956,
"learning_rate": 4.8958389369464156e-05,
"loss": 0.0251,
"step": 15940
},
{
"grad_norm": 0.5967833399772644,
"learning_rate": 4.890328631373666e-05,
"loss": 0.0219,
"step": 15950
},
{
"grad_norm": 0.5021195411682129,
"learning_rate": 4.88481845906197e-05,
"loss": 0.0251,
"step": 15960
},
{
"grad_norm": 0.6424903273582458,
"learning_rate": 4.879308426706707e-05,
"loss": 0.0277,
"step": 15970
},
{
"grad_norm": 0.555564284324646,
"learning_rate": 4.873798541003084e-05,
"loss": 0.0214,
"step": 15980
},
{
"grad_norm": 0.5983558297157288,
"learning_rate": 4.868288808646136e-05,
"loss": 0.0294,
"step": 15990
},
{
"grad_norm": 0.5757767558097839,
"learning_rate": 4.862779236330705e-05,
"loss": 0.0248,
"step": 16000
},
{
"grad_norm": 0.5398096442222595,
"learning_rate": 4.8572698307514395e-05,
"loss": 0.0275,
"step": 16010
},
{
"grad_norm": 0.49710050225257874,
"learning_rate": 4.85176059860279e-05,
"loss": 0.024,
"step": 16020
},
{
"grad_norm": 0.5954288840293884,
"learning_rate": 4.846251546578989e-05,
"loss": 0.0262,
"step": 16030
},
{
"grad_norm": 0.5458298325538635,
"learning_rate": 4.840742681374058e-05,
"loss": 0.0259,
"step": 16040
},
{
"grad_norm": 0.5595499277114868,
"learning_rate": 4.835234009681787e-05,
"loss": 0.0237,
"step": 16050
},
{
"grad_norm": 0.46070730686187744,
"learning_rate": 4.8297255381957296e-05,
"loss": 0.0262,
"step": 16060
},
{
"grad_norm": 0.4513572156429291,
"learning_rate": 4.824217273609198e-05,
"loss": 0.0244,
"step": 16070
},
{
"grad_norm": 0.5764757394790649,
"learning_rate": 4.8187092226152555e-05,
"loss": 0.0254,
"step": 16080
},
{
"grad_norm": 0.556885838508606,
"learning_rate": 4.8132013919067016e-05,
"loss": 0.0234,
"step": 16090
},
{
"grad_norm": 0.547948956489563,
"learning_rate": 4.807693788176071e-05,
"loss": 0.025,
"step": 16100
},
{
"grad_norm": 0.5187769532203674,
"learning_rate": 4.8021864181156214e-05,
"loss": 0.0264,
"step": 16110
},
{
"grad_norm": 0.5158167481422424,
"learning_rate": 4.796679288417326e-05,
"loss": 0.034,
"step": 16120
},
{
"grad_norm": 0.6233332753181458,
"learning_rate": 4.791172405772866e-05,
"loss": 0.0296,
"step": 16130
},
{
"grad_norm": 0.6494264602661133,
"learning_rate": 4.785665776873625e-05,
"loss": 0.0235,
"step": 16140
},
{
"grad_norm": 0.4427860677242279,
"learning_rate": 4.7801594084106763e-05,
"loss": 0.03,
"step": 16150
},
{
"grad_norm": 0.4503864347934723,
"learning_rate": 4.774653307074775e-05,
"loss": 0.0258,
"step": 16160
},
{
"grad_norm": 0.5485597252845764,
"learning_rate": 4.769147479556355e-05,
"loss": 0.0227,
"step": 16170
},
{
"grad_norm": 0.5298015475273132,
"learning_rate": 4.763641932545515e-05,
"loss": 0.0186,
"step": 16180
},
{
"grad_norm": 0.633695125579834,
"learning_rate": 4.758136672732013e-05,
"loss": 0.0264,
"step": 16190
},
{
"grad_norm": 0.451376348733902,
"learning_rate": 4.752631706805261e-05,
"loss": 0.0228,
"step": 16200
},
{
"grad_norm": 0.5708262920379639,
"learning_rate": 4.7471270414543114e-05,
"loss": 0.0342,
"step": 16210
},
{
"grad_norm": 0.40477505326271057,
"learning_rate": 4.741622683367849e-05,
"loss": 0.0207,
"step": 16220
},
{
"grad_norm": 0.6360788345336914,
"learning_rate": 4.736118639234191e-05,
"loss": 0.035,
"step": 16230
},
{
"grad_norm": 0.5337653160095215,
"learning_rate": 4.730614915741267e-05,
"loss": 0.0293,
"step": 16240
},
{
"grad_norm": 0.5289518237113953,
"learning_rate": 4.7251115195766234e-05,
"loss": 0.026,
"step": 16250
},
{
"grad_norm": 0.5841615200042725,
"learning_rate": 4.719608457427404e-05,
"loss": 0.0247,
"step": 16260
},
{
"grad_norm": 0.5297431349754333,
"learning_rate": 4.714105735980347e-05,
"loss": 0.0233,
"step": 16270
},
{
"grad_norm": 0.5326185822486877,
"learning_rate": 4.70860336192178e-05,
"loss": 0.0272,
"step": 16280
},
{
"grad_norm": 0.49954742193222046,
"learning_rate": 4.7031013419376034e-05,
"loss": 0.0235,
"step": 16290
},
{
"grad_norm": 0.6447804570198059,
"learning_rate": 4.6975996827132926e-05,
"loss": 0.0203,
"step": 16300
},
{
"grad_norm": 0.5317043662071228,
"learning_rate": 4.692098390933883e-05,
"loss": 0.0301,
"step": 16310
},
{
"grad_norm": 0.6723054051399231,
"learning_rate": 4.6865974732839615e-05,
"loss": 0.021,
"step": 16320
},
{
"grad_norm": 0.520461916923523,
"learning_rate": 4.6810969364476615e-05,
"loss": 0.0334,
"step": 16330
},
{
"grad_norm": 0.6142176985740662,
"learning_rate": 4.675596787108653e-05,
"loss": 0.0226,
"step": 16340
},
{
"grad_norm": 0.4634625315666199,
"learning_rate": 4.670097031950139e-05,
"loss": 0.0192,
"step": 16350
},
{
"grad_norm": 0.549230694770813,
"learning_rate": 4.664597677654838e-05,
"loss": 0.0283,
"step": 16360
},
{
"grad_norm": 0.573807954788208,
"learning_rate": 4.659098730904986e-05,
"loss": 0.0318,
"step": 16370
},
{
"grad_norm": 0.4916703402996063,
"learning_rate": 4.6536001983823204e-05,
"loss": 0.0224,
"step": 16380
},
{
"grad_norm": 0.5298190116882324,
"learning_rate": 4.648102086768078e-05,
"loss": 0.0338,
"step": 16390
},
{
"grad_norm": 0.5629880428314209,
"learning_rate": 4.6426044027429786e-05,
"loss": 0.0258,
"step": 16400
},
{
"grad_norm": 0.454583078622818,
"learning_rate": 4.637107152987234e-05,
"loss": 0.0266,
"step": 16410
},
{
"grad_norm": 0.5239219665527344,
"learning_rate": 4.631610344180515e-05,
"loss": 0.0223,
"step": 16420
},
{
"grad_norm": 0.6724321246147156,
"learning_rate": 4.626113983001965e-05,
"loss": 0.0276,
"step": 16430
},
{
"grad_norm": 0.5417946577072144,
"learning_rate": 4.620618076130182e-05,
"loss": 0.0271,
"step": 16440
},
{
"grad_norm": 0.653780996799469,
"learning_rate": 4.6151226302432074e-05,
"loss": 0.0229,
"step": 16450
},
{
"grad_norm": 0.5942651629447937,
"learning_rate": 4.6096276520185304e-05,
"loss": 0.0262,
"step": 16460
},
{
"grad_norm": 0.5365739464759827,
"learning_rate": 4.604133148133066e-05,
"loss": 0.0239,
"step": 16470
},
{
"grad_norm": 0.6195734143257141,
"learning_rate": 4.598639125263155e-05,
"loss": 0.0305,
"step": 16480
},
{
"grad_norm": 0.49016281962394714,
"learning_rate": 4.593145590084553e-05,
"loss": 0.0296,
"step": 16490
},
{
"grad_norm": 0.6089203953742981,
"learning_rate": 4.58765254927242e-05,
"loss": 0.0224,
"step": 16500
},
{
"grad_norm": 0.5796332955360413,
"learning_rate": 4.582160009501323e-05,
"loss": 0.023,
"step": 16510
},
{
"grad_norm": 0.5386127233505249,
"learning_rate": 4.5766679774452143e-05,
"loss": 0.0284,
"step": 16520
},
{
"grad_norm": 0.6628113389015198,
"learning_rate": 4.571176459777431e-05,
"loss": 0.0282,
"step": 16530
},
{
"grad_norm": 0.6224722862243652,
"learning_rate": 4.5656854631706846e-05,
"loss": 0.0237,
"step": 16540
},
{
"grad_norm": 0.6392751336097717,
"learning_rate": 4.560194994297054e-05,
"loss": 0.0242,
"step": 16550
},
{
"grad_norm": 0.6053057312965393,
"learning_rate": 4.554705059827974e-05,
"loss": 0.0306,
"step": 16560
},
{
"grad_norm": 0.5170279145240784,
"learning_rate": 4.5492156664342365e-05,
"loss": 0.0293,
"step": 16570
},
{
"grad_norm": 0.5439031720161438,
"learning_rate": 4.5437268207859695e-05,
"loss": 0.0337,
"step": 16580
},
{
"grad_norm": 0.4208448529243469,
"learning_rate": 4.538238529552641e-05,
"loss": 0.0187,
"step": 16590
},
{
"grad_norm": 0.4092481732368469,
"learning_rate": 4.5327507994030394e-05,
"loss": 0.0195,
"step": 16600
},
{
"grad_norm": 0.48292505741119385,
"learning_rate": 4.527263637005273e-05,
"loss": 0.0182,
"step": 16610
},
{
"grad_norm": 0.565443754196167,
"learning_rate": 4.521777049026766e-05,
"loss": 0.022,
"step": 16620
},
{
"grad_norm": 0.529554545879364,
"learning_rate": 4.5162910421342386e-05,
"loss": 0.0204,
"step": 16630
},
{
"grad_norm": 0.6223469376564026,
"learning_rate": 4.510805622993706e-05,
"loss": 0.0305,
"step": 16640
},
{
"grad_norm": 0.6829470992088318,
"learning_rate": 4.505320798270467e-05,
"loss": 0.0264,
"step": 16650
},
{
"grad_norm": 0.571816623210907,
"learning_rate": 4.499836574629105e-05,
"loss": 0.024,
"step": 16660
},
{
"grad_norm": 0.5117406845092773,
"learning_rate": 4.494352958733465e-05,
"loss": 0.0402,
"step": 16670
},
{
"grad_norm": 0.566963791847229,
"learning_rate": 4.488869957246663e-05,
"loss": 0.0235,
"step": 16680
},
{
"grad_norm": 0.6606296300888062,
"learning_rate": 4.4833875768310575e-05,
"loss": 0.0262,
"step": 16690
},
{
"grad_norm": 0.6570777297019958,
"learning_rate": 4.47790582414826e-05,
"loss": 0.0241,
"step": 16700
},
{
"grad_norm": 0.7603775858879089,
"learning_rate": 4.472424705859114e-05,
"loss": 0.0277,
"step": 16710
},
{
"grad_norm": 0.6441038250923157,
"learning_rate": 4.466944228623701e-05,
"loss": 0.0393,
"step": 16720
},
{
"grad_norm": 0.6399186849594116,
"learning_rate": 4.461464399101312e-05,
"loss": 0.0321,
"step": 16730
},
{
"grad_norm": 0.42669677734375,
"learning_rate": 4.45598522395046e-05,
"loss": 0.0203,
"step": 16740
},
{
"grad_norm": 0.5620223879814148,
"learning_rate": 4.4505067098288577e-05,
"loss": 0.0301,
"step": 16750
},
{
"grad_norm": 0.4847097098827362,
"learning_rate": 4.445028863393417e-05,
"loss": 0.022,
"step": 16760
},
{
"grad_norm": 0.48778265714645386,
"learning_rate": 4.439551691300236e-05,
"loss": 0.0225,
"step": 16770
},
{
"grad_norm": 0.5547480583190918,
"learning_rate": 4.434075200204599e-05,
"loss": 0.0272,
"step": 16780
},
{
"grad_norm": 0.4842349588871002,
"learning_rate": 4.428599396760956e-05,
"loss": 0.0209,
"step": 16790
},
{
"grad_norm": 0.6635875701904297,
"learning_rate": 4.423124287622926e-05,
"loss": 0.0259,
"step": 16800
},
{
"grad_norm": 0.5265995264053345,
"learning_rate": 4.417649879443282e-05,
"loss": 0.0211,
"step": 16810
},
{
"grad_norm": 0.5802075266838074,
"learning_rate": 4.4121761788739444e-05,
"loss": 0.0225,
"step": 16820
},
{
"grad_norm": 0.4265805184841156,
"learning_rate": 4.4067031925659805e-05,
"loss": 0.0222,
"step": 16830
},
{
"grad_norm": 0.5119280815124512,
"learning_rate": 4.4012309271695816e-05,
"loss": 0.021,
"step": 16840
},
{
"grad_norm": 0.5057207942008972,
"learning_rate": 4.395759389334067e-05,
"loss": 0.0223,
"step": 16850
},
{
"grad_norm": 0.5928848385810852,
"learning_rate": 4.3902885857078685e-05,
"loss": 0.0283,
"step": 16860
},
{
"grad_norm": 0.5164433121681213,
"learning_rate": 4.384818522938531e-05,
"loss": 0.0201,
"step": 16870
},
{
"grad_norm": 0.6173688173294067,
"learning_rate": 4.379349207672695e-05,
"loss": 0.0221,
"step": 16880
},
{
"grad_norm": 0.44588935375213623,
"learning_rate": 4.373880646556098e-05,
"loss": 0.0226,
"step": 16890
},
{
"grad_norm": 0.6115625500679016,
"learning_rate": 4.3684128462335533e-05,
"loss": 0.0188,
"step": 16900
},
{
"grad_norm": 0.5122357606887817,
"learning_rate": 4.362945813348955e-05,
"loss": 0.0209,
"step": 16910
},
{
"grad_norm": 0.5762780904769897,
"learning_rate": 4.357479554545262e-05,
"loss": 0.0252,
"step": 16920
},
{
"grad_norm": 0.5047693848609924,
"learning_rate": 4.352014076464499e-05,
"loss": 0.0278,
"step": 16930
},
{
"grad_norm": 0.5419317483901978,
"learning_rate": 4.346549385747734e-05,
"loss": 0.0236,
"step": 16940
},
{
"grad_norm": 0.4581110179424286,
"learning_rate": 4.34108548903508e-05,
"loss": 0.0354,
"step": 16950
},
{
"grad_norm": 0.602109968662262,
"learning_rate": 4.335622392965689e-05,
"loss": 0.0253,
"step": 16960
},
{
"grad_norm": 0.5886765122413635,
"learning_rate": 4.330160104177738e-05,
"loss": 0.0284,
"step": 16970
},
{
"grad_norm": 0.4764963984489441,
"learning_rate": 4.32469862930842e-05,
"loss": 0.0267,
"step": 16980
},
{
"grad_norm": 0.6563339829444885,
"learning_rate": 4.319237974993946e-05,
"loss": 0.0348,
"step": 16990
},
{
"grad_norm": 0.6705303192138672,
"learning_rate": 4.3137781478695236e-05,
"loss": 0.0308,
"step": 17000
},
{
"grad_norm": 0.4346845746040344,
"learning_rate": 4.308319154569358e-05,
"loss": 0.0359,
"step": 17010
},
{
"grad_norm": 0.5366098284721375,
"learning_rate": 4.302861001726641e-05,
"loss": 0.0319,
"step": 17020
},
{
"grad_norm": 0.4016347825527191,
"learning_rate": 4.2974036959735415e-05,
"loss": 0.0224,
"step": 17030
},
{
"grad_norm": 0.5177744030952454,
"learning_rate": 4.2919472439412034e-05,
"loss": 0.0229,
"step": 17040
},
{
"grad_norm": 0.6793135404586792,
"learning_rate": 4.286491652259729e-05,
"loss": 0.0272,
"step": 17050
},
{
"grad_norm": 0.5031121969223022,
"learning_rate": 4.281036927558178e-05,
"loss": 0.0385,
"step": 17060
},
{
"grad_norm": 0.46996191143989563,
"learning_rate": 4.275583076464552e-05,
"loss": 0.0206,
"step": 17070
},
{
"grad_norm": 0.4901029169559479,
"learning_rate": 4.270130105605794e-05,
"loss": 0.0211,
"step": 17080
},
{
"grad_norm": 0.4968283772468567,
"learning_rate": 4.264678021607782e-05,
"loss": 0.0297,
"step": 17090
},
{
"grad_norm": 0.4733715057373047,
"learning_rate": 4.25922683109531e-05,
"loss": 0.0243,
"step": 17100
},
{
"grad_norm": 0.5152900218963623,
"learning_rate": 4.25377654069209e-05,
"loss": 0.0272,
"step": 17110
},
{
"grad_norm": 0.6853880286216736,
"learning_rate": 4.248327157020737e-05,
"loss": 0.0234,
"step": 17120
},
{
"grad_norm": 0.5927631855010986,
"learning_rate": 4.242878686702763e-05,
"loss": 0.023,
"step": 17130
},
{
"grad_norm": 0.44483640789985657,
"learning_rate": 4.23743113635858e-05,
"loss": 0.0194,
"step": 17140
},
{
"grad_norm": 0.5842398405075073,
"learning_rate": 4.2319845126074705e-05,
"loss": 0.0295,
"step": 17150
},
{
"grad_norm": 0.49468865990638733,
"learning_rate": 4.226538822067598e-05,
"loss": 0.0272,
"step": 17160
},
{
"grad_norm": 0.5152892470359802,
"learning_rate": 4.2210940713559896e-05,
"loss": 0.0267,
"step": 17170
},
{
"grad_norm": 0.5590308308601379,
"learning_rate": 4.21565026708853e-05,
"loss": 0.0257,
"step": 17180
},
{
"grad_norm": 0.5924208760261536,
"learning_rate": 4.2102074158799535e-05,
"loss": 0.0223,
"step": 17190
},
{
"grad_norm": 0.5641340017318726,
"learning_rate": 4.204765524343841e-05,
"loss": 0.0298,
"step": 17200
},
{
"grad_norm": 0.51411372423172,
"learning_rate": 4.1993245990926036e-05,
"loss": 0.0263,
"step": 17210
},
{
"grad_norm": 0.5247602462768555,
"learning_rate": 4.193884646737474e-05,
"loss": 0.0214,
"step": 17220
},
{
"grad_norm": 0.4974265396595001,
"learning_rate": 4.1884456738885124e-05,
"loss": 0.0231,
"step": 17230
},
{
"grad_norm": 0.6074849963188171,
"learning_rate": 4.1830076871545795e-05,
"loss": 0.0226,
"step": 17240
},
{
"grad_norm": 0.5272232890129089,
"learning_rate": 4.177570693143346e-05,
"loss": 0.0351,
"step": 17250
},
{
"grad_norm": 0.5512498021125793,
"learning_rate": 4.172134698461271e-05,
"loss": 0.0236,
"step": 17260
},
{
"grad_norm": 0.5167386531829834,
"learning_rate": 4.166699709713599e-05,
"loss": 0.0217,
"step": 17270
},
{
"grad_norm": 0.5466392040252686,
"learning_rate": 4.1612657335043555e-05,
"loss": 0.0206,
"step": 17280
},
{
"grad_norm": 0.4369988739490509,
"learning_rate": 4.155832776436331e-05,
"loss": 0.0262,
"step": 17290
},
{
"grad_norm": 0.7579670548439026,
"learning_rate": 4.1504008451110845e-05,
"loss": 0.0231,
"step": 17300
},
{
"grad_norm": 0.5160486102104187,
"learning_rate": 4.1449699461289225e-05,
"loss": 0.0235,
"step": 17310
},
{
"grad_norm": 0.6330518126487732,
"learning_rate": 4.139540086088901e-05,
"loss": 0.0195,
"step": 17320
},
{
"grad_norm": 0.45424696803092957,
"learning_rate": 4.1341112715888106e-05,
"loss": 0.0233,
"step": 17330
},
{
"grad_norm": 0.5968734622001648,
"learning_rate": 4.128683509225172e-05,
"loss": 0.0261,
"step": 17340
},
{
"grad_norm": 0.5784403681755066,
"learning_rate": 4.123256805593231e-05,
"loss": 0.0233,
"step": 17350
},
{
"grad_norm": 0.6575821042060852,
"learning_rate": 4.117831167286943e-05,
"loss": 0.0306,
"step": 17360
},
{
"grad_norm": 0.7276219129562378,
"learning_rate": 4.1124066008989685e-05,
"loss": 0.0256,
"step": 17370
},
{
"grad_norm": 0.5619909167289734,
"learning_rate": 4.106983113020669e-05,
"loss": 0.0238,
"step": 17380
},
{
"grad_norm": 0.47952577471733093,
"learning_rate": 4.101560710242094e-05,
"loss": 0.0213,
"step": 17390
},
{
"grad_norm": 0.5935850739479065,
"learning_rate": 4.096139399151971e-05,
"loss": 0.0263,
"step": 17400
},
{
"grad_norm": 0.5553542971611023,
"learning_rate": 4.090719186337709e-05,
"loss": 0.017,
"step": 17410
},
{
"grad_norm": 0.5359089970588684,
"learning_rate": 4.0853000783853746e-05,
"loss": 0.0256,
"step": 17420
},
{
"grad_norm": 0.5041059851646423,
"learning_rate": 4.0798820818796956e-05,
"loss": 0.024,
"step": 17430
},
{
"grad_norm": 0.6335511803627014,
"learning_rate": 4.074465203404048e-05,
"loss": 0.031,
"step": 17440
},
{
"grad_norm": 0.5870668888092041,
"learning_rate": 4.0690494495404495e-05,
"loss": 0.0283,
"step": 17450
},
{
"grad_norm": 0.5247173309326172,
"learning_rate": 4.0636348268695535e-05,
"loss": 0.0226,
"step": 17460
},
{
"grad_norm": 0.5615478157997131,
"learning_rate": 4.058221341970637e-05,
"loss": 0.0219,
"step": 17470
},
{
"grad_norm": 0.4425387382507324,
"learning_rate": 4.0528090014215945e-05,
"loss": 0.0235,
"step": 17480
},
{
"grad_norm": 0.48552796244621277,
"learning_rate": 4.047397811798929e-05,
"loss": 0.0199,
"step": 17490
},
{
"grad_norm": 0.48763856291770935,
"learning_rate": 4.041987779677745e-05,
"loss": 0.0193,
"step": 17500
},
{
"grad_norm": 0.5235400199890137,
"learning_rate": 4.036578911631745e-05,
"loss": 0.0276,
"step": 17510
},
{
"grad_norm": 0.5607326626777649,
"learning_rate": 4.031171214233211e-05,
"loss": 0.0225,
"step": 17520
},
{
"grad_norm": 0.5337385535240173,
"learning_rate": 4.025764694053008e-05,
"loss": 0.0219,
"step": 17530
},
{
"grad_norm": 0.4675140678882599,
"learning_rate": 4.020359357660566e-05,
"loss": 0.0316,
"step": 17540
},
{
"grad_norm": 0.5932027697563171,
"learning_rate": 4.014955211623875e-05,
"loss": 0.0212,
"step": 17550
},
{
"grad_norm": 0.5992843508720398,
"learning_rate": 4.0095522625094895e-05,
"loss": 0.039,
"step": 17560
},
{
"grad_norm": 0.6216781139373779,
"learning_rate": 4.004150516882497e-05,
"loss": 0.0293,
"step": 17570
},
{
"grad_norm": 0.6554511785507202,
"learning_rate": 3.99874998130653e-05,
"loss": 0.0257,
"step": 17580
},
{
"grad_norm": 0.663788914680481,
"learning_rate": 3.9933506623437455e-05,
"loss": 0.0268,
"step": 17590
},
{
"grad_norm": 0.48164787888526917,
"learning_rate": 3.9879525665548276e-05,
"loss": 0.0361,
"step": 17600
},
{
"grad_norm": 0.5452975034713745,
"learning_rate": 3.9825557004989704e-05,
"loss": 0.0282,
"step": 17610
},
{
"grad_norm": 0.44786766171455383,
"learning_rate": 3.977160070733877e-05,
"loss": 0.023,
"step": 17620
},
{
"grad_norm": 0.48572060465812683,
"learning_rate": 3.971765683815746e-05,
"loss": 0.0177,
"step": 17630
},
{
"grad_norm": 0.49858593940734863,
"learning_rate": 3.966372546299265e-05,
"loss": 0.0248,
"step": 17640
},
{
"grad_norm": 0.4942566156387329,
"learning_rate": 3.960980664737604e-05,
"loss": 0.0194,
"step": 17650
},
{
"grad_norm": 0.5670414566993713,
"learning_rate": 3.955590045682408e-05,
"loss": 0.0352,
"step": 17660
},
{
"grad_norm": 0.5952290892601013,
"learning_rate": 3.950200695683788e-05,
"loss": 0.0254,
"step": 17670
},
{
"grad_norm": 0.7364068627357483,
"learning_rate": 3.944812621290314e-05,
"loss": 0.0254,
"step": 17680
},
{
"grad_norm": 0.5314580798149109,
"learning_rate": 3.939425829049002e-05,
"loss": 0.0236,
"step": 17690
},
{
"grad_norm": 0.4048086106777191,
"learning_rate": 3.9340403255053135e-05,
"loss": 0.0289,
"step": 17700
},
{
"grad_norm": 0.5209341049194336,
"learning_rate": 3.9286561172031407e-05,
"loss": 0.0273,
"step": 17710
},
{
"grad_norm": 0.5154175758361816,
"learning_rate": 3.923273210684809e-05,
"loss": 0.0214,
"step": 17720
},
{
"grad_norm": 0.38446804881095886,
"learning_rate": 3.9178916124910555e-05,
"loss": 0.0196,
"step": 17730
},
{
"grad_norm": 0.5245950818061829,
"learning_rate": 3.9125113291610274e-05,
"loss": 0.0315,
"step": 17740
},
{
"grad_norm": 0.5565282106399536,
"learning_rate": 3.907132367232278e-05,
"loss": 0.0361,
"step": 17750
},
{
"grad_norm": 0.5533303022384644,
"learning_rate": 3.901754733240752e-05,
"loss": 0.0296,
"step": 17760
},
{
"grad_norm": 0.5845425724983215,
"learning_rate": 3.896378433720785e-05,
"loss": 0.0196,
"step": 17770
},
{
"grad_norm": 0.6214548349380493,
"learning_rate": 3.891003475205086e-05,
"loss": 0.0354,
"step": 17780
},
{
"grad_norm": 0.4492555260658264,
"learning_rate": 3.885629864224736e-05,
"loss": 0.015,
"step": 17790
},
{
"grad_norm": 0.4099266231060028,
"learning_rate": 3.880257607309178e-05,
"loss": 0.021,
"step": 17800
},
{
"grad_norm": 0.5713179111480713,
"learning_rate": 3.874886710986213e-05,
"loss": 0.0283,
"step": 17810
},
{
"grad_norm": 0.4541378319263458,
"learning_rate": 3.8695171817819834e-05,
"loss": 0.0225,
"step": 17820
},
{
"grad_norm": 0.5824601650238037,
"learning_rate": 3.8641490262209774e-05,
"loss": 0.0216,
"step": 17830
},
{
"grad_norm": 0.5253449082374573,
"learning_rate": 3.8587822508260084e-05,
"loss": 0.0225,
"step": 17840
},
{
"grad_norm": 0.5966276526451111,
"learning_rate": 3.853416862118214e-05,
"loss": 0.0236,
"step": 17850
},
{
"grad_norm": 0.6945421099662781,
"learning_rate": 3.848052866617049e-05,
"loss": 0.0239,
"step": 17860
},
{
"grad_norm": 0.6163303852081299,
"learning_rate": 3.84269027084027e-05,
"loss": 0.0236,
"step": 17870
},
{
"grad_norm": 0.5455150008201599,
"learning_rate": 3.83732908130394e-05,
"loss": 0.0212,
"step": 17880
},
{
"grad_norm": 0.5464550852775574,
"learning_rate": 3.831969304522409e-05,
"loss": 0.0232,
"step": 17890
},
{
"grad_norm": 0.5543922781944275,
"learning_rate": 3.826610947008312e-05,
"loss": 0.0192,
"step": 17900
},
{
"grad_norm": 0.5998354554176331,
"learning_rate": 3.821254015272559e-05,
"loss": 0.0278,
"step": 17910
},
{
"grad_norm": 0.657037079334259,
"learning_rate": 3.8158985158243214e-05,
"loss": 0.0221,
"step": 17920
},
{
"grad_norm": 0.43966445326805115,
"learning_rate": 3.810544455171044e-05,
"loss": 0.0243,
"step": 17930
},
{
"grad_norm": 0.6159999370574951,
"learning_rate": 3.8051918398184114e-05,
"loss": 0.0209,
"step": 17940
},
{
"grad_norm": 0.45470353960990906,
"learning_rate": 3.799840676270356e-05,
"loss": 0.0286,
"step": 17950
},
{
"grad_norm": 0.5227033495903015,
"learning_rate": 3.794490971029048e-05,
"loss": 0.0204,
"step": 17960
},
{
"grad_norm": 0.4781654179096222,
"learning_rate": 3.789142730594881e-05,
"loss": 0.0171,
"step": 17970
},
{
"grad_norm": 0.5481382608413696,
"learning_rate": 3.783795961466471e-05,
"loss": 0.0342,
"step": 17980
},
{
"grad_norm": 0.5448956489562988,
"learning_rate": 3.778450670140651e-05,
"loss": 0.0174,
"step": 17990
},
{
"grad_norm": 0.44784048199653625,
"learning_rate": 3.773106863112451e-05,
"loss": 0.0235,
"step": 18000
},
{
"grad_norm": 0.4857816696166992,
"learning_rate": 3.7677645468751e-05,
"loss": 0.0352,
"step": 18010
},
{
"grad_norm": 0.5251076221466064,
"learning_rate": 3.762423727920018e-05,
"loss": 0.0233,
"step": 18020
},
{
"grad_norm": 0.47576624155044556,
"learning_rate": 3.7570844127367995e-05,
"loss": 0.0295,
"step": 18030
},
{
"grad_norm": 0.5805069208145142,
"learning_rate": 3.751746607813221e-05,
"loss": 0.0239,
"step": 18040
},
{
"grad_norm": 0.5968995690345764,
"learning_rate": 3.746410319635217e-05,
"loss": 0.0256,
"step": 18050
},
{
"grad_norm": 0.48391106724739075,
"learning_rate": 3.7410755546868806e-05,
"loss": 0.0272,
"step": 18060
},
{
"grad_norm": 0.6305971145629883,
"learning_rate": 3.735742319450454e-05,
"loss": 0.0257,
"step": 18070
},
{
"grad_norm": 0.46259912848472595,
"learning_rate": 3.7304106204063184e-05,
"loss": 0.0202,
"step": 18080
},
{
"grad_norm": 0.6493246555328369,
"learning_rate": 3.725080464032996e-05,
"loss": 0.0239,
"step": 18090
},
{
"grad_norm": 0.5006504654884338,
"learning_rate": 3.719751856807125e-05,
"loss": 0.033,
"step": 18100
},
{
"grad_norm": 0.5965682864189148,
"learning_rate": 3.71442480520347e-05,
"loss": 0.0198,
"step": 18110
},
{
"grad_norm": 0.6105927228927612,
"learning_rate": 3.709099315694897e-05,
"loss": 0.0222,
"step": 18120
},
{
"grad_norm": 0.5933615565299988,
"learning_rate": 3.703775394752378e-05,
"loss": 0.0253,
"step": 18130
},
{
"grad_norm": 0.39776700735092163,
"learning_rate": 3.698453048844983e-05,
"loss": 0.0182,
"step": 18140
},
{
"grad_norm": 0.5643149018287659,
"learning_rate": 3.693132284439861e-05,
"loss": 0.0226,
"step": 18150
},
{
"grad_norm": 0.4948100745677948,
"learning_rate": 3.6878131080022414e-05,
"loss": 0.0213,
"step": 18160
},
{
"grad_norm": 0.42133617401123047,
"learning_rate": 3.682495525995429e-05,
"loss": 0.0217,
"step": 18170
},
{
"grad_norm": 0.6469938158988953,
"learning_rate": 3.6771795448807846e-05,
"loss": 0.0203,
"step": 18180
},
{
"grad_norm": 0.4272308945655823,
"learning_rate": 3.671865171117724e-05,
"loss": 0.0227,
"step": 18190
},
{
"grad_norm": 0.5460634231567383,
"learning_rate": 3.666552411163718e-05,
"loss": 0.0194,
"step": 18200
},
{
"grad_norm": 0.5482089519500732,
"learning_rate": 3.661241271474269e-05,
"loss": 0.0223,
"step": 18210
},
{
"grad_norm": 0.5195192098617554,
"learning_rate": 3.6559317585029116e-05,
"loss": 0.0185,
"step": 18220
},
{
"grad_norm": 0.6348391175270081,
"learning_rate": 3.650623878701204e-05,
"loss": 0.0205,
"step": 18230
},
{
"grad_norm": 0.4638287425041199,
"learning_rate": 3.645317638518721e-05,
"loss": 0.02,
"step": 18240
},
{
"grad_norm": 0.6452730298042297,
"learning_rate": 3.640013044403046e-05,
"loss": 0.0285,
"step": 18250
},
{
"grad_norm": 0.6006781458854675,
"learning_rate": 3.634710102799761e-05,
"loss": 0.0214,
"step": 18260
},
{
"grad_norm": 0.5962496399879456,
"learning_rate": 3.6294088201524394e-05,
"loss": 0.0249,
"step": 18270
},
{
"grad_norm": 0.45764797925949097,
"learning_rate": 3.62410920290264e-05,
"loss": 0.018,
"step": 18280
},
{
"grad_norm": 0.5583778023719788,
"learning_rate": 3.6188112574898954e-05,
"loss": 0.021,
"step": 18290
},
{
"grad_norm": 0.6189075112342834,
"learning_rate": 3.6135149903517115e-05,
"loss": 0.0203,
"step": 18300
},
{
"grad_norm": 0.41629523038864136,
"learning_rate": 3.608220407923552e-05,
"loss": 0.0217,
"step": 18310
},
{
"grad_norm": 0.4896578788757324,
"learning_rate": 3.602927516638833e-05,
"loss": 0.0247,
"step": 18320
},
{
"grad_norm": 0.45864930748939514,
"learning_rate": 3.5976363229289165e-05,
"loss": 0.0207,
"step": 18330
},
{
"grad_norm": 0.7242967486381531,
"learning_rate": 3.5923468332231e-05,
"loss": 0.0245,
"step": 18340
},
{
"grad_norm": 0.5299853682518005,
"learning_rate": 3.587059053948616e-05,
"loss": 0.0217,
"step": 18350
},
{
"grad_norm": 0.5516331791877747,
"learning_rate": 3.5817729915306134e-05,
"loss": 0.0216,
"step": 18360
},
{
"grad_norm": 0.4872133135795593,
"learning_rate": 3.5764886523921566e-05,
"loss": 0.0174,
"step": 18370
},
{
"grad_norm": 0.6239660382270813,
"learning_rate": 3.571206042954214e-05,
"loss": 0.0222,
"step": 18380
},
{
"grad_norm": 0.43020114302635193,
"learning_rate": 3.5659251696356566e-05,
"loss": 0.0183,
"step": 18390
},
{
"grad_norm": 0.4811807870864868,
"learning_rate": 3.56064603885324e-05,
"loss": 0.0278,
"step": 18400
},
{
"grad_norm": 0.593210756778717,
"learning_rate": 3.5553686570216115e-05,
"loss": 0.0244,
"step": 18410
},
{
"grad_norm": 0.4324674606323242,
"learning_rate": 3.550093030553284e-05,
"loss": 0.0257,
"step": 18420
},
{
"grad_norm": 0.5001581907272339,
"learning_rate": 3.544819165858642e-05,
"loss": 0.0216,
"step": 18430
},
{
"grad_norm": 0.49375879764556885,
"learning_rate": 3.539547069345926e-05,
"loss": 0.0194,
"step": 18440
},
{
"grad_norm": 0.4514032304286957,
"learning_rate": 3.5342767474212344e-05,
"loss": 0.0261,
"step": 18450
},
{
"grad_norm": 0.5542451739311218,
"learning_rate": 3.529008206488502e-05,
"loss": 0.0209,
"step": 18460
},
{
"grad_norm": 0.49974653124809265,
"learning_rate": 3.5237414529495055e-05,
"loss": 0.029,
"step": 18470
},
{
"grad_norm": 0.4737391769886017,
"learning_rate": 3.5184764932038454e-05,
"loss": 0.0207,
"step": 18480
},
{
"grad_norm": 0.5834820866584778,
"learning_rate": 3.513213333648945e-05,
"loss": 0.0199,
"step": 18490
},
{
"grad_norm": 0.4612288475036621,
"learning_rate": 3.5079519806800374e-05,
"loss": 0.0323,
"step": 18500
},
{
"grad_norm": 0.5284826159477234,
"learning_rate": 3.502692440690165e-05,
"loss": 0.0219,
"step": 18510
},
{
"grad_norm": 0.46197372674942017,
"learning_rate": 3.497434720070165e-05,
"loss": 0.0209,
"step": 18520
},
{
"grad_norm": 0.49360838532447815,
"learning_rate": 3.4921788252086616e-05,
"loss": 0.0235,
"step": 18530
},
{
"grad_norm": 0.3846152126789093,
"learning_rate": 3.486924762492065e-05,
"loss": 0.0185,
"step": 18540
},
{
"grad_norm": 0.5600433945655823,
"learning_rate": 3.4816725383045534e-05,
"loss": 0.0191,
"step": 18550
},
{
"grad_norm": 0.4408324062824249,
"learning_rate": 3.476422159028079e-05,
"loss": 0.0364,
"step": 18560
},
{
"grad_norm": 0.5423922538757324,
"learning_rate": 3.471173631042345e-05,
"loss": 0.0213,
"step": 18570
},
{
"grad_norm": 0.5474420785903931,
"learning_rate": 3.465926960724808e-05,
"loss": 0.0194,
"step": 18580
},
{
"grad_norm": 0.60235995054245,
"learning_rate": 3.460682154450666e-05,
"loss": 0.0218,
"step": 18590
},
{
"grad_norm": 0.47215646505355835,
"learning_rate": 3.4554392185928564e-05,
"loss": 0.0403,
"step": 18600
},
{
"grad_norm": 0.6440649032592773,
"learning_rate": 3.450198159522037e-05,
"loss": 0.0182,
"step": 18610
},
{
"grad_norm": 0.5458577275276184,
"learning_rate": 3.444958983606592e-05,
"loss": 0.0387,
"step": 18620
},
{
"grad_norm": 0.7085891962051392,
"learning_rate": 3.439721697212612e-05,
"loss": 0.0477,
"step": 18630
},
{
"grad_norm": 0.5197232961654663,
"learning_rate": 3.4344863067038954e-05,
"loss": 0.0257,
"step": 18640
},
{
"grad_norm": 0.5722362399101257,
"learning_rate": 3.429252818441935e-05,
"loss": 0.0235,
"step": 18650
},
{
"grad_norm": 0.5304712057113647,
"learning_rate": 3.4240212387859094e-05,
"loss": 0.0244,
"step": 18660
},
{
"grad_norm": 0.6161213517189026,
"learning_rate": 3.418791574092686e-05,
"loss": 0.0241,
"step": 18670
},
{
"grad_norm": 0.5692610144615173,
"learning_rate": 3.413563830716796e-05,
"loss": 0.024,
"step": 18680
},
{
"grad_norm": 0.49639931321144104,
"learning_rate": 3.408338015010445e-05,
"loss": 0.0211,
"step": 18690
},
{
"grad_norm": 0.40382716059684753,
"learning_rate": 3.40311413332349e-05,
"loss": 0.0183,
"step": 18700
},
{
"grad_norm": 0.4314069449901581,
"learning_rate": 3.3978921920034365e-05,
"loss": 0.0214,
"step": 18710
},
{
"grad_norm": 0.44402173161506653,
"learning_rate": 3.392672197395441e-05,
"loss": 0.0181,
"step": 18720
},
{
"grad_norm": 0.5538862943649292,
"learning_rate": 3.387454155842287e-05,
"loss": 0.0476,
"step": 18730
},
{
"grad_norm": 0.4348220229148865,
"learning_rate": 3.382238073684386e-05,
"loss": 0.0181,
"step": 18740
},
{
"grad_norm": 0.45472294092178345,
"learning_rate": 3.377023957259771e-05,
"loss": 0.0169,
"step": 18750
},
{
"grad_norm": 0.5167591571807861,
"learning_rate": 3.3718118129040835e-05,
"loss": 0.0206,
"step": 18760
},
{
"grad_norm": 0.4869343340396881,
"learning_rate": 3.3666016469505724e-05,
"loss": 0.0387,
"step": 18770
},
{
"grad_norm": 0.5092122554779053,
"learning_rate": 3.361393465730079e-05,
"loss": 0.0326,
"step": 18780
},
{
"grad_norm": 0.5253457427024841,
"learning_rate": 3.3561872755710366e-05,
"loss": 0.0246,
"step": 18790
},
{
"grad_norm": 0.591752827167511,
"learning_rate": 3.350983082799456e-05,
"loss": 0.0207,
"step": 18800
},
{
"grad_norm": 0.40793463587760925,
"learning_rate": 3.34578089373892e-05,
"loss": 0.0209,
"step": 18810
},
{
"grad_norm": 0.4727259576320648,
"learning_rate": 3.340580714710581e-05,
"loss": 0.0204,
"step": 18820
},
{
"grad_norm": 0.5018488168716431,
"learning_rate": 3.3353825520331466e-05,
"loss": 0.0209,
"step": 18830
},
{
"grad_norm": 0.4784148037433624,
"learning_rate": 3.330186412022876e-05,
"loss": 0.0219,
"step": 18840
},
{
"grad_norm": 0.5628935694694519,
"learning_rate": 3.324992300993568e-05,
"loss": 0.0212,
"step": 18850
},
{
"grad_norm": 0.5655996799468994,
"learning_rate": 3.319800225256556e-05,
"loss": 0.0281,
"step": 18860
},
{
"grad_norm": 0.6220544576644897,
"learning_rate": 3.314610191120702e-05,
"loss": 0.024,
"step": 18870
},
{
"grad_norm": 0.3660975396633148,
"learning_rate": 3.30942220489239e-05,
"loss": 0.0194,
"step": 18880
},
{
"grad_norm": 0.4676853120326996,
"learning_rate": 3.3042362728755086e-05,
"loss": 0.0187,
"step": 18890
},
{
"grad_norm": 0.500323474407196,
"learning_rate": 3.299052401371456e-05,
"loss": 0.034,
"step": 18900
},
{
"grad_norm": 0.5766549110412598,
"learning_rate": 3.293870596679125e-05,
"loss": 0.0193,
"step": 18910
},
{
"grad_norm": 0.537203311920166,
"learning_rate": 3.288690865094895e-05,
"loss": 0.0197,
"step": 18920
},
{
"grad_norm": 0.6068326234817505,
"learning_rate": 3.283513212912632e-05,
"loss": 0.0227,
"step": 18930
},
{
"grad_norm": 0.5406763553619385,
"learning_rate": 3.2783376464236684e-05,
"loss": 0.0189,
"step": 18940
},
{
"grad_norm": 0.48711177706718445,
"learning_rate": 3.273164171916806e-05,
"loss": 0.0168,
"step": 18950
},
{
"grad_norm": 0.5679087042808533,
"learning_rate": 3.267992795678306e-05,
"loss": 0.0255,
"step": 18960
},
{
"grad_norm": 0.5519567728042603,
"learning_rate": 3.2628235239918744e-05,
"loss": 0.0215,
"step": 18970
},
{
"grad_norm": 0.5147620439529419,
"learning_rate": 3.2576563631386695e-05,
"loss": 0.0224,
"step": 18980
},
{
"grad_norm": 0.5254803895950317,
"learning_rate": 3.252491319397275e-05,
"loss": 0.0197,
"step": 18990
},
{
"grad_norm": 0.5681838393211365,
"learning_rate": 3.247328399043706e-05,
"loss": 0.0246,
"step": 19000
},
{
"grad_norm": 0.5036089420318604,
"learning_rate": 3.242167608351399e-05,
"loss": 0.0265,
"step": 19010
},
{
"grad_norm": 0.43626856803894043,
"learning_rate": 3.2370089535911986e-05,
"loss": 0.0221,
"step": 19020
},
{
"grad_norm": 0.4659767150878906,
"learning_rate": 3.23185244103136e-05,
"loss": 0.0187,
"step": 19030
},
{
"grad_norm": 0.4092032313346863,
"learning_rate": 3.22669807693753e-05,
"loss": 0.0266,
"step": 19040
},
{
"grad_norm": 0.5183829665184021,
"learning_rate": 3.2215458675727495e-05,
"loss": 0.0234,
"step": 19050
},
{
"grad_norm": 0.5866113305091858,
"learning_rate": 3.216395819197438e-05,
"loss": 0.022,
"step": 19060
},
{
"grad_norm": 0.439840167760849,
"learning_rate": 3.211247938069387e-05,
"loss": 0.0287,
"step": 19070
},
{
"grad_norm": 0.5565215349197388,
"learning_rate": 3.206102230443759e-05,
"loss": 0.0192,
"step": 19080
},
{
"grad_norm": 0.4933163821697235,
"learning_rate": 3.2009587025730764e-05,
"loss": 0.0306,
"step": 19090
},
{
"grad_norm": 0.45949244499206543,
"learning_rate": 3.195817360707207e-05,
"loss": 0.0295,
"step": 19100
},
{
"grad_norm": 0.5221502184867859,
"learning_rate": 3.19067821109337e-05,
"loss": 0.0229,
"step": 19110
},
{
"grad_norm": 0.47195959091186523,
"learning_rate": 3.185541259976114e-05,
"loss": 0.019,
"step": 19120
},
{
"grad_norm": 0.5452802777290344,
"learning_rate": 3.180406513597316e-05,
"loss": 0.0326,
"step": 19130
},
{
"grad_norm": 0.4659096896648407,
"learning_rate": 3.1752739781961835e-05,
"loss": 0.0202,
"step": 19140
},
{
"grad_norm": 0.44940969347953796,
"learning_rate": 3.170143660009228e-05,
"loss": 0.0145,
"step": 19150
},
{
"grad_norm": 0.48898056149482727,
"learning_rate": 3.16501556527027e-05,
"loss": 0.0237,
"step": 19160
},
{
"grad_norm": 0.4681195914745331,
"learning_rate": 3.1598897002104265e-05,
"loss": 0.0189,
"step": 19170
},
{
"grad_norm": 0.483635276556015,
"learning_rate": 3.154766071058108e-05,
"loss": 0.0213,
"step": 19180
},
{
"grad_norm": 0.5602278709411621,
"learning_rate": 3.149644684039008e-05,
"loss": 0.0381,
"step": 19190
},
{
"grad_norm": 0.46904876828193665,
"learning_rate": 3.144525545376095e-05,
"loss": 0.0212,
"step": 19200
},
{
"grad_norm": 0.4116206169128418,
"learning_rate": 3.139408661289603e-05,
"loss": 0.0191,
"step": 19210
},
{
"grad_norm": 0.4551697075366974,
"learning_rate": 3.134294037997032e-05,
"loss": 0.0169,
"step": 19220
},
{
"grad_norm": 0.5584338903427124,
"learning_rate": 3.129181681713127e-05,
"loss": 0.0235,
"step": 19230
},
{
"grad_norm": 0.37518104910850525,
"learning_rate": 3.1240715986498855e-05,
"loss": 0.0216,
"step": 19240
},
{
"grad_norm": 0.511990487575531,
"learning_rate": 3.1189637950165394e-05,
"loss": 0.022,
"step": 19250
},
{
"grad_norm": 0.4982094168663025,
"learning_rate": 3.1138582770195544e-05,
"loss": 0.0177,
"step": 19260
},
{
"grad_norm": 0.46236148476600647,
"learning_rate": 3.108755050862615e-05,
"loss": 0.0161,
"step": 19270
},
{
"grad_norm": 0.5215582847595215,
"learning_rate": 3.1036541227466204e-05,
"loss": 0.0187,
"step": 19280
},
{
"grad_norm": 0.6022099256515503,
"learning_rate": 3.0985554988696786e-05,
"loss": 0.0277,
"step": 19290
},
{
"grad_norm": 0.46802854537963867,
"learning_rate": 3.093459185427102e-05,
"loss": 0.0205,
"step": 19300
},
{
"grad_norm": 0.47440963983535767,
"learning_rate": 3.0883651886113905e-05,
"loss": 0.0185,
"step": 19310
},
{
"grad_norm": 0.4941540062427521,
"learning_rate": 3.0832735146122294e-05,
"loss": 0.0278,
"step": 19320
},
{
"grad_norm": 0.4535170793533325,
"learning_rate": 3.078184169616485e-05,
"loss": 0.0347,
"step": 19330
},
{
"grad_norm": 0.5314710140228271,
"learning_rate": 3.073097159808187e-05,
"loss": 0.0389,
"step": 19340
},
{
"grad_norm": 0.44979357719421387,
"learning_rate": 3.068012491368537e-05,
"loss": 0.0275,
"step": 19350
},
{
"grad_norm": 0.4830862581729889,
"learning_rate": 3.062930170475885e-05,
"loss": 0.0173,
"step": 19360
},
{
"grad_norm": 0.5388796329498291,
"learning_rate": 3.057850203305729e-05,
"loss": 0.021,
"step": 19370
},
{
"grad_norm": 0.5919861197471619,
"learning_rate": 3.052772596030708e-05,
"loss": 0.0278,
"step": 19380
},
{
"grad_norm": 0.4525998532772064,
"learning_rate": 3.0476973548205943e-05,
"loss": 0.0225,
"step": 19390
},
{
"grad_norm": 0.47802862524986267,
"learning_rate": 3.0426244858422847e-05,
"loss": 0.0175,
"step": 19400
},
{
"grad_norm": 0.5216492414474487,
"learning_rate": 3.0375539952597943e-05,
"loss": 0.0174,
"step": 19410
},
{
"grad_norm": 0.539806067943573,
"learning_rate": 3.0324858892342468e-05,
"loss": 0.0181,
"step": 19420
},
{
"grad_norm": 0.5692950487136841,
"learning_rate": 3.0274201739238672e-05,
"loss": 0.0344,
"step": 19430
},
{
"grad_norm": 0.6144890189170837,
"learning_rate": 3.0223568554839786e-05,
"loss": 0.0184,
"step": 19440
},
{
"grad_norm": 0.5481205582618713,
"learning_rate": 3.0172959400669886e-05,
"loss": 0.0199,
"step": 19450
},
{
"grad_norm": 0.4743034839630127,
"learning_rate": 3.0122374338223902e-05,
"loss": 0.0211,
"step": 19460
},
{
"grad_norm": 0.44647786021232605,
"learning_rate": 3.0071813428967427e-05,
"loss": 0.0288,
"step": 19470
},
{
"grad_norm": 0.5574972629547119,
"learning_rate": 3.0021276734336746e-05,
"loss": 0.0172,
"step": 19480
},
{
"grad_norm": 0.49921804666519165,
"learning_rate": 2.997076431573871e-05,
"loss": 0.034,
"step": 19490
},
{
"grad_norm": 0.5282748341560364,
"learning_rate": 2.9920276234550636e-05,
"loss": 0.0194,
"step": 19500
},
{
"grad_norm": 0.596801221370697,
"learning_rate": 2.986981255212035e-05,
"loss": 0.0325,
"step": 19510
},
{
"grad_norm": 0.42456284165382385,
"learning_rate": 2.9819373329765977e-05,
"loss": 0.0159,
"step": 19520
},
{
"grad_norm": 0.5352315306663513,
"learning_rate": 2.9768958628775902e-05,
"loss": 0.02,
"step": 19530
},
{
"grad_norm": 0.5637463927268982,
"learning_rate": 2.9718568510408762e-05,
"loss": 0.0195,
"step": 19540
},
{
"grad_norm": 0.513916552066803,
"learning_rate": 2.9668203035893272e-05,
"loss": 0.0206,
"step": 19550
},
{
"grad_norm": 0.4865305721759796,
"learning_rate": 2.9617862266428287e-05,
"loss": 0.0318,
"step": 19560
},
{
"grad_norm": 0.5254250764846802,
"learning_rate": 2.9567546263182556e-05,
"loss": 0.025,
"step": 19570
},
{
"grad_norm": 0.4951058328151703,
"learning_rate": 2.951725508729476e-05,
"loss": 0.024,
"step": 19580
},
{
"grad_norm": 0.41028282046318054,
"learning_rate": 2.946698879987344e-05,
"loss": 0.0195,
"step": 19590
},
{
"grad_norm": 0.6185815930366516,
"learning_rate": 2.9416747461996853e-05,
"loss": 0.0234,
"step": 19600
},
{
"grad_norm": 0.543910026550293,
"learning_rate": 2.9366531134712973e-05,
"loss": 0.0214,
"step": 19610
},
{
"grad_norm": 0.5623244643211365,
"learning_rate": 2.9316339879039367e-05,
"loss": 0.0205,
"step": 19620
},
{
"grad_norm": 0.5943134427070618,
"learning_rate": 2.9266173755963167e-05,
"loss": 0.0252,
"step": 19630
},
{
"grad_norm": 0.4547593593597412,
"learning_rate": 2.9216032826440926e-05,
"loss": 0.0169,
"step": 19640
},
{
"grad_norm": 0.5999636650085449,
"learning_rate": 2.9165917151398592e-05,
"loss": 0.0219,
"step": 19650
},
{
"grad_norm": 0.4831770062446594,
"learning_rate": 2.9115826791731428e-05,
"loss": 0.0185,
"step": 19660
},
{
"grad_norm": 0.4928132891654968,
"learning_rate": 2.906576180830398e-05,
"loss": 0.0183,
"step": 19670
},
{
"grad_norm": 0.4807361960411072,
"learning_rate": 2.9015722261949917e-05,
"loss": 0.0207,
"step": 19680
},
{
"grad_norm": 0.45469969511032104,
"learning_rate": 2.8965708213471986e-05,
"loss": 0.0254,
"step": 19690
},
{
"grad_norm": 0.47452518343925476,
"learning_rate": 2.8915719723641975e-05,
"loss": 0.0211,
"step": 19700
},
{
"grad_norm": 0.45091211795806885,
"learning_rate": 2.8865756853200604e-05,
"loss": 0.0253,
"step": 19710
},
{
"grad_norm": 0.4465250074863434,
"learning_rate": 2.8815819662857503e-05,
"loss": 0.0227,
"step": 19720
},
{
"grad_norm": 0.5750977396965027,
"learning_rate": 2.876590821329105e-05,
"loss": 0.0203,
"step": 19730
},
{
"grad_norm": 0.457687109708786,
"learning_rate": 2.871602256514836e-05,
"loss": 0.0181,
"step": 19740
},
{
"grad_norm": 0.5974989533424377,
"learning_rate": 2.8666162779045203e-05,
"loss": 0.0181,
"step": 19750
},
{
"grad_norm": 0.47144079208374023,
"learning_rate": 2.8616328915565904e-05,
"loss": 0.0244,
"step": 19760
},
{
"grad_norm": 0.466925710439682,
"learning_rate": 2.856652103526334e-05,
"loss": 0.0201,
"step": 19770
},
{
"grad_norm": 0.5244625806808472,
"learning_rate": 2.8516739198658755e-05,
"loss": 0.0234,
"step": 19780
},
{
"grad_norm": 0.48802387714385986,
"learning_rate": 2.846698346624177e-05,
"loss": 0.0269,
"step": 19790
},
{
"grad_norm": 0.5073035359382629,
"learning_rate": 2.841725389847032e-05,
"loss": 0.019,
"step": 19800
},
{
"grad_norm": 0.6250134706497192,
"learning_rate": 2.8367550555770506e-05,
"loss": 0.0216,
"step": 19810
},
{
"grad_norm": 0.413722962141037,
"learning_rate": 2.831787349853655e-05,
"loss": 0.0189,
"step": 19820
},
{
"grad_norm": 0.4726727306842804,
"learning_rate": 2.8268222787130806e-05,
"loss": 0.0138,
"step": 19830
},
{
"grad_norm": 0.5057432651519775,
"learning_rate": 2.821859848188355e-05,
"loss": 0.0177,
"step": 19840
},
{
"grad_norm": 0.6094271540641785,
"learning_rate": 2.816900064309299e-05,
"loss": 0.0335,
"step": 19850
},
{
"grad_norm": 0.5934154987335205,
"learning_rate": 2.811942933102517e-05,
"loss": 0.0301,
"step": 19860
},
{
"grad_norm": 0.5700088143348694,
"learning_rate": 2.806988460591391e-05,
"loss": 0.0179,
"step": 19870
},
{
"grad_norm": 0.5206426978111267,
"learning_rate": 2.802036652796074e-05,
"loss": 0.0283,
"step": 19880
},
{
"grad_norm": 0.4623904526233673,
"learning_rate": 2.797087515733478e-05,
"loss": 0.0199,
"step": 19890
},
{
"grad_norm": 0.47188711166381836,
"learning_rate": 2.7921410554172723e-05,
"loss": 0.037,
"step": 19900
},
{
"grad_norm": 0.45474207401275635,
"learning_rate": 2.787197277857871e-05,
"loss": 0.0238,
"step": 19910
},
{
"grad_norm": 0.536911129951477,
"learning_rate": 2.782256189062429e-05,
"loss": 0.0205,
"step": 19920
},
{
"grad_norm": 0.4885838031768799,
"learning_rate": 2.777317795034839e-05,
"loss": 0.0225,
"step": 19930
},
{
"grad_norm": 0.46294090151786804,
"learning_rate": 2.7723821017757112e-05,
"loss": 0.0232,
"step": 19940
},
{
"grad_norm": 0.6234087944030762,
"learning_rate": 2.7674491152823822e-05,
"loss": 0.0227,
"step": 19950
},
{
"grad_norm": 0.5919604897499084,
"learning_rate": 2.7625188415488944e-05,
"loss": 0.0208,
"step": 19960
},
{
"grad_norm": 0.6507205963134766,
"learning_rate": 2.7575912865659924e-05,
"loss": 0.0241,
"step": 19970
},
{
"grad_norm": 0.5505840182304382,
"learning_rate": 2.7526664563211245e-05,
"loss": 0.0226,
"step": 19980
},
{
"grad_norm": 0.49218544363975525,
"learning_rate": 2.7477443567984224e-05,
"loss": 0.0185,
"step": 19990
},
{
"grad_norm": 0.6263723969459534,
"learning_rate": 2.7428249939787e-05,
"loss": 0.0232,
"step": 20000
},
{
"grad_norm": 0.5072548985481262,
"learning_rate": 2.7379083738394484e-05,
"loss": 0.0183,
"step": 20010
},
{
"grad_norm": 0.49480533599853516,
"learning_rate": 2.732994502354823e-05,
"loss": 0.0196,
"step": 20020
},
{
"grad_norm": 0.5257173180580139,
"learning_rate": 2.72808338549564e-05,
"loss": 0.0191,
"step": 20030
},
{
"grad_norm": 0.52174311876297,
"learning_rate": 2.723175029229374e-05,
"loss": 0.0321,
"step": 20040
},
{
"grad_norm": 0.507576048374176,
"learning_rate": 2.718269439520138e-05,
"loss": 0.0249,
"step": 20050
},
{
"grad_norm": 0.3947443664073944,
"learning_rate": 2.7133666223286858e-05,
"loss": 0.018,
"step": 20060
},
{
"grad_norm": 0.5754436254501343,
"learning_rate": 2.7084665836124006e-05,
"loss": 0.0303,
"step": 20070
},
{
"grad_norm": 0.5143978595733643,
"learning_rate": 2.703569329325296e-05,
"loss": 0.0188,
"step": 20080
},
{
"grad_norm": 0.5967579483985901,
"learning_rate": 2.698674865417994e-05,
"loss": 0.0284,
"step": 20090
},
{
"grad_norm": 0.4717925786972046,
"learning_rate": 2.6937831978377325e-05,
"loss": 0.0192,
"step": 20100
},
{
"grad_norm": 0.4029403030872345,
"learning_rate": 2.6888943325283485e-05,
"loss": 0.0208,
"step": 20110
},
{
"grad_norm": 0.3853141963481903,
"learning_rate": 2.6840082754302733e-05,
"loss": 0.0203,
"step": 20120
},
{
"grad_norm": 0.44168299436569214,
"learning_rate": 2.679125032480525e-05,
"loss": 0.0228,
"step": 20130
},
{
"grad_norm": 0.4995706379413605,
"learning_rate": 2.674244609612708e-05,
"loss": 0.0172,
"step": 20140
},
{
"grad_norm": 0.4393904209136963,
"learning_rate": 2.6693670127569958e-05,
"loss": 0.0204,
"step": 20150
},
{
"grad_norm": 0.38267719745635986,
"learning_rate": 2.6644922478401268e-05,
"loss": 0.0194,
"step": 20160
},
{
"grad_norm": 0.5428462624549866,
"learning_rate": 2.6596203207854004e-05,
"loss": 0.028,
"step": 20170
},
{
"grad_norm": 0.4508622884750366,
"learning_rate": 2.654751237512666e-05,
"loss": 0.0273,
"step": 20180
},
{
"grad_norm": 0.502054750919342,
"learning_rate": 2.649885003938323e-05,
"loss": 0.0177,
"step": 20190
},
{
"grad_norm": 0.5550505518913269,
"learning_rate": 2.6450216259753e-05,
"loss": 0.0178,
"step": 20200
},
{
"grad_norm": 0.38846614956855774,
"learning_rate": 2.640161109533063e-05,
"loss": 0.0196,
"step": 20210
},
{
"grad_norm": 0.4677925705909729,
"learning_rate": 2.6353034605175936e-05,
"loss": 0.0197,
"step": 20220
},
{
"grad_norm": 0.4839683771133423,
"learning_rate": 2.630448684831398e-05,
"loss": 0.0263,
"step": 20230
},
{
"grad_norm": 0.5065863132476807,
"learning_rate": 2.625596788373482e-05,
"loss": 0.0171,
"step": 20240
},
{
"grad_norm": 0.5300456285476685,
"learning_rate": 2.620747777039363e-05,
"loss": 0.0162,
"step": 20250
},
{
"grad_norm": 0.4515978991985321,
"learning_rate": 2.6159016567210424e-05,
"loss": 0.0156,
"step": 20260
},
{
"grad_norm": 0.42838728427886963,
"learning_rate": 2.6110584333070153e-05,
"loss": 0.0179,
"step": 20270
},
{
"grad_norm": 0.45876210927963257,
"learning_rate": 2.6062181126822537e-05,
"loss": 0.0159,
"step": 20280
},
{
"grad_norm": 0.42384597659111023,
"learning_rate": 2.601380700728203e-05,
"loss": 0.0231,
"step": 20290
},
{
"grad_norm": 0.3705308437347412,
"learning_rate": 2.596546203322777e-05,
"loss": 0.0176,
"step": 20300
},
{
"grad_norm": 0.4720335602760315,
"learning_rate": 2.5917146263403457e-05,
"loss": 0.0238,
"step": 20310
},
{
"grad_norm": 0.368024080991745,
"learning_rate": 2.5868859756517296e-05,
"loss": 0.0196,
"step": 20320
},
{
"grad_norm": 0.49936455488204956,
"learning_rate": 2.5820602571241947e-05,
"loss": 0.0165,
"step": 20330
},
{
"grad_norm": 0.4847131371498108,
"learning_rate": 2.577237476621442e-05,
"loss": 0.0218,
"step": 20340
},
{
"grad_norm": 0.5033816695213318,
"learning_rate": 2.572417640003609e-05,
"loss": 0.0236,
"step": 20350
},
{
"grad_norm": 0.43895280361175537,
"learning_rate": 2.5676007531272473e-05,
"loss": 0.0178,
"step": 20360
},
{
"grad_norm": 0.44195061922073364,
"learning_rate": 2.5627868218453332e-05,
"loss": 0.031,
"step": 20370
},
{
"grad_norm": 0.47605100274086,
"learning_rate": 2.5579758520072443e-05,
"loss": 0.0337,
"step": 20380
},
{
"grad_norm": 0.5062282681465149,
"learning_rate": 2.5531678494587615e-05,
"loss": 0.0184,
"step": 20390
},
{
"grad_norm": 0.3553285002708435,
"learning_rate": 2.5483628200420644e-05,
"loss": 0.0162,
"step": 20400
},
{
"grad_norm": 0.5397069454193115,
"learning_rate": 2.5435607695957154e-05,
"loss": 0.0231,
"step": 20410
},
{
"grad_norm": 0.44820672273635864,
"learning_rate": 2.5387617039546586e-05,
"loss": 0.019,
"step": 20420
},
{
"grad_norm": 0.5218031406402588,
"learning_rate": 2.5339656289502106e-05,
"loss": 0.0183,
"step": 20430
},
{
"grad_norm": 0.4480111002922058,
"learning_rate": 2.529172550410056e-05,
"loss": 0.0238,
"step": 20440
},
{
"grad_norm": 0.48136335611343384,
"learning_rate": 2.524382474158234e-05,
"loss": 0.0228,
"step": 20450
},
{
"grad_norm": 0.39881566166877747,
"learning_rate": 2.5195954060151433e-05,
"loss": 0.047,
"step": 20460
},
{
"grad_norm": 0.5013149976730347,
"learning_rate": 2.5148113517975212e-05,
"loss": 0.0341,
"step": 20470
},
{
"grad_norm": 0.5401383638381958,
"learning_rate": 2.5100303173184447e-05,
"loss": 0.0196,
"step": 20480
},
{
"grad_norm": 0.47416049242019653,
"learning_rate": 2.5052523083873196e-05,
"loss": 0.0172,
"step": 20490
},
{
"grad_norm": 0.4311991035938263,
"learning_rate": 2.500477330809881e-05,
"loss": 0.0187,
"step": 20500
},
{
"grad_norm": 0.6335447430610657,
"learning_rate": 2.4957053903881733e-05,
"loss": 0.0199,
"step": 20510
},
{
"grad_norm": 0.4306323230266571,
"learning_rate": 2.4909364929205576e-05,
"loss": 0.0201,
"step": 20520
},
{
"grad_norm": 0.39984506368637085,
"learning_rate": 2.4861706442016923e-05,
"loss": 0.0235,
"step": 20530
},
{
"grad_norm": 0.4253261089324951,
"learning_rate": 2.481407850022533e-05,
"loss": 0.0152,
"step": 20540
},
{
"grad_norm": 0.4918616712093353,
"learning_rate": 2.4766481161703213e-05,
"loss": 0.0217,
"step": 20550
},
{
"grad_norm": 0.5328853130340576,
"learning_rate": 2.4718914484285876e-05,
"loss": 0.0211,
"step": 20560
},
{
"grad_norm": 0.5082404017448425,
"learning_rate": 2.467137852577129e-05,
"loss": 0.0172,
"step": 20570
},
{
"grad_norm": 0.45362117886543274,
"learning_rate": 2.4623873343920123e-05,
"loss": 0.0172,
"step": 20580
},
{
"grad_norm": 0.4228593111038208,
"learning_rate": 2.4576398996455656e-05,
"loss": 0.0199,
"step": 20590
},
{
"grad_norm": 0.59024977684021,
"learning_rate": 2.452895554106368e-05,
"loss": 0.0231,
"step": 20600
},
{
"grad_norm": 0.5722029805183411,
"learning_rate": 2.4481543035392506e-05,
"loss": 0.0264,
"step": 20610
},
{
"grad_norm": 0.35365307331085205,
"learning_rate": 2.4434161537052774e-05,
"loss": 0.0184,
"step": 20620
},
{
"grad_norm": 0.44225165247917175,
"learning_rate": 2.4386811103617474e-05,
"loss": 0.0204,
"step": 20630
},
{
"grad_norm": 0.47855260968208313,
"learning_rate": 2.4339491792621833e-05,
"loss": 0.0231,
"step": 20640
},
{
"grad_norm": 0.45429977774620056,
"learning_rate": 2.4292203661563312e-05,
"loss": 0.0154,
"step": 20650
},
{
"grad_norm": 0.5466557741165161,
"learning_rate": 2.424494676790141e-05,
"loss": 0.0162,
"step": 20660
},
{
"grad_norm": 0.4670475125312805,
"learning_rate": 2.4197721169057745e-05,
"loss": 0.0159,
"step": 20670
},
{
"grad_norm": 0.4130552113056183,
"learning_rate": 2.4150526922415857e-05,
"loss": 0.0184,
"step": 20680
},
{
"grad_norm": 0.5850991606712341,
"learning_rate": 2.41033640853212e-05,
"loss": 0.0183,
"step": 20690
},
{
"grad_norm": 0.5106390118598938,
"learning_rate": 2.405623271508108e-05,
"loss": 0.0219,
"step": 20700
},
{
"grad_norm": 0.4806405007839203,
"learning_rate": 2.4009132868964522e-05,
"loss": 0.0164,
"step": 20710
},
{
"grad_norm": 0.5156205296516418,
"learning_rate": 2.3962064604202327e-05,
"loss": 0.0275,
"step": 20720
},
{
"grad_norm": 0.5104268193244934,
"learning_rate": 2.3915027977986858e-05,
"loss": 0.029,
"step": 20730
},
{
"grad_norm": 0.456497460603714,
"learning_rate": 2.386802304747205e-05,
"loss": 0.0151,
"step": 20740
},
{
"grad_norm": 0.5398134589195251,
"learning_rate": 2.382104986977332e-05,
"loss": 0.0249,
"step": 20750
},
{
"grad_norm": 0.5466711521148682,
"learning_rate": 2.377410850196749e-05,
"loss": 0.0221,
"step": 20760
},
{
"grad_norm": 0.5715405941009521,
"learning_rate": 2.3727199001092797e-05,
"loss": 0.0295,
"step": 20770
},
{
"grad_norm": 0.6230849027633667,
"learning_rate": 2.3680321424148677e-05,
"loss": 0.0249,
"step": 20780
},
{
"grad_norm": 0.4281168282032013,
"learning_rate": 2.3633475828095793e-05,
"loss": 0.0144,
"step": 20790
},
{
"grad_norm": 0.4277052879333496,
"learning_rate": 2.358666226985599e-05,
"loss": 0.0254,
"step": 20800
},
{
"grad_norm": 0.4484339654445648,
"learning_rate": 2.353988080631213e-05,
"loss": 0.0174,
"step": 20810
},
{
"grad_norm": 0.5863062143325806,
"learning_rate": 2.3493131494308142e-05,
"loss": 0.0283,
"step": 20820
},
{
"grad_norm": 0.4466593265533447,
"learning_rate": 2.344641439064881e-05,
"loss": 0.0155,
"step": 20830
},
{
"grad_norm": 0.5129503607749939,
"learning_rate": 2.339972955209984e-05,
"loss": 0.0264,
"step": 20840
},
{
"grad_norm": 0.422333300113678,
"learning_rate": 2.335307703538771e-05,
"loss": 0.0197,
"step": 20850
},
{
"grad_norm": 0.3431125283241272,
"learning_rate": 2.330645689719962e-05,
"loss": 0.024,
"step": 20860
},
{
"grad_norm": 0.48317667841911316,
"learning_rate": 2.3259869194183414e-05,
"loss": 0.0196,
"step": 20870
},
{
"grad_norm": 0.43366193771362305,
"learning_rate": 2.321331398294759e-05,
"loss": 0.0212,
"step": 20880
},
{
"grad_norm": 0.5344362258911133,
"learning_rate": 2.3166791320061094e-05,
"loss": 0.0225,
"step": 20890
},
{
"grad_norm": 0.4723087549209595,
"learning_rate": 2.312030126205335e-05,
"loss": 0.0153,
"step": 20900
},
{
"grad_norm": 0.45322805643081665,
"learning_rate": 2.3073843865414162e-05,
"loss": 0.0201,
"step": 20910
},
{
"grad_norm": 0.4788106679916382,
"learning_rate": 2.302741918659363e-05,
"loss": 0.0174,
"step": 20920
},
{
"grad_norm": 0.3352571129798889,
"learning_rate": 2.2981027282002154e-05,
"loss": 0.0168,
"step": 20930
},
{
"grad_norm": 0.3777853846549988,
"learning_rate": 2.2934668208010236e-05,
"loss": 0.0193,
"step": 20940
},
{
"grad_norm": 0.5887193083763123,
"learning_rate": 2.2888342020948555e-05,
"loss": 0.0206,
"step": 20950
},
{
"grad_norm": 0.412389874458313,
"learning_rate": 2.2842048777107783e-05,
"loss": 0.0198,
"step": 20960
},
{
"grad_norm": 0.5225442051887512,
"learning_rate": 2.2795788532738553e-05,
"loss": 0.0188,
"step": 20970
},
{
"grad_norm": 0.46139100193977356,
"learning_rate": 2.2749561344051468e-05,
"loss": 0.0169,
"step": 20980
},
{
"grad_norm": 0.4591045379638672,
"learning_rate": 2.2703367267216896e-05,
"loss": 0.0169,
"step": 20990
},
{
"grad_norm": 0.4579628109931946,
"learning_rate": 2.2657206358364997e-05,
"loss": 0.0215,
"step": 21000
},
{
"grad_norm": 0.4154827296733856,
"learning_rate": 2.261107867358563e-05,
"loss": 0.0257,
"step": 21010
},
{
"grad_norm": 0.5045063495635986,
"learning_rate": 2.2564984268928263e-05,
"loss": 0.0174,
"step": 21020
},
{
"grad_norm": 0.39838239550590515,
"learning_rate": 2.251892320040198e-05,
"loss": 0.0227,
"step": 21030
},
{
"grad_norm": 0.5831874012947083,
"learning_rate": 2.2472895523975312e-05,
"loss": 0.0219,
"step": 21040
},
{
"grad_norm": 0.5415370464324951,
"learning_rate": 2.2426901295576212e-05,
"loss": 0.0225,
"step": 21050
},
{
"grad_norm": 0.46655938029289246,
"learning_rate": 2.2380940571092013e-05,
"loss": 0.0299,
"step": 21060
},
{
"grad_norm": 0.4753180742263794,
"learning_rate": 2.2335013406369302e-05,
"loss": 0.0205,
"step": 21070
},
{
"grad_norm": 0.4852846562862396,
"learning_rate": 2.228911985721397e-05,
"loss": 0.0174,
"step": 21080
},
{
"grad_norm": 0.4415545165538788,
"learning_rate": 2.224325997939095e-05,
"loss": 0.0231,
"step": 21090
},
{
"grad_norm": 0.5533166527748108,
"learning_rate": 2.219743382862437e-05,
"loss": 0.028,
"step": 21100
},
{
"grad_norm": 0.5202670693397522,
"learning_rate": 2.2151641460597294e-05,
"loss": 0.0203,
"step": 21110
},
{
"grad_norm": 0.4873451590538025,
"learning_rate": 2.210588293095177e-05,
"loss": 0.018,
"step": 21120
},
{
"grad_norm": 0.47083383798599243,
"learning_rate": 2.2060158295288714e-05,
"loss": 0.0273,
"step": 21130
},
{
"grad_norm": 0.48801010847091675,
"learning_rate": 2.2014467609167905e-05,
"loss": 0.0202,
"step": 21140
},
{
"grad_norm": 0.4932781457901001,
"learning_rate": 2.1968810928107806e-05,
"loss": 0.0179,
"step": 21150
},
{
"grad_norm": 0.3951832056045532,
"learning_rate": 2.1923188307585606e-05,
"loss": 0.0182,
"step": 21160
},
{
"grad_norm": 0.4791716933250427,
"learning_rate": 2.187759980303708e-05,
"loss": 0.02,
"step": 21170
},
{
"grad_norm": 0.45618584752082825,
"learning_rate": 2.1832045469856545e-05,
"loss": 0.0188,
"step": 21180
},
{
"grad_norm": 0.5678226351737976,
"learning_rate": 2.1786525363396844e-05,
"loss": 0.0198,
"step": 21190
},
{
"grad_norm": 0.5570819973945618,
"learning_rate": 2.1741039538969183e-05,
"loss": 0.0185,
"step": 21200
},
{
"grad_norm": 0.5135821104049683,
"learning_rate": 2.169558805184313e-05,
"loss": 0.0179,
"step": 21210
},
{
"grad_norm": 0.4030076563358307,
"learning_rate": 2.165017095724651e-05,
"loss": 0.0192,
"step": 21220
},
{
"grad_norm": 0.41680169105529785,
"learning_rate": 2.1604788310365405e-05,
"loss": 0.0228,
"step": 21230
},
{
"grad_norm": 0.558972954750061,
"learning_rate": 2.155944016634398e-05,
"loss": 0.0168,
"step": 21240
},
{
"grad_norm": 0.4544019401073456,
"learning_rate": 2.1514126580284538e-05,
"loss": 0.0163,
"step": 21250
},
{
"grad_norm": 0.574776291847229,
"learning_rate": 2.1468847607247344e-05,
"loss": 0.0232,
"step": 21260
},
{
"grad_norm": 0.3728034794330597,
"learning_rate": 2.1423603302250623e-05,
"loss": 0.0236,
"step": 21270
},
{
"grad_norm": 0.45261743664741516,
"learning_rate": 2.1378393720270468e-05,
"loss": 0.0207,
"step": 21280
},
{
"grad_norm": 0.4471754729747772,
"learning_rate": 2.133321891624076e-05,
"loss": 0.0182,
"step": 21290
},
{
"grad_norm": 0.4105958342552185,
"learning_rate": 2.1288078945053193e-05,
"loss": 0.0181,
"step": 21300
},
{
"grad_norm": 0.5151408910751343,
"learning_rate": 2.1242973861557064e-05,
"loss": 0.021,
"step": 21310
},
{
"grad_norm": 0.3717570900917053,
"learning_rate": 2.1197903720559304e-05,
"loss": 0.0165,
"step": 21320
},
{
"grad_norm": 0.6129958629608154,
"learning_rate": 2.115286857682438e-05,
"loss": 0.0282,
"step": 21330
},
{
"grad_norm": 0.4468870759010315,
"learning_rate": 2.1107868485074228e-05,
"loss": 0.0169,
"step": 21340
},
{
"grad_norm": 0.48474934697151184,
"learning_rate": 2.1062903499988234e-05,
"loss": 0.0203,
"step": 21350
},
{
"grad_norm": 0.6202364563941956,
"learning_rate": 2.101797367620308e-05,
"loss": 0.0267,
"step": 21360
},
{
"grad_norm": 0.4930534064769745,
"learning_rate": 2.0973079068312713e-05,
"loss": 0.0184,
"step": 21370
},
{
"grad_norm": 0.4943956434726715,
"learning_rate": 2.0928219730868358e-05,
"loss": 0.0162,
"step": 21380
},
{
"grad_norm": 0.47068357467651367,
"learning_rate": 2.0883395718378302e-05,
"loss": 0.0159,
"step": 21390
},
{
"grad_norm": 0.5048300623893738,
"learning_rate": 2.083860708530798e-05,
"loss": 0.0247,
"step": 21400
},
{
"grad_norm": 0.4881231188774109,
"learning_rate": 2.0793853886079796e-05,
"loss": 0.0242,
"step": 21410
},
{
"grad_norm": 0.4068640470504761,
"learning_rate": 2.074913617507309e-05,
"loss": 0.0173,
"step": 21420
},
{
"grad_norm": 0.4410274624824524,
"learning_rate": 2.070445400662411e-05,
"loss": 0.0175,
"step": 21430
},
{
"grad_norm": 0.2760695517063141,
"learning_rate": 2.0659807435025907e-05,
"loss": 0.0165,
"step": 21440
},
{
"grad_norm": 0.45781752467155457,
"learning_rate": 2.061519651452825e-05,
"loss": 0.0304,
"step": 21450
},
{
"grad_norm": 0.44387370347976685,
"learning_rate": 2.0570621299337656e-05,
"loss": 0.0162,
"step": 21460
},
{
"grad_norm": 0.4815578758716583,
"learning_rate": 2.0526081843617183e-05,
"loss": 0.0261,
"step": 21470
},
{
"grad_norm": 0.5162806510925293,
"learning_rate": 2.0481578201486485e-05,
"loss": 0.0167,
"step": 21480
},
{
"grad_norm": 0.4850384593009949,
"learning_rate": 2.0437110427021677e-05,
"loss": 0.0189,
"step": 21490
},
{
"grad_norm": 0.5275582075119019,
"learning_rate": 2.0392678574255282e-05,
"loss": 0.0187,
"step": 21500
},
{
"grad_norm": 0.5352576375007629,
"learning_rate": 2.0348282697176217e-05,
"loss": 0.0222,
"step": 21510
},
{
"grad_norm": 0.4415205121040344,
"learning_rate": 2.030392284972964e-05,
"loss": 0.0167,
"step": 21520
},
{
"grad_norm": 0.44696685671806335,
"learning_rate": 2.0259599085816973e-05,
"loss": 0.0162,
"step": 21530
},
{
"grad_norm": 0.47797292470932007,
"learning_rate": 2.0215311459295755e-05,
"loss": 0.0176,
"step": 21540
},
{
"grad_norm": 0.499439537525177,
"learning_rate": 2.0171060023979606e-05,
"loss": 0.0213,
"step": 21550
},
{
"grad_norm": 0.5209983587265015,
"learning_rate": 2.012684483363823e-05,
"loss": 0.0214,
"step": 21560
},
{
"grad_norm": 0.41028502583503723,
"learning_rate": 2.0082665941997237e-05,
"loss": 0.0188,
"step": 21570
},
{
"grad_norm": 0.3497350811958313,
"learning_rate": 2.0038523402738146e-05,
"loss": 0.0163,
"step": 21580
},
{
"grad_norm": 0.5028554797172546,
"learning_rate": 1.99944172694983e-05,
"loss": 0.0283,
"step": 21590
},
{
"grad_norm": 0.3954516649246216,
"learning_rate": 1.99503475958708e-05,
"loss": 0.0182,
"step": 21600
},
{
"grad_norm": 0.46891850233078003,
"learning_rate": 1.9906314435404484e-05,
"loss": 0.0279,
"step": 21610
},
{
"grad_norm": 0.4374026358127594,
"learning_rate": 1.986231784160378e-05,
"loss": 0.0171,
"step": 21620
},
{
"grad_norm": 0.4695480167865753,
"learning_rate": 1.9818357867928693e-05,
"loss": 0.0219,
"step": 21630
},
{
"grad_norm": 0.48793113231658936,
"learning_rate": 1.977443456779474e-05,
"loss": 0.0166,
"step": 21640
},
{
"grad_norm": 0.4849835932254791,
"learning_rate": 1.9730547994572858e-05,
"loss": 0.0158,
"step": 21650
},
{
"grad_norm": 0.4435117542743683,
"learning_rate": 1.9686698201589393e-05,
"loss": 0.0201,
"step": 21660
},
{
"grad_norm": 0.3900175988674164,
"learning_rate": 1.9642885242125964e-05,
"loss": 0.0174,
"step": 21670
},
{
"grad_norm": 0.4640091359615326,
"learning_rate": 1.9599109169419466e-05,
"loss": 0.0158,
"step": 21680
},
{
"grad_norm": 0.4023323059082031,
"learning_rate": 1.9555370036661948e-05,
"loss": 0.0198,
"step": 21690
},
{
"grad_norm": 0.461020827293396,
"learning_rate": 1.9511667897000574e-05,
"loss": 0.0175,
"step": 21700
},
{
"grad_norm": 0.5135725736618042,
"learning_rate": 1.946800280353755e-05,
"loss": 0.0165,
"step": 21710
},
{
"grad_norm": 0.4760790467262268,
"learning_rate": 1.9424374809330114e-05,
"loss": 0.0239,
"step": 21720
},
{
"grad_norm": 0.5040029287338257,
"learning_rate": 1.9380783967390382e-05,
"loss": 0.0152,
"step": 21730
},
{
"grad_norm": 0.4679045081138611,
"learning_rate": 1.933723033068533e-05,
"loss": 0.0185,
"step": 21740
},
{
"grad_norm": 0.5323666334152222,
"learning_rate": 1.929371395213674e-05,
"loss": 0.0162,
"step": 21750
},
{
"grad_norm": 0.4035722315311432,
"learning_rate": 1.9250234884621092e-05,
"loss": 0.0297,
"step": 21760
},
{
"grad_norm": 0.45118260383605957,
"learning_rate": 1.920679318096959e-05,
"loss": 0.0195,
"step": 21770
},
{
"grad_norm": 0.41953006386756897,
"learning_rate": 1.9163388893967982e-05,
"loss": 0.0322,
"step": 21780
},
{
"grad_norm": 0.49583613872528076,
"learning_rate": 1.9120022076356575e-05,
"loss": 0.0163,
"step": 21790
},
{
"grad_norm": 0.46589797735214233,
"learning_rate": 1.9076692780830114e-05,
"loss": 0.0209,
"step": 21800
},
{
"grad_norm": 0.394694447517395,
"learning_rate": 1.903340106003782e-05,
"loss": 0.0141,
"step": 21810
},
{
"grad_norm": 0.3844307065010071,
"learning_rate": 1.8990146966583183e-05,
"loss": 0.0166,
"step": 21820
},
{
"grad_norm": 0.4677315652370453,
"learning_rate": 1.8946930553024035e-05,
"loss": 0.0245,
"step": 21830
},
{
"grad_norm": 0.38595500588417053,
"learning_rate": 1.8903751871872378e-05,
"loss": 0.0183,
"step": 21840
},
{
"grad_norm": 0.41674989461898804,
"learning_rate": 1.8860610975594382e-05,
"loss": 0.0163,
"step": 21850
},
{
"grad_norm": 0.4303787648677826,
"learning_rate": 1.8817507916610307e-05,
"loss": 0.0252,
"step": 21860
},
{
"grad_norm": 0.42328909039497375,
"learning_rate": 1.8774442747294407e-05,
"loss": 0.0189,
"step": 21870
},
{
"grad_norm": 0.41215699911117554,
"learning_rate": 1.8731415519974965e-05,
"loss": 0.0144,
"step": 21880
},
{
"grad_norm": 0.4369277358055115,
"learning_rate": 1.86884262869341e-05,
"loss": 0.0164,
"step": 21890
},
{
"grad_norm": 0.45102688670158386,
"learning_rate": 1.8645475100407788e-05,
"loss": 0.0153,
"step": 21900
},
{
"grad_norm": 0.45941609144210815,
"learning_rate": 1.8602562012585768e-05,
"loss": 0.0212,
"step": 21910
},
{
"grad_norm": 0.4472567141056061,
"learning_rate": 1.8559687075611466e-05,
"loss": 0.0162,
"step": 21920
},
{
"grad_norm": 0.3982315957546234,
"learning_rate": 1.8516850341582017e-05,
"loss": 0.0148,
"step": 21930
},
{
"grad_norm": 0.6508101224899292,
"learning_rate": 1.847405186254807e-05,
"loss": 0.0178,
"step": 21940
},
{
"grad_norm": 0.40403786301612854,
"learning_rate": 1.843129169051379e-05,
"loss": 0.0123,
"step": 21950
},
{
"grad_norm": 0.5047371983528137,
"learning_rate": 1.8388569877436863e-05,
"loss": 0.0183,
"step": 21960
},
{
"grad_norm": 0.4385465383529663,
"learning_rate": 1.834588647522828e-05,
"loss": 0.0173,
"step": 21970
},
{
"grad_norm": 0.43770596385002136,
"learning_rate": 1.8303241535752435e-05,
"loss": 0.0157,
"step": 21980
},
{
"grad_norm": 0.4961800277233124,
"learning_rate": 1.8260635110826934e-05,
"loss": 0.0161,
"step": 21990
},
{
"grad_norm": 0.5195525884628296,
"learning_rate": 1.8218067252222597e-05,
"loss": 0.0164,
"step": 22000
},
{
"grad_norm": 0.48302003741264343,
"learning_rate": 1.817553801166339e-05,
"loss": 0.0153,
"step": 22010
},
{
"grad_norm": 0.4569348394870758,
"learning_rate": 1.8133047440826333e-05,
"loss": 0.0226,
"step": 22020
},
{
"grad_norm": 0.4503457546234131,
"learning_rate": 1.809059559134151e-05,
"loss": 0.0209,
"step": 22030
},
{
"grad_norm": 0.4744955003261566,
"learning_rate": 1.80481825147919e-05,
"loss": 0.0212,
"step": 22040
},
{
"grad_norm": 0.323397696018219,
"learning_rate": 1.8005808262713396e-05,
"loss": 0.0184,
"step": 22050
},
{
"grad_norm": 0.4371536672115326,
"learning_rate": 1.7963472886594714e-05,
"loss": 0.0157,
"step": 22060
},
{
"grad_norm": 0.40253961086273193,
"learning_rate": 1.7921176437877302e-05,
"loss": 0.0175,
"step": 22070
},
{
"grad_norm": 0.4644511640071869,
"learning_rate": 1.7878918967955366e-05,
"loss": 0.0168,
"step": 22080
},
{
"grad_norm": 0.4838891625404358,
"learning_rate": 1.7836700528175693e-05,
"loss": 0.0232,
"step": 22090
},
{
"grad_norm": 0.4842541515827179,
"learning_rate": 1.779452116983769e-05,
"loss": 0.018,
"step": 22100
},
{
"grad_norm": 0.4711235463619232,
"learning_rate": 1.775238094419325e-05,
"loss": 0.0251,
"step": 22110
},
{
"grad_norm": 0.41985833644866943,
"learning_rate": 1.7710279902446708e-05,
"loss": 0.0334,
"step": 22120
},
{
"grad_norm": 0.5116223692893982,
"learning_rate": 1.7668218095754795e-05,
"loss": 0.0219,
"step": 22130
},
{
"grad_norm": 0.334643691778183,
"learning_rate": 1.7626195575226594e-05,
"loss": 0.0194,
"step": 22140
},
{
"grad_norm": 0.36953485012054443,
"learning_rate": 1.758421239192343e-05,
"loss": 0.0171,
"step": 22150
},
{
"grad_norm": 0.2993636429309845,
"learning_rate": 1.7542268596858814e-05,
"loss": 0.0162,
"step": 22160
},
{
"grad_norm": 0.307218462228775,
"learning_rate": 1.750036424099841e-05,
"loss": 0.0133,
"step": 22170
},
{
"grad_norm": 0.4816868305206299,
"learning_rate": 1.7458499375259956e-05,
"loss": 0.0189,
"step": 22180
},
{
"grad_norm": 0.4074622690677643,
"learning_rate": 1.7416674050513245e-05,
"loss": 0.0195,
"step": 22190
},
{
"grad_norm": 0.3723468780517578,
"learning_rate": 1.737488831757997e-05,
"loss": 0.0155,
"step": 22200
},
{
"grad_norm": 0.548004686832428,
"learning_rate": 1.733314222723373e-05,
"loss": 0.0246,
"step": 22210
},
{
"grad_norm": 0.39833149313926697,
"learning_rate": 1.7291435830199954e-05,
"loss": 0.0168,
"step": 22220
},
{
"grad_norm": 0.3179417550563812,
"learning_rate": 1.7249769177155878e-05,
"loss": 0.0137,
"step": 22230
},
{
"grad_norm": 0.5118088722229004,
"learning_rate": 1.720814231873038e-05,
"loss": 0.0226,
"step": 22240
},
{
"grad_norm": 0.4195244312286377,
"learning_rate": 1.716655530550405e-05,
"loss": 0.0132,
"step": 22250
},
{
"grad_norm": 0.4862250089645386,
"learning_rate": 1.7125008188009017e-05,
"loss": 0.0211,
"step": 22260
},
{
"grad_norm": 0.42236393690109253,
"learning_rate": 1.708350101672894e-05,
"loss": 0.0158,
"step": 22270
},
{
"grad_norm": 0.4264114201068878,
"learning_rate": 1.7042033842098957e-05,
"loss": 0.0213,
"step": 22280
},
{
"grad_norm": 0.4667860269546509,
"learning_rate": 1.7000606714505567e-05,
"loss": 0.0173,
"step": 22290
},
{
"grad_norm": 0.5642156600952148,
"learning_rate": 1.695921968428668e-05,
"loss": 0.0175,
"step": 22300
},
{
"grad_norm": 0.427795946598053,
"learning_rate": 1.6917872801731416e-05,
"loss": 0.0198,
"step": 22310
},
{
"grad_norm": 0.42504897713661194,
"learning_rate": 1.687656611708014e-05,
"loss": 0.0218,
"step": 22320
},
{
"grad_norm": 0.3730849325656891,
"learning_rate": 1.6835299680524368e-05,
"loss": 0.0165,
"step": 22330
},
{
"grad_norm": 0.34781599044799805,
"learning_rate": 1.679407354220669e-05,
"loss": 0.0237,
"step": 22340
},
{
"grad_norm": 0.4106484055519104,
"learning_rate": 1.6752887752220793e-05,
"loss": 0.0194,
"step": 22350
},
{
"grad_norm": 0.35251057147979736,
"learning_rate": 1.6711742360611277e-05,
"loss": 0.0165,
"step": 22360
},
{
"grad_norm": 0.4298963248729706,
"learning_rate": 1.6670637417373652e-05,
"loss": 0.0286,
"step": 22370
},
{
"grad_norm": 0.42003342509269714,
"learning_rate": 1.6629572972454333e-05,
"loss": 0.0179,
"step": 22380
},
{
"grad_norm": 0.4516446590423584,
"learning_rate": 1.6588549075750464e-05,
"loss": 0.0146,
"step": 22390
},
{
"grad_norm": 0.48380357027053833,
"learning_rate": 1.6547565777109976e-05,
"loss": 0.0219,
"step": 22400
},
{
"grad_norm": 0.4990187883377075,
"learning_rate": 1.650662312633143e-05,
"loss": 0.0292,
"step": 22410
},
{
"grad_norm": 0.5426790118217468,
"learning_rate": 1.6465721173164002e-05,
"loss": 0.0213,
"step": 22420
},
{
"grad_norm": 0.5388015508651733,
"learning_rate": 1.6424859967307428e-05,
"loss": 0.0161,
"step": 22430
},
{
"grad_norm": 0.5149369239807129,
"learning_rate": 1.6384039558411903e-05,
"loss": 0.0176,
"step": 22440
},
{
"grad_norm": 0.5007680654525757,
"learning_rate": 1.634325999607811e-05,
"loss": 0.0209,
"step": 22450
},
{
"grad_norm": 0.4838658571243286,
"learning_rate": 1.6302521329857045e-05,
"loss": 0.0178,
"step": 22460
},
{
"grad_norm": 0.42054763436317444,
"learning_rate": 1.6261823609250027e-05,
"loss": 0.0177,
"step": 22470
},
{
"grad_norm": 0.4496299624443054,
"learning_rate": 1.6221166883708627e-05,
"loss": 0.0177,
"step": 22480
},
{
"grad_norm": 0.41475439071655273,
"learning_rate": 1.6180551202634603e-05,
"loss": 0.0175,
"step": 22490
},
{
"grad_norm": 0.5037750005722046,
"learning_rate": 1.613997661537981e-05,
"loss": 0.0156,
"step": 22500
},
{
"grad_norm": 0.578210175037384,
"learning_rate": 1.609944317124624e-05,
"loss": 0.0146,
"step": 22510
},
{
"grad_norm": 0.37438493967056274,
"learning_rate": 1.605895091948582e-05,
"loss": 0.0198,
"step": 22520
},
{
"grad_norm": 0.5041359066963196,
"learning_rate": 1.6018499909300477e-05,
"loss": 0.0258,
"step": 22530
},
{
"grad_norm": 0.3793189227581024,
"learning_rate": 1.597809018984199e-05,
"loss": 0.0139,
"step": 22540
},
{
"grad_norm": 0.2790461778640747,
"learning_rate": 1.593772181021196e-05,
"loss": 0.0236,
"step": 22550
},
{
"grad_norm": 0.4782765507698059,
"learning_rate": 1.5897394819461813e-05,
"loss": 0.0275,
"step": 22560
},
{
"grad_norm": 0.4679233431816101,
"learning_rate": 1.5857109266592625e-05,
"loss": 0.0158,
"step": 22570
},
{
"grad_norm": 0.3592627942562103,
"learning_rate": 1.581686520055514e-05,
"loss": 0.0135,
"step": 22580
},
{
"grad_norm": 0.3648861348628998,
"learning_rate": 1.5776662670249704e-05,
"loss": 0.0147,
"step": 22590
},
{
"grad_norm": 0.39617666602134705,
"learning_rate": 1.573650172452615e-05,
"loss": 0.0144,
"step": 22600
},
{
"grad_norm": 0.5906565189361572,
"learning_rate": 1.5696382412183852e-05,
"loss": 0.0149,
"step": 22610
},
{
"grad_norm": 0.37735745310783386,
"learning_rate": 1.565630478197155e-05,
"loss": 0.0149,
"step": 22620
},
{
"grad_norm": 0.3903099298477173,
"learning_rate": 1.5616268882587332e-05,
"loss": 0.0165,
"step": 22630
},
{
"grad_norm": 0.5556117296218872,
"learning_rate": 1.5576274762678594e-05,
"loss": 0.0226,
"step": 22640
},
{
"grad_norm": 0.5623263120651245,
"learning_rate": 1.5536322470841952e-05,
"loss": 0.0217,
"step": 22650
},
{
"grad_norm": 0.4955753982067108,
"learning_rate": 1.549641205562324e-05,
"loss": 0.0191,
"step": 22660
},
{
"grad_norm": 0.3689819276332855,
"learning_rate": 1.5456543565517334e-05,
"loss": 0.0151,
"step": 22670
},
{
"grad_norm": 0.4275532066822052,
"learning_rate": 1.541671704896825e-05,
"loss": 0.0178,
"step": 22680
},
{
"grad_norm": 0.40928807854652405,
"learning_rate": 1.5376932554368938e-05,
"loss": 0.0156,
"step": 22690
},
{
"grad_norm": 0.5950973033905029,
"learning_rate": 1.5337190130061318e-05,
"loss": 0.0161,
"step": 22700
},
{
"grad_norm": 0.5135419964790344,
"learning_rate": 1.5297489824336165e-05,
"loss": 0.0287,
"step": 22710
},
{
"grad_norm": 0.5599364042282104,
"learning_rate": 1.5257831685433121e-05,
"loss": 0.0137,
"step": 22720
},
{
"grad_norm": 0.4178580343723297,
"learning_rate": 1.521821576154055e-05,
"loss": 0.0167,
"step": 22730
},
{
"grad_norm": 0.35852137207984924,
"learning_rate": 1.5178642100795542e-05,
"loss": 0.0147,
"step": 22740
},
{
"grad_norm": 0.47794297337532043,
"learning_rate": 1.5139110751283819e-05,
"loss": 0.0219,
"step": 22750
},
{
"grad_norm": 0.5335997343063354,
"learning_rate": 1.5099621761039683e-05,
"loss": 0.0231,
"step": 22760
},
{
"grad_norm": 0.48440736532211304,
"learning_rate": 1.5060175178046016e-05,
"loss": 0.0168,
"step": 22770
},
{
"grad_norm": 0.42115530371665955,
"learning_rate": 1.5020771050234117e-05,
"loss": 0.015,
"step": 22780
},
{
"grad_norm": 0.5304067134857178,
"learning_rate": 1.4981409425483716e-05,
"loss": 0.013,
"step": 22790
},
{
"grad_norm": 0.3685724139213562,
"learning_rate": 1.4942090351622883e-05,
"loss": 0.0134,
"step": 22800
},
{
"grad_norm": 0.4468485713005066,
"learning_rate": 1.490281387642804e-05,
"loss": 0.0176,
"step": 22810
},
{
"grad_norm": 0.389216810464859,
"learning_rate": 1.4863580047623765e-05,
"loss": 0.0234,
"step": 22820
},
{
"grad_norm": 0.49556997418403625,
"learning_rate": 1.4824388912882897e-05,
"loss": 0.0151,
"step": 22830
},
{
"grad_norm": 0.36740437150001526,
"learning_rate": 1.4785240519826344e-05,
"loss": 0.0187,
"step": 22840
},
{
"grad_norm": 0.5557162165641785,
"learning_rate": 1.4746134916023097e-05,
"loss": 0.0325,
"step": 22850
},
{
"grad_norm": 0.39118531346321106,
"learning_rate": 1.4707072148990141e-05,
"loss": 0.0139,
"step": 22860
},
{
"grad_norm": 0.509096086025238,
"learning_rate": 1.4668052266192423e-05,
"loss": 0.0156,
"step": 22870
},
{
"grad_norm": 0.48011890053749084,
"learning_rate": 1.4629075315042795e-05,
"loss": 0.0228,
"step": 22880
},
{
"grad_norm": 0.39678987860679626,
"learning_rate": 1.4590141342901925e-05,
"loss": 0.0154,
"step": 22890
},
{
"grad_norm": 0.49110984802246094,
"learning_rate": 1.4551250397078253e-05,
"loss": 0.0297,
"step": 22900
},
{
"grad_norm": 0.5438308715820312,
"learning_rate": 1.4512402524827945e-05,
"loss": 0.0208,
"step": 22910
},
{
"grad_norm": 0.4456293284893036,
"learning_rate": 1.447359777335482e-05,
"loss": 0.0242,
"step": 22920
},
{
"grad_norm": 0.5340110063552856,
"learning_rate": 1.4434836189810335e-05,
"loss": 0.0159,
"step": 22930
},
{
"grad_norm": 0.40285205841064453,
"learning_rate": 1.4396117821293453e-05,
"loss": 0.0116,
"step": 22940
},
{
"grad_norm": 0.4606822431087494,
"learning_rate": 1.4357442714850633e-05,
"loss": 0.0299,
"step": 22950
},
{
"grad_norm": 0.55609130859375,
"learning_rate": 1.43188109174758e-05,
"loss": 0.0152,
"step": 22960
},
{
"grad_norm": 0.4163595139980316,
"learning_rate": 1.4280222476110206e-05,
"loss": 0.0146,
"step": 22970
},
{
"grad_norm": 0.44007793068885803,
"learning_rate": 1.4241677437642469e-05,
"loss": 0.0192,
"step": 22980
},
{
"grad_norm": 0.4690817594528198,
"learning_rate": 1.420317584890844e-05,
"loss": 0.0189,
"step": 22990
},
{
"grad_norm": 0.4701695740222931,
"learning_rate": 1.4164717756691176e-05,
"loss": 0.0142,
"step": 23000
},
{
"grad_norm": 0.3212604522705078,
"learning_rate": 1.4126303207720882e-05,
"loss": 0.0141,
"step": 23010
},
{
"grad_norm": 0.5325477719306946,
"learning_rate": 1.408793224867484e-05,
"loss": 0.0165,
"step": 23020
},
{
"grad_norm": 0.3951583206653595,
"learning_rate": 1.4049604926177423e-05,
"loss": 0.0198,
"step": 23030
},
{
"grad_norm": 0.4383741319179535,
"learning_rate": 1.4011321286799916e-05,
"loss": 0.0157,
"step": 23040
},
{
"grad_norm": 0.4338151812553406,
"learning_rate": 1.3973081377060565e-05,
"loss": 0.0256,
"step": 23050
},
{
"grad_norm": 0.34984585642814636,
"learning_rate": 1.393488524342445e-05,
"loss": 0.0151,
"step": 23060
},
{
"grad_norm": 0.4914739429950714,
"learning_rate": 1.3896732932303486e-05,
"loss": 0.0132,
"step": 23070
},
{
"grad_norm": 0.4527621567249298,
"learning_rate": 1.3858624490056304e-05,
"loss": 0.0148,
"step": 23080
},
{
"grad_norm": 0.426383376121521,
"learning_rate": 1.3820559962988299e-05,
"loss": 0.0162,
"step": 23090
},
{
"grad_norm": 0.37581759691238403,
"learning_rate": 1.3782539397351418e-05,
"loss": 0.0209,
"step": 23100
},
{
"grad_norm": 0.3409337103366852,
"learning_rate": 1.3744562839344266e-05,
"loss": 0.0216,
"step": 23110
},
{
"grad_norm": 0.4652528464794159,
"learning_rate": 1.370663033511193e-05,
"loss": 0.0178,
"step": 23120
},
{
"grad_norm": 0.37954452633857727,
"learning_rate": 1.3668741930745964e-05,
"loss": 0.0141,
"step": 23130
},
{
"grad_norm": 0.4364206790924072,
"learning_rate": 1.3630897672284382e-05,
"loss": 0.0117,
"step": 23140
},
{
"grad_norm": 0.48842477798461914,
"learning_rate": 1.3593097605711508e-05,
"loss": 0.0304,
"step": 23150
},
{
"grad_norm": 0.3097720444202423,
"learning_rate": 1.3555341776957992e-05,
"loss": 0.0142,
"step": 23160
},
{
"grad_norm": 0.41316497325897217,
"learning_rate": 1.3517630231900724e-05,
"loss": 0.017,
"step": 23170
},
{
"grad_norm": 0.32613134384155273,
"learning_rate": 1.3479963016362767e-05,
"loss": 0.0122,
"step": 23180
},
{
"grad_norm": 0.5037158727645874,
"learning_rate": 1.3442340176113377e-05,
"loss": 0.0205,
"step": 23190
},
{
"grad_norm": 0.4213935136795044,
"learning_rate": 1.340476175686784e-05,
"loss": 0.0159,
"step": 23200
},
{
"grad_norm": 0.34387701749801636,
"learning_rate": 1.3367227804287469e-05,
"loss": 0.0131,
"step": 23210
},
{
"grad_norm": 0.44235503673553467,
"learning_rate": 1.332973836397956e-05,
"loss": 0.014,
"step": 23220
},
{
"grad_norm": 0.4285106956958771,
"learning_rate": 1.329229348149731e-05,
"loss": 0.0189,
"step": 23230
},
{
"grad_norm": 0.40131789445877075,
"learning_rate": 1.3254893202339796e-05,
"loss": 0.0148,
"step": 23240
},
{
"grad_norm": 0.2989444434642792,
"learning_rate": 1.3217537571951872e-05,
"loss": 0.0191,
"step": 23250
},
{
"grad_norm": 0.2872249484062195,
"learning_rate": 1.3180226635724169e-05,
"loss": 0.0162,
"step": 23260
},
{
"grad_norm": 0.5776388049125671,
"learning_rate": 1.314296043899298e-05,
"loss": 0.0167,
"step": 23270
},
{
"grad_norm": 0.3660447895526886,
"learning_rate": 1.3105739027040248e-05,
"loss": 0.0168,
"step": 23280
},
{
"grad_norm": 0.5000708699226379,
"learning_rate": 1.3068562445093491e-05,
"loss": 0.0282,
"step": 23290
},
{
"grad_norm": 0.5894431471824646,
"learning_rate": 1.3031430738325778e-05,
"loss": 0.0186,
"step": 23300
},
{
"grad_norm": 0.4551120698451996,
"learning_rate": 1.299434395185563e-05,
"loss": 0.0229,
"step": 23310
},
{
"grad_norm": 0.6477834582328796,
"learning_rate": 1.295730213074699e-05,
"loss": 0.0171,
"step": 23320
},
{
"grad_norm": 0.40606406331062317,
"learning_rate": 1.2920305320009152e-05,
"loss": 0.0182,
"step": 23330
},
{
"grad_norm": 0.4402155578136444,
"learning_rate": 1.2883353564596729e-05,
"loss": 0.0185,
"step": 23340
},
{
"grad_norm": 0.4886578917503357,
"learning_rate": 1.2846446909409604e-05,
"loss": 0.0177,
"step": 23350
},
{
"grad_norm": 0.4952391982078552,
"learning_rate": 1.280958539929284e-05,
"loss": 0.0185,
"step": 23360
},
{
"grad_norm": 0.40972909331321716,
"learning_rate": 1.277276907903664e-05,
"loss": 0.0198,
"step": 23370
},
{
"grad_norm": 0.3662271797657013,
"learning_rate": 1.2735997993376298e-05,
"loss": 0.0202,
"step": 23380
},
{
"grad_norm": 0.5101296305656433,
"learning_rate": 1.2699272186992166e-05,
"loss": 0.0282,
"step": 23390
},
{
"grad_norm": 0.3920280635356903,
"learning_rate": 1.2662591704509547e-05,
"loss": 0.0247,
"step": 23400
},
{
"grad_norm": 0.4300827085971832,
"learning_rate": 1.2625956590498711e-05,
"loss": 0.0228,
"step": 23410
},
{
"grad_norm": 0.4248438775539398,
"learning_rate": 1.2589366889474757e-05,
"loss": 0.0119,
"step": 23420
},
{
"grad_norm": 0.3799787163734436,
"learning_rate": 1.2552822645897621e-05,
"loss": 0.0169,
"step": 23430
},
{
"grad_norm": 0.38060128688812256,
"learning_rate": 1.2516323904172e-05,
"loss": 0.0152,
"step": 23440
},
{
"grad_norm": 0.5116733908653259,
"learning_rate": 1.2479870708647324e-05,
"loss": 0.0189,
"step": 23450
},
{
"grad_norm": 0.3416810929775238,
"learning_rate": 1.2443463103617658e-05,
"loss": 0.0145,
"step": 23460
},
{
"grad_norm": 0.42973792552948,
"learning_rate": 1.240710113332167e-05,
"loss": 0.0154,
"step": 23470
},
{
"grad_norm": 0.4214729368686676,
"learning_rate": 1.2370784841942584e-05,
"loss": 0.0155,
"step": 23480
},
{
"grad_norm": 0.4657494127750397,
"learning_rate": 1.2334514273608117e-05,
"loss": 0.0184,
"step": 23490
},
{
"grad_norm": 0.35942745208740234,
"learning_rate": 1.2298289472390417e-05,
"loss": 0.0303,
"step": 23500
},
{
"grad_norm": 0.35730093717575073,
"learning_rate": 1.226211048230606e-05,
"loss": 0.0175,
"step": 23510
},
{
"grad_norm": 0.5626978278160095,
"learning_rate": 1.2225977347315921e-05,
"loss": 0.0372,
"step": 23520
},
{
"grad_norm": 0.43064674735069275,
"learning_rate": 1.2189890111325148e-05,
"loss": 0.0151,
"step": 23530
},
{
"grad_norm": 0.426013320684433,
"learning_rate": 1.2153848818183161e-05,
"loss": 0.0203,
"step": 23540
},
{
"grad_norm": 0.4776923954486847,
"learning_rate": 1.2117853511683507e-05,
"loss": 0.0171,
"step": 23550
},
{
"grad_norm": 0.4227933883666992,
"learning_rate": 1.2081904235563906e-05,
"loss": 0.0161,
"step": 23560
},
{
"grad_norm": 0.3704429566860199,
"learning_rate": 1.20460010335061e-05,
"loss": 0.016,
"step": 23570
},
{
"grad_norm": 0.3829001188278198,
"learning_rate": 1.2010143949135865e-05,
"loss": 0.0182,
"step": 23580
},
{
"grad_norm": 0.4521525800228119,
"learning_rate": 1.1974333026022938e-05,
"loss": 0.0259,
"step": 23590
},
{
"grad_norm": 0.46864110231399536,
"learning_rate": 1.1938568307680964e-05,
"loss": 0.0203,
"step": 23600
},
{
"grad_norm": 0.3084005117416382,
"learning_rate": 1.1902849837567464e-05,
"loss": 0.0206,
"step": 23610
},
{
"grad_norm": 0.35645031929016113,
"learning_rate": 1.1867177659083739e-05,
"loss": 0.0165,
"step": 23620
},
{
"grad_norm": 0.3938805162906647,
"learning_rate": 1.1831551815574848e-05,
"loss": 0.0147,
"step": 23630
},
{
"grad_norm": 0.5711188316345215,
"learning_rate": 1.1795972350329554e-05,
"loss": 0.0219,
"step": 23640
},
{
"grad_norm": 0.5373600721359253,
"learning_rate": 1.176043930658024e-05,
"loss": 0.019,
"step": 23650
},
{
"grad_norm": 0.29622259736061096,
"learning_rate": 1.1724952727502942e-05,
"loss": 0.0163,
"step": 23660
},
{
"grad_norm": 0.38981613516807556,
"learning_rate": 1.1689512656217178e-05,
"loss": 0.0156,
"step": 23670
},
{
"grad_norm": 0.3548983633518219,
"learning_rate": 1.1654119135785962e-05,
"loss": 0.0207,
"step": 23680
},
{
"grad_norm": 0.42422252893447876,
"learning_rate": 1.1618772209215794e-05,
"loss": 0.0153,
"step": 23690
},
{
"grad_norm": 0.4568403363227844,
"learning_rate": 1.1583471919456506e-05,
"loss": 0.0181,
"step": 23700
},
{
"grad_norm": 0.4619245231151581,
"learning_rate": 1.1548218309401266e-05,
"loss": 0.0152,
"step": 23710
},
{
"grad_norm": 0.3855067491531372,
"learning_rate": 1.1513011421886555e-05,
"loss": 0.0168,
"step": 23720
},
{
"grad_norm": 0.4894396662712097,
"learning_rate": 1.1477851299692056e-05,
"loss": 0.0234,
"step": 23730
},
{
"grad_norm": 0.4241814613342285,
"learning_rate": 1.144273798554063e-05,
"loss": 0.0141,
"step": 23740
},
{
"grad_norm": 0.44133204221725464,
"learning_rate": 1.1407671522098262e-05,
"loss": 0.0184,
"step": 23750
},
{
"grad_norm": 0.43295231461524963,
"learning_rate": 1.1372651951974e-05,
"loss": 0.0163,
"step": 23760
},
{
"grad_norm": 0.4684405028820038,
"learning_rate": 1.1337679317719952e-05,
"loss": 0.0161,
"step": 23770
},
{
"grad_norm": 0.5486006736755371,
"learning_rate": 1.1302753661831151e-05,
"loss": 0.0164,
"step": 23780
},
{
"grad_norm": 0.43995240330696106,
"learning_rate": 1.1267875026745561e-05,
"loss": 0.0159,
"step": 23790
},
{
"grad_norm": 0.4496006965637207,
"learning_rate": 1.1233043454844016e-05,
"loss": 0.017,
"step": 23800
},
{
"grad_norm": 0.42049458622932434,
"learning_rate": 1.1198258988450144e-05,
"loss": 0.0184,
"step": 23810
},
{
"grad_norm": 0.47422635555267334,
"learning_rate": 1.1163521669830368e-05,
"loss": 0.0147,
"step": 23820
},
{
"grad_norm": 0.41377052664756775,
"learning_rate": 1.1128831541193818e-05,
"loss": 0.0134,
"step": 23830
},
{
"grad_norm": 0.4919155240058899,
"learning_rate": 1.1094188644692255e-05,
"loss": 0.0176,
"step": 23840
},
{
"grad_norm": 0.2737899124622345,
"learning_rate": 1.1059593022420067e-05,
"loss": 0.012,
"step": 23850
},
{
"grad_norm": 0.3316786289215088,
"learning_rate": 1.1025044716414184e-05,
"loss": 0.0146,
"step": 23860
},
{
"grad_norm": 0.5062804222106934,
"learning_rate": 1.0990543768654082e-05,
"loss": 0.0191,
"step": 23870
},
{
"grad_norm": 0.377379447221756,
"learning_rate": 1.0956090221061655e-05,
"loss": 0.0136,
"step": 23880
},
{
"grad_norm": 0.3727629482746124,
"learning_rate": 1.0921684115501208e-05,
"loss": 0.019,
"step": 23890
},
{
"grad_norm": 0.3582492470741272,
"learning_rate": 1.0887325493779405e-05,
"loss": 0.0151,
"step": 23900
},
{
"grad_norm": 0.3322191536426544,
"learning_rate": 1.0853014397645207e-05,
"loss": 0.0143,
"step": 23910
},
{
"grad_norm": 0.4699956476688385,
"learning_rate": 1.0818750868789829e-05,
"loss": 0.0226,
"step": 23920
},
{
"grad_norm": 0.39576420187950134,
"learning_rate": 1.0784534948846703e-05,
"loss": 0.0155,
"step": 23930
},
{
"grad_norm": 0.39792147278785706,
"learning_rate": 1.0750366679391393e-05,
"loss": 0.0111,
"step": 23940
},
{
"grad_norm": 0.393524706363678,
"learning_rate": 1.0716246101941557e-05,
"loss": 0.0141,
"step": 23950
},
{
"grad_norm": 0.48502394556999207,
"learning_rate": 1.0682173257956934e-05,
"loss": 0.029,
"step": 23960
},
{
"grad_norm": 0.3700765073299408,
"learning_rate": 1.064814818883922e-05,
"loss": 0.0176,
"step": 23970
},
{
"grad_norm": 0.3032940924167633,
"learning_rate": 1.06141709359321e-05,
"loss": 0.0151,
"step": 23980
},
{
"grad_norm": 0.35760876536369324,
"learning_rate": 1.058024154052114e-05,
"loss": 0.02,
"step": 23990
},
{
"grad_norm": 0.49229052662849426,
"learning_rate": 1.0546360043833747e-05,
"loss": 0.0144,
"step": 24000
},
{
"grad_norm": 0.3013880252838135,
"learning_rate": 1.0512526487039138e-05,
"loss": 0.0119,
"step": 24010
},
{
"grad_norm": 0.39208975434303284,
"learning_rate": 1.0478740911248258e-05,
"loss": 0.0134,
"step": 24020
},
{
"grad_norm": 0.3787972331047058,
"learning_rate": 1.0445003357513793e-05,
"loss": 0.0137,
"step": 24030
},
{
"grad_norm": 0.5072857141494751,
"learning_rate": 1.0411313866830042e-05,
"loss": 0.0164,
"step": 24040
},
{
"grad_norm": 0.393159955739975,
"learning_rate": 1.0377672480132915e-05,
"loss": 0.0165,
"step": 24050
},
{
"grad_norm": 0.5104449987411499,
"learning_rate": 1.0344079238299864e-05,
"loss": 0.0188,
"step": 24060
},
{
"grad_norm": 0.32074877619743347,
"learning_rate": 1.0310534182149833e-05,
"loss": 0.0118,
"step": 24070
},
{
"grad_norm": 0.47053951025009155,
"learning_rate": 1.0277037352443258e-05,
"loss": 0.0224,
"step": 24080
},
{
"grad_norm": 0.31820032000541687,
"learning_rate": 1.024358878988193e-05,
"loss": 0.0112,
"step": 24090
},
{
"grad_norm": 0.37667524814605713,
"learning_rate": 1.0210188535108995e-05,
"loss": 0.0174,
"step": 24100
},
{
"grad_norm": 0.4278700053691864,
"learning_rate": 1.0176836628708935e-05,
"loss": 0.0153,
"step": 24110
},
{
"grad_norm": 0.584918737411499,
"learning_rate": 1.0143533111207454e-05,
"loss": 0.0139,
"step": 24120
},
{
"grad_norm": 0.416545033454895,
"learning_rate": 1.0110278023071446e-05,
"loss": 0.0257,
"step": 24130
},
{
"grad_norm": 0.34062397480010986,
"learning_rate": 1.007707140470901e-05,
"loss": 0.0191,
"step": 24140
},
{
"grad_norm": 0.4329567551612854,
"learning_rate": 1.0043913296469298e-05,
"loss": 0.0172,
"step": 24150
},
{
"grad_norm": 0.3700951933860779,
"learning_rate": 1.001080373864255e-05,
"loss": 0.0168,
"step": 24160
},
{
"grad_norm": 0.44920849800109863,
"learning_rate": 9.977742771459991e-06,
"loss": 0.0134,
"step": 24170
},
{
"grad_norm": 0.43868061900138855,
"learning_rate": 9.944730435093802e-06,
"loss": 0.0154,
"step": 24180
},
{
"grad_norm": 0.3852888345718384,
"learning_rate": 9.911766769657115e-06,
"loss": 0.0136,
"step": 24190
},
{
"grad_norm": 0.35018423199653625,
"learning_rate": 9.878851815203882e-06,
"loss": 0.0121,
"step": 24200
},
{
"grad_norm": 0.4435025751590729,
"learning_rate": 9.845985611728864e-06,
"loss": 0.0154,
"step": 24210
},
{
"grad_norm": 0.40852636098861694,
"learning_rate": 9.813168199167606e-06,
"loss": 0.0185,
"step": 24220
},
{
"grad_norm": 0.6129499673843384,
"learning_rate": 9.780399617396341e-06,
"loss": 0.0199,
"step": 24230
},
{
"grad_norm": 0.39691588282585144,
"learning_rate": 9.747679906232016e-06,
"loss": 0.0134,
"step": 24240
},
{
"grad_norm": 0.4939647614955902,
"learning_rate": 9.715009105432138e-06,
"loss": 0.0175,
"step": 24250
},
{
"grad_norm": 0.4715416133403778,
"learning_rate": 9.682387254694835e-06,
"loss": 0.015,
"step": 24260
},
{
"grad_norm": 0.465074360370636,
"learning_rate": 9.649814393658723e-06,
"loss": 0.0164,
"step": 24270
},
{
"grad_norm": 0.45065033435821533,
"learning_rate": 9.617290561902881e-06,
"loss": 0.016,
"step": 24280
},
{
"grad_norm": 0.576227605342865,
"learning_rate": 9.584815798946861e-06,
"loss": 0.0208,
"step": 24290
},
{
"grad_norm": 0.36152735352516174,
"learning_rate": 9.55239014425055e-06,
"loss": 0.0232,
"step": 24300
},
{
"grad_norm": 0.5241812467575073,
"learning_rate": 9.520013637214176e-06,
"loss": 0.022,
"step": 24310
},
{
"grad_norm": 0.4062443971633911,
"learning_rate": 9.48768631717824e-06,
"loss": 0.0144,
"step": 24320
},
{
"grad_norm": 0.3344646096229553,
"learning_rate": 9.455408223423496e-06,
"loss": 0.0123,
"step": 24330
},
{
"grad_norm": 0.3734626770019531,
"learning_rate": 9.423179395170845e-06,
"loss": 0.0187,
"step": 24340
},
{
"grad_norm": 0.34105339646339417,
"learning_rate": 9.390999871581391e-06,
"loss": 0.0191,
"step": 24350
},
{
"grad_norm": 0.4400784373283386,
"learning_rate": 9.358869691756273e-06,
"loss": 0.0175,
"step": 24360
},
{
"grad_norm": 0.40192151069641113,
"learning_rate": 9.326788894736688e-06,
"loss": 0.0152,
"step": 24370
},
{
"grad_norm": 0.4199756979942322,
"learning_rate": 9.294757519503811e-06,
"loss": 0.0125,
"step": 24380
},
{
"grad_norm": 0.31322163343429565,
"learning_rate": 9.262775604978819e-06,
"loss": 0.0129,
"step": 24390
},
{
"grad_norm": 0.4799484312534332,
"learning_rate": 9.230843190022725e-06,
"loss": 0.0158,
"step": 24400
},
{
"grad_norm": 0.3553149402141571,
"learning_rate": 9.198960313436445e-06,
"loss": 0.0132,
"step": 24410
},
{
"grad_norm": 0.4179014265537262,
"learning_rate": 9.16712701396067e-06,
"loss": 0.027,
"step": 24420
},
{
"grad_norm": 0.4681735932826996,
"learning_rate": 9.135343330275864e-06,
"loss": 0.0177,
"step": 24430
},
{
"grad_norm": 0.37707069516181946,
"learning_rate": 9.10360930100218e-06,
"loss": 0.0106,
"step": 24440
},
{
"grad_norm": 0.42087286710739136,
"learning_rate": 9.07192496469949e-06,
"loss": 0.0178,
"step": 24450
},
{
"grad_norm": 0.4371490478515625,
"learning_rate": 9.040290359867231e-06,
"loss": 0.0175,
"step": 24460
},
{
"grad_norm": 0.3750118017196655,
"learning_rate": 9.008705524944438e-06,
"loss": 0.0165,
"step": 24470
},
{
"grad_norm": 0.4406755566596985,
"learning_rate": 8.977170498309651e-06,
"loss": 0.0163,
"step": 24480
},
{
"grad_norm": 0.5126590132713318,
"learning_rate": 8.945685318280916e-06,
"loss": 0.0149,
"step": 24490
},
{
"grad_norm": 0.5802885890007019,
"learning_rate": 8.914250023115672e-06,
"loss": 0.0204,
"step": 24500
},
{
"grad_norm": 0.4918789267539978,
"learning_rate": 8.882864651010798e-06,
"loss": 0.0118,
"step": 24510
},
{
"grad_norm": 0.4564034342765808,
"learning_rate": 8.851529240102462e-06,
"loss": 0.0188,
"step": 24520
},
{
"grad_norm": 0.33692312240600586,
"learning_rate": 8.820243828466134e-06,
"loss": 0.0139,
"step": 24530
},
{
"grad_norm": 0.47377723455429077,
"learning_rate": 8.789008454116566e-06,
"loss": 0.014,
"step": 24540
},
{
"grad_norm": 0.4408530294895172,
"learning_rate": 8.757823155007655e-06,
"loss": 0.0182,
"step": 24550
},
{
"grad_norm": 0.40919235348701477,
"learning_rate": 8.72668796903251e-06,
"loss": 0.0123,
"step": 24560
},
{
"grad_norm": 0.3448387682437897,
"learning_rate": 8.695602934023301e-06,
"loss": 0.0148,
"step": 24570
},
{
"grad_norm": 0.49360695481300354,
"learning_rate": 8.664568087751274e-06,
"loss": 0.0165,
"step": 24580
},
{
"grad_norm": 0.38249117136001587,
"learning_rate": 8.633583467926697e-06,
"loss": 0.0144,
"step": 24590
},
{
"grad_norm": 0.3204295337200165,
"learning_rate": 8.602649112198796e-06,
"loss": 0.0139,
"step": 24600
},
{
"grad_norm": 0.4225796163082123,
"learning_rate": 8.571765058155745e-06,
"loss": 0.0218,
"step": 24610
},
{
"grad_norm": 0.36309465765953064,
"learning_rate": 8.540931343324583e-06,
"loss": 0.0117,
"step": 24620
},
{
"grad_norm": 0.3495940864086151,
"learning_rate": 8.51014800517117e-06,
"loss": 0.0236,
"step": 24630
},
{
"grad_norm": 0.39253878593444824,
"learning_rate": 8.479415081100167e-06,
"loss": 0.0197,
"step": 24640
},
{
"grad_norm": 0.3417019248008728,
"learning_rate": 8.448732608454967e-06,
"loss": 0.0156,
"step": 24650
},
{
"grad_norm": 0.38767996430397034,
"learning_rate": 8.418100624517688e-06,
"loss": 0.0169,
"step": 24660
},
{
"grad_norm": 0.33748486638069153,
"learning_rate": 8.387519166509062e-06,
"loss": 0.0192,
"step": 24670
},
{
"grad_norm": 0.45077845454216003,
"learning_rate": 8.356988271588445e-06,
"loss": 0.0145,
"step": 24680
},
{
"grad_norm": 0.3968607783317566,
"learning_rate": 8.326507976853765e-06,
"loss": 0.0189,
"step": 24690
},
{
"grad_norm": 0.48317936062812805,
"learning_rate": 8.296078319341443e-06,
"loss": 0.02,
"step": 24700
},
{
"grad_norm": 0.4443443715572357,
"learning_rate": 8.265699336026383e-06,
"loss": 0.0169,
"step": 24710
},
{
"grad_norm": 0.4841650128364563,
"learning_rate": 8.235371063821923e-06,
"loss": 0.0172,
"step": 24720
},
{
"grad_norm": 0.555178701877594,
"learning_rate": 8.205093539579767e-06,
"loss": 0.0199,
"step": 24730
},
{
"grad_norm": 0.32998302578926086,
"learning_rate": 8.174866800089963e-06,
"loss": 0.0183,
"step": 24740
},
{
"grad_norm": 0.4238419830799103,
"learning_rate": 8.144690882080853e-06,
"loss": 0.0186,
"step": 24750
},
{
"grad_norm": 0.36790022253990173,
"learning_rate": 8.114565822219006e-06,
"loss": 0.023,
"step": 24760
},
{
"grad_norm": 0.44996270537376404,
"learning_rate": 8.084491657109233e-06,
"loss": 0.018,
"step": 24770
},
{
"grad_norm": 0.47348320484161377,
"learning_rate": 8.054468423294476e-06,
"loss": 0.0215,
"step": 24780
},
{
"grad_norm": 0.458516001701355,
"learning_rate": 8.024496157255785e-06,
"loss": 0.0192,
"step": 24790
},
{
"grad_norm": 0.27791330218315125,
"learning_rate": 7.994574895412293e-06,
"loss": 0.0125,
"step": 24800
},
{
"grad_norm": 0.4677662253379822,
"learning_rate": 7.964704674121149e-06,
"loss": 0.0177,
"step": 24810
},
{
"grad_norm": 0.49106836318969727,
"learning_rate": 7.934885529677505e-06,
"loss": 0.0158,
"step": 24820
},
{
"grad_norm": 0.3729526102542877,
"learning_rate": 7.905117498314412e-06,
"loss": 0.0196,
"step": 24830
},
{
"grad_norm": 0.4097064733505249,
"learning_rate": 7.87540061620286e-06,
"loss": 0.0158,
"step": 24840
},
{
"grad_norm": 0.537936806678772,
"learning_rate": 7.845734919451647e-06,
"loss": 0.0138,
"step": 24850
},
{
"grad_norm": 0.3379949927330017,
"learning_rate": 7.816120444107383e-06,
"loss": 0.0131,
"step": 24860
},
{
"grad_norm": 0.4597718119621277,
"learning_rate": 7.786557226154473e-06,
"loss": 0.0242,
"step": 24870
},
{
"grad_norm": 0.33503758907318115,
"learning_rate": 7.757045301514998e-06,
"loss": 0.0145,
"step": 24880
},
{
"grad_norm": 0.534487783908844,
"learning_rate": 7.727584706048735e-06,
"loss": 0.0178,
"step": 24890
},
{
"grad_norm": 0.5116600394248962,
"learning_rate": 7.698175475553077e-06,
"loss": 0.0219,
"step": 24900
},
{
"grad_norm": 0.4854741096496582,
"learning_rate": 7.668817645763021e-06,
"loss": 0.0148,
"step": 24910
},
{
"grad_norm": 0.40737321972846985,
"learning_rate": 7.639511252351088e-06,
"loss": 0.0121,
"step": 24920
},
{
"grad_norm": 0.3625161945819855,
"learning_rate": 7.610256330927324e-06,
"loss": 0.016,
"step": 24930
},
{
"grad_norm": 0.2848629951477051,
"learning_rate": 7.581052917039211e-06,
"loss": 0.0175,
"step": 24940
},
{
"grad_norm": 0.4281764030456543,
"learning_rate": 7.551901046171645e-06,
"loss": 0.017,
"step": 24950
},
{
"grad_norm": 0.48112642765045166,
"learning_rate": 7.522800753746895e-06,
"loss": 0.0169,
"step": 24960
},
{
"grad_norm": 0.2691130042076111,
"learning_rate": 7.493752075124577e-06,
"loss": 0.0175,
"step": 24970
},
{
"grad_norm": 0.43648824095726013,
"learning_rate": 7.464755045601557e-06,
"loss": 0.0144,
"step": 24980
},
{
"grad_norm": 0.425437331199646,
"learning_rate": 7.435809700411972e-06,
"loss": 0.0314,
"step": 24990
},
{
"grad_norm": 0.5220417976379395,
"learning_rate": 7.40691607472715e-06,
"loss": 0.0186,
"step": 25000
},
{
"grad_norm": 0.4314422011375427,
"learning_rate": 7.3780742036555606e-06,
"loss": 0.0189,
"step": 25010
},
{
"grad_norm": 0.34729334712028503,
"learning_rate": 7.349284122242783e-06,
"loss": 0.0171,
"step": 25020
},
{
"grad_norm": 0.4264434278011322,
"learning_rate": 7.320545865471512e-06,
"loss": 0.0273,
"step": 25030
},
{
"grad_norm": 0.3189747631549835,
"learning_rate": 7.291859468261425e-06,
"loss": 0.0141,
"step": 25040
},
{
"grad_norm": 0.41960686445236206,
"learning_rate": 7.263224965469195e-06,
"loss": 0.0194,
"step": 25050
},
{
"grad_norm": 0.4310552477836609,
"learning_rate": 7.234642391888446e-06,
"loss": 0.0182,
"step": 25060
},
{
"grad_norm": 0.48914992809295654,
"learning_rate": 7.206111782249697e-06,
"loss": 0.0156,
"step": 25070
},
{
"grad_norm": 0.327129989862442,
"learning_rate": 7.1776331712203385e-06,
"loss": 0.0163,
"step": 25080
},
{
"grad_norm": 0.41288188099861145,
"learning_rate": 7.149206593404561e-06,
"loss": 0.0204,
"step": 25090
},
{
"grad_norm": 0.40793749690055847,
"learning_rate": 7.120832083343337e-06,
"loss": 0.0173,
"step": 25100
},
{
"grad_norm": 0.40860715508461,
"learning_rate": 7.092509675514369e-06,
"loss": 0.0156,
"step": 25110
},
{
"grad_norm": 0.3447897136211395,
"learning_rate": 7.064239404332063e-06,
"loss": 0.0165,
"step": 25120
},
{
"grad_norm": 0.39715713262557983,
"learning_rate": 7.03602130414745e-06,
"loss": 0.0316,
"step": 25130
},
{
"grad_norm": 0.267139732837677,
"learning_rate": 7.007855409248198e-06,
"loss": 0.0185,
"step": 25140
},
{
"grad_norm": 0.4673466384410858,
"learning_rate": 6.979741753858521e-06,
"loss": 0.0148,
"step": 25150
},
{
"grad_norm": 0.3346693217754364,
"learning_rate": 6.951680372139158e-06,
"loss": 0.0154,
"step": 25160
},
{
"grad_norm": 0.5334688425064087,
"learning_rate": 6.923671298187334e-06,
"loss": 0.0126,
"step": 25170
},
{
"grad_norm": 0.35860922932624817,
"learning_rate": 6.895714566036704e-06,
"loss": 0.0147,
"step": 25180
},
{
"grad_norm": 0.3603096306324005,
"learning_rate": 6.86781020965736e-06,
"loss": 0.0182,
"step": 25190
},
{
"grad_norm": 0.30324238538742065,
"learning_rate": 6.839958262955709e-06,
"loss": 0.0133,
"step": 25200
},
{
"grad_norm": 0.41404154896736145,
"learning_rate": 6.8121587597744894e-06,
"loss": 0.0178,
"step": 25210
},
{
"grad_norm": 0.3475189507007599,
"learning_rate": 6.784411733892732e-06,
"loss": 0.0126,
"step": 25220
},
{
"grad_norm": 0.4086148738861084,
"learning_rate": 6.756717219025666e-06,
"loss": 0.0207,
"step": 25230
},
{
"grad_norm": 0.31809866428375244,
"learning_rate": 6.729075248824762e-06,
"loss": 0.0139,
"step": 25240
},
{
"grad_norm": 0.4038105905056,
"learning_rate": 6.701485856877615e-06,
"loss": 0.0108,
"step": 25250
},
{
"grad_norm": 0.3611947000026703,
"learning_rate": 6.673949076707925e-06,
"loss": 0.0125,
"step": 25260
},
{
"grad_norm": 0.32640787959098816,
"learning_rate": 6.646464941775499e-06,
"loss": 0.0157,
"step": 25270
},
{
"grad_norm": 0.37438204884529114,
"learning_rate": 6.619033485476128e-06,
"loss": 0.0171,
"step": 25280
},
{
"grad_norm": 0.30280420184135437,
"learning_rate": 6.59165474114164e-06,
"loss": 0.0143,
"step": 25290
},
{
"grad_norm": 0.3627287745475769,
"learning_rate": 6.564328742039783e-06,
"loss": 0.017,
"step": 25300
},
{
"grad_norm": 0.40442222356796265,
"learning_rate": 6.537055521374219e-06,
"loss": 0.0177,
"step": 25310
},
{
"grad_norm": 0.3775210380554199,
"learning_rate": 6.509835112284485e-06,
"loss": 0.0114,
"step": 25320
},
{
"grad_norm": 0.4335203170776367,
"learning_rate": 6.482667547845944e-06,
"loss": 0.0223,
"step": 25330
},
{
"grad_norm": 0.3852187991142273,
"learning_rate": 6.455552861069736e-06,
"loss": 0.0123,
"step": 25340
},
{
"grad_norm": 0.46274450421333313,
"learning_rate": 6.428491084902788e-06,
"loss": 0.0161,
"step": 25350
},
{
"grad_norm": 0.43703699111938477,
"learning_rate": 6.401482252227698e-06,
"loss": 0.0148,
"step": 25360
},
{
"grad_norm": 0.48273104429244995,
"learning_rate": 6.37452639586274e-06,
"loss": 0.0166,
"step": 25370
},
{
"grad_norm": 0.29464346170425415,
"learning_rate": 6.347623548561826e-06,
"loss": 0.0121,
"step": 25380
},
{
"grad_norm": 0.3679550290107727,
"learning_rate": 6.320773743014441e-06,
"loss": 0.0143,
"step": 25390
},
{
"grad_norm": 0.3420891761779785,
"learning_rate": 6.293977011845648e-06,
"loss": 0.0114,
"step": 25400
},
{
"grad_norm": 0.4843360483646393,
"learning_rate": 6.2672333876159835e-06,
"loss": 0.0143,
"step": 25410
},
{
"grad_norm": 0.44745635986328125,
"learning_rate": 6.2405429028215e-06,
"loss": 0.0116,
"step": 25420
},
{
"grad_norm": 0.588222861289978,
"learning_rate": 6.213905589893631e-06,
"loss": 0.0217,
"step": 25430
},
{
"grad_norm": 0.34517425298690796,
"learning_rate": 6.187321481199221e-06,
"loss": 0.014,
"step": 25440
},
{
"grad_norm": 0.37210360169410706,
"learning_rate": 6.1607906090404895e-06,
"loss": 0.0236,
"step": 25450
},
{
"grad_norm": 0.4113438129425049,
"learning_rate": 6.134313005654929e-06,
"loss": 0.0148,
"step": 25460
},
{
"grad_norm": 0.5271087884902954,
"learning_rate": 6.107888703215337e-06,
"loss": 0.0207,
"step": 25470
},
{
"grad_norm": 0.4223916828632355,
"learning_rate": 6.081517733829722e-06,
"loss": 0.0132,
"step": 25480
},
{
"grad_norm": 0.32123517990112305,
"learning_rate": 6.055200129541294e-06,
"loss": 0.0129,
"step": 25490
},
{
"grad_norm": 0.4004266560077667,
"learning_rate": 6.0289359223284434e-06,
"loss": 0.0179,
"step": 25500
},
{
"grad_norm": 0.3302522301673889,
"learning_rate": 6.002725144104648e-06,
"loss": 0.0141,
"step": 25510
},
{
"grad_norm": 0.2889688014984131,
"learning_rate": 5.976567826718477e-06,
"loss": 0.0128,
"step": 25520
},
{
"grad_norm": 0.39661434292793274,
"learning_rate": 5.9504640019535315e-06,
"loss": 0.0207,
"step": 25530
},
{
"grad_norm": 0.4181612432003021,
"learning_rate": 5.924413701528414e-06,
"loss": 0.0165,
"step": 25540
},
{
"grad_norm": 0.3915834426879883,
"learning_rate": 5.898416957096703e-06,
"loss": 0.0147,
"step": 25550
},
{
"grad_norm": 0.38000690937042236,
"learning_rate": 5.872473800246913e-06,
"loss": 0.0166,
"step": 25560
},
{
"grad_norm": 0.5581942200660706,
"learning_rate": 5.846584262502402e-06,
"loss": 0.0164,
"step": 25570
},
{
"grad_norm": 0.3469103276729584,
"learning_rate": 5.820748375321411e-06,
"loss": 0.0145,
"step": 25580
},
{
"grad_norm": 0.2949829697608948,
"learning_rate": 5.794966170096977e-06,
"loss": 0.016,
"step": 25590
},
{
"grad_norm": 0.40998974442481995,
"learning_rate": 5.769237678156897e-06,
"loss": 0.0118,
"step": 25600
},
{
"grad_norm": 0.2569010853767395,
"learning_rate": 5.743562930763735e-06,
"loss": 0.0167,
"step": 25610
},
{
"grad_norm": 0.5085321068763733,
"learning_rate": 5.717941959114726e-06,
"loss": 0.0179,
"step": 25620
},
{
"grad_norm": 0.41701844334602356,
"learning_rate": 5.69237479434176e-06,
"loss": 0.0172,
"step": 25630
},
{
"grad_norm": 0.44365638494491577,
"learning_rate": 5.666861467511353e-06,
"loss": 0.0157,
"step": 25640
},
{
"grad_norm": 0.4292283058166504,
"learning_rate": 5.641402009624591e-06,
"loss": 0.0207,
"step": 25650
},
{
"grad_norm": 0.3052199184894562,
"learning_rate": 5.615996451617145e-06,
"loss": 0.0117,
"step": 25660
},
{
"grad_norm": 0.40409615635871887,
"learning_rate": 5.590644824359148e-06,
"loss": 0.0285,
"step": 25670
},
{
"grad_norm": 0.4622977375984192,
"learning_rate": 5.56534715865521e-06,
"loss": 0.014,
"step": 25680
},
{
"grad_norm": 0.44835177063941956,
"learning_rate": 5.5401034852443965e-06,
"loss": 0.0159,
"step": 25690
},
{
"grad_norm": 0.587017834186554,
"learning_rate": 5.514913834800134e-06,
"loss": 0.0148,
"step": 25700
},
{
"grad_norm": 0.37277692556381226,
"learning_rate": 5.489778237930238e-06,
"loss": 0.0188,
"step": 25710
},
{
"grad_norm": 0.3703053891658783,
"learning_rate": 5.464696725176827e-06,
"loss": 0.0132,
"step": 25720
},
{
"grad_norm": 0.4304894208908081,
"learning_rate": 5.439669327016294e-06,
"loss": 0.0207,
"step": 25730
},
{
"grad_norm": 0.46178969740867615,
"learning_rate": 5.4146960738592985e-06,
"loss": 0.0136,
"step": 25740
},
{
"grad_norm": 0.3891173005104065,
"learning_rate": 5.389776996050694e-06,
"loss": 0.015,
"step": 25750
},
{
"grad_norm": 0.31939613819122314,
"learning_rate": 5.364912123869492e-06,
"loss": 0.0167,
"step": 25760
},
{
"grad_norm": 0.34931299090385437,
"learning_rate": 5.3401014875288864e-06,
"loss": 0.0175,
"step": 25770
},
{
"grad_norm": 0.38726136088371277,
"learning_rate": 5.315345117176129e-06,
"loss": 0.0112,
"step": 25780
},
{
"grad_norm": 0.41092386841773987,
"learning_rate": 5.2906430428925405e-06,
"loss": 0.0148,
"step": 25790
},
{
"grad_norm": 0.4061776399612427,
"learning_rate": 5.265995294693471e-06,
"loss": 0.0157,
"step": 25800
},
{
"grad_norm": 0.4403809607028961,
"learning_rate": 5.241401902528254e-06,
"loss": 0.013,
"step": 25810
},
{
"grad_norm": 0.22249527275562286,
"learning_rate": 5.216862896280189e-06,
"loss": 0.0111,
"step": 25820
},
{
"grad_norm": 0.30791324377059937,
"learning_rate": 5.192378305766471e-06,
"loss": 0.0138,
"step": 25830
},
{
"grad_norm": 0.3978344798088074,
"learning_rate": 5.1679481607382065e-06,
"loss": 0.0213,
"step": 25840
},
{
"grad_norm": 0.2977312207221985,
"learning_rate": 5.1435724908803105e-06,
"loss": 0.0168,
"step": 25850
},
{
"grad_norm": 0.3598964512348175,
"learning_rate": 5.119251325811514e-06,
"loss": 0.0154,
"step": 25860
},
{
"grad_norm": 0.3217323124408722,
"learning_rate": 5.094984695084348e-06,
"loss": 0.0164,
"step": 25870
},
{
"grad_norm": 0.4409118592739105,
"learning_rate": 5.070772628185039e-06,
"loss": 0.025,
"step": 25880
},
{
"grad_norm": 0.30168431997299194,
"learning_rate": 5.046615154533535e-06,
"loss": 0.0155,
"step": 25890
},
{
"grad_norm": 0.33657553791999817,
"learning_rate": 5.022512303483451e-06,
"loss": 0.0131,
"step": 25900
},
{
"grad_norm": 0.2526300847530365,
"learning_rate": 4.998464104322015e-06,
"loss": 0.0216,
"step": 25910
},
{
"grad_norm": 0.48157042264938354,
"learning_rate": 4.974470586270047e-06,
"loss": 0.0161,
"step": 25920
},
{
"grad_norm": 0.32674258947372437,
"learning_rate": 4.950531778481965e-06,
"loss": 0.0167,
"step": 25930
},
{
"grad_norm": 0.40056464076042175,
"learning_rate": 4.926647710045651e-06,
"loss": 0.012,
"step": 25940
},
{
"grad_norm": 0.2650813162326813,
"learning_rate": 4.9028184099825125e-06,
"loss": 0.0158,
"step": 25950
},
{
"grad_norm": 0.2787788212299347,
"learning_rate": 4.879043907247382e-06,
"loss": 0.0179,
"step": 25960
},
{
"grad_norm": 0.34599676728248596,
"learning_rate": 4.8553242307285414e-06,
"loss": 0.0142,
"step": 25970
},
{
"grad_norm": 0.5007773637771606,
"learning_rate": 4.8316594092476185e-06,
"loss": 0.0173,
"step": 25980
},
{
"grad_norm": 0.2606089413166046,
"learning_rate": 4.808049471559617e-06,
"loss": 0.0153,
"step": 25990
},
{
"grad_norm": 0.4318283200263977,
"learning_rate": 4.784494446352833e-06,
"loss": 0.0164,
"step": 26000
},
{
"grad_norm": 0.5407612919807434,
"learning_rate": 4.760994362248833e-06,
"loss": 0.0179,
"step": 26010
},
{
"grad_norm": 0.41632673144340515,
"learning_rate": 4.737549247802442e-06,
"loss": 0.0149,
"step": 26020
},
{
"grad_norm": 0.4203053116798401,
"learning_rate": 4.714159131501689e-06,
"loss": 0.017,
"step": 26030
},
{
"grad_norm": 0.4959763288497925,
"learning_rate": 4.690824041767766e-06,
"loss": 0.0198,
"step": 26040
},
{
"grad_norm": 0.2840794026851654,
"learning_rate": 4.6675440069550105e-06,
"loss": 0.0198,
"step": 26050
},
{
"grad_norm": 0.3874311149120331,
"learning_rate": 4.64431905535086e-06,
"loss": 0.0147,
"step": 26060
},
{
"grad_norm": 0.3414832651615143,
"learning_rate": 4.621149215175808e-06,
"loss": 0.0194,
"step": 26070
},
{
"grad_norm": 0.44545161724090576,
"learning_rate": 4.5980345145834154e-06,
"loss": 0.0273,
"step": 26080
},
{
"grad_norm": 0.3311502933502197,
"learning_rate": 4.574974981660213e-06,
"loss": 0.0218,
"step": 26090
},
{
"grad_norm": 0.37866833806037903,
"learning_rate": 4.551970644425707e-06,
"loss": 0.0202,
"step": 26100
},
{
"grad_norm": 0.26023873686790466,
"learning_rate": 4.529021530832328e-06,
"loss": 0.0129,
"step": 26110
},
{
"grad_norm": 0.34336647391319275,
"learning_rate": 4.5061276687654306e-06,
"loss": 0.0175,
"step": 26120
},
{
"grad_norm": 0.33116036653518677,
"learning_rate": 4.4832890860431965e-06,
"loss": 0.0192,
"step": 26130
},
{
"grad_norm": 0.3552340269088745,
"learning_rate": 4.460505810416682e-06,
"loss": 0.015,
"step": 26140
},
{
"grad_norm": 0.415071040391922,
"learning_rate": 4.4377778695696984e-06,
"loss": 0.0227,
"step": 26150
},
{
"grad_norm": 0.3329204022884369,
"learning_rate": 4.415105291118843e-06,
"loss": 0.0131,
"step": 26160
},
{
"grad_norm": 0.30578961968421936,
"learning_rate": 4.392488102613435e-06,
"loss": 0.015,
"step": 26170
},
{
"grad_norm": 0.4557114839553833,
"learning_rate": 4.369926331535473e-06,
"loss": 0.0151,
"step": 26180
},
{
"grad_norm": 0.4233812689781189,
"learning_rate": 4.347420005299668e-06,
"loss": 0.0162,
"step": 26190
},
{
"grad_norm": 0.332687109708786,
"learning_rate": 4.324969151253317e-06,
"loss": 0.0186,
"step": 26200
},
{
"grad_norm": 0.3804143965244293,
"learning_rate": 4.302573796676312e-06,
"loss": 0.0187,
"step": 26210
},
{
"grad_norm": 0.5833508372306824,
"learning_rate": 4.2802339687811385e-06,
"loss": 0.0184,
"step": 26220
},
{
"grad_norm": 0.4092487692832947,
"learning_rate": 4.257949694712771e-06,
"loss": 0.0166,
"step": 26230
},
{
"grad_norm": 0.45519745349884033,
"learning_rate": 4.235721001548726e-06,
"loss": 0.015,
"step": 26240
},
{
"grad_norm": 0.25106358528137207,
"learning_rate": 4.21354791629896e-06,
"loss": 0.0201,
"step": 26250
},
{
"grad_norm": 0.5040042996406555,
"learning_rate": 4.191430465905843e-06,
"loss": 0.0202,
"step": 26260
},
{
"grad_norm": 0.283476322889328,
"learning_rate": 4.169368677244184e-06,
"loss": 0.0125,
"step": 26270
},
{
"grad_norm": 0.3535626232624054,
"learning_rate": 4.147362577121122e-06,
"loss": 0.0255,
"step": 26280
},
{
"grad_norm": 0.2842586934566498,
"learning_rate": 4.125412192276157e-06,
"loss": 0.0253,
"step": 26290
},
{
"grad_norm": 0.3407583236694336,
"learning_rate": 4.10351754938107e-06,
"loss": 0.0185,
"step": 26300
},
{
"grad_norm": 0.3250633478164673,
"learning_rate": 4.081678675039913e-06,
"loss": 0.0285,
"step": 26310
},
{
"grad_norm": 0.43948566913604736,
"learning_rate": 4.0598955957889876e-06,
"loss": 0.0139,
"step": 26320
},
{
"grad_norm": 0.35457393527030945,
"learning_rate": 4.038168338096776e-06,
"loss": 0.0148,
"step": 26330
},
{
"grad_norm": 0.42895036935806274,
"learning_rate": 4.016496928363944e-06,
"loss": 0.0138,
"step": 26340
},
{
"grad_norm": 0.3466548025608063,
"learning_rate": 3.994881392923317e-06,
"loss": 0.0128,
"step": 26350
},
{
"grad_norm": 0.30872589349746704,
"learning_rate": 3.9733217580397945e-06,
"loss": 0.017,
"step": 26360
},
{
"grad_norm": 0.22679181396961212,
"learning_rate": 3.951818049910366e-06,
"loss": 0.0122,
"step": 26370
},
{
"grad_norm": 0.38675087690353394,
"learning_rate": 3.930370294664071e-06,
"loss": 0.0144,
"step": 26380
},
{
"grad_norm": 0.3481721878051758,
"learning_rate": 3.908978518361938e-06,
"loss": 0.0191,
"step": 26390
},
{
"grad_norm": 0.45811954140663147,
"learning_rate": 3.887642746997017e-06,
"loss": 0.0146,
"step": 26400
},
{
"grad_norm": 0.2942539155483246,
"learning_rate": 3.866363006494255e-06,
"loss": 0.0136,
"step": 26410
},
{
"grad_norm": 0.39050185680389404,
"learning_rate": 3.8451393227105725e-06,
"loss": 0.0131,
"step": 26420
},
{
"grad_norm": 0.4212814271450043,
"learning_rate": 3.823971721434727e-06,
"loss": 0.0185,
"step": 26430
},
{
"grad_norm": 0.30072999000549316,
"learning_rate": 3.8028602283873505e-06,
"loss": 0.0154,
"step": 26440
},
{
"grad_norm": 0.3415946364402771,
"learning_rate": 3.781804869220912e-06,
"loss": 0.0155,
"step": 26450
},
{
"grad_norm": 0.2959311306476593,
"learning_rate": 3.760805669519646e-06,
"loss": 0.0145,
"step": 26460
},
{
"grad_norm": 0.41133913397789,
"learning_rate": 3.7398626547995587e-06,
"loss": 0.0135,
"step": 26470
},
{
"grad_norm": 0.48112842440605164,
"learning_rate": 3.7189758505083995e-06,
"loss": 0.014,
"step": 26480
},
{
"grad_norm": 0.3386521637439728,
"learning_rate": 3.6981452820255837e-06,
"loss": 0.0149,
"step": 26490
},
{
"grad_norm": 0.4371505081653595,
"learning_rate": 3.6773709746622363e-06,
"loss": 0.0164,
"step": 26500
},
{
"grad_norm": 0.24571537971496582,
"learning_rate": 3.6566529536610995e-06,
"loss": 0.0253,
"step": 26510
},
{
"grad_norm": 0.2777116000652313,
"learning_rate": 3.635991244196513e-06,
"loss": 0.0159,
"step": 26520
},
{
"grad_norm": 0.3428117036819458,
"learning_rate": 3.615385871374405e-06,
"loss": 0.0174,
"step": 26530
},
{
"grad_norm": 0.4435311555862427,
"learning_rate": 3.5948368602322367e-06,
"loss": 0.0148,
"step": 26540
},
{
"grad_norm": 0.6348838806152344,
"learning_rate": 3.5743442357390187e-06,
"loss": 0.0144,
"step": 26550
},
{
"grad_norm": 0.4783121347427368,
"learning_rate": 3.5539080227951937e-06,
"loss": 0.0148,
"step": 26560
},
{
"grad_norm": 0.2093973010778427,
"learning_rate": 3.533528246232709e-06,
"loss": 0.0115,
"step": 26570
},
{
"grad_norm": 0.3697986602783203,
"learning_rate": 3.5132049308149117e-06,
"loss": 0.0113,
"step": 26580
},
{
"grad_norm": 0.4290424883365631,
"learning_rate": 3.4929381012365425e-06,
"loss": 0.0145,
"step": 26590
},
{
"grad_norm": 0.32248929142951965,
"learning_rate": 3.472727782123697e-06,
"loss": 0.0177,
"step": 26600
},
{
"grad_norm": 0.28691598773002625,
"learning_rate": 3.4525739980338425e-06,
"loss": 0.0129,
"step": 26610
},
{
"grad_norm": 0.30068615078926086,
"learning_rate": 3.432476773455712e-06,
"loss": 0.0204,
"step": 26620
},
{
"grad_norm": 0.36392855644226074,
"learning_rate": 3.4124361328093378e-06,
"loss": 0.0156,
"step": 26630
},
{
"grad_norm": 0.3685069680213928,
"learning_rate": 3.392452100445975e-06,
"loss": 0.0136,
"step": 26640
},
{
"grad_norm": 0.26294007897377014,
"learning_rate": 3.3725247006481163e-06,
"loss": 0.0109,
"step": 26650
},
{
"grad_norm": 0.3966827392578125,
"learning_rate": 3.3526539576294313e-06,
"loss": 0.0187,
"step": 26660
},
{
"grad_norm": 0.35300779342651367,
"learning_rate": 3.332839895534745e-06,
"loss": 0.0141,
"step": 26670
},
{
"grad_norm": 0.3720976710319519,
"learning_rate": 3.3130825384400155e-06,
"loss": 0.0147,
"step": 26680
},
{
"grad_norm": 0.30224716663360596,
"learning_rate": 3.293381910352278e-06,
"loss": 0.0155,
"step": 26690
},
{
"grad_norm": 0.4918513000011444,
"learning_rate": 3.273738035209678e-06,
"loss": 0.025,
"step": 26700
},
{
"grad_norm": 0.430819034576416,
"learning_rate": 3.254150936881356e-06,
"loss": 0.0185,
"step": 26710
},
{
"grad_norm": 0.3752150535583496,
"learning_rate": 3.234620639167496e-06,
"loss": 0.012,
"step": 26720
},
{
"grad_norm": 0.3229324519634247,
"learning_rate": 3.2151471657992482e-06,
"loss": 0.0112,
"step": 26730
},
{
"grad_norm": 0.40031006932258606,
"learning_rate": 3.195730540438718e-06,
"loss": 0.014,
"step": 26740
},
{
"grad_norm": 0.42470359802246094,
"learning_rate": 3.1763707866789336e-06,
"loss": 0.019,
"step": 26750
},
{
"grad_norm": 0.2860279679298401,
"learning_rate": 3.1570679280438165e-06,
"loss": 0.0155,
"step": 26760
},
{
"grad_norm": 0.3643791675567627,
"learning_rate": 3.1378219879881767e-06,
"loss": 0.0133,
"step": 26770
},
{
"grad_norm": 0.2748326361179352,
"learning_rate": 3.11863298989763e-06,
"loss": 0.0169,
"step": 26780
},
{
"grad_norm": 0.4636596143245697,
"learning_rate": 3.0995009570886303e-06,
"loss": 0.017,
"step": 26790
},
{
"grad_norm": 0.396058052778244,
"learning_rate": 3.0804259128083936e-06,
"loss": 0.0141,
"step": 26800
},
{
"grad_norm": 0.21432220935821533,
"learning_rate": 3.06140788023489e-06,
"loss": 0.0124,
"step": 26810
},
{
"grad_norm": 0.5208573937416077,
"learning_rate": 3.042446882476846e-06,
"loss": 0.0136,
"step": 26820
},
{
"grad_norm": 0.3299054205417633,
"learning_rate": 3.023542942573643e-06,
"loss": 0.0137,
"step": 26830
},
{
"grad_norm": 0.4935120940208435,
"learning_rate": 3.004696083495351e-06,
"loss": 0.019,
"step": 26840
},
{
"grad_norm": 0.4263279438018799,
"learning_rate": 2.98590632814269e-06,
"loss": 0.0163,
"step": 26850
},
{
"grad_norm": 0.2607741355895996,
"learning_rate": 2.9671736993469702e-06,
"loss": 0.0133,
"step": 26860
},
{
"grad_norm": 0.5244705080986023,
"learning_rate": 2.9484982198701216e-06,
"loss": 0.0128,
"step": 26870
},
{
"grad_norm": 0.4485233724117279,
"learning_rate": 2.929879912404604e-06,
"loss": 0.0165,
"step": 26880
},
{
"grad_norm": 0.2098061591386795,
"learning_rate": 2.9113187995734147e-06,
"loss": 0.0134,
"step": 26890
},
{
"grad_norm": 0.5017186999320984,
"learning_rate": 2.8928149039300524e-06,
"loss": 0.0129,
"step": 26900
},
{
"grad_norm": 0.25786396861076355,
"learning_rate": 2.8743682479584975e-06,
"loss": 0.0184,
"step": 26910
},
{
"grad_norm": 0.44982466101646423,
"learning_rate": 2.8559788540731824e-06,
"loss": 0.0169,
"step": 26920
},
{
"grad_norm": 0.4634377062320709,
"learning_rate": 2.8376467446189493e-06,
"loss": 0.0163,
"step": 26930
},
{
"grad_norm": 0.5419193506240845,
"learning_rate": 2.8193719418710406e-06,
"loss": 0.0285,
"step": 26940
},
{
"grad_norm": 0.21374577283859253,
"learning_rate": 2.8011544680350667e-06,
"loss": 0.0166,
"step": 26950
},
{
"grad_norm": 0.28059569001197815,
"learning_rate": 2.7829943452469753e-06,
"loss": 0.0129,
"step": 26960
},
{
"grad_norm": 0.381965309381485,
"learning_rate": 2.7648915955730213e-06,
"loss": 0.0179,
"step": 26970
},
{
"grad_norm": 0.20380960404872894,
"learning_rate": 2.746846241009765e-06,
"loss": 0.0095,
"step": 26980
},
{
"grad_norm": 0.4158795475959778,
"learning_rate": 2.7288583034839945e-06,
"loss": 0.015,
"step": 26990
},
{
"grad_norm": 0.4324040412902832,
"learning_rate": 2.7109278048527752e-06,
"loss": 0.0119,
"step": 27000
},
{
"grad_norm": 0.3038398027420044,
"learning_rate": 2.6930547669033413e-06,
"loss": 0.0103,
"step": 27010
},
{
"grad_norm": 0.38014689087867737,
"learning_rate": 2.675239211353109e-06,
"loss": 0.0139,
"step": 27020
},
{
"grad_norm": 0.2803562581539154,
"learning_rate": 2.6574811598496786e-06,
"loss": 0.011,
"step": 27030
},
{
"grad_norm": 0.2650359570980072,
"learning_rate": 2.6397806339707456e-06,
"loss": 0.0224,
"step": 27040
},
{
"grad_norm": 0.5116207599639893,
"learning_rate": 2.6221376552241217e-06,
"loss": 0.0122,
"step": 27050
},
{
"grad_norm": 0.26263681054115295,
"learning_rate": 2.604552245047681e-06,
"loss": 0.0113,
"step": 27060
},
{
"grad_norm": 0.28292402625083923,
"learning_rate": 2.587024424809359e-06,
"loss": 0.0124,
"step": 27070
},
{
"grad_norm": 0.3794916868209839,
"learning_rate": 2.5695542158071186e-06,
"loss": 0.0147,
"step": 27080
},
{
"grad_norm": 0.21366499364376068,
"learning_rate": 2.5521416392689068e-06,
"loss": 0.0107,
"step": 27090
},
{
"grad_norm": 0.3498448431491852,
"learning_rate": 2.5347867163526384e-06,
"loss": 0.014,
"step": 27100
},
{
"grad_norm": 0.468476802110672,
"learning_rate": 2.517489468146189e-06,
"loss": 0.0175,
"step": 27110
},
{
"grad_norm": 0.4153870940208435,
"learning_rate": 2.500249915667341e-06,
"loss": 0.0153,
"step": 27120
},
{
"grad_norm": 0.36520376801490784,
"learning_rate": 2.4830680798637817e-06,
"loss": 0.0113,
"step": 27130
},
{
"grad_norm": 0.441702663898468,
"learning_rate": 2.465943981613056e-06,
"loss": 0.0209,
"step": 27140
},
{
"grad_norm": 0.3706674575805664,
"learning_rate": 2.448877641722569e-06,
"loss": 0.0124,
"step": 27150
},
{
"grad_norm": 0.24800175428390503,
"learning_rate": 2.431869080929522e-06,
"loss": 0.0145,
"step": 27160
},
{
"grad_norm": 0.343250036239624,
"learning_rate": 2.4149183199009216e-06,
"loss": 0.0165,
"step": 27170
},
{
"grad_norm": 0.561669111251831,
"learning_rate": 2.3980253792335427e-06,
"loss": 0.021,
"step": 27180
},
{
"grad_norm": 0.33100470900535583,
"learning_rate": 2.381190279453899e-06,
"loss": 0.0168,
"step": 27190
},
{
"grad_norm": 0.46022120118141174,
"learning_rate": 2.364413041018232e-06,
"loss": 0.0204,
"step": 27200
},
{
"grad_norm": 0.29592642188072205,
"learning_rate": 2.347693684312463e-06,
"loss": 0.0124,
"step": 27210
},
{
"grad_norm": 0.3268638253211975,
"learning_rate": 2.331032229652186e-06,
"loss": 0.0113,
"step": 27220
},
{
"grad_norm": 0.46540743112564087,
"learning_rate": 2.314428697282628e-06,
"loss": 0.0137,
"step": 27230
},
{
"grad_norm": 0.292393296957016,
"learning_rate": 2.297883107378673e-06,
"loss": 0.0127,
"step": 27240
},
{
"grad_norm": 0.5551952123641968,
"learning_rate": 2.2813954800447513e-06,
"loss": 0.0209,
"step": 27250
},
{
"grad_norm": 0.345243901014328,
"learning_rate": 2.2649658353148974e-06,
"loss": 0.0102,
"step": 27260
},
{
"grad_norm": 0.3113568425178528,
"learning_rate": 2.2485941931526645e-06,
"loss": 0.0096,
"step": 27270
},
{
"grad_norm": 0.41139549016952515,
"learning_rate": 2.232280573451151e-06,
"loss": 0.0132,
"step": 27280
},
{
"grad_norm": 0.32451415061950684,
"learning_rate": 2.216024996032945e-06,
"loss": 0.0104,
"step": 27290
},
{
"grad_norm": 0.3578490614891052,
"learning_rate": 2.1998274806501074e-06,
"loss": 0.0107,
"step": 27300
},
{
"grad_norm": 0.5116119384765625,
"learning_rate": 2.183688046984139e-06,
"loss": 0.016,
"step": 27310
},
{
"grad_norm": 0.31192484498023987,
"learning_rate": 2.167606714645981e-06,
"loss": 0.0125,
"step": 27320
},
{
"grad_norm": 0.31950926780700684,
"learning_rate": 2.151583503175958e-06,
"loss": 0.0089,
"step": 27330
},
{
"grad_norm": 0.3920804560184479,
"learning_rate": 2.1356184320437957e-06,
"loss": 0.0127,
"step": 27340
},
{
"grad_norm": 0.3101117014884949,
"learning_rate": 2.119711520648554e-06,
"loss": 0.0155,
"step": 27350
},
{
"grad_norm": 0.32201501727104187,
"learning_rate": 2.103862788318628e-06,
"loss": 0.0149,
"step": 27360
},
{
"grad_norm": 0.2699914872646332,
"learning_rate": 2.088072254311729e-06,
"loss": 0.0118,
"step": 27370
},
{
"grad_norm": 0.3497477173805237,
"learning_rate": 2.0723399378148435e-06,
"loss": 0.0137,
"step": 27380
},
{
"grad_norm": 0.36691927909851074,
"learning_rate": 2.0566658579442065e-06,
"loss": 0.0132,
"step": 27390
},
{
"grad_norm": 0.3882042467594147,
"learning_rate": 2.0410500337453176e-06,
"loss": 0.0105,
"step": 27400
},
{
"grad_norm": 0.1790439486503601,
"learning_rate": 2.0254924841928645e-06,
"loss": 0.0168,
"step": 27410
},
{
"grad_norm": 0.4932402968406677,
"learning_rate": 2.009993228190754e-06,
"loss": 0.0127,
"step": 27420
},
{
"grad_norm": 0.5637838840484619,
"learning_rate": 1.9945522845720323e-06,
"loss": 0.0145,
"step": 27430
},
{
"grad_norm": 0.38749492168426514,
"learning_rate": 1.9791696720988963e-06,
"loss": 0.0205,
"step": 27440
},
{
"grad_norm": 0.37772607803344727,
"learning_rate": 1.9638454094626833e-06,
"loss": 0.0145,
"step": 27450
},
{
"grad_norm": 0.2821347117424011,
"learning_rate": 1.9485795152838104e-06,
"loss": 0.014,
"step": 27460
},
{
"grad_norm": 0.33716022968292236,
"learning_rate": 1.933372008111778e-06,
"loss": 0.0217,
"step": 27470
},
{
"grad_norm": 0.4063624143600464,
"learning_rate": 1.918222906425143e-06,
"loss": 0.0156,
"step": 27480
},
{
"grad_norm": 0.4477507770061493,
"learning_rate": 1.903132228631488e-06,
"loss": 0.0151,
"step": 27490
},
{
"grad_norm": 0.31133583188056946,
"learning_rate": 1.8880999930674215e-06,
"loss": 0.0219,
"step": 27500
},
{
"grad_norm": 0.4767483174800873,
"learning_rate": 1.8731262179985166e-06,
"loss": 0.0195,
"step": 27510
},
{
"grad_norm": 0.3435777425765991,
"learning_rate": 1.8582109216193244e-06,
"loss": 0.0203,
"step": 27520
},
{
"grad_norm": 0.41874992847442627,
"learning_rate": 1.8433541220533368e-06,
"loss": 0.0169,
"step": 27530
},
{
"grad_norm": 0.30250054597854614,
"learning_rate": 1.8285558373529577e-06,
"loss": 0.011,
"step": 27540
},
{
"grad_norm": 0.3175620138645172,
"learning_rate": 1.8138160854995145e-06,
"loss": 0.0165,
"step": 27550
},
{
"grad_norm": 0.231672003865242,
"learning_rate": 1.7991348844031864e-06,
"loss": 0.0116,
"step": 27560
},
{
"grad_norm": 0.48176318407058716,
"learning_rate": 1.78451225190302e-06,
"loss": 0.0139,
"step": 27570
},
{
"grad_norm": 0.22513213753700256,
"learning_rate": 1.7699482057668914e-06,
"loss": 0.0119,
"step": 27580
},
{
"grad_norm": 0.37919580936431885,
"learning_rate": 1.7554427636914895e-06,
"loss": 0.0126,
"step": 27590
},
{
"grad_norm": 0.3356527090072632,
"learning_rate": 1.7409959433022871e-06,
"loss": 0.0165,
"step": 27600
},
{
"grad_norm": 0.45707541704177856,
"learning_rate": 1.7266077621535481e-06,
"loss": 0.0197,
"step": 27610
},
{
"grad_norm": 0.40554526448249817,
"learning_rate": 1.7122782377282597e-06,
"loss": 0.0136,
"step": 27620
},
{
"grad_norm": 0.27840882539749146,
"learning_rate": 1.6980073874381497e-06,
"loss": 0.0114,
"step": 27630
},
{
"grad_norm": 0.3260006606578827,
"learning_rate": 1.6837952286236413e-06,
"loss": 0.0143,
"step": 27640
},
{
"grad_norm": 0.3506390452384949,
"learning_rate": 1.6696417785538488e-06,
"loss": 0.0248,
"step": 27650
},
{
"grad_norm": 0.3581506907939911,
"learning_rate": 1.6555470544265538e-06,
"loss": 0.0238,
"step": 27660
},
{
"grad_norm": 0.3052370548248291,
"learning_rate": 1.6415110733681738e-06,
"loss": 0.0205,
"step": 27670
},
{
"grad_norm": 0.40939122438430786,
"learning_rate": 1.6275338524337436e-06,
"loss": 0.0155,
"step": 27680
},
{
"grad_norm": 0.3108097016811371,
"learning_rate": 1.6136154086069056e-06,
"loss": 0.0116,
"step": 27690
},
{
"grad_norm": 0.49243268370628357,
"learning_rate": 1.5997557587998868e-06,
"loss": 0.0173,
"step": 27700
},
{
"grad_norm": 0.4166024625301361,
"learning_rate": 1.5859549198534606e-06,
"loss": 0.0102,
"step": 27710
},
{
"grad_norm": 0.2429506629705429,
"learning_rate": 1.572212908536963e-06,
"loss": 0.0126,
"step": 27720
},
{
"grad_norm": 0.4066307842731476,
"learning_rate": 1.5585297415482202e-06,
"loss": 0.0114,
"step": 27730
},
{
"grad_norm": 0.24735142290592194,
"learning_rate": 1.5449054355135717e-06,
"loss": 0.0138,
"step": 27740
},
{
"grad_norm": 0.3906565010547638,
"learning_rate": 1.5313400069878415e-06,
"loss": 0.0144,
"step": 27750
},
{
"grad_norm": 0.354854017496109,
"learning_rate": 1.5178334724542887e-06,
"loss": 0.0131,
"step": 27760
},
{
"grad_norm": 0.46871986985206604,
"learning_rate": 1.504385848324641e-06,
"loss": 0.0252,
"step": 27770
},
{
"grad_norm": 0.4615026116371155,
"learning_rate": 1.490997150939033e-06,
"loss": 0.0163,
"step": 27780
},
{
"grad_norm": 0.41691991686820984,
"learning_rate": 1.4776673965659793e-06,
"loss": 0.018,
"step": 27790
},
{
"grad_norm": 0.4448513686656952,
"learning_rate": 1.4643966014023957e-06,
"loss": 0.0167,
"step": 27800
},
{
"grad_norm": 0.47015318274497986,
"learning_rate": 1.4511847815735503e-06,
"loss": 0.0205,
"step": 27810
},
{
"grad_norm": 0.3004237711429596,
"learning_rate": 1.438031953133051e-06,
"loss": 0.0105,
"step": 27820
},
{
"grad_norm": 0.35496529936790466,
"learning_rate": 1.4249381320628197e-06,
"loss": 0.0143,
"step": 27830
},
{
"grad_norm": 0.2744775712490082,
"learning_rate": 1.4119033342730902e-06,
"loss": 0.0182,
"step": 27840
},
{
"grad_norm": 0.2278398722410202,
"learning_rate": 1.3989275756023656e-06,
"loss": 0.011,
"step": 27850
},
{
"grad_norm": 0.24998895823955536,
"learning_rate": 1.386010871817417e-06,
"loss": 0.012,
"step": 27860
},
{
"grad_norm": 0.35584744811058044,
"learning_rate": 1.3731532386132616e-06,
"loss": 0.0111,
"step": 27870
},
{
"grad_norm": 0.5122441649436951,
"learning_rate": 1.360354691613136e-06,
"loss": 0.0317,
"step": 27880
},
{
"grad_norm": 0.33008939027786255,
"learning_rate": 1.3476152463684776e-06,
"loss": 0.0149,
"step": 27890
},
{
"grad_norm": 0.26000121235847473,
"learning_rate": 1.3349349183589155e-06,
"loss": 0.0118,
"step": 27900
},
{
"grad_norm": 0.3235805630683899,
"learning_rate": 1.3223137229922356e-06,
"loss": 0.0136,
"step": 27910
},
{
"grad_norm": 0.34234243631362915,
"learning_rate": 1.3097516756043981e-06,
"loss": 0.0144,
"step": 27920
},
{
"grad_norm": 0.38213762640953064,
"learning_rate": 1.29724879145946e-06,
"loss": 0.0158,
"step": 27930
},
{
"grad_norm": 0.2693699896335602,
"learning_rate": 1.284805085749613e-06,
"loss": 0.0149,
"step": 27940
},
{
"grad_norm": 0.38070452213287354,
"learning_rate": 1.2724205735951288e-06,
"loss": 0.0155,
"step": 27950
},
{
"grad_norm": 0.3319533169269562,
"learning_rate": 1.2600952700443591e-06,
"loss": 0.0244,
"step": 27960
},
{
"grad_norm": 0.43223685026168823,
"learning_rate": 1.247829190073707e-06,
"loss": 0.0148,
"step": 27970
},
{
"grad_norm": 0.30870532989501953,
"learning_rate": 1.2356223485876173e-06,
"loss": 0.0095,
"step": 27980
},
{
"grad_norm": 0.40724262595176697,
"learning_rate": 1.2234747604185526e-06,
"loss": 0.0147,
"step": 27990
},
{
"grad_norm": 0.3512396514415741,
"learning_rate": 1.2113864403269836e-06,
"loss": 0.0176,
"step": 28000
},
{
"grad_norm": 0.22693540155887604,
"learning_rate": 1.1993574030013555e-06,
"loss": 0.013,
"step": 28010
},
{
"grad_norm": 0.5310458540916443,
"learning_rate": 1.1873876630580816e-06,
"loss": 0.0189,
"step": 28020
},
{
"grad_norm": 0.5604068040847778,
"learning_rate": 1.1754772350415278e-06,
"loss": 0.0152,
"step": 28030
},
{
"grad_norm": 0.3594236671924591,
"learning_rate": 1.163626133423984e-06,
"loss": 0.0147,
"step": 28040
},
{
"grad_norm": 0.22398263216018677,
"learning_rate": 1.1518343726056591e-06,
"loss": 0.0121,
"step": 28050
},
{
"grad_norm": 0.3321385383605957,
"learning_rate": 1.1401019669146474e-06,
"loss": 0.0128,
"step": 28060
},
{
"grad_norm": 0.3970191776752472,
"learning_rate": 1.128428930606934e-06,
"loss": 0.0135,
"step": 28070
},
{
"grad_norm": 0.24080753326416016,
"learning_rate": 1.1168152778663621e-06,
"loss": 0.0111,
"step": 28080
},
{
"grad_norm": 0.40169256925582886,
"learning_rate": 1.10526102280461e-06,
"loss": 0.0152,
"step": 28090
},
{
"grad_norm": 0.2242729216814041,
"learning_rate": 1.0937661794611864e-06,
"loss": 0.0125,
"step": 28100
},
{
"grad_norm": 0.38252270221710205,
"learning_rate": 1.082330761803413e-06,
"loss": 0.0132,
"step": 28110
},
{
"grad_norm": 0.4267057478427887,
"learning_rate": 1.0709547837263966e-06,
"loss": 0.0146,
"step": 28120
},
{
"grad_norm": 0.24338865280151367,
"learning_rate": 1.059638259053025e-06,
"loss": 0.0121,
"step": 28130
},
{
"grad_norm": 0.4436602294445038,
"learning_rate": 1.0483812015339434e-06,
"loss": 0.0135,
"step": 28140
},
{
"grad_norm": 0.5249113440513611,
"learning_rate": 1.037183624847543e-06,
"loss": 0.0224,
"step": 28150
},
{
"grad_norm": 0.33951127529144287,
"learning_rate": 1.0260455425999238e-06,
"loss": 0.0143,
"step": 28160
},
{
"grad_norm": 0.28513088822364807,
"learning_rate": 1.0149669683249208e-06,
"loss": 0.0114,
"step": 28170
},
{
"grad_norm": 0.40154075622558594,
"learning_rate": 1.003947915484027e-06,
"loss": 0.0199,
"step": 28180
},
{
"grad_norm": 0.5109683275222778,
"learning_rate": 9.92988397466449e-07,
"loss": 0.0328,
"step": 28190
},
{
"grad_norm": 0.5050356388092041,
"learning_rate": 9.820884275890286e-07,
"loss": 0.0171,
"step": 28200
},
{
"grad_norm": 0.25841373205184937,
"learning_rate": 9.712480190962546e-07,
"loss": 0.0142,
"step": 28210
},
{
"grad_norm": 0.36717575788497925,
"learning_rate": 9.604671851602464e-07,
"loss": 0.0132,
"step": 28220
},
{
"grad_norm": 0.5295617580413818,
"learning_rate": 9.497459388807306e-07,
"loss": 0.0139,
"step": 28230
},
{
"grad_norm": 0.299816757440567,
"learning_rate": 9.390842932850364e-07,
"loss": 0.011,
"step": 28240
},
{
"grad_norm": 0.573762059211731,
"learning_rate": 9.28482261328073e-07,
"loss": 0.0216,
"step": 28250
},
{
"grad_norm": 0.42940637469291687,
"learning_rate": 9.179398558923025e-07,
"loss": 0.0179,
"step": 28260
},
{
"grad_norm": 0.3064630627632141,
"learning_rate": 9.074570897877388e-07,
"loss": 0.0116,
"step": 28270
},
{
"grad_norm": 0.4707602262496948,
"learning_rate": 8.970339757519375e-07,
"loss": 0.0162,
"step": 28280
},
{
"grad_norm": 0.24272644519805908,
"learning_rate": 8.866705264499619e-07,
"loss": 0.0119,
"step": 28290
},
{
"grad_norm": 0.3467409312725067,
"learning_rate": 8.763667544743836e-07,
"loss": 0.0126,
"step": 28300
},
{
"grad_norm": 0.4798063635826111,
"learning_rate": 8.661226723452543e-07,
"loss": 0.0159,
"step": 28310
},
{
"grad_norm": 0.3833737373352051,
"learning_rate": 8.559382925101001e-07,
"loss": 0.0131,
"step": 28320
},
{
"grad_norm": 0.27063193917274475,
"learning_rate": 8.458136273438943e-07,
"loss": 0.0181,
"step": 28330
},
{
"grad_norm": 0.35476040840148926,
"learning_rate": 8.357486891490795e-07,
"loss": 0.0154,
"step": 28340
},
{
"grad_norm": 0.4162757098674774,
"learning_rate": 8.257434901554895e-07,
"loss": 0.0145,
"step": 28350
},
{
"grad_norm": 0.2247340977191925,
"learning_rate": 8.157980425203937e-07,
"loss": 0.0113,
"step": 28360
},
{
"grad_norm": 0.3919512927532196,
"learning_rate": 8.059123583284367e-07,
"loss": 0.0156,
"step": 28370
},
{
"grad_norm": 0.30795955657958984,
"learning_rate": 7.960864495916654e-07,
"loss": 0.0142,
"step": 28380
},
{
"grad_norm": 0.31409621238708496,
"learning_rate": 7.863203282494847e-07,
"loss": 0.019,
"step": 28390
},
{
"grad_norm": 0.3321409225463867,
"learning_rate": 7.766140061686522e-07,
"loss": 0.0124,
"step": 28400
},
{
"grad_norm": 0.4122728407382965,
"learning_rate": 7.669674951432615e-07,
"loss": 0.0149,
"step": 28410
},
{
"grad_norm": 0.3834235370159149,
"learning_rate": 7.573808068947363e-07,
"loss": 0.0201,
"step": 28420
},
{
"grad_norm": 0.4651888310909271,
"learning_rate": 7.478539530718087e-07,
"loss": 0.0128,
"step": 28430
},
{
"grad_norm": 0.3024914562702179,
"learning_rate": 7.383869452504965e-07,
"loss": 0.013,
"step": 28440
},
{
"grad_norm": 0.2285594642162323,
"learning_rate": 7.289797949341204e-07,
"loss": 0.013,
"step": 28450
},
{
"grad_norm": 0.3532334268093109,
"learning_rate": 7.196325135532423e-07,
"loss": 0.0141,
"step": 28460
},
{
"grad_norm": 0.4982225298881531,
"learning_rate": 7.10345112465699e-07,
"loss": 0.0137,
"step": 28470
},
{
"grad_norm": 0.41040676832199097,
"learning_rate": 7.011176029565525e-07,
"loss": 0.0202,
"step": 28480
},
{
"grad_norm": 0.3807709515094757,
"learning_rate": 6.919499962381004e-07,
"loss": 0.01,
"step": 28490
},
{
"grad_norm": 0.33622366189956665,
"learning_rate": 6.828423034498488e-07,
"loss": 0.0146,
"step": 28500
},
{
"grad_norm": 0.41057056188583374,
"learning_rate": 6.737945356585007e-07,
"loss": 0.0133,
"step": 28510
},
{
"grad_norm": 0.2999885678291321,
"learning_rate": 6.648067038579508e-07,
"loss": 0.0188,
"step": 28520
},
{
"grad_norm": 0.4609304666519165,
"learning_rate": 6.558788189692578e-07,
"loss": 0.0141,
"step": 28530
},
{
"grad_norm": 0.19212481379508972,
"learning_rate": 6.470108918406492e-07,
"loss": 0.0132,
"step": 28540
},
{
"grad_norm": 0.4477176070213318,
"learning_rate": 6.382029332474893e-07,
"loss": 0.0194,
"step": 28550
},
{
"grad_norm": 0.3561260998249054,
"learning_rate": 6.294549538922778e-07,
"loss": 0.0168,
"step": 28560
},
{
"grad_norm": 0.33151739835739136,
"learning_rate": 6.207669644046344e-07,
"loss": 0.0231,
"step": 28570
},
{
"grad_norm": 0.2881300747394562,
"learning_rate": 6.121389753412865e-07,
"loss": 0.0191,
"step": 28580
},
{
"grad_norm": 0.2922115623950958,
"learning_rate": 6.035709971860592e-07,
"loss": 0.0119,
"step": 28590
},
{
"grad_norm": 0.19373643398284912,
"learning_rate": 5.950630403498469e-07,
"loss": 0.0112,
"step": 28600
},
{
"grad_norm": 0.36686837673187256,
"learning_rate": 5.866151151706189e-07,
"loss": 0.0268,
"step": 28610
},
{
"grad_norm": 0.305568128824234,
"learning_rate": 5.782272319134086e-07,
"loss": 0.017,
"step": 28620
},
{
"grad_norm": 0.20490911602973938,
"learning_rate": 5.698994007702796e-07,
"loss": 0.0112,
"step": 28630
},
{
"grad_norm": 0.17658255994319916,
"learning_rate": 5.616316318603321e-07,
"loss": 0.0153,
"step": 28640
},
{
"grad_norm": 0.22727838158607483,
"learning_rate": 5.534239352296799e-07,
"loss": 0.0147,
"step": 28650
},
{
"grad_norm": 0.45215991139411926,
"learning_rate": 5.452763208514621e-07,
"loss": 0.0164,
"step": 28660
},
{
"grad_norm": 0.37515363097190857,
"learning_rate": 5.371887986257873e-07,
"loss": 0.0117,
"step": 28670
},
{
"grad_norm": 0.3815527558326721,
"learning_rate": 5.291613783797611e-07,
"loss": 0.0155,
"step": 28680
},
{
"grad_norm": 0.3479291498661041,
"learning_rate": 5.211940698674534e-07,
"loss": 0.0156,
"step": 28690
},
{
"grad_norm": 0.27868950366973877,
"learning_rate": 5.132868827698978e-07,
"loss": 0.0161,
"step": 28700
},
{
"grad_norm": 0.3692083954811096,
"learning_rate": 5.054398266950755e-07,
"loss": 0.0135,
"step": 28710
},
{
"grad_norm": 0.23220574855804443,
"learning_rate": 4.976529111778872e-07,
"loss": 0.0169,
"step": 28720
},
{
"grad_norm": 0.30263152718544006,
"learning_rate": 4.899261456801862e-07,
"loss": 0.03,
"step": 28730
},
{
"grad_norm": 0.20539313554763794,
"learning_rate": 4.822595395907126e-07,
"loss": 0.0098,
"step": 28740
},
{
"grad_norm": 0.3712899088859558,
"learning_rate": 4.7465310222510886e-07,
"loss": 0.0139,
"step": 28750
},
{
"grad_norm": 0.23300428688526154,
"learning_rate": 4.6710684282593175e-07,
"loss": 0.0225,
"step": 28760
},
{
"grad_norm": 0.313564658164978,
"learning_rate": 4.5962077056257993e-07,
"loss": 0.0113,
"step": 28770
},
{
"grad_norm": 0.3722042441368103,
"learning_rate": 4.521948945313492e-07,
"loss": 0.0151,
"step": 28780
},
{
"grad_norm": 0.27745521068573,
"learning_rate": 4.4482922375537195e-07,
"loss": 0.0151,
"step": 28790
},
{
"grad_norm": 0.26778700947761536,
"learning_rate": 4.375237671846333e-07,
"loss": 0.0133,
"step": 28800
},
{
"grad_norm": 0.20752976834774017,
"learning_rate": 4.302785336959547e-07,
"loss": 0.0109,
"step": 28810
},
{
"grad_norm": 0.27620404958724976,
"learning_rate": 4.2309353209297744e-07,
"loss": 0.0155,
"step": 28820
},
{
"grad_norm": 0.3235064446926117,
"learning_rate": 4.159687711061566e-07,
"loss": 0.0125,
"step": 28830
},
{
"grad_norm": 0.43429240584373474,
"learning_rate": 4.0890425939275055e-07,
"loss": 0.0113,
"step": 28840
},
{
"grad_norm": 0.5757569670677185,
"learning_rate": 4.0190000553679827e-07,
"loss": 0.0185,
"step": 28850
},
{
"grad_norm": 0.30419468879699707,
"learning_rate": 3.9495601804913627e-07,
"loss": 0.0139,
"step": 28860
},
{
"grad_norm": 0.24808485805988312,
"learning_rate": 3.880723053673652e-07,
"loss": 0.0159,
"step": 28870
},
{
"grad_norm": 0.30856001377105713,
"learning_rate": 3.812488758558386e-07,
"loss": 0.0221,
"step": 28880
},
{
"grad_norm": 0.34450119733810425,
"learning_rate": 3.744857378056743e-07,
"loss": 0.013,
"step": 28890
},
{
"grad_norm": 0.3464754819869995,
"learning_rate": 3.677828994347154e-07,
"loss": 0.0131,
"step": 28900
},
{
"grad_norm": 0.2846100628376007,
"learning_rate": 3.61140368887547e-07,
"loss": 0.0139,
"step": 28910
},
{
"grad_norm": 0.2649233937263489,
"learning_rate": 3.545581542354681e-07,
"loss": 0.0117,
"step": 28920
},
{
"grad_norm": 0.32452788949012756,
"learning_rate": 3.480362634764922e-07,
"loss": 0.0126,
"step": 28930
},
{
"grad_norm": 0.4297107458114624,
"learning_rate": 3.4157470453533014e-07,
"loss": 0.015,
"step": 28940
},
{
"grad_norm": 0.2148803025484085,
"learning_rate": 3.3517348526339034e-07,
"loss": 0.0178,
"step": 28950
},
{
"grad_norm": 0.2583885192871094,
"learning_rate": 3.288326134387454e-07,
"loss": 0.0123,
"step": 28960
},
{
"grad_norm": 0.24354404211044312,
"learning_rate": 3.225520967661655e-07,
"loss": 0.0148,
"step": 28970
},
{
"grad_norm": 0.27437496185302734,
"learning_rate": 3.163319428770628e-07,
"loss": 0.0114,
"step": 28980
},
{
"grad_norm": 0.4009459316730499,
"learning_rate": 3.1017215932951374e-07,
"loss": 0.0133,
"step": 28990
},
{
"grad_norm": 0.42472630739212036,
"learning_rate": 3.040727536082366e-07,
"loss": 0.0146,
"step": 29000
},
{
"grad_norm": 0.2857051193714142,
"learning_rate": 2.980337331245864e-07,
"loss": 0.0144,
"step": 29010
},
{
"grad_norm": 0.1988508403301239,
"learning_rate": 2.9205510521653213e-07,
"loss": 0.0129,
"step": 29020
},
{
"grad_norm": 0.2253413200378418,
"learning_rate": 2.86136877148685e-07,
"loss": 0.014,
"step": 29030
},
{
"grad_norm": 0.44171422719955444,
"learning_rate": 2.8027905611223704e-07,
"loss": 0.019,
"step": 29040
},
{
"grad_norm": 0.2528877556324005,
"learning_rate": 2.7448164922500573e-07,
"loss": 0.0158,
"step": 29050
},
{
"grad_norm": 0.4307449758052826,
"learning_rate": 2.687446635313784e-07,
"loss": 0.0156,
"step": 29060
},
{
"grad_norm": 0.28007540106773376,
"learning_rate": 2.630681060023343e-07,
"loss": 0.0151,
"step": 29070
},
{
"grad_norm": 0.27351516485214233,
"learning_rate": 2.5745198353542833e-07,
"loss": 0.0126,
"step": 29080
},
{
"grad_norm": 0.384237676858902,
"learning_rate": 2.518963029547794e-07,
"loss": 0.0116,
"step": 29090
},
{
"grad_norm": 0.29920491576194763,
"learning_rate": 2.464010710110598e-07,
"loss": 0.015,
"step": 29100
},
{
"grad_norm": 0.22279126942157745,
"learning_rate": 2.4096629438150054e-07,
"loss": 0.0104,
"step": 29110
},
{
"grad_norm": 0.23430466651916504,
"learning_rate": 2.3559197966985802e-07,
"loss": 0.0124,
"step": 29120
},
{
"grad_norm": 0.3163418471813202,
"learning_rate": 2.3027813340644188e-07,
"loss": 0.0095,
"step": 29130
},
{
"grad_norm": 0.5211916565895081,
"learning_rate": 2.2502476204807055e-07,
"loss": 0.0155,
"step": 29140
},
{
"grad_norm": 0.29995298385620117,
"learning_rate": 2.1983187197808786e-07,
"loss": 0.0119,
"step": 29150
},
{
"grad_norm": 0.3520030975341797,
"learning_rate": 2.1469946950634644e-07,
"loss": 0.0181,
"step": 29160
},
{
"grad_norm": 0.38317179679870605,
"learning_rate": 2.096275608691911e-07,
"loss": 0.0209,
"step": 29170
},
{
"grad_norm": 0.3694935142993927,
"learning_rate": 2.046161522294754e-07,
"loss": 0.0152,
"step": 29180
},
{
"grad_norm": 0.3599371314048767,
"learning_rate": 1.9966524967653944e-07,
"loss": 0.014,
"step": 29190
},
{
"grad_norm": 0.32517603039741516,
"learning_rate": 1.9477485922618222e-07,
"loss": 0.0143,
"step": 29200
},
{
"grad_norm": 0.42375895380973816,
"learning_rate": 1.899449868207004e-07,
"loss": 0.0141,
"step": 29210
},
{
"grad_norm": 0.31157833337783813,
"learning_rate": 1.851756383288439e-07,
"loss": 0.0263,
"step": 29220
},
{
"grad_norm": 0.34806984663009644,
"learning_rate": 1.8046681954581035e-07,
"loss": 0.0115,
"step": 29230
},
{
"grad_norm": 0.2210974544286728,
"learning_rate": 1.7581853619327294e-07,
"loss": 0.0248,
"step": 29240
},
{
"grad_norm": 0.5340934991836548,
"learning_rate": 1.7123079391932472e-07,
"loss": 0.013,
"step": 29250
},
{
"grad_norm": 0.20636121928691864,
"learning_rate": 1.6670359829850657e-07,
"loss": 0.0139,
"step": 29260
},
{
"grad_norm": 0.2312215119600296,
"learning_rate": 1.6223695483179037e-07,
"loss": 0.0138,
"step": 29270
},
{
"grad_norm": 0.39863261580467224,
"learning_rate": 1.5783086894656795e-07,
"loss": 0.0155,
"step": 29280
},
{
"grad_norm": 0.4236014783382416,
"learning_rate": 1.5348534599665121e-07,
"loss": 0.0141,
"step": 29290
},
{
"grad_norm": 0.35660314559936523,
"learning_rate": 1.4920039126225527e-07,
"loss": 0.0148,
"step": 29300
},
{
"grad_norm": 0.2859340012073517,
"learning_rate": 1.449760099500097e-07,
"loss": 0.0219,
"step": 29310
},
{
"grad_norm": 0.33396115899086,
"learning_rate": 1.4081220719293075e-07,
"loss": 0.0157,
"step": 29320
},
{
"grad_norm": 0.4001111686229706,
"learning_rate": 1.3670898805043797e-07,
"loss": 0.0131,
"step": 29330
},
{
"grad_norm": 0.3623659610748291,
"learning_rate": 1.326663575083209e-07,
"loss": 0.0134,
"step": 29340
},
{
"grad_norm": 0.5002261400222778,
"learning_rate": 1.2868432047876688e-07,
"loss": 0.0171,
"step": 29350
},
{
"grad_norm": 0.29328837990760803,
"learning_rate": 1.2476288180032213e-07,
"loss": 0.0137,
"step": 29360
},
{
"grad_norm": 0.39149999618530273,
"learning_rate": 1.209020462379029e-07,
"loss": 0.0151,
"step": 29370
},
{
"grad_norm": 0.39905765652656555,
"learning_rate": 1.1710181848278435e-07,
"loss": 0.0127,
"step": 29380
},
{
"grad_norm": 0.29861074686050415,
"learning_rate": 1.133622031526116e-07,
"loss": 0.0099,
"step": 29390
},
{
"grad_norm": 0.25885435938835144,
"learning_rate": 1.0968320479136095e-07,
"loss": 0.0141,
"step": 29400
},
{
"grad_norm": 0.3761281967163086,
"learning_rate": 1.0606482786936767e-07,
"loss": 0.0138,
"step": 29410
},
{
"grad_norm": 0.2875429689884186,
"learning_rate": 1.0250707678329808e-07,
"loss": 0.0101,
"step": 29420
},
{
"grad_norm": 0.30535414814949036,
"learning_rate": 9.900995585615525e-08,
"loss": 0.0133,
"step": 29430
},
{
"grad_norm": 0.29741784930229187,
"learning_rate": 9.55734693372734e-08,
"loss": 0.0162,
"step": 29440
},
{
"grad_norm": 0.2132072001695633,
"learning_rate": 9.219762140231236e-08,
"loss": 0.0137,
"step": 29450
},
{
"grad_norm": 0.2586209177970886,
"learning_rate": 8.888241615322978e-08,
"loss": 0.009,
"step": 29460
},
{
"grad_norm": 0.34166112542152405,
"learning_rate": 8.562785761833114e-08,
"loss": 0.0132,
"step": 29470
},
{
"grad_norm": 0.2688508629798889,
"learning_rate": 8.243394975219753e-08,
"loss": 0.0189,
"step": 29480
},
{
"grad_norm": 0.45210134983062744,
"learning_rate": 7.930069643573568e-08,
"loss": 0.015,
"step": 29490
},
{
"grad_norm": 0.3290596306324005,
"learning_rate": 7.622810147614456e-08,
"loss": 0.0187,
"step": 29500
},
{
"grad_norm": 0.3801783621311188,
"learning_rate": 7.321616860690995e-08,
"loss": 0.012,
"step": 29510
},
{
"grad_norm": 0.3804666996002197,
"learning_rate": 7.026490148782095e-08,
"loss": 0.0119,
"step": 29520
},
{
"grad_norm": 0.24719037115573883,
"learning_rate": 6.737430370494236e-08,
"loss": 0.0148,
"step": 29530
},
{
"grad_norm": 0.18382184207439423,
"learning_rate": 6.454437877062569e-08,
"loss": 0.0244,
"step": 29540
},
{
"grad_norm": 0.25170573592185974,
"learning_rate": 6.177513012349256e-08,
"loss": 0.0134,
"step": 29550
},
{
"grad_norm": 0.2782489061355591,
"learning_rate": 5.9066561128445775e-08,
"loss": 0.0118,
"step": 29560
},
{
"grad_norm": 0.31158795952796936,
"learning_rate": 5.6418675076641556e-08,
"loss": 0.0161,
"step": 29570
},
{
"grad_norm": 0.2339198887348175,
"learning_rate": 5.383147518552845e-08,
"loss": 0.016,
"step": 29580
},
{
"grad_norm": 0.2501128613948822,
"learning_rate": 5.1304964598786196e-08,
"loss": 0.0142,
"step": 29590
},
{
"grad_norm": 0.36374998092651367,
"learning_rate": 4.883914638636467e-08,
"loss": 0.0108,
"step": 29600
},
{
"grad_norm": 0.1859276294708252,
"learning_rate": 4.643402354446713e-08,
"loss": 0.009,
"step": 29610
},
{
"grad_norm": 0.43017128109931946,
"learning_rate": 4.4089598995544766e-08,
"loss": 0.012,
"step": 29620
},
{
"grad_norm": 0.26247483491897583,
"learning_rate": 4.180587558829663e-08,
"loss": 0.0167,
"step": 29630
},
{
"grad_norm": 0.4845767021179199,
"learning_rate": 3.958285609765855e-08,
"loss": 0.0126,
"step": 29640
},
{
"grad_norm": 0.2297232449054718,
"learning_rate": 3.74205432248087e-08,
"loss": 0.0145,
"step": 29650
},
{
"grad_norm": 0.22535023093223572,
"learning_rate": 3.531893959716204e-08,
"loss": 0.0194,
"step": 29660
},
{
"grad_norm": 0.43984371423721313,
"learning_rate": 3.327804776837029e-08,
"loss": 0.0148,
"step": 29670
},
{
"grad_norm": 0.3021300435066223,
"learning_rate": 3.129787021829977e-08,
"loss": 0.016,
"step": 29680
},
{
"grad_norm": 0.2772665321826935,
"learning_rate": 2.9378409353059133e-08,
"loss": 0.0129,
"step": 29690
},
{
"grad_norm": 0.3452436029911041,
"learning_rate": 2.7519667504971593e-08,
"loss": 0.0115,
"step": 29700
},
{
"grad_norm": 0.49415692687034607,
"learning_rate": 2.572164693258605e-08,
"loss": 0.0153,
"step": 29710
},
{
"grad_norm": 0.37586814165115356,
"learning_rate": 2.3984349820665996e-08,
"loss": 0.0153,
"step": 29720
},
{
"grad_norm": 0.3782850205898285,
"learning_rate": 2.2307778280189485e-08,
"loss": 0.0252,
"step": 29730
},
{
"grad_norm": 0.2626226842403412,
"learning_rate": 2.0691934348354704e-08,
"loss": 0.0112,
"step": 29740
},
{
"grad_norm": 0.30373114347457886,
"learning_rate": 1.9136819988557763e-08,
"loss": 0.0172,
"step": 29750
},
{
"grad_norm": 0.30115896463394165,
"learning_rate": 1.7642437090414908e-08,
"loss": 0.0135,
"step": 29760
},
{
"grad_norm": 0.3171970546245575,
"learning_rate": 1.6208787469734755e-08,
"loss": 0.0143,
"step": 29770
},
{
"grad_norm": 0.35484132170677185,
"learning_rate": 1.483587286854604e-08,
"loss": 0.0143,
"step": 29780
},
{
"grad_norm": 0.21756920218467712,
"learning_rate": 1.3523694955064336e-08,
"loss": 0.0098,
"step": 29790
},
{
"grad_norm": 0.3546982407569885,
"learning_rate": 1.2272255323708681e-08,
"loss": 0.0144,
"step": 29800
},
{
"grad_norm": 0.264802485704422,
"learning_rate": 1.1081555495096042e-08,
"loss": 0.0111,
"step": 29810
},
{
"grad_norm": 0.23246298730373383,
"learning_rate": 9.951596916041306e-09,
"loss": 0.0094,
"step": 29820
},
{
"grad_norm": 0.44798821210861206,
"learning_rate": 8.882380959551739e-09,
"loss": 0.0198,
"step": 29830
},
{
"grad_norm": 0.3467652201652527,
"learning_rate": 7.873908924821428e-09,
"loss": 0.0202,
"step": 29840
},
{
"grad_norm": 0.42316746711730957,
"learning_rate": 6.926182037242379e-09,
"loss": 0.016,
"step": 29850
},
{
"grad_norm": 0.2708476185798645,
"learning_rate": 6.039201448387876e-09,
"loss": 0.0139,
"step": 29860
},
{
"grad_norm": 0.2635956406593323,
"learning_rate": 5.212968236029125e-09,
"loss": 0.0109,
"step": 29870
},
{
"grad_norm": 0.4002338647842407,
"learning_rate": 4.447483404118602e-09,
"loss": 0.0174,
"step": 29880
},
{
"grad_norm": 0.21138200163841248,
"learning_rate": 3.742747882784503e-09,
"loss": 0.0108,
"step": 29890
},
{
"grad_norm": 0.338915079832077,
"learning_rate": 3.0987625283473987e-09,
"loss": 0.0228,
"step": 29900
},
{
"grad_norm": 0.43735426664352417,
"learning_rate": 2.515528123320232e-09,
"loss": 0.0136,
"step": 29910
},
{
"grad_norm": 0.3527759313583374,
"learning_rate": 1.9930453763750133e-09,
"loss": 0.0143,
"step": 29920
},
{
"grad_norm": 0.41124871373176575,
"learning_rate": 1.5313149223872281e-09,
"loss": 0.023,
"step": 29930
},
{
"grad_norm": 0.36797401309013367,
"learning_rate": 1.1303373224025305e-09,
"loss": 0.016,
"step": 29940
},
{
"grad_norm": 0.339068740606308,
"learning_rate": 7.901130636367438e-10,
"loss": 0.0126,
"step": 29950
},
{
"grad_norm": 0.46259835362434387,
"learning_rate": 5.106425595036158e-10,
"loss": 0.0184,
"step": 29960
},
{
"grad_norm": 0.34556180238723755,
"learning_rate": 2.9192614958706286e-10,
"loss": 0.0131,
"step": 29970
},
{
"grad_norm": 0.3667891025543213,
"learning_rate": 1.3396409964117064e-10,
"loss": 0.0167,
"step": 29980
},
{
"grad_norm": 0.2845686078071594,
"learning_rate": 3.6756601606846574e-11,
"loss": 0.0171,
"step": 29990
},
{
"grad_norm": 0.38681939244270325,
"learning_rate": 3.0377360626943076e-13,
"loss": 0.0127,
"step": 30000
}
],
"logging_steps": 10,
"max_steps": 30000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 10000,
"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": 8,
"trial_name": null,
"trial_params": null
}