diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7951 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9696969696969697,
+  "eval_steps": 141,
+  "global_step": 1122,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0017825311942959,
+      "grad_norm": 13.759645462036133,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.5522,
+      "step": 1
+    },
+    {
+      "epoch": 0.0017825311942959,
+      "eval_loss": 1.6059316396713257,
+      "eval_runtime": 329.95,
+      "eval_samples_per_second": 8.917,
+      "eval_steps_per_second": 1.115,
+      "step": 1
+    },
+    {
+      "epoch": 0.0035650623885918,
+      "grad_norm": 14.121225357055664,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.6595,
+      "step": 2
+    },
+    {
+      "epoch": 0.0053475935828877,
+      "grad_norm": 11.482634544372559,
+      "learning_rate": 6e-06,
+      "loss": 1.6028,
+      "step": 3
+    },
+    {
+      "epoch": 0.0071301247771836,
+      "grad_norm": 9.531785011291504,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.4464,
+      "step": 4
+    },
+    {
+      "epoch": 0.008912655971479501,
+      "grad_norm": 9.846260070800781,
+      "learning_rate": 1e-05,
+      "loss": 1.3576,
+      "step": 5
+    },
+    {
+      "epoch": 0.0106951871657754,
+      "grad_norm": 7.445741653442383,
+      "learning_rate": 1.2e-05,
+      "loss": 1.2678,
+      "step": 6
+    },
+    {
+      "epoch": 0.012477718360071301,
+      "grad_norm": 5.630918979644775,
+      "learning_rate": 1.4e-05,
+      "loss": 1.299,
+      "step": 7
+    },
+    {
+      "epoch": 0.0142602495543672,
+      "grad_norm": 6.509588241577148,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2572,
+      "step": 8
+    },
+    {
+      "epoch": 0.016042780748663103,
+      "grad_norm": 6.023720741271973,
+      "learning_rate": 1.8e-05,
+      "loss": 1.2671,
+      "step": 9
+    },
+    {
+      "epoch": 0.017825311942959002,
+      "grad_norm": 4.982049465179443,
+      "learning_rate": 2e-05,
+      "loss": 1.1808,
+      "step": 10
+    },
+    {
+      "epoch": 0.0196078431372549,
+      "grad_norm": 4.719905853271484,
+      "learning_rate": 1.9999960092007093e-05,
+      "loss": 1.2005,
+      "step": 11
+    },
+    {
+      "epoch": 0.0213903743315508,
+      "grad_norm": 4.139680862426758,
+      "learning_rate": 1.9999840368346898e-05,
+      "loss": 1.1395,
+      "step": 12
+    },
+    {
+      "epoch": 0.023172905525846704,
+      "grad_norm": 3.6177799701690674,
+      "learning_rate": 1.9999640829975005e-05,
+      "loss": 1.2353,
+      "step": 13
+    },
+    {
+      "epoch": 0.024955436720142603,
+      "grad_norm": 3.3829057216644287,
+      "learning_rate": 1.9999361478484043e-05,
+      "loss": 1.1544,
+      "step": 14
+    },
+    {
+      "epoch": 0.026737967914438502,
+      "grad_norm": 3.5621323585510254,
+      "learning_rate": 1.9999002316103692e-05,
+      "loss": 1.2219,
+      "step": 15
+    },
+    {
+      "epoch": 0.0285204991087344,
+      "grad_norm": 3.8984572887420654,
+      "learning_rate": 1.9998563345700635e-05,
+      "loss": 1.131,
+      "step": 16
+    },
+    {
+      "epoch": 0.030303030303030304,
+      "grad_norm": 3.623622417449951,
+      "learning_rate": 1.999804457077856e-05,
+      "loss": 1.174,
+      "step": 17
+    },
+    {
+      "epoch": 0.03208556149732621,
+      "grad_norm": 3.9878575801849365,
+      "learning_rate": 1.999744599547812e-05,
+      "loss": 1.206,
+      "step": 18
+    },
+    {
+      "epoch": 0.0338680926916221,
+      "grad_norm": 3.3543357849121094,
+      "learning_rate": 1.9996767624576902e-05,
+      "loss": 1.1469,
+      "step": 19
+    },
+    {
+      "epoch": 0.035650623885918005,
+      "grad_norm": 3.7037160396575928,
+      "learning_rate": 1.9996009463489393e-05,
+      "loss": 1.1786,
+      "step": 20
+    },
+    {
+      "epoch": 0.0374331550802139,
+      "grad_norm": 3.6356801986694336,
+      "learning_rate": 1.9995171518266926e-05,
+      "loss": 1.1448,
+      "step": 21
+    },
+    {
+      "epoch": 0.0392156862745098,
+      "grad_norm": 3.112527370452881,
+      "learning_rate": 1.999425379559765e-05,
+      "loss": 1.137,
+      "step": 22
+    },
+    {
+      "epoch": 0.040998217468805706,
+      "grad_norm": 3.8700811862945557,
+      "learning_rate": 1.9993256302806452e-05,
+      "loss": 1.1646,
+      "step": 23
+    },
+    {
+      "epoch": 0.0427807486631016,
+      "grad_norm": 4.020349502563477,
+      "learning_rate": 1.9992179047854923e-05,
+      "loss": 1.1245,
+      "step": 24
+    },
+    {
+      "epoch": 0.044563279857397504,
+      "grad_norm": 3.30739426612854,
+      "learning_rate": 1.9991022039341278e-05,
+      "loss": 1.1164,
+      "step": 25
+    },
+    {
+      "epoch": 0.04634581105169341,
+      "grad_norm": 3.206475019454956,
+      "learning_rate": 1.9989785286500294e-05,
+      "loss": 1.2038,
+      "step": 26
+    },
+    {
+      "epoch": 0.0481283422459893,
+      "grad_norm": 3.0726990699768066,
+      "learning_rate": 1.998846879920324e-05,
+      "loss": 1.09,
+      "step": 27
+    },
+    {
+      "epoch": 0.049910873440285206,
+      "grad_norm": 3.2845897674560547,
+      "learning_rate": 1.9987072587957784e-05,
+      "loss": 1.143,
+      "step": 28
+    },
+    {
+      "epoch": 0.05169340463458111,
+      "grad_norm": 3.3358678817749023,
+      "learning_rate": 1.9985596663907924e-05,
+      "loss": 1.1128,
+      "step": 29
+    },
+    {
+      "epoch": 0.053475935828877004,
+      "grad_norm": 3.362863063812256,
+      "learning_rate": 1.99840410388339e-05,
+      "loss": 1.2224,
+      "step": 30
+    },
+    {
+      "epoch": 0.05525846702317291,
+      "grad_norm": 3.2610557079315186,
+      "learning_rate": 1.9982405725152073e-05,
+      "loss": 1.039,
+      "step": 31
+    },
+    {
+      "epoch": 0.0570409982174688,
+      "grad_norm": 2.9130852222442627,
+      "learning_rate": 1.998069073591488e-05,
+      "loss": 1.0769,
+      "step": 32
+    },
+    {
+      "epoch": 0.058823529411764705,
+      "grad_norm": 3.179290533065796,
+      "learning_rate": 1.9978896084810656e-05,
+      "loss": 1.1426,
+      "step": 33
+    },
+    {
+      "epoch": 0.06060606060606061,
+      "grad_norm": 2.628631591796875,
+      "learning_rate": 1.99770217861636e-05,
+      "loss": 1.0908,
+      "step": 34
+    },
+    {
+      "epoch": 0.062388591800356503,
+      "grad_norm": 3.04935884475708,
+      "learning_rate": 1.9975067854933607e-05,
+      "loss": 1.0795,
+      "step": 35
+    },
+    {
+      "epoch": 0.06417112299465241,
+      "grad_norm": 3.455918312072754,
+      "learning_rate": 1.9973034306716173e-05,
+      "loss": 1.1744,
+      "step": 36
+    },
+    {
+      "epoch": 0.0659536541889483,
+      "grad_norm": 3.023421049118042,
+      "learning_rate": 1.997092115774226e-05,
+      "loss": 1.1438,
+      "step": 37
+    },
+    {
+      "epoch": 0.0677361853832442,
+      "grad_norm": 2.92010760307312,
+      "learning_rate": 1.9968728424878178e-05,
+      "loss": 1.0627,
+      "step": 38
+    },
+    {
+      "epoch": 0.06951871657754011,
+      "grad_norm": 2.4813754558563232,
+      "learning_rate": 1.996645612562544e-05,
+      "loss": 1.1508,
+      "step": 39
+    },
+    {
+      "epoch": 0.07130124777183601,
+      "grad_norm": 4.026529788970947,
+      "learning_rate": 1.9964104278120624e-05,
+      "loss": 1.078,
+      "step": 40
+    },
+    {
+      "epoch": 0.07308377896613191,
+      "grad_norm": 3.1887829303741455,
+      "learning_rate": 1.9961672901135238e-05,
+      "loss": 1.0675,
+      "step": 41
+    },
+    {
+      "epoch": 0.0748663101604278,
+      "grad_norm": 3.5456881523132324,
+      "learning_rate": 1.9959162014075553e-05,
+      "loss": 0.993,
+      "step": 42
+    },
+    {
+      "epoch": 0.0766488413547237,
+      "grad_norm": 2.937511444091797,
+      "learning_rate": 1.9956571636982463e-05,
+      "loss": 1.1662,
+      "step": 43
+    },
+    {
+      "epoch": 0.0784313725490196,
+      "grad_norm": 2.6722967624664307,
+      "learning_rate": 1.9953901790531315e-05,
+      "loss": 1.0205,
+      "step": 44
+    },
+    {
+      "epoch": 0.08021390374331551,
+      "grad_norm": 2.8215572834014893,
+      "learning_rate": 1.9951152496031755e-05,
+      "loss": 1.066,
+      "step": 45
+    },
+    {
+      "epoch": 0.08199643493761141,
+      "grad_norm": 3.3156650066375732,
+      "learning_rate": 1.994832377542755e-05,
+      "loss": 1.1836,
+      "step": 46
+    },
+    {
+      "epoch": 0.08377896613190731,
+      "grad_norm": 2.6018829345703125,
+      "learning_rate": 1.9945415651296408e-05,
+      "loss": 1.1132,
+      "step": 47
+    },
+    {
+      "epoch": 0.0855614973262032,
+      "grad_norm": 2.6550025939941406,
+      "learning_rate": 1.994242814684981e-05,
+      "loss": 1.1259,
+      "step": 48
+    },
+    {
+      "epoch": 0.0873440285204991,
+      "grad_norm": 3.2128894329071045,
+      "learning_rate": 1.9939361285932818e-05,
+      "loss": 1.1379,
+      "step": 49
+    },
+    {
+      "epoch": 0.08912655971479501,
+      "grad_norm": 2.9681684970855713,
+      "learning_rate": 1.9936215093023884e-05,
+      "loss": 1.1592,
+      "step": 50
+    },
+    {
+      "epoch": 0.09090909090909091,
+      "grad_norm": 2.9430744647979736,
+      "learning_rate": 1.9932989593234656e-05,
+      "loss": 1.0492,
+      "step": 51
+    },
+    {
+      "epoch": 0.09269162210338681,
+      "grad_norm": 2.971011161804199,
+      "learning_rate": 1.9929684812309783e-05,
+      "loss": 1.1518,
+      "step": 52
+    },
+    {
+      "epoch": 0.0944741532976827,
+      "grad_norm": 3.132988691329956,
+      "learning_rate": 1.9926300776626694e-05,
+      "loss": 0.9236,
+      "step": 53
+    },
+    {
+      "epoch": 0.0962566844919786,
+      "grad_norm": 2.9208507537841797,
+      "learning_rate": 1.9922837513195406e-05,
+      "loss": 1.1269,
+      "step": 54
+    },
+    {
+      "epoch": 0.09803921568627451,
+      "grad_norm": 3.3337550163269043,
+      "learning_rate": 1.9919295049658296e-05,
+      "loss": 1.124,
+      "step": 55
+    },
+    {
+      "epoch": 0.09982174688057041,
+      "grad_norm": 3.489339828491211,
+      "learning_rate": 1.9915673414289885e-05,
+      "loss": 1.1692,
+      "step": 56
+    },
+    {
+      "epoch": 0.10160427807486631,
+      "grad_norm": 3.233980417251587,
+      "learning_rate": 1.991197263599662e-05,
+      "loss": 1.0572,
+      "step": 57
+    },
+    {
+      "epoch": 0.10338680926916222,
+      "grad_norm": 3.279974937438965,
+      "learning_rate": 1.990819274431662e-05,
+      "loss": 1.0731,
+      "step": 58
+    },
+    {
+      "epoch": 0.1051693404634581,
+      "grad_norm": 3.0058257579803467,
+      "learning_rate": 1.990433376941947e-05,
+      "loss": 1.0906,
+      "step": 59
+    },
+    {
+      "epoch": 0.10695187165775401,
+      "grad_norm": 2.8452155590057373,
+      "learning_rate": 1.9900395742105948e-05,
+      "loss": 1.0852,
+      "step": 60
+    },
+    {
+      "epoch": 0.10873440285204991,
+      "grad_norm": 2.801748037338257,
+      "learning_rate": 1.989637869380782e-05,
+      "loss": 1.0664,
+      "step": 61
+    },
+    {
+      "epoch": 0.11051693404634581,
+      "grad_norm": 2.682849168777466,
+      "learning_rate": 1.989228265658754e-05,
+      "loss": 1.1374,
+      "step": 62
+    },
+    {
+      "epoch": 0.11229946524064172,
+      "grad_norm": 2.923135995864868,
+      "learning_rate": 1.988810766313804e-05,
+      "loss": 1.0354,
+      "step": 63
+    },
+    {
+      "epoch": 0.1140819964349376,
+      "grad_norm": 2.826714277267456,
+      "learning_rate": 1.9883853746782447e-05,
+      "loss": 1.0734,
+      "step": 64
+    },
+    {
+      "epoch": 0.11586452762923351,
+      "grad_norm": 2.7044596672058105,
+      "learning_rate": 1.9879520941473804e-05,
+      "loss": 1.0588,
+      "step": 65
+    },
+    {
+      "epoch": 0.11764705882352941,
+      "grad_norm": 2.67295503616333,
+      "learning_rate": 1.9875109281794828e-05,
+      "loss": 1.1014,
+      "step": 66
+    },
+    {
+      "epoch": 0.11942959001782531,
+      "grad_norm": 2.5908753871917725,
+      "learning_rate": 1.9870618802957617e-05,
+      "loss": 1.1196,
+      "step": 67
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 2.461550712585449,
+      "learning_rate": 1.9866049540803364e-05,
+      "loss": 1.0968,
+      "step": 68
+    },
+    {
+      "epoch": 0.12299465240641712,
+      "grad_norm": 2.6429128646850586,
+      "learning_rate": 1.9861401531802093e-05,
+      "loss": 1.0237,
+      "step": 69
+    },
+    {
+      "epoch": 0.12477718360071301,
+      "grad_norm": 2.7659785747528076,
+      "learning_rate": 1.9856674813052345e-05,
+      "loss": 1.086,
+      "step": 70
+    },
+    {
+      "epoch": 0.1265597147950089,
+      "grad_norm": 2.5340003967285156,
+      "learning_rate": 1.9851869422280887e-05,
+      "loss": 1.0793,
+      "step": 71
+    },
+    {
+      "epoch": 0.12834224598930483,
+      "grad_norm": 2.8008666038513184,
+      "learning_rate": 1.9846985397842428e-05,
+      "loss": 1.1166,
+      "step": 72
+    },
+    {
+      "epoch": 0.13012477718360071,
+      "grad_norm": 3.1304714679718018,
+      "learning_rate": 1.9842022778719277e-05,
+      "loss": 1.0872,
+      "step": 73
+    },
+    {
+      "epoch": 0.1319073083778966,
+      "grad_norm": 2.276888608932495,
+      "learning_rate": 1.9836981604521077e-05,
+      "loss": 1.0454,
+      "step": 74
+    },
+    {
+      "epoch": 0.13368983957219252,
+      "grad_norm": 2.865384578704834,
+      "learning_rate": 1.9831861915484457e-05,
+      "loss": 1.1022,
+      "step": 75
+    },
+    {
+      "epoch": 0.1354723707664884,
+      "grad_norm": 2.901850938796997,
+      "learning_rate": 1.9826663752472716e-05,
+      "loss": 1.0293,
+      "step": 76
+    },
+    {
+      "epoch": 0.13725490196078433,
+      "grad_norm": 2.7473998069763184,
+      "learning_rate": 1.982138715697551e-05,
+      "loss": 1.1318,
+      "step": 77
+    },
+    {
+      "epoch": 0.13903743315508021,
+      "grad_norm": 2.841933488845825,
+      "learning_rate": 1.98160321711085e-05,
+      "loss": 1.0665,
+      "step": 78
+    },
+    {
+      "epoch": 0.1408199643493761,
+      "grad_norm": 2.8788764476776123,
+      "learning_rate": 1.9810598837613034e-05,
+      "loss": 1.1207,
+      "step": 79
+    },
+    {
+      "epoch": 0.14260249554367202,
+      "grad_norm": 2.6716983318328857,
+      "learning_rate": 1.9805087199855807e-05,
+      "loss": 1.1143,
+      "step": 80
+    },
+    {
+      "epoch": 0.1443850267379679,
+      "grad_norm": 2.925053834915161,
+      "learning_rate": 1.979949730182849e-05,
+      "loss": 1.1319,
+      "step": 81
+    },
+    {
+      "epoch": 0.14616755793226383,
+      "grad_norm": 2.625619411468506,
+      "learning_rate": 1.9793829188147406e-05,
+      "loss": 1.0684,
+      "step": 82
+    },
+    {
+      "epoch": 0.14795008912655971,
+      "grad_norm": 2.7794134616851807,
+      "learning_rate": 1.9788082904053168e-05,
+      "loss": 1.0967,
+      "step": 83
+    },
+    {
+      "epoch": 0.1497326203208556,
+      "grad_norm": 2.9480831623077393,
+      "learning_rate": 1.9782258495410306e-05,
+      "loss": 1.0883,
+      "step": 84
+    },
+    {
+      "epoch": 0.15151515151515152,
+      "grad_norm": 2.7370266914367676,
+      "learning_rate": 1.977635600870691e-05,
+      "loss": 1.1109,
+      "step": 85
+    },
+    {
+      "epoch": 0.1532976827094474,
+      "grad_norm": 2.5123989582061768,
+      "learning_rate": 1.9770375491054264e-05,
+      "loss": 1.0783,
+      "step": 86
+    },
+    {
+      "epoch": 0.15508021390374332,
+      "grad_norm": 2.7342569828033447,
+      "learning_rate": 1.976431699018646e-05,
+      "loss": 1.1205,
+      "step": 87
+    },
+    {
+      "epoch": 0.1568627450980392,
+      "grad_norm": 2.8322882652282715,
+      "learning_rate": 1.975818055446001e-05,
+      "loss": 1.0659,
+      "step": 88
+    },
+    {
+      "epoch": 0.1586452762923351,
+      "grad_norm": 2.488312244415283,
+      "learning_rate": 1.975196623285349e-05,
+      "loss": 1.1361,
+      "step": 89
+    },
+    {
+      "epoch": 0.16042780748663102,
+      "grad_norm": 3.0017895698547363,
+      "learning_rate": 1.974567407496712e-05,
+      "loss": 1.1512,
+      "step": 90
+    },
+    {
+      "epoch": 0.1622103386809269,
+      "grad_norm": 2.684353828430176,
+      "learning_rate": 1.9739304131022377e-05,
+      "loss": 1.0893,
+      "step": 91
+    },
+    {
+      "epoch": 0.16399286987522282,
+      "grad_norm": 2.4697201251983643,
+      "learning_rate": 1.9732856451861596e-05,
+      "loss": 0.986,
+      "step": 92
+    },
+    {
+      "epoch": 0.1657754010695187,
+      "grad_norm": 2.581225872039795,
+      "learning_rate": 1.9726331088947563e-05,
+      "loss": 1.0817,
+      "step": 93
+    },
+    {
+      "epoch": 0.16755793226381463,
+      "grad_norm": 2.7253096103668213,
+      "learning_rate": 1.9719728094363103e-05,
+      "loss": 1.0637,
+      "step": 94
+    },
+    {
+      "epoch": 0.16934046345811052,
+      "grad_norm": 2.6325371265411377,
+      "learning_rate": 1.9713047520810677e-05,
+      "loss": 1.0655,
+      "step": 95
+    },
+    {
+      "epoch": 0.1711229946524064,
+      "grad_norm": 3.3519203662872314,
+      "learning_rate": 1.970628942161193e-05,
+      "loss": 1.0285,
+      "step": 96
+    },
+    {
+      "epoch": 0.17290552584670232,
+      "grad_norm": 2.5676865577697754,
+      "learning_rate": 1.969945385070731e-05,
+      "loss": 1.1274,
+      "step": 97
+    },
+    {
+      "epoch": 0.1746880570409982,
+      "grad_norm": 2.7812039852142334,
+      "learning_rate": 1.9692540862655587e-05,
+      "loss": 1.1341,
+      "step": 98
+    },
+    {
+      "epoch": 0.17647058823529413,
+      "grad_norm": 2.6595091819763184,
+      "learning_rate": 1.9685550512633464e-05,
+      "loss": 1.1443,
+      "step": 99
+    },
+    {
+      "epoch": 0.17825311942959002,
+      "grad_norm": 2.484269142150879,
+      "learning_rate": 1.9678482856435107e-05,
+      "loss": 1.0319,
+      "step": 100
+    },
+    {
+      "epoch": 0.1800356506238859,
+      "grad_norm": 2.556206703186035,
+      "learning_rate": 1.9671337950471713e-05,
+      "loss": 1.0688,
+      "step": 101
+    },
+    {
+      "epoch": 0.18181818181818182,
+      "grad_norm": 2.646878480911255,
+      "learning_rate": 1.966411585177105e-05,
+      "loss": 1.1072,
+      "step": 102
+    },
+    {
+      "epoch": 0.1836007130124777,
+      "grad_norm": 2.8139848709106445,
+      "learning_rate": 1.9656816617977012e-05,
+      "loss": 1.0723,
+      "step": 103
+    },
+    {
+      "epoch": 0.18538324420677363,
+      "grad_norm": 2.6773009300231934,
+      "learning_rate": 1.9649440307349156e-05,
+      "loss": 1.1483,
+      "step": 104
+    },
+    {
+      "epoch": 0.18716577540106952,
+      "grad_norm": 2.9415299892425537,
+      "learning_rate": 1.9641986978762228e-05,
+      "loss": 1.1437,
+      "step": 105
+    },
+    {
+      "epoch": 0.1889483065953654,
+      "grad_norm": 2.647825241088867,
+      "learning_rate": 1.9634456691705705e-05,
+      "loss": 1.0377,
+      "step": 106
+    },
+    {
+      "epoch": 0.19073083778966132,
+      "grad_norm": 2.5940890312194824,
+      "learning_rate": 1.9626849506283316e-05,
+      "loss": 1.0183,
+      "step": 107
+    },
+    {
+      "epoch": 0.1925133689839572,
+      "grad_norm": 2.4696593284606934,
+      "learning_rate": 1.9619165483212565e-05,
+      "loss": 1.1108,
+      "step": 108
+    },
+    {
+      "epoch": 0.19429590017825313,
+      "grad_norm": 2.620328903198242,
+      "learning_rate": 1.9611404683824234e-05,
+      "loss": 1.0566,
+      "step": 109
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 2.8399813175201416,
+      "learning_rate": 1.9603567170061918e-05,
+      "loss": 1.0877,
+      "step": 110
+    },
+    {
+      "epoch": 0.19786096256684493,
+      "grad_norm": 2.864955425262451,
+      "learning_rate": 1.9595653004481493e-05,
+      "loss": 1.1072,
+      "step": 111
+    },
+    {
+      "epoch": 0.19964349376114082,
+      "grad_norm": 2.5178327560424805,
+      "learning_rate": 1.958766225025066e-05,
+      "loss": 1.052,
+      "step": 112
+    },
+    {
+      "epoch": 0.2014260249554367,
+      "grad_norm": 2.9980790615081787,
+      "learning_rate": 1.957959497114841e-05,
+      "loss": 1.1248,
+      "step": 113
+    },
+    {
+      "epoch": 0.20320855614973263,
+      "grad_norm": 2.472766637802124,
+      "learning_rate": 1.9571451231564523e-05,
+      "loss": 1.0617,
+      "step": 114
+    },
+    {
+      "epoch": 0.20499108734402852,
+      "grad_norm": 2.4862942695617676,
+      "learning_rate": 1.9563231096499066e-05,
+      "loss": 0.9894,
+      "step": 115
+    },
+    {
+      "epoch": 0.20677361853832443,
+      "grad_norm": 2.491788387298584,
+      "learning_rate": 1.955493463156185e-05,
+      "loss": 0.9946,
+      "step": 116
+    },
+    {
+      "epoch": 0.20855614973262032,
+      "grad_norm": 2.5384109020233154,
+      "learning_rate": 1.9546561902971935e-05,
+      "loss": 1.0207,
+      "step": 117
+    },
+    {
+      "epoch": 0.2103386809269162,
+      "grad_norm": 2.386254072189331,
+      "learning_rate": 1.9538112977557077e-05,
+      "loss": 1.0857,
+      "step": 118
+    },
+    {
+      "epoch": 0.21212121212121213,
+      "grad_norm": 2.649524688720703,
+      "learning_rate": 1.9529587922753205e-05,
+      "loss": 1.0267,
+      "step": 119
+    },
+    {
+      "epoch": 0.21390374331550802,
+      "grad_norm": 2.55436110496521,
+      "learning_rate": 1.9520986806603882e-05,
+      "loss": 1.1293,
+      "step": 120
+    },
+    {
+      "epoch": 0.21568627450980393,
+      "grad_norm": 2.5951426029205322,
+      "learning_rate": 1.951230969775977e-05,
+      "loss": 0.973,
+      "step": 121
+    },
+    {
+      "epoch": 0.21746880570409982,
+      "grad_norm": 3.4098243713378906,
+      "learning_rate": 1.9503556665478066e-05,
+      "loss": 1.1334,
+      "step": 122
+    },
+    {
+      "epoch": 0.2192513368983957,
+      "grad_norm": 2.6225473880767822,
+      "learning_rate": 1.949472777962196e-05,
+      "loss": 1.094,
+      "step": 123
+    },
+    {
+      "epoch": 0.22103386809269163,
+      "grad_norm": 2.8392271995544434,
+      "learning_rate": 1.948582311066008e-05,
+      "loss": 1.0163,
+      "step": 124
+    },
+    {
+      "epoch": 0.22281639928698752,
+      "grad_norm": 2.559697389602661,
+      "learning_rate": 1.9476842729665912e-05,
+      "loss": 1.0216,
+      "step": 125
+    },
+    {
+      "epoch": 0.22459893048128343,
+      "grad_norm": 2.266484498977661,
+      "learning_rate": 1.9467786708317257e-05,
+      "loss": 0.9825,
+      "step": 126
+    },
+    {
+      "epoch": 0.22638146167557932,
+      "grad_norm": 2.2157444953918457,
+      "learning_rate": 1.9458655118895634e-05,
+      "loss": 1.0281,
+      "step": 127
+    },
+    {
+      "epoch": 0.2281639928698752,
+      "grad_norm": 2.868452787399292,
+      "learning_rate": 1.9449448034285737e-05,
+      "loss": 1.148,
+      "step": 128
+    },
+    {
+      "epoch": 0.22994652406417113,
+      "grad_norm": 2.490839958190918,
+      "learning_rate": 1.9440165527974808e-05,
+      "loss": 1.0174,
+      "step": 129
+    },
+    {
+      "epoch": 0.23172905525846701,
+      "grad_norm": 2.3830602169036865,
+      "learning_rate": 1.9430807674052092e-05,
+      "loss": 1.0219,
+      "step": 130
+    },
+    {
+      "epoch": 0.23351158645276293,
+      "grad_norm": 2.284559488296509,
+      "learning_rate": 1.9421374547208223e-05,
+      "loss": 0.9775,
+      "step": 131
+    },
+    {
+      "epoch": 0.23529411764705882,
+      "grad_norm": 2.1950523853302,
+      "learning_rate": 1.941186622273463e-05,
+      "loss": 1.0513,
+      "step": 132
+    },
+    {
+      "epoch": 0.23707664884135474,
+      "grad_norm": 2.3674473762512207,
+      "learning_rate": 1.9402282776522943e-05,
+      "loss": 1.0465,
+      "step": 133
+    },
+    {
+      "epoch": 0.23885918003565063,
+      "grad_norm": 2.5275073051452637,
+      "learning_rate": 1.939262428506438e-05,
+      "loss": 1.0873,
+      "step": 134
+    },
+    {
+      "epoch": 0.24064171122994651,
+      "grad_norm": 2.838315486907959,
+      "learning_rate": 1.938289082544915e-05,
+      "loss": 1.0799,
+      "step": 135
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 2.5094432830810547,
+      "learning_rate": 1.937308247536582e-05,
+      "loss": 1.0696,
+      "step": 136
+    },
+    {
+      "epoch": 0.24420677361853832,
+      "grad_norm": 2.2962303161621094,
+      "learning_rate": 1.9363199313100693e-05,
+      "loss": 1.0703,
+      "step": 137
+    },
+    {
+      "epoch": 0.24598930481283424,
+      "grad_norm": 2.6103367805480957,
+      "learning_rate": 1.9353241417537216e-05,
+      "loss": 1.0648,
+      "step": 138
+    },
+    {
+      "epoch": 0.24777183600713013,
+      "grad_norm": 2.8506877422332764,
+      "learning_rate": 1.93432088681553e-05,
+      "loss": 1.1143,
+      "step": 139
+    },
+    {
+      "epoch": 0.24955436720142601,
+      "grad_norm": 2.5705273151397705,
+      "learning_rate": 1.9333101745030735e-05,
+      "loss": 1.1011,
+      "step": 140
+    },
+    {
+      "epoch": 0.25133689839572193,
+      "grad_norm": 2.8784642219543457,
+      "learning_rate": 1.9322920128834527e-05,
+      "loss": 0.9982,
+      "step": 141
+    },
+    {
+      "epoch": 0.25133689839572193,
+      "eval_loss": 1.0543876886367798,
+      "eval_runtime": 329.8866,
+      "eval_samples_per_second": 8.918,
+      "eval_steps_per_second": 1.116,
+      "step": 141
+    },
+    {
+      "epoch": 0.2531194295900178,
+      "grad_norm": 3.0359432697296143,
+      "learning_rate": 1.9312664100832236e-05,
+      "loss": 1.0954,
+      "step": 142
+    },
+    {
+      "epoch": 0.2549019607843137,
+      "grad_norm": 2.64568829536438,
+      "learning_rate": 1.930233374288337e-05,
+      "loss": 1.0921,
+      "step": 143
+    },
+    {
+      "epoch": 0.25668449197860965,
+      "grad_norm": 2.968031644821167,
+      "learning_rate": 1.929192913744069e-05,
+      "loss": 1.1665,
+      "step": 144
+    },
+    {
+      "epoch": 0.25846702317290554,
+      "grad_norm": 2.6442747116088867,
+      "learning_rate": 1.9281450367549593e-05,
+      "loss": 1.0702,
+      "step": 145
+    },
+    {
+      "epoch": 0.26024955436720143,
+      "grad_norm": 2.722597360610962,
+      "learning_rate": 1.9270897516847406e-05,
+      "loss": 1.0307,
+      "step": 146
+    },
+    {
+      "epoch": 0.2620320855614973,
+      "grad_norm": 2.6280815601348877,
+      "learning_rate": 1.9260270669562747e-05,
+      "loss": 1.0478,
+      "step": 147
+    },
+    {
+      "epoch": 0.2638146167557932,
+      "grad_norm": 2.531064033508301,
+      "learning_rate": 1.9249569910514846e-05,
+      "loss": 1.0316,
+      "step": 148
+    },
+    {
+      "epoch": 0.26559714795008915,
+      "grad_norm": 2.552273750305176,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 1.0622,
+      "step": 149
+    },
+    {
+      "epoch": 0.26737967914438504,
+      "grad_norm": 2.358672618865967,
+      "learning_rate": 1.9227946999355226e-05,
+      "loss": 1.0164,
+      "step": 150
+    },
+    {
+      "epoch": 0.26916221033868093,
+      "grad_norm": 2.6471526622772217,
+      "learning_rate": 1.9217025019828907e-05,
+      "loss": 1.0108,
+      "step": 151
+    },
+    {
+      "epoch": 0.2709447415329768,
+      "grad_norm": 2.467954397201538,
+      "learning_rate": 1.920602947370876e-05,
+      "loss": 1.0493,
+      "step": 152
+    },
+    {
+      "epoch": 0.2727272727272727,
+      "grad_norm": 2.6386890411376953,
+      "learning_rate": 1.9194960448756824e-05,
+      "loss": 1.0464,
+      "step": 153
+    },
+    {
+      "epoch": 0.27450980392156865,
+      "grad_norm": 2.795897960662842,
+      "learning_rate": 1.9183818033321612e-05,
+      "loss": 1.0289,
+      "step": 154
+    },
+    {
+      "epoch": 0.27629233511586454,
+      "grad_norm": 2.4060723781585693,
+      "learning_rate": 1.9172602316337413e-05,
+      "loss": 1.0661,
+      "step": 155
+    },
+    {
+      "epoch": 0.27807486631016043,
+      "grad_norm": 2.434025526046753,
+      "learning_rate": 1.9161313387323574e-05,
+      "loss": 1.0468,
+      "step": 156
+    },
+    {
+      "epoch": 0.2798573975044563,
+      "grad_norm": 2.5258090496063232,
+      "learning_rate": 1.9149951336383798e-05,
+      "loss": 1.0225,
+      "step": 157
+    },
+    {
+      "epoch": 0.2816399286987522,
+      "grad_norm": 2.409041166305542,
+      "learning_rate": 1.9138516254205416e-05,
+      "loss": 0.9958,
+      "step": 158
+    },
+    {
+      "epoch": 0.28342245989304815,
+      "grad_norm": 2.4306955337524414,
+      "learning_rate": 1.912700823205866e-05,
+      "loss": 1.0155,
+      "step": 159
+    },
+    {
+      "epoch": 0.28520499108734404,
+      "grad_norm": 2.49703311920166,
+      "learning_rate": 1.911542736179594e-05,
+      "loss": 1.0727,
+      "step": 160
+    },
+    {
+      "epoch": 0.28698752228163993,
+      "grad_norm": 2.567338228225708,
+      "learning_rate": 1.910377373585113e-05,
+      "loss": 1.0044,
+      "step": 161
+    },
+    {
+      "epoch": 0.2887700534759358,
+      "grad_norm": 2.35740327835083,
+      "learning_rate": 1.9092047447238775e-05,
+      "loss": 1.0163,
+      "step": 162
+    },
+    {
+      "epoch": 0.2905525846702317,
+      "grad_norm": 2.319448709487915,
+      "learning_rate": 1.908024858955341e-05,
+      "loss": 0.9863,
+      "step": 163
+    },
+    {
+      "epoch": 0.29233511586452765,
+      "grad_norm": 2.6530685424804688,
+      "learning_rate": 1.9068377256968782e-05,
+      "loss": 1.0428,
+      "step": 164
+    },
+    {
+      "epoch": 0.29411764705882354,
+      "grad_norm": 2.4042210578918457,
+      "learning_rate": 1.9056433544237106e-05,
+      "loss": 1.026,
+      "step": 165
+    },
+    {
+      "epoch": 0.29590017825311943,
+      "grad_norm": 2.575970411300659,
+      "learning_rate": 1.9044417546688295e-05,
+      "loss": 1.0751,
+      "step": 166
+    },
+    {
+      "epoch": 0.2976827094474153,
+      "grad_norm": 2.477625608444214,
+      "learning_rate": 1.9032329360229225e-05,
+      "loss": 1.0335,
+      "step": 167
+    },
+    {
+      "epoch": 0.2994652406417112,
+      "grad_norm": 2.2049059867858887,
+      "learning_rate": 1.9020169081342942e-05,
+      "loss": 0.9878,
+      "step": 168
+    },
+    {
+      "epoch": 0.30124777183600715,
+      "grad_norm": 2.736179828643799,
+      "learning_rate": 1.900793680708792e-05,
+      "loss": 1.0221,
+      "step": 169
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 2.3086302280426025,
+      "learning_rate": 1.899563263509725e-05,
+      "loss": 1.023,
+      "step": 170
+    },
+    {
+      "epoch": 0.3048128342245989,
+      "grad_norm": 2.41434645652771,
+      "learning_rate": 1.8983256663577898e-05,
+      "loss": 1.0016,
+      "step": 171
+    },
+    {
+      "epoch": 0.3065953654188948,
+      "grad_norm": 2.048405647277832,
+      "learning_rate": 1.8970808991309905e-05,
+      "loss": 0.9322,
+      "step": 172
+    },
+    {
+      "epoch": 0.3083778966131907,
+      "grad_norm": 2.6779351234436035,
+      "learning_rate": 1.895828971764559e-05,
+      "loss": 1.06,
+      "step": 173
+    },
+    {
+      "epoch": 0.31016042780748665,
+      "grad_norm": 2.5748863220214844,
+      "learning_rate": 1.894569894250877e-05,
+      "loss": 1.0392,
+      "step": 174
+    },
+    {
+      "epoch": 0.31194295900178254,
+      "grad_norm": 2.1274595260620117,
+      "learning_rate": 1.8933036766393962e-05,
+      "loss": 1.0284,
+      "step": 175
+    },
+    {
+      "epoch": 0.3137254901960784,
+      "grad_norm": 2.525455951690674,
+      "learning_rate": 1.8920303290365568e-05,
+      "loss": 1.0964,
+      "step": 176
+    },
+    {
+      "epoch": 0.3155080213903743,
+      "grad_norm": 2.5733320713043213,
+      "learning_rate": 1.8907498616057084e-05,
+      "loss": 1.032,
+      "step": 177
+    },
+    {
+      "epoch": 0.3172905525846702,
+      "grad_norm": 2.2275173664093018,
+      "learning_rate": 1.8894622845670282e-05,
+      "loss": 1.0473,
+      "step": 178
+    },
+    {
+      "epoch": 0.31907308377896615,
+      "grad_norm": 2.2616934776306152,
+      "learning_rate": 1.888167608197439e-05,
+      "loss": 1.0048,
+      "step": 179
+    },
+    {
+      "epoch": 0.32085561497326204,
+      "grad_norm": 2.0974690914154053,
+      "learning_rate": 1.886865842830528e-05,
+      "loss": 1.0566,
+      "step": 180
+    },
+    {
+      "epoch": 0.3226381461675579,
+      "grad_norm": 2.5724925994873047,
+      "learning_rate": 1.8855569988564636e-05,
+      "loss": 1.0125,
+      "step": 181
+    },
+    {
+      "epoch": 0.3244206773618538,
+      "grad_norm": 2.321138620376587,
+      "learning_rate": 1.8842410867219137e-05,
+      "loss": 1.0347,
+      "step": 182
+    },
+    {
+      "epoch": 0.32620320855614976,
+      "grad_norm": 2.638758420944214,
+      "learning_rate": 1.8829181169299596e-05,
+      "loss": 1.095,
+      "step": 183
+    },
+    {
+      "epoch": 0.32798573975044565,
+      "grad_norm": 2.51639986038208,
+      "learning_rate": 1.8815881000400164e-05,
+      "loss": 1.0083,
+      "step": 184
+    },
+    {
+      "epoch": 0.32976827094474154,
+      "grad_norm": 2.4949166774749756,
+      "learning_rate": 1.880251046667744e-05,
+      "loss": 1.0022,
+      "step": 185
+    },
+    {
+      "epoch": 0.3315508021390374,
+      "grad_norm": 2.2100119590759277,
+      "learning_rate": 1.878906967484966e-05,
+      "loss": 1.0869,
+      "step": 186
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 2.4192895889282227,
+      "learning_rate": 1.877555873219583e-05,
+      "loss": 0.9759,
+      "step": 187
+    },
+    {
+      "epoch": 0.33511586452762926,
+      "grad_norm": 2.273297071456909,
+      "learning_rate": 1.876197774655487e-05,
+      "loss": 1.0539,
+      "step": 188
+    },
+    {
+      "epoch": 0.33689839572192515,
+      "grad_norm": 2.259113311767578,
+      "learning_rate": 1.874832682632476e-05,
+      "loss": 0.9344,
+      "step": 189
+    },
+    {
+      "epoch": 0.33868092691622104,
+      "grad_norm": 2.2211596965789795,
+      "learning_rate": 1.8734606080461657e-05,
+      "loss": 1.0609,
+      "step": 190
+    },
+    {
+      "epoch": 0.3404634581105169,
+      "grad_norm": 2.131674289703369,
+      "learning_rate": 1.8720815618479053e-05,
+      "loss": 0.999,
+      "step": 191
+    },
+    {
+      "epoch": 0.3422459893048128,
+      "grad_norm": 2.7799980640411377,
+      "learning_rate": 1.870695555044688e-05,
+      "loss": 1.1119,
+      "step": 192
+    },
+    {
+      "epoch": 0.34402852049910876,
+      "grad_norm": 2.1310036182403564,
+      "learning_rate": 1.869302598699063e-05,
+      "loss": 1.0332,
+      "step": 193
+    },
+    {
+      "epoch": 0.34581105169340465,
+      "grad_norm": 2.501399040222168,
+      "learning_rate": 1.86790270392905e-05,
+      "loss": 0.971,
+      "step": 194
+    },
+    {
+      "epoch": 0.34759358288770054,
+      "grad_norm": 2.521914482116699,
+      "learning_rate": 1.866495881908046e-05,
+      "loss": 0.9663,
+      "step": 195
+    },
+    {
+      "epoch": 0.3493761140819964,
+      "grad_norm": 2.6178207397460938,
+      "learning_rate": 1.86508214386474e-05,
+      "loss": 1.0189,
+      "step": 196
+    },
+    {
+      "epoch": 0.3511586452762923,
+      "grad_norm": 2.4940268993377686,
+      "learning_rate": 1.8636615010830216e-05,
+      "loss": 0.9989,
+      "step": 197
+    },
+    {
+      "epoch": 0.35294117647058826,
+      "grad_norm": 2.109654188156128,
+      "learning_rate": 1.8622339649018907e-05,
+      "loss": 1.0024,
+      "step": 198
+    },
+    {
+      "epoch": 0.35472370766488415,
+      "grad_norm": 2.862342596054077,
+      "learning_rate": 1.8607995467153692e-05,
+      "loss": 1.0465,
+      "step": 199
+    },
+    {
+      "epoch": 0.35650623885918004,
+      "grad_norm": 2.5327987670898438,
+      "learning_rate": 1.8593582579724062e-05,
+      "loss": 0.9383,
+      "step": 200
+    },
+    {
+      "epoch": 0.3582887700534759,
+      "grad_norm": 2.031452178955078,
+      "learning_rate": 1.8579101101767904e-05,
+      "loss": 1.0371,
+      "step": 201
+    },
+    {
+      "epoch": 0.3600713012477718,
+      "grad_norm": 2.711760997772217,
+      "learning_rate": 1.856455114887056e-05,
+      "loss": 1.0148,
+      "step": 202
+    },
+    {
+      "epoch": 0.36185383244206776,
+      "grad_norm": 2.1266770362854004,
+      "learning_rate": 1.8549932837163917e-05,
+      "loss": 0.9999,
+      "step": 203
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 2.262907028198242,
+      "learning_rate": 1.8535246283325468e-05,
+      "loss": 0.9768,
+      "step": 204
+    },
+    {
+      "epoch": 0.36541889483065954,
+      "grad_norm": 2.307704210281372,
+      "learning_rate": 1.8520491604577388e-05,
+      "loss": 1.0673,
+      "step": 205
+    },
+    {
+      "epoch": 0.3672014260249554,
+      "grad_norm": 2.203144073486328,
+      "learning_rate": 1.8505668918685603e-05,
+      "loss": 1.0154,
+      "step": 206
+    },
+    {
+      "epoch": 0.3689839572192513,
+      "grad_norm": 2.0945725440979004,
+      "learning_rate": 1.849077834395884e-05,
+      "loss": 1.0539,
+      "step": 207
+    },
+    {
+      "epoch": 0.37076648841354726,
+      "grad_norm": 2.8280982971191406,
+      "learning_rate": 1.8475819999247694e-05,
+      "loss": 1.0071,
+      "step": 208
+    },
+    {
+      "epoch": 0.37254901960784315,
+      "grad_norm": 2.0130624771118164,
+      "learning_rate": 1.8460794003943658e-05,
+      "loss": 0.9843,
+      "step": 209
+    },
+    {
+      "epoch": 0.37433155080213903,
+      "grad_norm": 2.186617612838745,
+      "learning_rate": 1.8445700477978207e-05,
+      "loss": 1.0136,
+      "step": 210
+    },
+    {
+      "epoch": 0.3761140819964349,
+      "grad_norm": 2.1577677726745605,
+      "learning_rate": 1.8430539541821795e-05,
+      "loss": 1.0273,
+      "step": 211
+    },
+    {
+      "epoch": 0.3778966131907308,
+      "grad_norm": 2.9277737140655518,
+      "learning_rate": 1.8415311316482937e-05,
+      "loss": 0.9436,
+      "step": 212
+    },
+    {
+      "epoch": 0.37967914438502676,
+      "grad_norm": 2.1575350761413574,
+      "learning_rate": 1.840001592350721e-05,
+      "loss": 1.0307,
+      "step": 213
+    },
+    {
+      "epoch": 0.38146167557932265,
+      "grad_norm": 2.3168020248413086,
+      "learning_rate": 1.8384653484976305e-05,
+      "loss": 0.9642,
+      "step": 214
+    },
+    {
+      "epoch": 0.38324420677361853,
+      "grad_norm": 2.5893986225128174,
+      "learning_rate": 1.8369224123507035e-05,
+      "loss": 1.0545,
+      "step": 215
+    },
+    {
+      "epoch": 0.3850267379679144,
+      "grad_norm": 2.237194538116455,
+      "learning_rate": 1.835372796225037e-05,
+      "loss": 1.0229,
+      "step": 216
+    },
+    {
+      "epoch": 0.3868092691622103,
+      "grad_norm": 2.373006582260132,
+      "learning_rate": 1.8338165124890455e-05,
+      "loss": 1.0323,
+      "step": 217
+    },
+    {
+      "epoch": 0.38859180035650626,
+      "grad_norm": 2.0743327140808105,
+      "learning_rate": 1.8322535735643604e-05,
+      "loss": 1.0576,
+      "step": 218
+    },
+    {
+      "epoch": 0.39037433155080214,
+      "grad_norm": 2.5559821128845215,
+      "learning_rate": 1.830683991925733e-05,
+      "loss": 1.0307,
+      "step": 219
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 2.2722411155700684,
+      "learning_rate": 1.829107780100934e-05,
+      "loss": 0.9956,
+      "step": 220
+    },
+    {
+      "epoch": 0.3939393939393939,
+      "grad_norm": 2.3079795837402344,
+      "learning_rate": 1.8275249506706535e-05,
+      "loss": 1.0553,
+      "step": 221
+    },
+    {
+      "epoch": 0.39572192513368987,
+      "grad_norm": 2.027052402496338,
+      "learning_rate": 1.8259355162684e-05,
+      "loss": 1.0089,
+      "step": 222
+    },
+    {
+      "epoch": 0.39750445632798576,
+      "grad_norm": 2.2788703441619873,
+      "learning_rate": 1.8243394895804012e-05,
+      "loss": 0.9516,
+      "step": 223
+    },
+    {
+      "epoch": 0.39928698752228164,
+      "grad_norm": 2.1616568565368652,
+      "learning_rate": 1.8227368833455023e-05,
+      "loss": 0.9019,
+      "step": 224
+    },
+    {
+      "epoch": 0.40106951871657753,
+      "grad_norm": 2.5529351234436035,
+      "learning_rate": 1.821127710355062e-05,
+      "loss": 1.0444,
+      "step": 225
+    },
+    {
+      "epoch": 0.4028520499108734,
+      "grad_norm": 2.000537872314453,
+      "learning_rate": 1.8195119834528535e-05,
+      "loss": 0.9659,
+      "step": 226
+    },
+    {
+      "epoch": 0.40463458110516937,
+      "grad_norm": 2.050112724304199,
+      "learning_rate": 1.8178897155349598e-05,
+      "loss": 1.0564,
+      "step": 227
+    },
+    {
+      "epoch": 0.40641711229946526,
+      "grad_norm": 1.993181586265564,
+      "learning_rate": 1.816260919549673e-05,
+      "loss": 0.9761,
+      "step": 228
+    },
+    {
+      "epoch": 0.40819964349376114,
+      "grad_norm": 2.506268262863159,
+      "learning_rate": 1.814625608497389e-05,
+      "loss": 1.0261,
+      "step": 229
+    },
+    {
+      "epoch": 0.40998217468805703,
+      "grad_norm": 2.155731439590454,
+      "learning_rate": 1.8129837954305033e-05,
+      "loss": 1.0477,
+      "step": 230
+    },
+    {
+      "epoch": 0.4117647058823529,
+      "grad_norm": 2.631788969039917,
+      "learning_rate": 1.8113354934533086e-05,
+      "loss": 1.057,
+      "step": 231
+    },
+    {
+      "epoch": 0.41354723707664887,
+      "grad_norm": 2.3771722316741943,
+      "learning_rate": 1.809680715721891e-05,
+      "loss": 1.0031,
+      "step": 232
+    },
+    {
+      "epoch": 0.41532976827094475,
+      "grad_norm": 2.1514644622802734,
+      "learning_rate": 1.8080194754440205e-05,
+      "loss": 1.0661,
+      "step": 233
+    },
+    {
+      "epoch": 0.41711229946524064,
+      "grad_norm": 2.065730333328247,
+      "learning_rate": 1.8063517858790517e-05,
+      "loss": 0.963,
+      "step": 234
+    },
+    {
+      "epoch": 0.41889483065953653,
+      "grad_norm": 2.359872341156006,
+      "learning_rate": 1.804677660337812e-05,
+      "loss": 1.0196,
+      "step": 235
+    },
+    {
+      "epoch": 0.4206773618538324,
+      "grad_norm": 2.3731632232666016,
+      "learning_rate": 1.8029971121824997e-05,
+      "loss": 0.9878,
+      "step": 236
+    },
+    {
+      "epoch": 0.42245989304812837,
+      "grad_norm": 2.6104416847229004,
+      "learning_rate": 1.801310154826576e-05,
+      "loss": 1.0736,
+      "step": 237
+    },
+    {
+      "epoch": 0.42424242424242425,
+      "grad_norm": 2.411689043045044,
+      "learning_rate": 1.799616801734657e-05,
+      "loss": 1.0152,
+      "step": 238
+    },
+    {
+      "epoch": 0.42602495543672014,
+      "grad_norm": 2.3272037506103516,
+      "learning_rate": 1.7979170664224078e-05,
+      "loss": 1.0123,
+      "step": 239
+    },
+    {
+      "epoch": 0.42780748663101603,
+      "grad_norm": 2.3518447875976562,
+      "learning_rate": 1.7962109624564324e-05,
+      "loss": 1.0213,
+      "step": 240
+    },
+    {
+      "epoch": 0.4295900178253119,
+      "grad_norm": 2.7915849685668945,
+      "learning_rate": 1.794498503454169e-05,
+      "loss": 1.1125,
+      "step": 241
+    },
+    {
+      "epoch": 0.43137254901960786,
+      "grad_norm": 2.335583448410034,
+      "learning_rate": 1.792779703083777e-05,
+      "loss": 0.935,
+      "step": 242
+    },
+    {
+      "epoch": 0.43315508021390375,
+      "grad_norm": 2.5337862968444824,
+      "learning_rate": 1.7910545750640317e-05,
+      "loss": 1.1122,
+      "step": 243
+    },
+    {
+      "epoch": 0.43493761140819964,
+      "grad_norm": 2.29020357131958,
+      "learning_rate": 1.7893231331642118e-05,
+      "loss": 0.9865,
+      "step": 244
+    },
+    {
+      "epoch": 0.43672014260249553,
+      "grad_norm": 2.0491623878479004,
+      "learning_rate": 1.7875853912039915e-05,
+      "loss": 0.9985,
+      "step": 245
+    },
+    {
+      "epoch": 0.4385026737967914,
+      "grad_norm": 2.5333423614501953,
+      "learning_rate": 1.7858413630533305e-05,
+      "loss": 1.0376,
+      "step": 246
+    },
+    {
+      "epoch": 0.44028520499108736,
+      "grad_norm": 2.1056642532348633,
+      "learning_rate": 1.7840910626323603e-05,
+      "loss": 0.9189,
+      "step": 247
+    },
+    {
+      "epoch": 0.44206773618538325,
+      "grad_norm": 2.0595109462738037,
+      "learning_rate": 1.7823345039112772e-05,
+      "loss": 1.0123,
+      "step": 248
+    },
+    {
+      "epoch": 0.44385026737967914,
+      "grad_norm": 2.0415713787078857,
+      "learning_rate": 1.780571700910227e-05,
+      "loss": 0.9969,
+      "step": 249
+    },
+    {
+      "epoch": 0.44563279857397503,
+      "grad_norm": 2.055589437484741,
+      "learning_rate": 1.778802667699196e-05,
+      "loss": 1.0098,
+      "step": 250
+    },
+    {
+      "epoch": 0.4474153297682709,
+      "grad_norm": 1.972589373588562,
+      "learning_rate": 1.7770274183978975e-05,
+      "loss": 0.9739,
+      "step": 251
+    },
+    {
+      "epoch": 0.44919786096256686,
+      "grad_norm": 1.9704911708831787,
+      "learning_rate": 1.775245967175658e-05,
+      "loss": 0.9221,
+      "step": 252
+    },
+    {
+      "epoch": 0.45098039215686275,
+      "grad_norm": 2.3809704780578613,
+      "learning_rate": 1.773458328251307e-05,
+      "loss": 1.0125,
+      "step": 253
+    },
+    {
+      "epoch": 0.45276292335115864,
+      "grad_norm": 2.554600715637207,
+      "learning_rate": 1.77166451589306e-05,
+      "loss": 1.1072,
+      "step": 254
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 2.169062614440918,
+      "learning_rate": 1.7698645444184074e-05,
+      "loss": 1.0075,
+      "step": 255
+    },
+    {
+      "epoch": 0.4563279857397504,
+      "grad_norm": 2.2745189666748047,
+      "learning_rate": 1.768058428193999e-05,
+      "loss": 1.0474,
+      "step": 256
+    },
+    {
+      "epoch": 0.45811051693404636,
+      "grad_norm": 2.583153247833252,
+      "learning_rate": 1.76624618163553e-05,
+      "loss": 1.0531,
+      "step": 257
+    },
+    {
+      "epoch": 0.45989304812834225,
+      "grad_norm": 2.389552116394043,
+      "learning_rate": 1.764427819207624e-05,
+      "loss": 0.9571,
+      "step": 258
+    },
+    {
+      "epoch": 0.46167557932263814,
+      "grad_norm": 2.1886773109436035,
+      "learning_rate": 1.762603355423721e-05,
+      "loss": 0.9814,
+      "step": 259
+    },
+    {
+      "epoch": 0.46345811051693403,
+      "grad_norm": 2.6514391899108887,
+      "learning_rate": 1.7607728048459572e-05,
+      "loss": 1.1205,
+      "step": 260
+    },
+    {
+      "epoch": 0.46524064171123,
+      "grad_norm": 2.1365182399749756,
+      "learning_rate": 1.758936182085054e-05,
+      "loss": 0.9833,
+      "step": 261
+    },
+    {
+      "epoch": 0.46702317290552586,
+      "grad_norm": 2.6089439392089844,
+      "learning_rate": 1.757093501800196e-05,
+      "loss": 1.0578,
+      "step": 262
+    },
+    {
+      "epoch": 0.46880570409982175,
+      "grad_norm": 2.455092668533325,
+      "learning_rate": 1.755244778698918e-05,
+      "loss": 1.0678,
+      "step": 263
+    },
+    {
+      "epoch": 0.47058823529411764,
+      "grad_norm": 2.013976573944092,
+      "learning_rate": 1.7533900275369857e-05,
+      "loss": 1.0032,
+      "step": 264
+    },
+    {
+      "epoch": 0.47237076648841353,
+      "grad_norm": 2.275420904159546,
+      "learning_rate": 1.7515292631182782e-05,
+      "loss": 0.973,
+      "step": 265
+    },
+    {
+      "epoch": 0.4741532976827095,
+      "grad_norm": 2.0543277263641357,
+      "learning_rate": 1.7496625002946702e-05,
+      "loss": 1.031,
+      "step": 266
+    },
+    {
+      "epoch": 0.47593582887700536,
+      "grad_norm": 2.4391744136810303,
+      "learning_rate": 1.747789753965913e-05,
+      "loss": 1.0207,
+      "step": 267
+    },
+    {
+      "epoch": 0.47771836007130125,
+      "grad_norm": 2.3594958782196045,
+      "learning_rate": 1.745911039079516e-05,
+      "loss": 1.0113,
+      "step": 268
+    },
+    {
+      "epoch": 0.47950089126559714,
+      "grad_norm": 2.252363681793213,
+      "learning_rate": 1.744026370630628e-05,
+      "loss": 0.9894,
+      "step": 269
+    },
+    {
+      "epoch": 0.48128342245989303,
+      "grad_norm": 2.4105823040008545,
+      "learning_rate": 1.7421357636619153e-05,
+      "loss": 1.0012,
+      "step": 270
+    },
+    {
+      "epoch": 0.483065953654189,
+      "grad_norm": 2.206552028656006,
+      "learning_rate": 1.7402392332634442e-05,
+      "loss": 1.0078,
+      "step": 271
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 2.089542865753174,
+      "learning_rate": 1.7383367945725584e-05,
+      "loss": 1.0526,
+      "step": 272
+    },
+    {
+      "epoch": 0.48663101604278075,
+      "grad_norm": 2.173814058303833,
+      "learning_rate": 1.7364284627737603e-05,
+      "loss": 0.9653,
+      "step": 273
+    },
+    {
+      "epoch": 0.48841354723707664,
+      "grad_norm": 2.486128568649292,
+      "learning_rate": 1.734514253098589e-05,
+      "loss": 1.0228,
+      "step": 274
+    },
+    {
+      "epoch": 0.49019607843137253,
+      "grad_norm": 2.1726722717285156,
+      "learning_rate": 1.732594180825496e-05,
+      "loss": 0.9509,
+      "step": 275
+    },
+    {
+      "epoch": 0.4919786096256685,
+      "grad_norm": 1.9146912097930908,
+      "learning_rate": 1.730668261279729e-05,
+      "loss": 0.9773,
+      "step": 276
+    },
+    {
+      "epoch": 0.49376114081996436,
+      "grad_norm": 2.7147133350372314,
+      "learning_rate": 1.7287365098332042e-05,
+      "loss": 1.0312,
+      "step": 277
+    },
+    {
+      "epoch": 0.49554367201426025,
+      "grad_norm": 2.2072091102600098,
+      "learning_rate": 1.726798941904386e-05,
+      "loss": 1.0854,
+      "step": 278
+    },
+    {
+      "epoch": 0.49732620320855614,
+      "grad_norm": 2.2345147132873535,
+      "learning_rate": 1.724855572958164e-05,
+      "loss": 0.9868,
+      "step": 279
+    },
+    {
+      "epoch": 0.49910873440285203,
+      "grad_norm": 2.1517088413238525,
+      "learning_rate": 1.722906418505729e-05,
+      "loss": 0.9413,
+      "step": 280
+    },
+    {
+      "epoch": 0.5008912655971479,
+      "grad_norm": 2.3915605545043945,
+      "learning_rate": 1.7209514941044494e-05,
+      "loss": 0.9697,
+      "step": 281
+    },
+    {
+      "epoch": 0.5026737967914439,
+      "grad_norm": 2.306183099746704,
+      "learning_rate": 1.7189908153577473e-05,
+      "loss": 1.0336,
+      "step": 282
+    },
+    {
+      "epoch": 0.5026737967914439,
+      "eval_loss": 1.0013747215270996,
+      "eval_runtime": 329.9461,
+      "eval_samples_per_second": 8.917,
+      "eval_steps_per_second": 1.115,
+      "step": 282
+    },
+    {
+      "epoch": 0.5044563279857398,
+      "grad_norm": 2.168407917022705,
+      "learning_rate": 1.717024397914973e-05,
+      "loss": 0.932,
+      "step": 283
+    },
+    {
+      "epoch": 0.5062388591800356,
+      "grad_norm": 2.4041762351989746,
+      "learning_rate": 1.7150522574712815e-05,
+      "loss": 1.0129,
+      "step": 284
+    },
+    {
+      "epoch": 0.5080213903743316,
+      "grad_norm": 2.6760098934173584,
+      "learning_rate": 1.7130744097675058e-05,
+      "loss": 1.0125,
+      "step": 285
+    },
+    {
+      "epoch": 0.5098039215686274,
+      "grad_norm": 2.7730419635772705,
+      "learning_rate": 1.7110908705900322e-05,
+      "loss": 1.021,
+      "step": 286
+    },
+    {
+      "epoch": 0.5115864527629234,
+      "grad_norm": 2.342268943786621,
+      "learning_rate": 1.7091016557706747e-05,
+      "loss": 1.0381,
+      "step": 287
+    },
+    {
+      "epoch": 0.5133689839572193,
+      "grad_norm": 2.5511789321899414,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 1.0157,
+      "step": 288
+    },
+    {
+      "epoch": 0.5151515151515151,
+      "grad_norm": 2.5863358974456787,
+      "learning_rate": 1.7051062627599385e-05,
+      "loss": 1.0438,
+      "step": 289
+    },
+    {
+      "epoch": 0.5169340463458111,
+      "grad_norm": 2.2332570552825928,
+      "learning_rate": 1.7031001164581828e-05,
+      "loss": 1.0236,
+      "step": 290
+    },
+    {
+      "epoch": 0.5187165775401069,
+      "grad_norm": 1.9721466302871704,
+      "learning_rate": 1.701088358293535e-05,
+      "loss": 0.9639,
+      "step": 291
+    },
+    {
+      "epoch": 0.5204991087344029,
+      "grad_norm": 2.1957123279571533,
+      "learning_rate": 1.6990710043230407e-05,
+      "loss": 0.9288,
+      "step": 292
+    },
+    {
+      "epoch": 0.5222816399286988,
+      "grad_norm": 2.659257411956787,
+      "learning_rate": 1.69704807064841e-05,
+      "loss": 1.1005,
+      "step": 293
+    },
+    {
+      "epoch": 0.5240641711229946,
+      "grad_norm": 2.452608346939087,
+      "learning_rate": 1.6950195734158874e-05,
+      "loss": 0.9854,
+      "step": 294
+    },
+    {
+      "epoch": 0.5258467023172906,
+      "grad_norm": 2.114299774169922,
+      "learning_rate": 1.6929855288161234e-05,
+      "loss": 0.9477,
+      "step": 295
+    },
+    {
+      "epoch": 0.5276292335115864,
+      "grad_norm": 2.170616865158081,
+      "learning_rate": 1.6909459530840457e-05,
+      "loss": 1.0347,
+      "step": 296
+    },
+    {
+      "epoch": 0.5294117647058824,
+      "grad_norm": 2.016688823699951,
+      "learning_rate": 1.6889008624987284e-05,
+      "loss": 0.9756,
+      "step": 297
+    },
+    {
+      "epoch": 0.5311942959001783,
+      "grad_norm": 2.1537399291992188,
+      "learning_rate": 1.6868502733832647e-05,
+      "loss": 1.104,
+      "step": 298
+    },
+    {
+      "epoch": 0.5329768270944741,
+      "grad_norm": 2.0217056274414062,
+      "learning_rate": 1.684794202104633e-05,
+      "loss": 0.9949,
+      "step": 299
+    },
+    {
+      "epoch": 0.5347593582887701,
+      "grad_norm": 1.9496427774429321,
+      "learning_rate": 1.682732665073569e-05,
+      "loss": 1.0041,
+      "step": 300
+    },
+    {
+      "epoch": 0.5365418894830659,
+      "grad_norm": 2.2652711868286133,
+      "learning_rate": 1.6806656787444338e-05,
+      "loss": 0.985,
+      "step": 301
+    },
+    {
+      "epoch": 0.5383244206773619,
+      "grad_norm": 2.223661184310913,
+      "learning_rate": 1.6785932596150827e-05,
+      "loss": 0.9828,
+      "step": 302
+    },
+    {
+      "epoch": 0.5401069518716578,
+      "grad_norm": 2.382096767425537,
+      "learning_rate": 1.6765154242267328e-05,
+      "loss": 1.0278,
+      "step": 303
+    },
+    {
+      "epoch": 0.5418894830659536,
+      "grad_norm": 2.034925699234009,
+      "learning_rate": 1.6744321891638328e-05,
+      "loss": 0.9006,
+      "step": 304
+    },
+    {
+      "epoch": 0.5436720142602496,
+      "grad_norm": 2.47273588180542,
+      "learning_rate": 1.6723435710539286e-05,
+      "loss": 1.0393,
+      "step": 305
+    },
+    {
+      "epoch": 0.5454545454545454,
+      "grad_norm": 2.252715587615967,
+      "learning_rate": 1.670249586567531e-05,
+      "loss": 0.9549,
+      "step": 306
+    },
+    {
+      "epoch": 0.5472370766488414,
+      "grad_norm": 2.026567220687866,
+      "learning_rate": 1.668150252417984e-05,
+      "loss": 1.0656,
+      "step": 307
+    },
+    {
+      "epoch": 0.5490196078431373,
+      "grad_norm": 2.188436508178711,
+      "learning_rate": 1.6660455853613303e-05,
+      "loss": 0.9633,
+      "step": 308
+    },
+    {
+      "epoch": 0.5508021390374331,
+      "grad_norm": 2.008063316345215,
+      "learning_rate": 1.6639356021961767e-05,
+      "loss": 0.9901,
+      "step": 309
+    },
+    {
+      "epoch": 0.5525846702317291,
+      "grad_norm": 2.1796090602874756,
+      "learning_rate": 1.6618203197635624e-05,
+      "loss": 0.983,
+      "step": 310
+    },
+    {
+      "epoch": 0.5543672014260249,
+      "grad_norm": 2.1821656227111816,
+      "learning_rate": 1.659699754946823e-05,
+      "loss": 0.9757,
+      "step": 311
+    },
+    {
+      "epoch": 0.5561497326203209,
+      "grad_norm": 2.1550426483154297,
+      "learning_rate": 1.657573924671455e-05,
+      "loss": 1.0422,
+      "step": 312
+    },
+    {
+      "epoch": 0.5579322638146168,
+      "grad_norm": 2.2479467391967773,
+      "learning_rate": 1.6554428459049826e-05,
+      "loss": 0.9551,
+      "step": 313
+    },
+    {
+      "epoch": 0.5597147950089126,
+      "grad_norm": 2.107180595397949,
+      "learning_rate": 1.6533065356568206e-05,
+      "loss": 1.0978,
+      "step": 314
+    },
+    {
+      "epoch": 0.5614973262032086,
+      "grad_norm": 2.159715175628662,
+      "learning_rate": 1.651165010978141e-05,
+      "loss": 0.9501,
+      "step": 315
+    },
+    {
+      "epoch": 0.5632798573975044,
+      "grad_norm": 2.454437494277954,
+      "learning_rate": 1.6490182889617326e-05,
+      "loss": 0.9584,
+      "step": 316
+    },
+    {
+      "epoch": 0.5650623885918004,
+      "grad_norm": 2.113189458847046,
+      "learning_rate": 1.6468663867418705e-05,
+      "loss": 0.9517,
+      "step": 317
+    },
+    {
+      "epoch": 0.5668449197860963,
+      "grad_norm": 2.0066540241241455,
+      "learning_rate": 1.6447093214941727e-05,
+      "loss": 0.9769,
+      "step": 318
+    },
+    {
+      "epoch": 0.5686274509803921,
+      "grad_norm": 2.08366322517395,
+      "learning_rate": 1.6425471104354694e-05,
+      "loss": 0.9421,
+      "step": 319
+    },
+    {
+      "epoch": 0.5704099821746881,
+      "grad_norm": 1.937745213508606,
+      "learning_rate": 1.6403797708236603e-05,
+      "loss": 0.9922,
+      "step": 320
+    },
+    {
+      "epoch": 0.5721925133689839,
+      "grad_norm": 1.8456331491470337,
+      "learning_rate": 1.6382073199575816e-05,
+      "loss": 0.9445,
+      "step": 321
+    },
+    {
+      "epoch": 0.5739750445632799,
+      "grad_norm": 2.1327157020568848,
+      "learning_rate": 1.636029775176862e-05,
+      "loss": 1.0098,
+      "step": 322
+    },
+    {
+      "epoch": 0.5757575757575758,
+      "grad_norm": 2.1541051864624023,
+      "learning_rate": 1.6338471538617918e-05,
+      "loss": 1.0162,
+      "step": 323
+    },
+    {
+      "epoch": 0.5775401069518716,
+      "grad_norm": 1.9894920587539673,
+      "learning_rate": 1.631659473433177e-05,
+      "loss": 1.0404,
+      "step": 324
+    },
+    {
+      "epoch": 0.5793226381461676,
+      "grad_norm": 2.2389776706695557,
+      "learning_rate": 1.6294667513522054e-05,
+      "loss": 1.0277,
+      "step": 325
+    },
+    {
+      "epoch": 0.5811051693404634,
+      "grad_norm": 2.7732322216033936,
+      "learning_rate": 1.627269005120304e-05,
+      "loss": 0.9646,
+      "step": 326
+    },
+    {
+      "epoch": 0.5828877005347594,
+      "grad_norm": 2.0119757652282715,
+      "learning_rate": 1.625066252279001e-05,
+      "loss": 1.0093,
+      "step": 327
+    },
+    {
+      "epoch": 0.5846702317290553,
+      "grad_norm": 1.7421563863754272,
+      "learning_rate": 1.622858510409785e-05,
+      "loss": 0.9558,
+      "step": 328
+    },
+    {
+      "epoch": 0.5864527629233511,
+      "grad_norm": 2.3108606338500977,
+      "learning_rate": 1.620645797133966e-05,
+      "loss": 0.9656,
+      "step": 329
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 2.1272060871124268,
+      "learning_rate": 1.618428130112533e-05,
+      "loss": 1.0287,
+      "step": 330
+    },
+    {
+      "epoch": 0.5900178253119429,
+      "grad_norm": 2.0039730072021484,
+      "learning_rate": 1.616205527046014e-05,
+      "loss": 0.9728,
+      "step": 331
+    },
+    {
+      "epoch": 0.5918003565062389,
+      "grad_norm": 2.106395721435547,
+      "learning_rate": 1.6139780056743343e-05,
+      "loss": 1.0319,
+      "step": 332
+    },
+    {
+      "epoch": 0.5935828877005348,
+      "grad_norm": 1.868335247039795,
+      "learning_rate": 1.6117455837766752e-05,
+      "loss": 0.9216,
+      "step": 333
+    },
+    {
+      "epoch": 0.5953654188948306,
+      "grad_norm": 2.4309539794921875,
+      "learning_rate": 1.6095082791713322e-05,
+      "loss": 1.0407,
+      "step": 334
+    },
+    {
+      "epoch": 0.5971479500891266,
+      "grad_norm": 2.3109192848205566,
+      "learning_rate": 1.6072661097155732e-05,
+      "loss": 1.0216,
+      "step": 335
+    },
+    {
+      "epoch": 0.5989304812834224,
+      "grad_norm": 1.9943416118621826,
+      "learning_rate": 1.6050190933054937e-05,
+      "loss": 0.9701,
+      "step": 336
+    },
+    {
+      "epoch": 0.6007130124777184,
+      "grad_norm": 2.3711092472076416,
+      "learning_rate": 1.6027672478758776e-05,
+      "loss": 0.98,
+      "step": 337
+    },
+    {
+      "epoch": 0.6024955436720143,
+      "grad_norm": 2.296501874923706,
+      "learning_rate": 1.6005105914000508e-05,
+      "loss": 0.9921,
+      "step": 338
+    },
+    {
+      "epoch": 0.6042780748663101,
+      "grad_norm": 2.1762170791625977,
+      "learning_rate": 1.5982491418897393e-05,
+      "loss": 0.9573,
+      "step": 339
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 2.417280673980713,
+      "learning_rate": 1.5959829173949256e-05,
+      "loss": 1.0293,
+      "step": 340
+    },
+    {
+      "epoch": 0.6078431372549019,
+      "grad_norm": 1.763693928718567,
+      "learning_rate": 1.5937119360037035e-05,
+      "loss": 0.9643,
+      "step": 341
+    },
+    {
+      "epoch": 0.6096256684491979,
+      "grad_norm": 2.0555365085601807,
+      "learning_rate": 1.5914362158421352e-05,
+      "loss": 1.0429,
+      "step": 342
+    },
+    {
+      "epoch": 0.6114081996434938,
+      "grad_norm": 2.018656015396118,
+      "learning_rate": 1.5891557750741054e-05,
+      "loss": 1.0253,
+      "step": 343
+    },
+    {
+      "epoch": 0.6131907308377896,
+      "grad_norm": 2.1717021465301514,
+      "learning_rate": 1.586870631901177e-05,
+      "loss": 0.9776,
+      "step": 344
+    },
+    {
+      "epoch": 0.6149732620320856,
+      "grad_norm": 2.066257953643799,
+      "learning_rate": 1.5845808045624456e-05,
+      "loss": 0.9581,
+      "step": 345
+    },
+    {
+      "epoch": 0.6167557932263814,
+      "grad_norm": 1.9698491096496582,
+      "learning_rate": 1.5822863113343934e-05,
+      "loss": 0.8828,
+      "step": 346
+    },
+    {
+      "epoch": 0.6185383244206774,
+      "grad_norm": 2.027034282684326,
+      "learning_rate": 1.5799871705307447e-05,
+      "loss": 1.041,
+      "step": 347
+    },
+    {
+      "epoch": 0.6203208556149733,
+      "grad_norm": 2.147871255874634,
+      "learning_rate": 1.5776834005023184e-05,
+      "loss": 1.0008,
+      "step": 348
+    },
+    {
+      "epoch": 0.6221033868092691,
+      "grad_norm": 2.168610095977783,
+      "learning_rate": 1.5753750196368822e-05,
+      "loss": 0.9446,
+      "step": 349
+    },
+    {
+      "epoch": 0.6238859180035651,
+      "grad_norm": 1.9053958654403687,
+      "learning_rate": 1.5730620463590052e-05,
+      "loss": 0.9966,
+      "step": 350
+    },
+    {
+      "epoch": 0.6256684491978609,
+      "grad_norm": 2.295462131500244,
+      "learning_rate": 1.5707444991299116e-05,
+      "loss": 0.9764,
+      "step": 351
+    },
+    {
+      "epoch": 0.6274509803921569,
+      "grad_norm": 1.9814562797546387,
+      "learning_rate": 1.5684223964473338e-05,
+      "loss": 0.9882,
+      "step": 352
+    },
+    {
+      "epoch": 0.6292335115864528,
+      "grad_norm": 2.3943731784820557,
+      "learning_rate": 1.566095756845362e-05,
+      "loss": 1.0871,
+      "step": 353
+    },
+    {
+      "epoch": 0.6310160427807486,
+      "grad_norm": 1.8611518144607544,
+      "learning_rate": 1.5637645988943008e-05,
+      "loss": 1.0166,
+      "step": 354
+    },
+    {
+      "epoch": 0.6327985739750446,
+      "grad_norm": 2.168264389038086,
+      "learning_rate": 1.5614289412005164e-05,
+      "loss": 0.9882,
+      "step": 355
+    },
+    {
+      "epoch": 0.6345811051693404,
+      "grad_norm": 1.9876874685287476,
+      "learning_rate": 1.559088802406292e-05,
+      "loss": 0.9658,
+      "step": 356
+    },
+    {
+      "epoch": 0.6363636363636364,
+      "grad_norm": 2.154973268508911,
+      "learning_rate": 1.5567442011896748e-05,
+      "loss": 0.9549,
+      "step": 357
+    },
+    {
+      "epoch": 0.6381461675579323,
+      "grad_norm": 1.892314076423645,
+      "learning_rate": 1.554395156264331e-05,
+      "loss": 0.9583,
+      "step": 358
+    },
+    {
+      "epoch": 0.6399286987522281,
+      "grad_norm": 2.019479274749756,
+      "learning_rate": 1.5520416863793942e-05,
+      "loss": 1.0063,
+      "step": 359
+    },
+    {
+      "epoch": 0.6417112299465241,
+      "grad_norm": 2.3134050369262695,
+      "learning_rate": 1.549683810319317e-05,
+      "loss": 0.9719,
+      "step": 360
+    },
+    {
+      "epoch": 0.64349376114082,
+      "grad_norm": 2.2110652923583984,
+      "learning_rate": 1.5473215469037187e-05,
+      "loss": 0.9595,
+      "step": 361
+    },
+    {
+      "epoch": 0.6452762923351159,
+      "grad_norm": 1.9033259153366089,
+      "learning_rate": 1.544954914987238e-05,
+      "loss": 0.9726,
+      "step": 362
+    },
+    {
+      "epoch": 0.6470588235294118,
+      "grad_norm": 1.9950685501098633,
+      "learning_rate": 1.5425839334593803e-05,
+      "loss": 0.8797,
+      "step": 363
+    },
+    {
+      "epoch": 0.6488413547237076,
+      "grad_norm": 2.0263285636901855,
+      "learning_rate": 1.5402086212443694e-05,
+      "loss": 0.9327,
+      "step": 364
+    },
+    {
+      "epoch": 0.6506238859180036,
+      "grad_norm": 2.092421531677246,
+      "learning_rate": 1.5378289973009933e-05,
+      "loss": 0.9811,
+      "step": 365
+    },
+    {
+      "epoch": 0.6524064171122995,
+      "grad_norm": 2.227092981338501,
+      "learning_rate": 1.5354450806224553e-05,
+      "loss": 0.9588,
+      "step": 366
+    },
+    {
+      "epoch": 0.6541889483065954,
+      "grad_norm": 2.0348732471466064,
+      "learning_rate": 1.533056890236221e-05,
+      "loss": 0.9289,
+      "step": 367
+    },
+    {
+      "epoch": 0.6559714795008913,
+      "grad_norm": 2.1439108848571777,
+      "learning_rate": 1.5306644452038682e-05,
+      "loss": 1.0311,
+      "step": 368
+    },
+    {
+      "epoch": 0.6577540106951871,
+      "grad_norm": 2.0412769317626953,
+      "learning_rate": 1.528267764620932e-05,
+      "loss": 1.0458,
+      "step": 369
+    },
+    {
+      "epoch": 0.6595365418894831,
+      "grad_norm": 1.9768530130386353,
+      "learning_rate": 1.5258668676167548e-05,
+      "loss": 1.0051,
+      "step": 370
+    },
+    {
+      "epoch": 0.661319073083779,
+      "grad_norm": 1.908537745475769,
+      "learning_rate": 1.5234617733543334e-05,
+      "loss": 0.9829,
+      "step": 371
+    },
+    {
+      "epoch": 0.6631016042780749,
+      "grad_norm": 2.1109814643859863,
+      "learning_rate": 1.5210525010301638e-05,
+      "loss": 0.8928,
+      "step": 372
+    },
+    {
+      "epoch": 0.6648841354723708,
+      "grad_norm": 2.2573094367980957,
+      "learning_rate": 1.5186390698740909e-05,
+      "loss": 0.9743,
+      "step": 373
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 1.938982367515564,
+      "learning_rate": 1.516221499149154e-05,
+      "loss": 0.9388,
+      "step": 374
+    },
+    {
+      "epoch": 0.6684491978609626,
+      "grad_norm": 2.141972780227661,
+      "learning_rate": 1.5137998081514313e-05,
+      "loss": 0.972,
+      "step": 375
+    },
+    {
+      "epoch": 0.6702317290552585,
+      "grad_norm": 2.0705463886260986,
+      "learning_rate": 1.5113740162098887e-05,
+      "loss": 0.9987,
+      "step": 376
+    },
+    {
+      "epoch": 0.6720142602495544,
+      "grad_norm": 2.0516183376312256,
+      "learning_rate": 1.5089441426862239e-05,
+      "loss": 0.9494,
+      "step": 377
+    },
+    {
+      "epoch": 0.6737967914438503,
+      "grad_norm": 1.9151005744934082,
+      "learning_rate": 1.5065102069747117e-05,
+      "loss": 0.9614,
+      "step": 378
+    },
+    {
+      "epoch": 0.6755793226381461,
+      "grad_norm": 2.014390468597412,
+      "learning_rate": 1.5040722285020497e-05,
+      "loss": 0.9896,
+      "step": 379
+    },
+    {
+      "epoch": 0.6773618538324421,
+      "grad_norm": 2.075321912765503,
+      "learning_rate": 1.501630226727204e-05,
+      "loss": 0.9529,
+      "step": 380
+    },
+    {
+      "epoch": 0.679144385026738,
+      "grad_norm": 1.912244200706482,
+      "learning_rate": 1.499184221141252e-05,
+      "loss": 0.9753,
+      "step": 381
+    },
+    {
+      "epoch": 0.6809269162210339,
+      "grad_norm": 1.9480012655258179,
+      "learning_rate": 1.4967342312672283e-05,
+      "loss": 0.9064,
+      "step": 382
+    },
+    {
+      "epoch": 0.6827094474153298,
+      "grad_norm": 2.0167205333709717,
+      "learning_rate": 1.4942802766599694e-05,
+      "loss": 0.898,
+      "step": 383
+    },
+    {
+      "epoch": 0.6844919786096256,
+      "grad_norm": 2.0487241744995117,
+      "learning_rate": 1.491822376905955e-05,
+      "loss": 0.9681,
+      "step": 384
+    },
+    {
+      "epoch": 0.6862745098039216,
+      "grad_norm": 2.0675833225250244,
+      "learning_rate": 1.489360551623155e-05,
+      "loss": 1.0658,
+      "step": 385
+    },
+    {
+      "epoch": 0.6880570409982175,
+      "grad_norm": 2.2603049278259277,
+      "learning_rate": 1.48689482046087e-05,
+      "loss": 1.0167,
+      "step": 386
+    },
+    {
+      "epoch": 0.6898395721925134,
+      "grad_norm": 2.2933084964752197,
+      "learning_rate": 1.4844252030995768e-05,
+      "loss": 0.9918,
+      "step": 387
+    },
+    {
+      "epoch": 0.6916221033868093,
+      "grad_norm": 2.080951690673828,
+      "learning_rate": 1.4819517192507698e-05,
+      "loss": 0.9779,
+      "step": 388
+    },
+    {
+      "epoch": 0.6934046345811051,
+      "grad_norm": 2.087494134902954,
+      "learning_rate": 1.4794743886568034e-05,
+      "loss": 1.0108,
+      "step": 389
+    },
+    {
+      "epoch": 0.6951871657754011,
+      "grad_norm": 2.1779730319976807,
+      "learning_rate": 1.4769932310907372e-05,
+      "loss": 0.9769,
+      "step": 390
+    },
+    {
+      "epoch": 0.696969696969697,
+      "grad_norm": 2.1749205589294434,
+      "learning_rate": 1.4745082663561741e-05,
+      "loss": 0.9973,
+      "step": 391
+    },
+    {
+      "epoch": 0.6987522281639929,
+      "grad_norm": 2.1819114685058594,
+      "learning_rate": 1.4720195142871054e-05,
+      "loss": 0.9246,
+      "step": 392
+    },
+    {
+      "epoch": 0.7005347593582888,
+      "grad_norm": 2.2912583351135254,
+      "learning_rate": 1.4695269947477506e-05,
+      "loss": 1.0427,
+      "step": 393
+    },
+    {
+      "epoch": 0.7023172905525846,
+      "grad_norm": 2.0110158920288086,
+      "learning_rate": 1.467030727632401e-05,
+      "loss": 0.9688,
+      "step": 394
+    },
+    {
+      "epoch": 0.7040998217468806,
+      "grad_norm": 2.1311464309692383,
+      "learning_rate": 1.4645307328652578e-05,
+      "loss": 0.9511,
+      "step": 395
+    },
+    {
+      "epoch": 0.7058823529411765,
+      "grad_norm": 2.2430477142333984,
+      "learning_rate": 1.4620270304002762e-05,
+      "loss": 1.0001,
+      "step": 396
+    },
+    {
+      "epoch": 0.7076648841354723,
+      "grad_norm": 2.282311201095581,
+      "learning_rate": 1.459519640221004e-05,
+      "loss": 1.0278,
+      "step": 397
+    },
+    {
+      "epoch": 0.7094474153297683,
+      "grad_norm": 2.0738134384155273,
+      "learning_rate": 1.4570085823404232e-05,
+      "loss": 0.9887,
+      "step": 398
+    },
+    {
+      "epoch": 0.7112299465240641,
+      "grad_norm": 2.2161929607391357,
+      "learning_rate": 1.45449387680079e-05,
+      "loss": 0.954,
+      "step": 399
+    },
+    {
+      "epoch": 0.7130124777183601,
+      "grad_norm": 1.9803860187530518,
+      "learning_rate": 1.4519755436734744e-05,
+      "loss": 0.9361,
+      "step": 400
+    },
+    {
+      "epoch": 0.714795008912656,
+      "grad_norm": 2.0878264904022217,
+      "learning_rate": 1.4494536030588003e-05,
+      "loss": 0.9482,
+      "step": 401
+    },
+    {
+      "epoch": 0.7165775401069518,
+      "grad_norm": 2.086885690689087,
+      "learning_rate": 1.4469280750858854e-05,
+      "loss": 0.9556,
+      "step": 402
+    },
+    {
+      "epoch": 0.7183600713012478,
+      "grad_norm": 1.8085094690322876,
+      "learning_rate": 1.4443989799124807e-05,
+      "loss": 0.9211,
+      "step": 403
+    },
+    {
+      "epoch": 0.7201426024955436,
+      "grad_norm": 1.8291845321655273,
+      "learning_rate": 1.4418663377248078e-05,
+      "loss": 1.0208,
+      "step": 404
+    },
+    {
+      "epoch": 0.7219251336898396,
+      "grad_norm": 2.2089219093322754,
+      "learning_rate": 1.4393301687374009e-05,
+      "loss": 0.9355,
+      "step": 405
+    },
+    {
+      "epoch": 0.7237076648841355,
+      "grad_norm": 1.7623047828674316,
+      "learning_rate": 1.4367904931929422e-05,
+      "loss": 0.9629,
+      "step": 406
+    },
+    {
+      "epoch": 0.7254901960784313,
+      "grad_norm": 2.1797492504119873,
+      "learning_rate": 1.4342473313621026e-05,
+      "loss": 0.9792,
+      "step": 407
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 1.8126072883605957,
+      "learning_rate": 1.4317007035433788e-05,
+      "loss": 0.9567,
+      "step": 408
+    },
+    {
+      "epoch": 0.7290552584670231,
+      "grad_norm": 2.481426239013672,
+      "learning_rate": 1.4291506300629322e-05,
+      "loss": 1.0067,
+      "step": 409
+    },
+    {
+      "epoch": 0.7308377896613191,
+      "grad_norm": 2.2379016876220703,
+      "learning_rate": 1.4265971312744252e-05,
+      "loss": 0.9958,
+      "step": 410
+    },
+    {
+      "epoch": 0.732620320855615,
+      "grad_norm": 1.8787038326263428,
+      "learning_rate": 1.4240402275588602e-05,
+      "loss": 0.9115,
+      "step": 411
+    },
+    {
+      "epoch": 0.7344028520499108,
+      "grad_norm": 1.802093267440796,
+      "learning_rate": 1.4214799393244167e-05,
+      "loss": 0.9095,
+      "step": 412
+    },
+    {
+      "epoch": 0.7361853832442068,
+      "grad_norm": 2.0813779830932617,
+      "learning_rate": 1.4189162870062869e-05,
+      "loss": 0.8838,
+      "step": 413
+    },
+    {
+      "epoch": 0.7379679144385026,
+      "grad_norm": 1.891223669052124,
+      "learning_rate": 1.4163492910665153e-05,
+      "loss": 0.8615,
+      "step": 414
+    },
+    {
+      "epoch": 0.7397504456327986,
+      "grad_norm": 1.8859976530075073,
+      "learning_rate": 1.4137789719938324e-05,
+      "loss": 0.9456,
+      "step": 415
+    },
+    {
+      "epoch": 0.7415329768270945,
+      "grad_norm": 1.8638348579406738,
+      "learning_rate": 1.4112053503034937e-05,
+      "loss": 0.8893,
+      "step": 416
+    },
+    {
+      "epoch": 0.7433155080213903,
+      "grad_norm": 2.4459497928619385,
+      "learning_rate": 1.4086284465371144e-05,
+      "loss": 0.9707,
+      "step": 417
+    },
+    {
+      "epoch": 0.7450980392156863,
+      "grad_norm": 2.1890289783477783,
+      "learning_rate": 1.4060482812625055e-05,
+      "loss": 1.0272,
+      "step": 418
+    },
+    {
+      "epoch": 0.7468805704099821,
+      "grad_norm": 2.002814769744873,
+      "learning_rate": 1.4034648750735109e-05,
+      "loss": 0.9049,
+      "step": 419
+    },
+    {
+      "epoch": 0.7486631016042781,
+      "grad_norm": 1.784708023071289,
+      "learning_rate": 1.4008782485898419e-05,
+      "loss": 0.8982,
+      "step": 420
+    },
+    {
+      "epoch": 0.750445632798574,
+      "grad_norm": 2.3620011806488037,
+      "learning_rate": 1.3982884224569121e-05,
+      "loss": 0.9832,
+      "step": 421
+    },
+    {
+      "epoch": 0.7522281639928698,
+      "grad_norm": 1.8406814336776733,
+      "learning_rate": 1.395695417345675e-05,
+      "loss": 0.9249,
+      "step": 422
+    },
+    {
+      "epoch": 0.7540106951871658,
+      "grad_norm": 2.3838629722595215,
+      "learning_rate": 1.393099253952456e-05,
+      "loss": 1.0131,
+      "step": 423
+    },
+    {
+      "epoch": 0.7540106951871658,
+      "eval_loss": 0.9591041803359985,
+      "eval_runtime": 329.7127,
+      "eval_samples_per_second": 8.923,
+      "eval_steps_per_second": 1.116,
+      "step": 423
+    },
+    {
+      "epoch": 0.7557932263814616,
+      "grad_norm": 1.7104860544204712,
+      "learning_rate": 1.390499952998789e-05,
+      "loss": 0.9101,
+      "step": 424
+    },
+    {
+      "epoch": 0.7575757575757576,
+      "grad_norm": 1.9570362567901611,
+      "learning_rate": 1.3878975352312511e-05,
+      "loss": 0.9648,
+      "step": 425
+    },
+    {
+      "epoch": 0.7593582887700535,
+      "grad_norm": 2.0794196128845215,
+      "learning_rate": 1.3852920214212966e-05,
+      "loss": 0.9525,
+      "step": 426
+    },
+    {
+      "epoch": 0.7611408199643493,
+      "grad_norm": 1.821881890296936,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 1.0086,
+      "step": 427
+    },
+    {
+      "epoch": 0.7629233511586453,
+      "grad_norm": 2.062831401824951,
+      "learning_rate": 1.3800717888833423e-05,
+      "loss": 0.8474,
+      "step": 428
+    },
+    {
+      "epoch": 0.7647058823529411,
+      "grad_norm": 2.1243581771850586,
+      "learning_rate": 1.3774571118211441e-05,
+      "loss": 0.9913,
+      "step": 429
+    },
+    {
+      "epoch": 0.7664884135472371,
+      "grad_norm": 1.9317386150360107,
+      "learning_rate": 1.3748394220477972e-05,
+      "loss": 0.9431,
+      "step": 430
+    },
+    {
+      "epoch": 0.768270944741533,
+      "grad_norm": 2.076014518737793,
+      "learning_rate": 1.3722187404566508e-05,
+      "loss": 0.9975,
+      "step": 431
+    },
+    {
+      "epoch": 0.7700534759358288,
+      "grad_norm": 2.021941900253296,
+      "learning_rate": 1.3695950879649337e-05,
+      "loss": 1.0482,
+      "step": 432
+    },
+    {
+      "epoch": 0.7718360071301248,
+      "grad_norm": 1.8548147678375244,
+      "learning_rate": 1.3669684855135868e-05,
+      "loss": 0.9462,
+      "step": 433
+    },
+    {
+      "epoch": 0.7736185383244206,
+      "grad_norm": 1.9138809442520142,
+      "learning_rate": 1.3643389540670963e-05,
+      "loss": 0.9553,
+      "step": 434
+    },
+    {
+      "epoch": 0.7754010695187166,
+      "grad_norm": 2.0975797176361084,
+      "learning_rate": 1.361706514613327e-05,
+      "loss": 1.0501,
+      "step": 435
+    },
+    {
+      "epoch": 0.7771836007130125,
+      "grad_norm": 1.9075837135314941,
+      "learning_rate": 1.3590711881633535e-05,
+      "loss": 0.9563,
+      "step": 436
+    },
+    {
+      "epoch": 0.7789661319073083,
+      "grad_norm": 2.0391204357147217,
+      "learning_rate": 1.3564329957512941e-05,
+      "loss": 0.9746,
+      "step": 437
+    },
+    {
+      "epoch": 0.7807486631016043,
+      "grad_norm": 1.8339051008224487,
+      "learning_rate": 1.3537919584341413e-05,
+      "loss": 0.9903,
+      "step": 438
+    },
+    {
+      "epoch": 0.7825311942959001,
+      "grad_norm": 2.02894926071167,
+      "learning_rate": 1.3511480972915946e-05,
+      "loss": 1.0049,
+      "step": 439
+    },
+    {
+      "epoch": 0.7843137254901961,
+      "grad_norm": 1.8192859888076782,
+      "learning_rate": 1.348501433425893e-05,
+      "loss": 0.9277,
+      "step": 440
+    },
+    {
+      "epoch": 0.786096256684492,
+      "grad_norm": 2.4397940635681152,
+      "learning_rate": 1.3458519879616447e-05,
+      "loss": 1.0003,
+      "step": 441
+    },
+    {
+      "epoch": 0.7878787878787878,
+      "grad_norm": 1.9915246963500977,
+      "learning_rate": 1.3431997820456592e-05,
+      "loss": 1.0435,
+      "step": 442
+    },
+    {
+      "epoch": 0.7896613190730838,
+      "grad_norm": 1.7727062702178955,
+      "learning_rate": 1.3405448368467808e-05,
+      "loss": 0.9235,
+      "step": 443
+    },
+    {
+      "epoch": 0.7914438502673797,
+      "grad_norm": 2.1558029651641846,
+      "learning_rate": 1.3378871735557156e-05,
+      "loss": 1.047,
+      "step": 444
+    },
+    {
+      "epoch": 0.7932263814616756,
+      "grad_norm": 2.189209222793579,
+      "learning_rate": 1.335226813384865e-05,
+      "loss": 1.013,
+      "step": 445
+    },
+    {
+      "epoch": 0.7950089126559715,
+      "grad_norm": 2.119126319885254,
+      "learning_rate": 1.3325637775681561e-05,
+      "loss": 0.9492,
+      "step": 446
+    },
+    {
+      "epoch": 0.7967914438502673,
+      "grad_norm": 1.9351035356521606,
+      "learning_rate": 1.329898087360872e-05,
+      "loss": 0.97,
+      "step": 447
+    },
+    {
+      "epoch": 0.7985739750445633,
+      "grad_norm": 1.7888542413711548,
+      "learning_rate": 1.3272297640394818e-05,
+      "loss": 0.9156,
+      "step": 448
+    },
+    {
+      "epoch": 0.8003565062388592,
+      "grad_norm": 2.022596836090088,
+      "learning_rate": 1.3245588289014712e-05,
+      "loss": 0.9857,
+      "step": 449
+    },
+    {
+      "epoch": 0.8021390374331551,
+      "grad_norm": 1.9733895063400269,
+      "learning_rate": 1.3218853032651719e-05,
+      "loss": 0.9402,
+      "step": 450
+    },
+    {
+      "epoch": 0.803921568627451,
+      "grad_norm": 2.3745057582855225,
+      "learning_rate": 1.3192092084695927e-05,
+      "loss": 0.9005,
+      "step": 451
+    },
+    {
+      "epoch": 0.8057040998217468,
+      "grad_norm": 2.057950496673584,
+      "learning_rate": 1.316530565874248e-05,
+      "loss": 0.933,
+      "step": 452
+    },
+    {
+      "epoch": 0.8074866310160428,
+      "grad_norm": 2.111551284790039,
+      "learning_rate": 1.3138493968589875e-05,
+      "loss": 0.9787,
+      "step": 453
+    },
+    {
+      "epoch": 0.8092691622103387,
+      "grad_norm": 2.0102903842926025,
+      "learning_rate": 1.3111657228238263e-05,
+      "loss": 0.9461,
+      "step": 454
+    },
+    {
+      "epoch": 0.8110516934046346,
+      "grad_norm": 1.9631766080856323,
+      "learning_rate": 1.3084795651887734e-05,
+      "loss": 0.9222,
+      "step": 455
+    },
+    {
+      "epoch": 0.8128342245989305,
+      "grad_norm": 2.0150415897369385,
+      "learning_rate": 1.3057909453936604e-05,
+      "loss": 0.9728,
+      "step": 456
+    },
+    {
+      "epoch": 0.8146167557932263,
+      "grad_norm": 2.039376735687256,
+      "learning_rate": 1.3030998848979714e-05,
+      "loss": 0.9848,
+      "step": 457
+    },
+    {
+      "epoch": 0.8163992869875223,
+      "grad_norm": 1.98569917678833,
+      "learning_rate": 1.3004064051806712e-05,
+      "loss": 0.9544,
+      "step": 458
+    },
+    {
+      "epoch": 0.8181818181818182,
+      "grad_norm": 1.67859947681427,
+      "learning_rate": 1.2977105277400331e-05,
+      "loss": 0.9606,
+      "step": 459
+    },
+    {
+      "epoch": 0.8199643493761141,
+      "grad_norm": 1.8734769821166992,
+      "learning_rate": 1.2950122740934691e-05,
+      "loss": 0.9627,
+      "step": 460
+    },
+    {
+      "epoch": 0.82174688057041,
+      "grad_norm": 2.455686330795288,
+      "learning_rate": 1.2923116657773571e-05,
+      "loss": 1.0661,
+      "step": 461
+    },
+    {
+      "epoch": 0.8235294117647058,
+      "grad_norm": 1.980596899986267,
+      "learning_rate": 1.2896087243468673e-05,
+      "loss": 0.9292,
+      "step": 462
+    },
+    {
+      "epoch": 0.8253119429590018,
+      "grad_norm": 2.062181234359741,
+      "learning_rate": 1.2869034713757949e-05,
+      "loss": 0.9011,
+      "step": 463
+    },
+    {
+      "epoch": 0.8270944741532977,
+      "grad_norm": 1.7144403457641602,
+      "learning_rate": 1.2841959284563818e-05,
+      "loss": 0.9019,
+      "step": 464
+    },
+    {
+      "epoch": 0.8288770053475936,
+      "grad_norm": 2.134111166000366,
+      "learning_rate": 1.2814861171991495e-05,
+      "loss": 0.9295,
+      "step": 465
+    },
+    {
+      "epoch": 0.8306595365418895,
+      "grad_norm": 2.0311472415924072,
+      "learning_rate": 1.2787740592327232e-05,
+      "loss": 0.8818,
+      "step": 466
+    },
+    {
+      "epoch": 0.8324420677361853,
+      "grad_norm": 1.967252492904663,
+      "learning_rate": 1.2760597762036614e-05,
+      "loss": 0.8939,
+      "step": 467
+    },
+    {
+      "epoch": 0.8342245989304813,
+      "grad_norm": 1.8478994369506836,
+      "learning_rate": 1.2733432897762814e-05,
+      "loss": 0.9583,
+      "step": 468
+    },
+    {
+      "epoch": 0.8360071301247772,
+      "grad_norm": 1.8894236087799072,
+      "learning_rate": 1.270624621632487e-05,
+      "loss": 0.8987,
+      "step": 469
+    },
+    {
+      "epoch": 0.8377896613190731,
+      "grad_norm": 2.0267107486724854,
+      "learning_rate": 1.267903793471597e-05,
+      "loss": 0.8838,
+      "step": 470
+    },
+    {
+      "epoch": 0.839572192513369,
+      "grad_norm": 2.0982789993286133,
+      "learning_rate": 1.2651808270101688e-05,
+      "loss": 0.9397,
+      "step": 471
+    },
+    {
+      "epoch": 0.8413547237076648,
+      "grad_norm": 2.369551420211792,
+      "learning_rate": 1.2624557439818277e-05,
+      "loss": 0.9967,
+      "step": 472
+    },
+    {
+      "epoch": 0.8431372549019608,
+      "grad_norm": 2.0669097900390625,
+      "learning_rate": 1.2597285661370928e-05,
+      "loss": 0.9919,
+      "step": 473
+    },
+    {
+      "epoch": 0.8449197860962567,
+      "grad_norm": 1.8921197652816772,
+      "learning_rate": 1.2569993152432028e-05,
+      "loss": 0.8554,
+      "step": 474
+    },
+    {
+      "epoch": 0.8467023172905526,
+      "grad_norm": 2.012568235397339,
+      "learning_rate": 1.254268013083943e-05,
+      "loss": 0.8641,
+      "step": 475
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 2.1181392669677734,
+      "learning_rate": 1.2515346814594704e-05,
+      "loss": 0.9415,
+      "step": 476
+    },
+    {
+      "epoch": 0.8502673796791443,
+      "grad_norm": 2.0450236797332764,
+      "learning_rate": 1.248799342186141e-05,
+      "loss": 0.8715,
+      "step": 477
+    },
+    {
+      "epoch": 0.8520499108734403,
+      "grad_norm": 2.0089316368103027,
+      "learning_rate": 1.2460620170963353e-05,
+      "loss": 0.9163,
+      "step": 478
+    },
+    {
+      "epoch": 0.8538324420677362,
+      "grad_norm": 2.2603275775909424,
+      "learning_rate": 1.2433227280382827e-05,
+      "loss": 0.8885,
+      "step": 479
+    },
+    {
+      "epoch": 0.8556149732620321,
+      "grad_norm": 2.2340140342712402,
+      "learning_rate": 1.2405814968758889e-05,
+      "loss": 0.9835,
+      "step": 480
+    },
+    {
+      "epoch": 0.857397504456328,
+      "grad_norm": 1.7993637323379517,
+      "learning_rate": 1.2378383454885614e-05,
+      "loss": 0.9201,
+      "step": 481
+    },
+    {
+      "epoch": 0.8591800356506238,
+      "grad_norm": 1.803969383239746,
+      "learning_rate": 1.2350932957710322e-05,
+      "loss": 0.9015,
+      "step": 482
+    },
+    {
+      "epoch": 0.8609625668449198,
+      "grad_norm": 2.1099390983581543,
+      "learning_rate": 1.2323463696331873e-05,
+      "loss": 0.9213,
+      "step": 483
+    },
+    {
+      "epoch": 0.8627450980392157,
+      "grad_norm": 2.0957653522491455,
+      "learning_rate": 1.2295975889998882e-05,
+      "loss": 0.9323,
+      "step": 484
+    },
+    {
+      "epoch": 0.8645276292335116,
+      "grad_norm": 1.8833292722702026,
+      "learning_rate": 1.226846975810798e-05,
+      "loss": 0.997,
+      "step": 485
+    },
+    {
+      "epoch": 0.8663101604278075,
+      "grad_norm": 1.7665544748306274,
+      "learning_rate": 1.2240945520202079e-05,
+      "loss": 0.9256,
+      "step": 486
+    },
+    {
+      "epoch": 0.8680926916221033,
+      "grad_norm": 1.7884703874588013,
+      "learning_rate": 1.2213403395968593e-05,
+      "loss": 0.9072,
+      "step": 487
+    },
+    {
+      "epoch": 0.8698752228163993,
+      "grad_norm": 1.8856773376464844,
+      "learning_rate": 1.2185843605237698e-05,
+      "loss": 0.9515,
+      "step": 488
+    },
+    {
+      "epoch": 0.8716577540106952,
+      "grad_norm": 1.8611061573028564,
+      "learning_rate": 1.2158266367980584e-05,
+      "loss": 0.8748,
+      "step": 489
+    },
+    {
+      "epoch": 0.8734402852049911,
+      "grad_norm": 1.7856934070587158,
+      "learning_rate": 1.2130671904307692e-05,
+      "loss": 0.838,
+      "step": 490
+    },
+    {
+      "epoch": 0.875222816399287,
+      "grad_norm": 1.7748630046844482,
+      "learning_rate": 1.2103060434466946e-05,
+      "loss": 0.9826,
+      "step": 491
+    },
+    {
+      "epoch": 0.8770053475935828,
+      "grad_norm": 2.0026159286499023,
+      "learning_rate": 1.2075432178842022e-05,
+      "loss": 0.9543,
+      "step": 492
+    },
+    {
+      "epoch": 0.8787878787878788,
+      "grad_norm": 2.055417537689209,
+      "learning_rate": 1.2047787357950561e-05,
+      "loss": 0.9127,
+      "step": 493
+    },
+    {
+      "epoch": 0.8805704099821747,
+      "grad_norm": 2.038173198699951,
+      "learning_rate": 1.202012619244243e-05,
+      "loss": 0.9169,
+      "step": 494
+    },
+    {
+      "epoch": 0.8823529411764706,
+      "grad_norm": 1.8826080560684204,
+      "learning_rate": 1.1992448903097948e-05,
+      "loss": 0.8814,
+      "step": 495
+    },
+    {
+      "epoch": 0.8841354723707665,
+      "grad_norm": 1.911270260810852,
+      "learning_rate": 1.1964755710826125e-05,
+      "loss": 0.8788,
+      "step": 496
+    },
+    {
+      "epoch": 0.8859180035650623,
+      "grad_norm": 2.083873987197876,
+      "learning_rate": 1.1937046836662906e-05,
+      "loss": 0.9855,
+      "step": 497
+    },
+    {
+      "epoch": 0.8877005347593583,
+      "grad_norm": 1.874510645866394,
+      "learning_rate": 1.1909322501769407e-05,
+      "loss": 0.841,
+      "step": 498
+    },
+    {
+      "epoch": 0.8894830659536542,
+      "grad_norm": 2.250204563140869,
+      "learning_rate": 1.1881582927430135e-05,
+      "loss": 0.968,
+      "step": 499
+    },
+    {
+      "epoch": 0.8912655971479501,
+      "grad_norm": 1.907017707824707,
+      "learning_rate": 1.1853828335051236e-05,
+      "loss": 0.8956,
+      "step": 500
+    },
+    {
+      "epoch": 0.893048128342246,
+      "grad_norm": 2.0900657176971436,
+      "learning_rate": 1.182605894615873e-05,
+      "loss": 0.9568,
+      "step": 501
+    },
+    {
+      "epoch": 0.8948306595365418,
+      "grad_norm": 1.7974416017532349,
+      "learning_rate": 1.1798274982396728e-05,
+      "loss": 0.9507,
+      "step": 502
+    },
+    {
+      "epoch": 0.8966131907308378,
+      "grad_norm": 1.8360202312469482,
+      "learning_rate": 1.1770476665525673e-05,
+      "loss": 0.9114,
+      "step": 503
+    },
+    {
+      "epoch": 0.8983957219251337,
+      "grad_norm": 2.1357533931732178,
+      "learning_rate": 1.1742664217420584e-05,
+      "loss": 0.965,
+      "step": 504
+    },
+    {
+      "epoch": 0.9001782531194296,
+      "grad_norm": 2.0421371459960938,
+      "learning_rate": 1.1714837860069243e-05,
+      "loss": 0.9259,
+      "step": 505
+    },
+    {
+      "epoch": 0.9019607843137255,
+      "grad_norm": 1.7875169515609741,
+      "learning_rate": 1.1686997815570473e-05,
+      "loss": 0.9225,
+      "step": 506
+    },
+    {
+      "epoch": 0.9037433155080213,
+      "grad_norm": 1.9849542379379272,
+      "learning_rate": 1.1659144306132332e-05,
+      "loss": 0.9419,
+      "step": 507
+    },
+    {
+      "epoch": 0.9055258467023173,
+      "grad_norm": 2.0346877574920654,
+      "learning_rate": 1.163127755407035e-05,
+      "loss": 0.9155,
+      "step": 508
+    },
+    {
+      "epoch": 0.9073083778966132,
+      "grad_norm": 1.891310453414917,
+      "learning_rate": 1.1603397781805754e-05,
+      "loss": 0.8802,
+      "step": 509
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 1.672937273979187,
+      "learning_rate": 1.15755052118637e-05,
+      "loss": 0.8794,
+      "step": 510
+    },
+    {
+      "epoch": 0.910873440285205,
+      "grad_norm": 2.178529977798462,
+      "learning_rate": 1.154760006687148e-05,
+      "loss": 0.9578,
+      "step": 511
+    },
+    {
+      "epoch": 0.9126559714795008,
+      "grad_norm": 2.18163800239563,
+      "learning_rate": 1.1519682569556758e-05,
+      "loss": 0.9322,
+      "step": 512
+    },
+    {
+      "epoch": 0.9144385026737968,
+      "grad_norm": 1.9352266788482666,
+      "learning_rate": 1.14917529427458e-05,
+      "loss": 0.8904,
+      "step": 513
+    },
+    {
+      "epoch": 0.9162210338680927,
+      "grad_norm": 2.227802038192749,
+      "learning_rate": 1.1463811409361667e-05,
+      "loss": 1.0417,
+      "step": 514
+    },
+    {
+      "epoch": 0.9180035650623886,
+      "grad_norm": 1.9616986513137817,
+      "learning_rate": 1.1435858192422464e-05,
+      "loss": 0.8749,
+      "step": 515
+    },
+    {
+      "epoch": 0.9197860962566845,
+      "grad_norm": 2.0140492916107178,
+      "learning_rate": 1.1407893515039555e-05,
+      "loss": 0.9186,
+      "step": 516
+    },
+    {
+      "epoch": 0.9215686274509803,
+      "grad_norm": 1.7624683380126953,
+      "learning_rate": 1.1379917600415758e-05,
+      "loss": 0.9289,
+      "step": 517
+    },
+    {
+      "epoch": 0.9233511586452763,
+      "grad_norm": 1.8611254692077637,
+      "learning_rate": 1.13519306718436e-05,
+      "loss": 0.924,
+      "step": 518
+    },
+    {
+      "epoch": 0.9251336898395722,
+      "grad_norm": 1.6258656978607178,
+      "learning_rate": 1.1323932952703512e-05,
+      "loss": 0.9176,
+      "step": 519
+    },
+    {
+      "epoch": 0.9269162210338681,
+      "grad_norm": 2.0411124229431152,
+      "learning_rate": 1.1295924666462042e-05,
+      "loss": 0.9044,
+      "step": 520
+    },
+    {
+      "epoch": 0.928698752228164,
+      "grad_norm": 2.3665778636932373,
+      "learning_rate": 1.1267906036670093e-05,
+      "loss": 1.0121,
+      "step": 521
+    },
+    {
+      "epoch": 0.93048128342246,
+      "grad_norm": 2.205308675765991,
+      "learning_rate": 1.1239877286961123e-05,
+      "loss": 0.9616,
+      "step": 522
+    },
+    {
+      "epoch": 0.9322638146167558,
+      "grad_norm": 1.706140398979187,
+      "learning_rate": 1.1211838641049355e-05,
+      "loss": 0.9018,
+      "step": 523
+    },
+    {
+      "epoch": 0.9340463458110517,
+      "grad_norm": 1.945457935333252,
+      "learning_rate": 1.1183790322728012e-05,
+      "loss": 1.0168,
+      "step": 524
+    },
+    {
+      "epoch": 0.9358288770053476,
+      "grad_norm": 1.9941941499710083,
+      "learning_rate": 1.1155732555867504e-05,
+      "loss": 0.9059,
+      "step": 525
+    },
+    {
+      "epoch": 0.9376114081996435,
+      "grad_norm": 2.1140549182891846,
+      "learning_rate": 1.112766556441367e-05,
+      "loss": 0.9084,
+      "step": 526
+    },
+    {
+      "epoch": 0.9393939393939394,
+      "grad_norm": 1.8418564796447754,
+      "learning_rate": 1.1099589572385968e-05,
+      "loss": 0.9572,
+      "step": 527
+    },
+    {
+      "epoch": 0.9411764705882353,
+      "grad_norm": 1.7444286346435547,
+      "learning_rate": 1.1071504803875692e-05,
+      "loss": 0.9027,
+      "step": 528
+    },
+    {
+      "epoch": 0.9429590017825312,
+      "grad_norm": 2.1305134296417236,
+      "learning_rate": 1.1043411483044193e-05,
+      "loss": 0.9695,
+      "step": 529
+    },
+    {
+      "epoch": 0.9447415329768271,
+      "grad_norm": 1.7305046319961548,
+      "learning_rate": 1.1015309834121083e-05,
+      "loss": 0.8617,
+      "step": 530
+    },
+    {
+      "epoch": 0.946524064171123,
+      "grad_norm": 1.9054243564605713,
+      "learning_rate": 1.098720008140244e-05,
+      "loss": 0.8649,
+      "step": 531
+    },
+    {
+      "epoch": 0.948306595365419,
+      "grad_norm": 1.9462075233459473,
+      "learning_rate": 1.0959082449249025e-05,
+      "loss": 0.9458,
+      "step": 532
+    },
+    {
+      "epoch": 0.9500891265597148,
+      "grad_norm": 2.21990966796875,
+      "learning_rate": 1.0930957162084496e-05,
+      "loss": 0.9427,
+      "step": 533
+    },
+    {
+      "epoch": 0.9518716577540107,
+      "grad_norm": 1.8306422233581543,
+      "learning_rate": 1.0902824444393602e-05,
+      "loss": 0.9289,
+      "step": 534
+    },
+    {
+      "epoch": 0.9536541889483066,
+      "grad_norm": 1.9472652673721313,
+      "learning_rate": 1.0874684520720405e-05,
+      "loss": 0.937,
+      "step": 535
+    },
+    {
+      "epoch": 0.9554367201426025,
+      "grad_norm": 1.8183170557022095,
+      "learning_rate": 1.0846537615666477e-05,
+      "loss": 0.9081,
+      "step": 536
+    },
+    {
+      "epoch": 0.9572192513368984,
+      "grad_norm": 1.9509567022323608,
+      "learning_rate": 1.0818383953889118e-05,
+      "loss": 0.9423,
+      "step": 537
+    },
+    {
+      "epoch": 0.9590017825311943,
+      "grad_norm": 1.9274513721466064,
+      "learning_rate": 1.079022376009955e-05,
+      "loss": 0.9819,
+      "step": 538
+    },
+    {
+      "epoch": 0.9607843137254902,
+      "grad_norm": 2.0134758949279785,
+      "learning_rate": 1.0762057259061143e-05,
+      "loss": 0.9518,
+      "step": 539
+    },
+    {
+      "epoch": 0.9625668449197861,
+      "grad_norm": 1.5931446552276611,
+      "learning_rate": 1.0733884675587595e-05,
+      "loss": 0.9256,
+      "step": 540
+    },
+    {
+      "epoch": 0.964349376114082,
+      "grad_norm": 1.8069127798080444,
+      "learning_rate": 1.0705706234541163e-05,
+      "loss": 0.9294,
+      "step": 541
+    },
+    {
+      "epoch": 0.966131907308378,
+      "grad_norm": 2.059453010559082,
+      "learning_rate": 1.067752216083085e-05,
+      "loss": 0.9317,
+      "step": 542
+    },
+    {
+      "epoch": 0.9679144385026738,
+      "grad_norm": 1.7917944192886353,
+      "learning_rate": 1.0649332679410615e-05,
+      "loss": 0.8465,
+      "step": 543
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 1.9388025999069214,
+      "learning_rate": 1.062113801527759e-05,
+      "loss": 0.8946,
+      "step": 544
+    },
+    {
+      "epoch": 0.9714795008912656,
+      "grad_norm": 1.9177924394607544,
+      "learning_rate": 1.0592938393470267e-05,
+      "loss": 0.9512,
+      "step": 545
+    },
+    {
+      "epoch": 0.9732620320855615,
+      "grad_norm": 2.0775253772735596,
+      "learning_rate": 1.05647340390667e-05,
+      "loss": 0.9381,
+      "step": 546
+    },
+    {
+      "epoch": 0.9750445632798574,
+      "grad_norm": 2.1442859172821045,
+      "learning_rate": 1.0536525177182728e-05,
+      "loss": 0.8385,
+      "step": 547
+    },
+    {
+      "epoch": 0.9768270944741533,
+      "grad_norm": 2.0453238487243652,
+      "learning_rate": 1.0508312032970165e-05,
+      "loss": 0.8533,
+      "step": 548
+    },
+    {
+      "epoch": 0.9786096256684492,
+      "grad_norm": 2.2256951332092285,
+      "learning_rate": 1.0480094831614998e-05,
+      "loss": 0.9253,
+      "step": 549
+    },
+    {
+      "epoch": 0.9803921568627451,
+      "grad_norm": 1.551547884941101,
+      "learning_rate": 1.0451873798335605e-05,
+      "loss": 0.8915,
+      "step": 550
+    },
+    {
+      "epoch": 0.982174688057041,
+      "grad_norm": 1.9802758693695068,
+      "learning_rate": 1.0423649158380947e-05,
+      "loss": 0.9275,
+      "step": 551
+    },
+    {
+      "epoch": 0.983957219251337,
+      "grad_norm": 2.0591816902160645,
+      "learning_rate": 1.0395421137028761e-05,
+      "loss": 0.929,
+      "step": 552
+    },
+    {
+      "epoch": 0.9857397504456328,
+      "grad_norm": 2.0869827270507812,
+      "learning_rate": 1.0367189959583791e-05,
+      "loss": 0.9333,
+      "step": 553
+    },
+    {
+      "epoch": 0.9875222816399287,
+      "grad_norm": 1.6645945310592651,
+      "learning_rate": 1.0338955851375962e-05,
+      "loss": 0.8624,
+      "step": 554
+    },
+    {
+      "epoch": 0.9893048128342246,
+      "grad_norm": 1.5197809934616089,
+      "learning_rate": 1.031071903775859e-05,
+      "loss": 0.8751,
+      "step": 555
+    },
+    {
+      "epoch": 0.9910873440285205,
+      "grad_norm": 1.6992182731628418,
+      "learning_rate": 1.0282479744106589e-05,
+      "loss": 0.8991,
+      "step": 556
+    },
+    {
+      "epoch": 0.9928698752228164,
+      "grad_norm": 1.7648831605911255,
+      "learning_rate": 1.0254238195814659e-05,
+      "loss": 0.9318,
+      "step": 557
+    },
+    {
+      "epoch": 0.9946524064171123,
+      "grad_norm": 1.7822153568267822,
+      "learning_rate": 1.0225994618295507e-05,
+      "loss": 1.0017,
+      "step": 558
+    },
+    {
+      "epoch": 0.9964349376114082,
+      "grad_norm": 1.8108640909194946,
+      "learning_rate": 1.0197749236978034e-05,
+      "loss": 0.8772,
+      "step": 559
+    },
+    {
+      "epoch": 0.9982174688057041,
+      "grad_norm": 1.8217051029205322,
+      "learning_rate": 1.0169502277305528e-05,
+      "loss": 0.9624,
+      "step": 560
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.1218109130859375,
+      "learning_rate": 1.0141253964733886e-05,
+      "loss": 0.9463,
+      "step": 561
+    },
+    {
+      "epoch": 1.0017825311942958,
+      "grad_norm": 2.0523557662963867,
+      "learning_rate": 1.01130045247298e-05,
+      "loss": 0.9625,
+      "step": 562
+    },
+    {
+      "epoch": 1.0035650623885919,
+      "grad_norm": 1.754342794418335,
+      "learning_rate": 1.0084754182768959e-05,
+      "loss": 0.9668,
+      "step": 563
+    },
+    {
+      "epoch": 1.0053475935828877,
+      "grad_norm": 1.6300525665283203,
+      "learning_rate": 1.0056503164334252e-05,
+      "loss": 0.8519,
+      "step": 564
+    },
+    {
+      "epoch": 1.0053475935828877,
+      "eval_loss": 0.9207794070243835,
+      "eval_runtime": 329.8091,
+      "eval_samples_per_second": 8.92,
+      "eval_steps_per_second": 1.116,
+      "step": 564
+    },
+    {
+      "epoch": 1.0071301247771836,
+      "grad_norm": 1.831165075302124,
+      "learning_rate": 1.0028251694913971e-05,
+      "loss": 0.9349,
+      "step": 565
+    },
+    {
+      "epoch": 1.0089126559714796,
+      "grad_norm": 1.8127673864364624,
+      "learning_rate": 1e-05,
+      "loss": 0.9269,
+      "step": 566
+    },
+    {
+      "epoch": 1.0106951871657754,
+      "grad_norm": 2.10707426071167,
+      "learning_rate": 9.97174830508603e-06,
+      "loss": 1.0066,
+      "step": 567
+    },
+    {
+      "epoch": 1.0124777183600713,
+      "grad_norm": 1.9743505716323853,
+      "learning_rate": 9.943496835665751e-06,
+      "loss": 0.88,
+      "step": 568
+    },
+    {
+      "epoch": 1.014260249554367,
+      "grad_norm": 2.391684055328369,
+      "learning_rate": 9.915245817231044e-06,
+      "loss": 0.89,
+      "step": 569
+    },
+    {
+      "epoch": 1.0160427807486632,
+      "grad_norm": 2.0105912685394287,
+      "learning_rate": 9.886995475270205e-06,
+      "loss": 0.8331,
+      "step": 570
+    },
+    {
+      "epoch": 1.017825311942959,
+      "grad_norm": 1.8959972858428955,
+      "learning_rate": 9.85874603526612e-06,
+      "loss": 0.9467,
+      "step": 571
+    },
+    {
+      "epoch": 1.0196078431372548,
+      "grad_norm": 1.8537402153015137,
+      "learning_rate": 9.830497722694477e-06,
+      "loss": 0.902,
+      "step": 572
+    },
+    {
+      "epoch": 1.0213903743315509,
+      "grad_norm": 1.906813383102417,
+      "learning_rate": 9.802250763021972e-06,
+      "loss": 0.9137,
+      "step": 573
+    },
+    {
+      "epoch": 1.0231729055258467,
+      "grad_norm": 1.8584880828857422,
+      "learning_rate": 9.774005381704498e-06,
+      "loss": 0.9084,
+      "step": 574
+    },
+    {
+      "epoch": 1.0249554367201426,
+      "grad_norm": 1.7670100927352905,
+      "learning_rate": 9.745761804185346e-06,
+      "loss": 0.9244,
+      "step": 575
+    },
+    {
+      "epoch": 1.0267379679144386,
+      "grad_norm": 1.6661604642868042,
+      "learning_rate": 9.717520255893415e-06,
+      "loss": 0.8834,
+      "step": 576
+    },
+    {
+      "epoch": 1.0285204991087344,
+      "grad_norm": 1.8599357604980469,
+      "learning_rate": 9.689280962241411e-06,
+      "loss": 0.9902,
+      "step": 577
+    },
+    {
+      "epoch": 1.0303030303030303,
+      "grad_norm": 1.8606723546981812,
+      "learning_rate": 9.661044148624038e-06,
+      "loss": 0.935,
+      "step": 578
+    },
+    {
+      "epoch": 1.0017825311942958,
+      "grad_norm": 1.840346336364746,
+      "learning_rate": 9.63281004041621e-06,
+      "loss": 0.7274,
+      "step": 579
+    },
+    {
+      "epoch": 1.0035650623885919,
+      "grad_norm": 1.8151859045028687,
+      "learning_rate": 9.60457886297124e-06,
+      "loss": 0.6942,
+      "step": 580
+    },
+    {
+      "epoch": 1.0053475935828877,
+      "grad_norm": 2.0746374130249023,
+      "learning_rate": 9.576350841619057e-06,
+      "loss": 0.7098,
+      "step": 581
+    },
+    {
+      "epoch": 1.0071301247771836,
+      "grad_norm": 2.0493171215057373,
+      "learning_rate": 9.548126201664398e-06,
+      "loss": 0.7328,
+      "step": 582
+    },
+    {
+      "epoch": 1.0089126559714796,
+      "grad_norm": 1.6757124662399292,
+      "learning_rate": 9.519905168385004e-06,
+      "loss": 0.6247,
+      "step": 583
+    },
+    {
+      "epoch": 1.0106951871657754,
+      "grad_norm": 1.8793476819992065,
+      "learning_rate": 9.491687967029839e-06,
+      "loss": 0.6553,
+      "step": 584
+    },
+    {
+      "epoch": 1.0124777183600713,
+      "grad_norm": 1.6432827711105347,
+      "learning_rate": 9.463474822817274e-06,
+      "loss": 0.6913,
+      "step": 585
+    },
+    {
+      "epoch": 1.014260249554367,
+      "grad_norm": 2.0046606063842773,
+      "learning_rate": 9.435265960933304e-06,
+      "loss": 0.6651,
+      "step": 586
+    },
+    {
+      "epoch": 1.0160427807486632,
+      "grad_norm": 2.0444467067718506,
+      "learning_rate": 9.407061606529736e-06,
+      "loss": 0.6596,
+      "step": 587
+    },
+    {
+      "epoch": 1.017825311942959,
+      "grad_norm": 2.309298515319824,
+      "learning_rate": 9.378861984722411e-06,
+      "loss": 0.7316,
+      "step": 588
+    },
+    {
+      "epoch": 1.0196078431372548,
+      "grad_norm": 2.2565701007843018,
+      "learning_rate": 9.350667320589387e-06,
+      "loss": 0.6892,
+      "step": 589
+    },
+    {
+      "epoch": 1.0213903743315509,
+      "grad_norm": 2.022109270095825,
+      "learning_rate": 9.322477839169156e-06,
+      "loss": 0.7209,
+      "step": 590
+    },
+    {
+      "epoch": 1.0231729055258467,
+      "grad_norm": 2.318474054336548,
+      "learning_rate": 9.294293765458844e-06,
+      "loss": 0.6836,
+      "step": 591
+    },
+    {
+      "epoch": 1.0249554367201426,
+      "grad_norm": 2.0525121688842773,
+      "learning_rate": 9.26611532441241e-06,
+      "loss": 0.6873,
+      "step": 592
+    },
+    {
+      "epoch": 1.0267379679144386,
+      "grad_norm": 2.1689891815185547,
+      "learning_rate": 9.237942740938862e-06,
+      "loss": 0.7309,
+      "step": 593
+    },
+    {
+      "epoch": 1.0285204991087344,
+      "grad_norm": 2.0695502758026123,
+      "learning_rate": 9.209776239900453e-06,
+      "loss": 0.6479,
+      "step": 594
+    },
+    {
+      "epoch": 1.0303030303030303,
+      "grad_norm": 2.0197293758392334,
+      "learning_rate": 9.181616046110889e-06,
+      "loss": 0.7159,
+      "step": 595
+    },
+    {
+      "epoch": 1.032085561497326,
+      "grad_norm": 1.8955740928649902,
+      "learning_rate": 9.153462384333525e-06,
+      "loss": 0.6547,
+      "step": 596
+    },
+    {
+      "epoch": 1.0338680926916222,
+      "grad_norm": 1.9541598558425903,
+      "learning_rate": 9.125315479279597e-06,
+      "loss": 0.6928,
+      "step": 597
+    },
+    {
+      "epoch": 1.035650623885918,
+      "grad_norm": 1.8442533016204834,
+      "learning_rate": 9.097175555606396e-06,
+      "loss": 0.6086,
+      "step": 598
+    },
+    {
+      "epoch": 1.0374331550802138,
+      "grad_norm": 1.706773042678833,
+      "learning_rate": 9.069042837915506e-06,
+      "loss": 0.6705,
+      "step": 599
+    },
+    {
+      "epoch": 1.0392156862745099,
+      "grad_norm": 1.8387624025344849,
+      "learning_rate": 9.040917550750976e-06,
+      "loss": 0.671,
+      "step": 600
+    },
+    {
+      "epoch": 1.0409982174688057,
+      "grad_norm": 1.6865826845169067,
+      "learning_rate": 9.012799918597562e-06,
+      "loss": 0.6913,
+      "step": 601
+    },
+    {
+      "epoch": 1.0427807486631016,
+      "grad_norm": 2.1576929092407227,
+      "learning_rate": 8.98469016587892e-06,
+      "loss": 0.6028,
+      "step": 602
+    },
+    {
+      "epoch": 1.0445632798573976,
+      "grad_norm": 1.873191475868225,
+      "learning_rate": 8.956588516955809e-06,
+      "loss": 0.6492,
+      "step": 603
+    },
+    {
+      "epoch": 1.0463458110516934,
+      "grad_norm": 1.9716717004776,
+      "learning_rate": 8.928495196124311e-06,
+      "loss": 0.6813,
+      "step": 604
+    },
+    {
+      "epoch": 1.0481283422459893,
+      "grad_norm": 1.963454008102417,
+      "learning_rate": 8.900410427614036e-06,
+      "loss": 0.6293,
+      "step": 605
+    },
+    {
+      "epoch": 1.049910873440285,
+      "grad_norm": 1.8422149419784546,
+      "learning_rate": 8.872334435586333e-06,
+      "loss": 0.6259,
+      "step": 606
+    },
+    {
+      "epoch": 1.0516934046345812,
+      "grad_norm": 2.425657272338867,
+      "learning_rate": 8.844267444132499e-06,
+      "loss": 0.7206,
+      "step": 607
+    },
+    {
+      "epoch": 1.053475935828877,
+      "grad_norm": 1.8881195783615112,
+      "learning_rate": 8.816209677271991e-06,
+      "loss": 0.6916,
+      "step": 608
+    },
+    {
+      "epoch": 1.0552584670231728,
+      "grad_norm": 2.1058132648468018,
+      "learning_rate": 8.788161358950649e-06,
+      "loss": 0.6965,
+      "step": 609
+    },
+    {
+      "epoch": 1.0570409982174689,
+      "grad_norm": 2.0532283782958984,
+      "learning_rate": 8.76012271303888e-06,
+      "loss": 0.6694,
+      "step": 610
+    },
+    {
+      "epoch": 1.0588235294117647,
+      "grad_norm": 1.8037793636322021,
+      "learning_rate": 8.732093963329908e-06,
+      "loss": 0.6497,
+      "step": 611
+    },
+    {
+      "epoch": 1.0606060606060606,
+      "grad_norm": 1.674838900566101,
+      "learning_rate": 8.704075333537963e-06,
+      "loss": 0.6706,
+      "step": 612
+    },
+    {
+      "epoch": 1.0623885918003566,
+      "grad_norm": 2.194537878036499,
+      "learning_rate": 8.676067047296495e-06,
+      "loss": 0.7005,
+      "step": 613
+    },
+    {
+      "epoch": 1.0641711229946524,
+      "grad_norm": 1.9183086156845093,
+      "learning_rate": 8.648069328156403e-06,
+      "loss": 0.6841,
+      "step": 614
+    },
+    {
+      "epoch": 1.0659536541889483,
+      "grad_norm": 1.9987599849700928,
+      "learning_rate": 8.620082399584247e-06,
+      "loss": 0.6381,
+      "step": 615
+    },
+    {
+      "epoch": 1.067736185383244,
+      "grad_norm": 1.9794397354125977,
+      "learning_rate": 8.592106484960447e-06,
+      "loss": 0.6248,
+      "step": 616
+    },
+    {
+      "epoch": 1.0695187165775402,
+      "grad_norm": 1.6956665515899658,
+      "learning_rate": 8.564141807577535e-06,
+      "loss": 0.671,
+      "step": 617
+    },
+    {
+      "epoch": 1.071301247771836,
+      "grad_norm": 1.9094434976577759,
+      "learning_rate": 8.536188590638334e-06,
+      "loss": 0.6372,
+      "step": 618
+    },
+    {
+      "epoch": 1.0730837789661318,
+      "grad_norm": 2.042654275894165,
+      "learning_rate": 8.5082470572542e-06,
+      "loss": 0.6743,
+      "step": 619
+    },
+    {
+      "epoch": 1.0748663101604279,
+      "grad_norm": 2.3108088970184326,
+      "learning_rate": 8.480317430443242e-06,
+      "loss": 0.7117,
+      "step": 620
+    },
+    {
+      "epoch": 1.0766488413547237,
+      "grad_norm": 2.5405666828155518,
+      "learning_rate": 8.452399933128523e-06,
+      "loss": 0.6924,
+      "step": 621
+    },
+    {
+      "epoch": 1.0784313725490196,
+      "grad_norm": 2.2125980854034424,
+      "learning_rate": 8.424494788136303e-06,
+      "loss": 0.6396,
+      "step": 622
+    },
+    {
+      "epoch": 1.0802139037433156,
+      "grad_norm": 1.8044298887252808,
+      "learning_rate": 8.396602218194248e-06,
+      "loss": 0.66,
+      "step": 623
+    },
+    {
+      "epoch": 1.0819964349376114,
+      "grad_norm": 2.01863169670105,
+      "learning_rate": 8.368722445929653e-06,
+      "loss": 0.6819,
+      "step": 624
+    },
+    {
+      "epoch": 1.0837789661319073,
+      "grad_norm": 2.040804386138916,
+      "learning_rate": 8.34085569386767e-06,
+      "loss": 0.7114,
+      "step": 625
+    },
+    {
+      "epoch": 1.085561497326203,
+      "grad_norm": 2.0602118968963623,
+      "learning_rate": 8.313002184429529e-06,
+      "loss": 0.6833,
+      "step": 626
+    },
+    {
+      "epoch": 1.0873440285204992,
+      "grad_norm": 1.781294345855713,
+      "learning_rate": 8.285162139930759e-06,
+      "loss": 0.6208,
+      "step": 627
+    },
+    {
+      "epoch": 1.089126559714795,
+      "grad_norm": 1.9819340705871582,
+      "learning_rate": 8.257335782579419e-06,
+      "loss": 0.6679,
+      "step": 628
+    },
+    {
+      "epoch": 1.0909090909090908,
+      "grad_norm": 2.2654504776000977,
+      "learning_rate": 8.229523334474328e-06,
+      "loss": 0.6833,
+      "step": 629
+    },
+    {
+      "epoch": 1.0926916221033869,
+      "grad_norm": 2.196784734725952,
+      "learning_rate": 8.201725017603277e-06,
+      "loss": 0.6615,
+      "step": 630
+    },
+    {
+      "epoch": 1.0944741532976827,
+      "grad_norm": 2.2365517616271973,
+      "learning_rate": 8.173941053841275e-06,
+      "loss": 0.6903,
+      "step": 631
+    },
+    {
+      "epoch": 1.0962566844919786,
+      "grad_norm": 2.070225954055786,
+      "learning_rate": 8.146171664948769e-06,
+      "loss": 0.6732,
+      "step": 632
+    },
+    {
+      "epoch": 1.0980392156862746,
+      "grad_norm": 1.6552367210388184,
+      "learning_rate": 8.118417072569871e-06,
+      "loss": 0.7107,
+      "step": 633
+    },
+    {
+      "epoch": 1.0998217468805704,
+      "grad_norm": 2.4717843532562256,
+      "learning_rate": 8.090677498230598e-06,
+      "loss": 0.733,
+      "step": 634
+    },
+    {
+      "epoch": 1.1016042780748663,
+      "grad_norm": 1.726193904876709,
+      "learning_rate": 8.062953163337097e-06,
+      "loss": 0.7136,
+      "step": 635
+    },
+    {
+      "epoch": 1.1033868092691623,
+      "grad_norm": 2.0040595531463623,
+      "learning_rate": 8.035244289173876e-06,
+      "loss": 0.6663,
+      "step": 636
+    },
+    {
+      "epoch": 1.1051693404634582,
+      "grad_norm": 1.8458000421524048,
+      "learning_rate": 8.007551096902055e-06,
+      "loss": 0.6524,
+      "step": 637
+    },
+    {
+      "epoch": 1.106951871657754,
+      "grad_norm": 1.8933156728744507,
+      "learning_rate": 7.97987380755757e-06,
+      "loss": 0.591,
+      "step": 638
+    },
+    {
+      "epoch": 1.1087344028520498,
+      "grad_norm": 1.9119715690612793,
+      "learning_rate": 7.95221264204944e-06,
+      "loss": 0.624,
+      "step": 639
+    },
+    {
+      "epoch": 1.1105169340463459,
+      "grad_norm": 2.19096302986145,
+      "learning_rate": 7.924567821157981e-06,
+      "loss": 0.7283,
+      "step": 640
+    },
+    {
+      "epoch": 1.1122994652406417,
+      "grad_norm": 1.7500660419464111,
+      "learning_rate": 7.896939565533056e-06,
+      "loss": 0.683,
+      "step": 641
+    },
+    {
+      "epoch": 1.1140819964349375,
+      "grad_norm": 2.377091407775879,
+      "learning_rate": 7.869328095692313e-06,
+      "loss": 0.6901,
+      "step": 642
+    },
+    {
+      "epoch": 1.1158645276292336,
+      "grad_norm": 2.178258180618286,
+      "learning_rate": 7.841733632019419e-06,
+      "loss": 0.6515,
+      "step": 643
+    },
+    {
+      "epoch": 1.1176470588235294,
+      "grad_norm": 2.0514190196990967,
+      "learning_rate": 7.814156394762306e-06,
+      "loss": 0.6545,
+      "step": 644
+    },
+    {
+      "epoch": 1.1194295900178253,
+      "grad_norm": 1.8169552087783813,
+      "learning_rate": 7.78659660403141e-06,
+      "loss": 0.6809,
+      "step": 645
+    },
+    {
+      "epoch": 1.121212121212121,
+      "grad_norm": 1.7938815355300903,
+      "learning_rate": 7.759054479797924e-06,
+      "loss": 0.5212,
+      "step": 646
+    },
+    {
+      "epoch": 1.1229946524064172,
+      "grad_norm": 1.9933950901031494,
+      "learning_rate": 7.731530241892021e-06,
+      "loss": 0.6459,
+      "step": 647
+    },
+    {
+      "epoch": 1.124777183600713,
+      "grad_norm": 1.9721087217330933,
+      "learning_rate": 7.70402411000112e-06,
+      "loss": 0.6865,
+      "step": 648
+    },
+    {
+      "epoch": 1.1265597147950088,
+      "grad_norm": 2.2990550994873047,
+      "learning_rate": 7.67653630366813e-06,
+      "loss": 0.7032,
+      "step": 649
+    },
+    {
+      "epoch": 1.1283422459893049,
+      "grad_norm": 1.9798555374145508,
+      "learning_rate": 7.649067042289681e-06,
+      "loss": 0.6512,
+      "step": 650
+    },
+    {
+      "epoch": 1.1301247771836007,
+      "grad_norm": 2.096360445022583,
+      "learning_rate": 7.621616545114392e-06,
+      "loss": 0.6178,
+      "step": 651
+    },
+    {
+      "epoch": 1.1319073083778965,
+      "grad_norm": 2.1116669178009033,
+      "learning_rate": 7.594185031241114e-06,
+      "loss": 0.5875,
+      "step": 652
+    },
+    {
+      "epoch": 1.1336898395721926,
+      "grad_norm": 1.7805746793746948,
+      "learning_rate": 7.5667727196171795e-06,
+      "loss": 0.6481,
+      "step": 653
+    },
+    {
+      "epoch": 1.1354723707664884,
+      "grad_norm": 2.0286173820495605,
+      "learning_rate": 7.539379829036652e-06,
+      "loss": 0.639,
+      "step": 654
+    },
+    {
+      "epoch": 1.1372549019607843,
+      "grad_norm": 2.071683406829834,
+      "learning_rate": 7.512006578138593e-06,
+      "loss": 0.646,
+      "step": 655
+    },
+    {
+      "epoch": 1.1390374331550803,
+      "grad_norm": 1.7731094360351562,
+      "learning_rate": 7.484653185405295e-06,
+      "loss": 0.621,
+      "step": 656
+    },
+    {
+      "epoch": 1.1408199643493762,
+      "grad_norm": 1.7044860124588013,
+      "learning_rate": 7.457319869160572e-06,
+      "loss": 0.6831,
+      "step": 657
+    },
+    {
+      "epoch": 1.142602495543672,
+      "grad_norm": 1.930659294128418,
+      "learning_rate": 7.430006847567972e-06,
+      "loss": 0.6447,
+      "step": 658
+    },
+    {
+      "epoch": 1.1443850267379678,
+      "grad_norm": 2.157649040222168,
+      "learning_rate": 7.402714338629072e-06,
+      "loss": 0.6754,
+      "step": 659
+    },
+    {
+      "epoch": 1.1461675579322639,
+      "grad_norm": 1.9548604488372803,
+      "learning_rate": 7.375442560181725e-06,
+      "loss": 0.6472,
+      "step": 660
+    },
+    {
+      "epoch": 1.1479500891265597,
+      "grad_norm": 2.099818706512451,
+      "learning_rate": 7.348191729898315e-06,
+      "loss": 0.6487,
+      "step": 661
+    },
+    {
+      "epoch": 1.1497326203208555,
+      "grad_norm": 1.9589630365371704,
+      "learning_rate": 7.320962065284032e-06,
+      "loss": 0.6708,
+      "step": 662
+    },
+    {
+      "epoch": 1.1515151515151516,
+      "grad_norm": 2.367982864379883,
+      "learning_rate": 7.293753783675132e-06,
+      "loss": 0.6582,
+      "step": 663
+    },
+    {
+      "epoch": 1.1532976827094474,
+      "grad_norm": 2.1937007904052734,
+      "learning_rate": 7.26656710223719e-06,
+      "loss": 0.641,
+      "step": 664
+    },
+    {
+      "epoch": 1.1550802139037433,
+      "grad_norm": 1.922019600868225,
+      "learning_rate": 7.239402237963389e-06,
+      "loss": 0.6647,
+      "step": 665
+    },
+    {
+      "epoch": 1.156862745098039,
+      "grad_norm": 2.0569257736206055,
+      "learning_rate": 7.2122594076727705e-06,
+      "loss": 0.6558,
+      "step": 666
+    },
+    {
+      "epoch": 1.1586452762923352,
+      "grad_norm": 1.9273213148117065,
+      "learning_rate": 7.185138828008509e-06,
+      "loss": 0.6673,
+      "step": 667
+    },
+    {
+      "epoch": 1.160427807486631,
+      "grad_norm": 2.145204544067383,
+      "learning_rate": 7.158040715436184e-06,
+      "loss": 0.639,
+      "step": 668
+    },
+    {
+      "epoch": 1.1622103386809268,
+      "grad_norm": 1.9361869096755981,
+      "learning_rate": 7.1309652862420546e-06,
+      "loss": 0.7118,
+      "step": 669
+    },
+    {
+      "epoch": 1.1639928698752229,
+      "grad_norm": 2.302577257156372,
+      "learning_rate": 7.1039127565313285e-06,
+      "loss": 0.6418,
+      "step": 670
+    },
+    {
+      "epoch": 1.1657754010695187,
+      "grad_norm": 1.851583480834961,
+      "learning_rate": 7.076883342226435e-06,
+      "loss": 0.6092,
+      "step": 671
+    },
+    {
+      "epoch": 1.1675579322638145,
+      "grad_norm": 1.97311532497406,
+      "learning_rate": 7.049877259065312e-06,
+      "loss": 0.6821,
+      "step": 672
+    },
+    {
+      "epoch": 1.1693404634581106,
+      "grad_norm": 1.8712797164916992,
+      "learning_rate": 7.022894722599673e-06,
+      "loss": 0.6464,
+      "step": 673
+    },
+    {
+      "epoch": 1.1711229946524064,
+      "grad_norm": 2.1860239505767822,
+      "learning_rate": 6.995935948193294e-06,
+      "loss": 0.6394,
+      "step": 674
+    },
+    {
+      "epoch": 1.1729055258467023,
+      "grad_norm": 2.164442539215088,
+      "learning_rate": 6.969001151020289e-06,
+      "loss": 0.665,
+      "step": 675
+    },
+    {
+      "epoch": 1.1746880570409983,
+      "grad_norm": 2.0195791721343994,
+      "learning_rate": 6.9420905460633955e-06,
+      "loss": 0.6232,
+      "step": 676
+    },
+    {
+      "epoch": 1.1764705882352942,
+      "grad_norm": 1.70195734500885,
+      "learning_rate": 6.915204348112268e-06,
+      "loss": 0.613,
+      "step": 677
+    },
+    {
+      "epoch": 1.17825311942959,
+      "grad_norm": 2.0594401359558105,
+      "learning_rate": 6.888342771761737e-06,
+      "loss": 0.6564,
+      "step": 678
+    },
+    {
+      "epoch": 1.1800356506238858,
+      "grad_norm": 1.9122473001480103,
+      "learning_rate": 6.861506031410125e-06,
+      "loss": 0.6624,
+      "step": 679
+    },
+    {
+      "epoch": 1.1818181818181819,
+      "grad_norm": 2.3307456970214844,
+      "learning_rate": 6.834694341257521e-06,
+      "loss": 0.6366,
+      "step": 680
+    },
+    {
+      "epoch": 1.1836007130124777,
+      "grad_norm": 1.771002173423767,
+      "learning_rate": 6.807907915304075e-06,
+      "loss": 0.637,
+      "step": 681
+    },
+    {
+      "epoch": 1.1853832442067735,
+      "grad_norm": 1.8012512922286987,
+      "learning_rate": 6.781146967348283e-06,
+      "loss": 0.5959,
+      "step": 682
+    },
+    {
+      "epoch": 1.1871657754010696,
+      "grad_norm": 1.9354641437530518,
+      "learning_rate": 6.754411710985291e-06,
+      "loss": 0.5782,
+      "step": 683
+    },
+    {
+      "epoch": 1.1889483065953654,
+      "grad_norm": 2.067774534225464,
+      "learning_rate": 6.727702359605185e-06,
+      "loss": 0.6635,
+      "step": 684
+    },
+    {
+      "epoch": 1.1907308377896613,
+      "grad_norm": 2.001398801803589,
+      "learning_rate": 6.701019126391282e-06,
+      "loss": 0.5945,
+      "step": 685
+    },
+    {
+      "epoch": 1.192513368983957,
+      "grad_norm": 1.8363351821899414,
+      "learning_rate": 6.6743622243184405e-06,
+      "loss": 0.6435,
+      "step": 686
+    },
+    {
+      "epoch": 1.1942959001782532,
+      "grad_norm": 1.7396736145019531,
+      "learning_rate": 6.647731866151355e-06,
+      "loss": 0.6195,
+      "step": 687
+    },
+    {
+      "epoch": 1.196078431372549,
+      "grad_norm": 2.132824659347534,
+      "learning_rate": 6.6211282644428475e-06,
+      "loss": 0.6538,
+      "step": 688
+    },
+    {
+      "epoch": 1.1978609625668448,
+      "grad_norm": 2.2380311489105225,
+      "learning_rate": 6.594551631532193e-06,
+      "loss": 0.6601,
+      "step": 689
+    },
+    {
+      "epoch": 1.1996434937611409,
+      "grad_norm": 1.9663583040237427,
+      "learning_rate": 6.568002179543409e-06,
+      "loss": 0.6255,
+      "step": 690
+    },
+    {
+      "epoch": 1.2014260249554367,
+      "grad_norm": 2.0859267711639404,
+      "learning_rate": 6.541480120383558e-06,
+      "loss": 0.6336,
+      "step": 691
+    },
+    {
+      "epoch": 1.2032085561497325,
+      "grad_norm": 1.7798055410385132,
+      "learning_rate": 6.514985665741074e-06,
+      "loss": 0.6164,
+      "step": 692
+    },
+    {
+      "epoch": 1.2049910873440286,
+      "grad_norm": 1.9800183773040771,
+      "learning_rate": 6.488519027084057e-06,
+      "loss": 0.6288,
+      "step": 693
+    },
+    {
+      "epoch": 1.2067736185383244,
+      "grad_norm": 2.3849520683288574,
+      "learning_rate": 6.462080415658591e-06,
+      "loss": 0.6915,
+      "step": 694
+    },
+    {
+      "epoch": 1.2085561497326203,
+      "grad_norm": 1.9202960729599,
+      "learning_rate": 6.435670042487063e-06,
+      "loss": 0.6595,
+      "step": 695
+    },
+    {
+      "epoch": 1.2103386809269163,
+      "grad_norm": 1.947290301322937,
+      "learning_rate": 6.409288118366465e-06,
+      "loss": 0.6351,
+      "step": 696
+    },
+    {
+      "epoch": 1.2121212121212122,
+      "grad_norm": 1.7837152481079102,
+      "learning_rate": 6.38293485386673e-06,
+      "loss": 0.5829,
+      "step": 697
+    },
+    {
+      "epoch": 1.213903743315508,
+      "grad_norm": 1.7423700094223022,
+      "learning_rate": 6.356610459329038e-06,
+      "loss": 0.6512,
+      "step": 698
+    },
+    {
+      "epoch": 1.215686274509804,
+      "grad_norm": 2.0267062187194824,
+      "learning_rate": 6.330315144864133e-06,
+      "loss": 0.6964,
+      "step": 699
+    },
+    {
+      "epoch": 1.2174688057040999,
+      "grad_norm": 2.088834762573242,
+      "learning_rate": 6.304049120350664e-06,
+      "loss": 0.6915,
+      "step": 700
+    },
+    {
+      "epoch": 1.2192513368983957,
+      "grad_norm": 2.190154790878296,
+      "learning_rate": 6.277812595433495e-06,
+      "loss": 0.6787,
+      "step": 701
+    },
+    {
+      "epoch": 1.2210338680926915,
+      "grad_norm": 2.1694934368133545,
+      "learning_rate": 6.251605779522032e-06,
+      "loss": 0.7416,
+      "step": 702
+    },
+    {
+      "epoch": 1.2228163992869876,
+      "grad_norm": 1.927525520324707,
+      "learning_rate": 6.225428881788562e-06,
+      "loss": 0.6981,
+      "step": 703
+    },
+    {
+      "epoch": 1.2245989304812834,
+      "grad_norm": 1.847321629524231,
+      "learning_rate": 6.199282111166578e-06,
+      "loss": 0.6869,
+      "step": 704
+    },
+    {
+      "epoch": 1.2263814616755793,
+      "grad_norm": 1.8871819972991943,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.5826,
+      "step": 705
+    },
+    {
+      "epoch": 1.2263814616755793,
+      "eval_loss": 0.9314715266227722,
+      "eval_runtime": 329.4034,
+      "eval_samples_per_second": 8.931,
+      "eval_steps_per_second": 1.117,
+      "step": 705
+    },
+    {
+      "epoch": 1.228163992869875,
+      "grad_norm": 1.7520660161972046,
+      "learning_rate": 6.147079785787038e-06,
+      "loss": 0.6601,
+      "step": 706
+    },
+    {
+      "epoch": 1.2299465240641712,
+      "grad_norm": 2.066180467605591,
+      "learning_rate": 6.121024647687491e-06,
+      "loss": 0.6879,
+      "step": 707
+    },
+    {
+      "epoch": 1.231729055258467,
+      "grad_norm": 1.692914366722107,
+      "learning_rate": 6.095000470012112e-06,
+      "loss": 0.6889,
+      "step": 708
+    },
+    {
+      "epoch": 1.2335115864527628,
+      "grad_norm": 1.7511357069015503,
+      "learning_rate": 6.069007460475443e-06,
+      "loss": 0.6463,
+      "step": 709
+    },
+    {
+      "epoch": 1.2352941176470589,
+      "grad_norm": 1.5913302898406982,
+      "learning_rate": 6.043045826543254e-06,
+      "loss": 0.6332,
+      "step": 710
+    },
+    {
+      "epoch": 1.2370766488413547,
+      "grad_norm": 1.9926722049713135,
+      "learning_rate": 6.017115775430882e-06,
+      "loss": 0.6227,
+      "step": 711
+    },
+    {
+      "epoch": 1.2388591800356505,
+      "grad_norm": 2.430294990539551,
+      "learning_rate": 5.991217514101586e-06,
+      "loss": 0.6776,
+      "step": 712
+    },
+    {
+      "epoch": 1.2406417112299466,
+      "grad_norm": 1.9795050621032715,
+      "learning_rate": 5.965351249264895e-06,
+      "loss": 0.6389,
+      "step": 713
+    },
+    {
+      "epoch": 1.2424242424242424,
+      "grad_norm": 1.7031593322753906,
+      "learning_rate": 5.93951718737495e-06,
+      "loss": 0.5936,
+      "step": 714
+    },
+    {
+      "epoch": 1.2442067736185383,
+      "grad_norm": 1.8579968214035034,
+      "learning_rate": 5.91371553462886e-06,
+      "loss": 0.6108,
+      "step": 715
+    },
+    {
+      "epoch": 1.2459893048128343,
+      "grad_norm": 1.869901180267334,
+      "learning_rate": 5.8879464969650645e-06,
+      "loss": 0.6401,
+      "step": 716
+    },
+    {
+      "epoch": 1.2477718360071302,
+      "grad_norm": 2.143773317337036,
+      "learning_rate": 5.862210280061676e-06,
+      "loss": 0.6346,
+      "step": 717
+    },
+    {
+      "epoch": 1.249554367201426,
+      "grad_norm": 2.0378427505493164,
+      "learning_rate": 5.836507089334849e-06,
+      "loss": 0.5339,
+      "step": 718
+    },
+    {
+      "epoch": 1.251336898395722,
+      "grad_norm": 2.2269880771636963,
+      "learning_rate": 5.81083712993713e-06,
+      "loss": 0.6808,
+      "step": 719
+    },
+    {
+      "epoch": 1.2531194295900179,
+      "grad_norm": 2.203113555908203,
+      "learning_rate": 5.785200606755839e-06,
+      "loss": 0.6719,
+      "step": 720
+    },
+    {
+      "epoch": 1.2549019607843137,
+      "grad_norm": 1.8317898511886597,
+      "learning_rate": 5.759597724411401e-06,
+      "loss": 0.6935,
+      "step": 721
+    },
+    {
+      "epoch": 1.2566844919786098,
+      "grad_norm": 1.8929369449615479,
+      "learning_rate": 5.7340286872557515e-06,
+      "loss": 0.5599,
+      "step": 722
+    },
+    {
+      "epoch": 1.2584670231729056,
+      "grad_norm": 1.679452896118164,
+      "learning_rate": 5.708493699370681e-06,
+      "loss": 0.6314,
+      "step": 723
+    },
+    {
+      "epoch": 1.2602495543672014,
+      "grad_norm": 2.099459409713745,
+      "learning_rate": 5.682992964566213e-06,
+      "loss": 0.6407,
+      "step": 724
+    },
+    {
+      "epoch": 1.2620320855614973,
+      "grad_norm": 2.3213915824890137,
+      "learning_rate": 5.657526686378975e-06,
+      "loss": 0.6363,
+      "step": 725
+    },
+    {
+      "epoch": 1.263814616755793,
+      "grad_norm": 2.0476975440979004,
+      "learning_rate": 5.6320950680705826e-06,
+      "loss": 0.7205,
+      "step": 726
+    },
+    {
+      "epoch": 1.2655971479500892,
+      "grad_norm": 2.2015089988708496,
+      "learning_rate": 5.606698312625995e-06,
+      "loss": 0.6329,
+      "step": 727
+    },
+    {
+      "epoch": 1.267379679144385,
+      "grad_norm": 1.7152899503707886,
+      "learning_rate": 5.581336622751924e-06,
+      "loss": 0.6307,
+      "step": 728
+    },
+    {
+      "epoch": 1.2691622103386808,
+      "grad_norm": 2.1059634685516357,
+      "learning_rate": 5.556010200875197e-06,
+      "loss": 0.5633,
+      "step": 729
+    },
+    {
+      "epoch": 1.2709447415329769,
+      "grad_norm": 2.0322976112365723,
+      "learning_rate": 5.530719249141148e-06,
+      "loss": 0.5963,
+      "step": 730
+    },
+    {
+      "epoch": 1.2727272727272727,
+      "grad_norm": 2.0814220905303955,
+      "learning_rate": 5.505463969412e-06,
+      "loss": 0.6359,
+      "step": 731
+    },
+    {
+      "epoch": 1.2745098039215685,
+      "grad_norm": 1.858677864074707,
+      "learning_rate": 5.480244563265263e-06,
+      "loss": 0.6106,
+      "step": 732
+    },
+    {
+      "epoch": 1.2762923351158646,
+      "grad_norm": 2.226047992706299,
+      "learning_rate": 5.4550612319921044e-06,
+      "loss": 0.6129,
+      "step": 733
+    },
+    {
+      "epoch": 1.2780748663101604,
+      "grad_norm": 2.1976687908172607,
+      "learning_rate": 5.429914176595772e-06,
+      "loss": 0.6231,
+      "step": 734
+    },
+    {
+      "epoch": 1.2798573975044563,
+      "grad_norm": 1.840461015701294,
+      "learning_rate": 5.4048035977899604e-06,
+      "loss": 0.6002,
+      "step": 735
+    },
+    {
+      "epoch": 1.2816399286987523,
+      "grad_norm": 1.7595787048339844,
+      "learning_rate": 5.3797296959972375e-06,
+      "loss": 0.6185,
+      "step": 736
+    },
+    {
+      "epoch": 1.2834224598930482,
+      "grad_norm": 2.267902374267578,
+      "learning_rate": 5.354692671347426e-06,
+      "loss": 0.6246,
+      "step": 737
+    },
+    {
+      "epoch": 1.285204991087344,
+      "grad_norm": 1.9644438028335571,
+      "learning_rate": 5.329692723675994e-06,
+      "loss": 0.6002,
+      "step": 738
+    },
+    {
+      "epoch": 1.28698752228164,
+      "grad_norm": 2.6038873195648193,
+      "learning_rate": 5.304730052522495e-06,
+      "loss": 0.683,
+      "step": 739
+    },
+    {
+      "epoch": 1.2887700534759359,
+      "grad_norm": 2.0842905044555664,
+      "learning_rate": 5.27980485712895e-06,
+      "loss": 0.6482,
+      "step": 740
+    },
+    {
+      "epoch": 1.2905525846702317,
+      "grad_norm": 1.7665643692016602,
+      "learning_rate": 5.254917336438261e-06,
+      "loss": 0.6249,
+      "step": 741
+    },
+    {
+      "epoch": 1.2923351158645278,
+      "grad_norm": 1.8952322006225586,
+      "learning_rate": 5.230067689092629e-06,
+      "loss": 0.6405,
+      "step": 742
+    },
+    {
+      "epoch": 1.2941176470588236,
+      "grad_norm": 1.8656741380691528,
+      "learning_rate": 5.205256113431969e-06,
+      "loss": 0.6718,
+      "step": 743
+    },
+    {
+      "epoch": 1.2959001782531194,
+      "grad_norm": 1.841829776763916,
+      "learning_rate": 5.180482807492309e-06,
+      "loss": 0.6472,
+      "step": 744
+    },
+    {
+      "epoch": 1.2976827094474153,
+      "grad_norm": 2.3330764770507812,
+      "learning_rate": 5.1557479690042355e-06,
+      "loss": 0.6351,
+      "step": 745
+    },
+    {
+      "epoch": 1.299465240641711,
+      "grad_norm": 2.0946245193481445,
+      "learning_rate": 5.131051795391302e-06,
+      "loss": 0.6327,
+      "step": 746
+    },
+    {
+      "epoch": 1.3012477718360071,
+      "grad_norm": 1.9958549737930298,
+      "learning_rate": 5.106394483768453e-06,
+      "loss": 0.5928,
+      "step": 747
+    },
+    {
+      "epoch": 1.303030303030303,
+      "grad_norm": 2.39805269241333,
+      "learning_rate": 5.081776230940452e-06,
+      "loss": 0.6207,
+      "step": 748
+    },
+    {
+      "epoch": 1.3048128342245988,
+      "grad_norm": 2.0922398567199707,
+      "learning_rate": 5.057197233400313e-06,
+      "loss": 0.696,
+      "step": 749
+    },
+    {
+      "epoch": 1.3065953654188949,
+      "grad_norm": 2.055389642715454,
+      "learning_rate": 5.03265768732772e-06,
+      "loss": 0.6663,
+      "step": 750
+    },
+    {
+      "epoch": 1.3083778966131907,
+      "grad_norm": 2.2379744052886963,
+      "learning_rate": 5.008157788587485e-06,
+      "loss": 0.6594,
+      "step": 751
+    },
+    {
+      "epoch": 1.3101604278074865,
+      "grad_norm": 2.141425609588623,
+      "learning_rate": 4.983697732727965e-06,
+      "loss": 0.6661,
+      "step": 752
+    },
+    {
+      "epoch": 1.3119429590017826,
+      "grad_norm": 2.284132480621338,
+      "learning_rate": 4.959277714979506e-06,
+      "loss": 0.6737,
+      "step": 753
+    },
+    {
+      "epoch": 1.3137254901960784,
+      "grad_norm": 1.9574843645095825,
+      "learning_rate": 4.934897930252887e-06,
+      "loss": 0.6836,
+      "step": 754
+    },
+    {
+      "epoch": 1.3155080213903743,
+      "grad_norm": 1.9436346292495728,
+      "learning_rate": 4.910558573137763e-06,
+      "loss": 0.6683,
+      "step": 755
+    },
+    {
+      "epoch": 1.3172905525846703,
+      "grad_norm": 1.832389235496521,
+      "learning_rate": 4.886259837901113e-06,
+      "loss": 0.6404,
+      "step": 756
+    },
+    {
+      "epoch": 1.3190730837789661,
+      "grad_norm": 1.9720847606658936,
+      "learning_rate": 4.8620019184856884e-06,
+      "loss": 0.6259,
+      "step": 757
+    },
+    {
+      "epoch": 1.320855614973262,
+      "grad_norm": 1.904499888420105,
+      "learning_rate": 4.837785008508462e-06,
+      "loss": 0.6508,
+      "step": 758
+    },
+    {
+      "epoch": 1.322638146167558,
+      "grad_norm": 1.9810409545898438,
+      "learning_rate": 4.813609301259091e-06,
+      "loss": 0.6935,
+      "step": 759
+    },
+    {
+      "epoch": 1.3244206773618539,
+      "grad_norm": 2.132432699203491,
+      "learning_rate": 4.789474989698368e-06,
+      "loss": 0.5755,
+      "step": 760
+    },
+    {
+      "epoch": 1.3262032085561497,
+      "grad_norm": 2.0095694065093994,
+      "learning_rate": 4.765382266456673e-06,
+      "loss": 0.6273,
+      "step": 761
+    },
+    {
+      "epoch": 1.3279857397504458,
+      "grad_norm": 2.2048168182373047,
+      "learning_rate": 4.7413313238324556e-06,
+      "loss": 0.6762,
+      "step": 762
+    },
+    {
+      "epoch": 1.3297682709447416,
+      "grad_norm": 1.8805530071258545,
+      "learning_rate": 4.717322353790684e-06,
+      "loss": 0.6606,
+      "step": 763
+    },
+    {
+      "epoch": 1.3315508021390374,
+      "grad_norm": 1.9276254177093506,
+      "learning_rate": 4.69335554796132e-06,
+      "loss": 0.6225,
+      "step": 764
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 1.8722336292266846,
+      "learning_rate": 4.669431097637789e-06,
+      "loss": 0.624,
+      "step": 765
+    },
+    {
+      "epoch": 1.3351158645276293,
+      "grad_norm": 1.9843535423278809,
+      "learning_rate": 4.645549193775452e-06,
+      "loss": 0.5885,
+      "step": 766
+    },
+    {
+      "epoch": 1.3368983957219251,
+      "grad_norm": 2.110941171646118,
+      "learning_rate": 4.6217100269900704e-06,
+      "loss": 0.5695,
+      "step": 767
+    },
+    {
+      "epoch": 1.338680926916221,
+      "grad_norm": 2.172638416290283,
+      "learning_rate": 4.597913787556308e-06,
+      "loss": 0.6313,
+      "step": 768
+    },
+    {
+      "epoch": 1.3404634581105168,
+      "grad_norm": 2.2388806343078613,
+      "learning_rate": 4.574160665406199e-06,
+      "loss": 0.5842,
+      "step": 769
+    },
+    {
+      "epoch": 1.3422459893048129,
+      "grad_norm": 2.3002452850341797,
+      "learning_rate": 4.550450850127626e-06,
+      "loss": 0.7563,
+      "step": 770
+    },
+    {
+      "epoch": 1.3440285204991087,
+      "grad_norm": 1.9640158414840698,
+      "learning_rate": 4.5267845309628164e-06,
+      "loss": 0.6389,
+      "step": 771
+    },
+    {
+      "epoch": 1.3458110516934045,
+      "grad_norm": 2.21022629737854,
+      "learning_rate": 4.503161896806832e-06,
+      "loss": 0.5932,
+      "step": 772
+    },
+    {
+      "epoch": 1.3475935828877006,
+      "grad_norm": 1.6235777139663696,
+      "learning_rate": 4.47958313620606e-06,
+      "loss": 0.5943,
+      "step": 773
+    },
+    {
+      "epoch": 1.3493761140819964,
+      "grad_norm": 2.571772575378418,
+      "learning_rate": 4.4560484373566945e-06,
+      "loss": 0.667,
+      "step": 774
+    },
+    {
+      "epoch": 1.3511586452762923,
+      "grad_norm": 1.9727287292480469,
+      "learning_rate": 4.4325579881032535e-06,
+      "loss": 0.59,
+      "step": 775
+    },
+    {
+      "epoch": 1.3529411764705883,
+      "grad_norm": 1.6672075986862183,
+      "learning_rate": 4.409111975937082e-06,
+      "loss": 0.5799,
+      "step": 776
+    },
+    {
+      "epoch": 1.3547237076648841,
+      "grad_norm": 1.9198261499404907,
+      "learning_rate": 4.385710587994836e-06,
+      "loss": 0.614,
+      "step": 777
+    },
+    {
+      "epoch": 1.35650623885918,
+      "grad_norm": 2.1884326934814453,
+      "learning_rate": 4.3623540110569935e-06,
+      "loss": 0.6769,
+      "step": 778
+    },
+    {
+      "epoch": 1.358288770053476,
+      "grad_norm": 2.1063883304595947,
+      "learning_rate": 4.339042431546381e-06,
+      "loss": 0.6494,
+      "step": 779
+    },
+    {
+      "epoch": 1.3600713012477719,
+      "grad_norm": 2.000030755996704,
+      "learning_rate": 4.315776035526666e-06,
+      "loss": 0.6632,
+      "step": 780
+    },
+    {
+      "epoch": 1.3618538324420677,
+      "grad_norm": 2.023421287536621,
+      "learning_rate": 4.292555008700885e-06,
+      "loss": 0.6351,
+      "step": 781
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 2.176358699798584,
+      "learning_rate": 4.26937953640995e-06,
+      "loss": 0.5972,
+      "step": 782
+    },
+    {
+      "epoch": 1.3654188948306596,
+      "grad_norm": 1.9683263301849365,
+      "learning_rate": 4.24624980363118e-06,
+      "loss": 0.602,
+      "step": 783
+    },
+    {
+      "epoch": 1.3672014260249554,
+      "grad_norm": 1.9342206716537476,
+      "learning_rate": 4.223165994976819e-06,
+      "loss": 0.6334,
+      "step": 784
+    },
+    {
+      "epoch": 1.3689839572192513,
+      "grad_norm": 1.9298633337020874,
+      "learning_rate": 4.200128294692555e-06,
+      "loss": 0.686,
+      "step": 785
+    },
+    {
+      "epoch": 1.3707664884135473,
+      "grad_norm": 2.0359370708465576,
+      "learning_rate": 4.177136886656067e-06,
+      "loss": 0.6508,
+      "step": 786
+    },
+    {
+      "epoch": 1.3725490196078431,
+      "grad_norm": 1.9444752931594849,
+      "learning_rate": 4.154191954375546e-06,
+      "loss": 0.6446,
+      "step": 787
+    },
+    {
+      "epoch": 1.374331550802139,
+      "grad_norm": 2.000025510787964,
+      "learning_rate": 4.13129368098823e-06,
+      "loss": 0.5784,
+      "step": 788
+    },
+    {
+      "epoch": 1.3761140819964348,
+      "grad_norm": 1.9926388263702393,
+      "learning_rate": 4.108442249258946e-06,
+      "loss": 0.5659,
+      "step": 789
+    },
+    {
+      "epoch": 1.3778966131907309,
+      "grad_norm": 2.011845350265503,
+      "learning_rate": 4.085637841578652e-06,
+      "loss": 0.6143,
+      "step": 790
+    },
+    {
+      "epoch": 1.3796791443850267,
+      "grad_norm": 2.0345590114593506,
+      "learning_rate": 4.062880639962969e-06,
+      "loss": 0.5627,
+      "step": 791
+    },
+    {
+      "epoch": 1.3814616755793225,
+      "grad_norm": 1.7945417165756226,
+      "learning_rate": 4.04017082605075e-06,
+      "loss": 0.6093,
+      "step": 792
+    },
+    {
+      "epoch": 1.3832442067736186,
+      "grad_norm": 2.027658700942993,
+      "learning_rate": 4.017508581102612e-06,
+      "loss": 0.5853,
+      "step": 793
+    },
+    {
+      "epoch": 1.3850267379679144,
+      "grad_norm": 2.184375047683716,
+      "learning_rate": 3.9948940859994964e-06,
+      "loss": 0.6131,
+      "step": 794
+    },
+    {
+      "epoch": 1.3868092691622103,
+      "grad_norm": 1.7773215770721436,
+      "learning_rate": 3.972327521241227e-06,
+      "loss": 0.6008,
+      "step": 795
+    },
+    {
+      "epoch": 1.3885918003565063,
+      "grad_norm": 1.9173908233642578,
+      "learning_rate": 3.949809066945064e-06,
+      "loss": 0.5749,
+      "step": 796
+    },
+    {
+      "epoch": 1.3903743315508021,
+      "grad_norm": 2.451683759689331,
+      "learning_rate": 3.927338902844271e-06,
+      "loss": 0.704,
+      "step": 797
+    },
+    {
+      "epoch": 1.392156862745098,
+      "grad_norm": 2.4320666790008545,
+      "learning_rate": 3.9049172082866786e-06,
+      "loss": 0.7077,
+      "step": 798
+    },
+    {
+      "epoch": 1.393939393939394,
+      "grad_norm": 2.23956561088562,
+      "learning_rate": 3.8825441622332496e-06,
+      "loss": 0.6582,
+      "step": 799
+    },
+    {
+      "epoch": 1.3957219251336899,
+      "grad_norm": 2.500241756439209,
+      "learning_rate": 3.8602199432566585e-06,
+      "loss": 0.6123,
+      "step": 800
+    },
+    {
+      "epoch": 1.3975044563279857,
+      "grad_norm": 1.892356276512146,
+      "learning_rate": 3.837944729539864e-06,
+      "loss": 0.6394,
+      "step": 801
+    },
+    {
+      "epoch": 1.3992869875222818,
+      "grad_norm": 2.28920841217041,
+      "learning_rate": 3.815718698874672e-06,
+      "loss": 0.59,
+      "step": 802
+    },
+    {
+      "epoch": 1.4010695187165776,
+      "grad_norm": 1.9007834196090698,
+      "learning_rate": 3.793542028660342e-06,
+      "loss": 0.6232,
+      "step": 803
+    },
+    {
+      "epoch": 1.4028520499108734,
+      "grad_norm": 2.0133168697357178,
+      "learning_rate": 3.7714148959021523e-06,
+      "loss": 0.6705,
+      "step": 804
+    },
+    {
+      "epoch": 1.4046345811051695,
+      "grad_norm": 2.347144603729248,
+      "learning_rate": 3.7493374772099944e-06,
+      "loss": 0.6269,
+      "step": 805
+    },
+    {
+      "epoch": 1.4064171122994653,
+      "grad_norm": 1.9117176532745361,
+      "learning_rate": 3.727309948796963e-06,
+      "loss": 0.6566,
+      "step": 806
+    },
+    {
+      "epoch": 1.4081996434937611,
+      "grad_norm": 2.026123046875,
+      "learning_rate": 3.7053324864779504e-06,
+      "loss": 0.6331,
+      "step": 807
+    },
+    {
+      "epoch": 1.409982174688057,
+      "grad_norm": 2.160001277923584,
+      "learning_rate": 3.6834052656682316e-06,
+      "loss": 0.615,
+      "step": 808
+    },
+    {
+      "epoch": 1.4117647058823528,
+      "grad_norm": 2.2837295532226562,
+      "learning_rate": 3.6615284613820845e-06,
+      "loss": 0.6356,
+      "step": 809
+    },
+    {
+      "epoch": 1.4135472370766489,
+      "grad_norm": 2.0919857025146484,
+      "learning_rate": 3.6397022482313804e-06,
+      "loss": 0.6751,
+      "step": 810
+    },
+    {
+      "epoch": 1.4153297682709447,
+      "grad_norm": 1.7666800022125244,
+      "learning_rate": 3.6179268004241906e-06,
+      "loss": 0.6264,
+      "step": 811
+    },
+    {
+      "epoch": 1.4171122994652405,
+      "grad_norm": 1.8978486061096191,
+      "learning_rate": 3.5962022917633977e-06,
+      "loss": 0.6581,
+      "step": 812
+    },
+    {
+      "epoch": 1.4188948306595366,
+      "grad_norm": 1.7779465913772583,
+      "learning_rate": 3.5745288956453126e-06,
+      "loss": 0.6369,
+      "step": 813
+    },
+    {
+      "epoch": 1.4206773618538324,
+      "grad_norm": 1.783486247062683,
+      "learning_rate": 3.552906785058278e-06,
+      "loss": 0.6153,
+      "step": 814
+    },
+    {
+      "epoch": 1.4224598930481283,
+      "grad_norm": 2.0954372882843018,
+      "learning_rate": 3.531336132581299e-06,
+      "loss": 0.6021,
+      "step": 815
+    },
+    {
+      "epoch": 1.4242424242424243,
+      "grad_norm": 2.061905860900879,
+      "learning_rate": 3.5098171103826716e-06,
+      "loss": 0.6079,
+      "step": 816
+    },
+    {
+      "epoch": 1.4260249554367201,
+      "grad_norm": 2.2032501697540283,
+      "learning_rate": 3.488349890218592e-06,
+      "loss": 0.6178,
+      "step": 817
+    },
+    {
+      "epoch": 1.427807486631016,
+      "grad_norm": 2.141941547393799,
+      "learning_rate": 3.466934643431795e-06,
+      "loss": 0.6102,
+      "step": 818
+    },
+    {
+      "epoch": 1.429590017825312,
+      "grad_norm": 2.2551662921905518,
+      "learning_rate": 3.4455715409501776e-06,
+      "loss": 0.639,
+      "step": 819
+    },
+    {
+      "epoch": 1.4313725490196079,
+      "grad_norm": 2.1851370334625244,
+      "learning_rate": 3.4242607532854533e-06,
+      "loss": 0.6023,
+      "step": 820
+    },
+    {
+      "epoch": 1.4331550802139037,
+      "grad_norm": 2.0705204010009766,
+      "learning_rate": 3.403002450531773e-06,
+      "loss": 0.5642,
+      "step": 821
+    },
+    {
+      "epoch": 1.4349376114081998,
+      "grad_norm": 2.305468797683716,
+      "learning_rate": 3.3817968023643766e-06,
+      "loss": 0.6493,
+      "step": 822
+    },
+    {
+      "epoch": 1.4367201426024956,
+      "grad_norm": 2.520127296447754,
+      "learning_rate": 3.3606439780382348e-06,
+      "loss": 0.7074,
+      "step": 823
+    },
+    {
+      "epoch": 1.4385026737967914,
+      "grad_norm": 2.0943031311035156,
+      "learning_rate": 3.339544146386704e-06,
+      "loss": 0.6028,
+      "step": 824
+    },
+    {
+      "epoch": 1.4402852049910875,
+      "grad_norm": 2.1441266536712646,
+      "learning_rate": 3.3184974758201637e-06,
+      "loss": 0.5351,
+      "step": 825
+    },
+    {
+      "epoch": 1.4420677361853833,
+      "grad_norm": 2.0502147674560547,
+      "learning_rate": 3.2975041343246937e-06,
+      "loss": 0.6036,
+      "step": 826
+    },
+    {
+      "epoch": 1.4438502673796791,
+      "grad_norm": 2.0448861122131348,
+      "learning_rate": 3.2765642894607186e-06,
+      "loss": 0.6573,
+      "step": 827
+    },
+    {
+      "epoch": 1.445632798573975,
+      "grad_norm": 1.9928828477859497,
+      "learning_rate": 3.2556781083616727e-06,
+      "loss": 0.5587,
+      "step": 828
+    },
+    {
+      "epoch": 1.4474153297682708,
+      "grad_norm": 1.7939866781234741,
+      "learning_rate": 3.234845757732673e-06,
+      "loss": 0.6953,
+      "step": 829
+    },
+    {
+      "epoch": 1.4491978609625669,
+      "grad_norm": 2.1059212684631348,
+      "learning_rate": 3.214067403849179e-06,
+      "loss": 0.6814,
+      "step": 830
+    },
+    {
+      "epoch": 1.4509803921568627,
+      "grad_norm": 2.4292821884155273,
+      "learning_rate": 3.193343212555667e-06,
+      "loss": 0.6433,
+      "step": 831
+    },
+    {
+      "epoch": 1.4527629233511585,
+      "grad_norm": 1.789070725440979,
+      "learning_rate": 3.1726733492643157e-06,
+      "loss": 0.5441,
+      "step": 832
+    },
+    {
+      "epoch": 1.4545454545454546,
+      "grad_norm": 1.9609915018081665,
+      "learning_rate": 3.1520579789536753e-06,
+      "loss": 0.6271,
+      "step": 833
+    },
+    {
+      "epoch": 1.4563279857397504,
+      "grad_norm": 1.9168120622634888,
+      "learning_rate": 3.1314972661673572e-06,
+      "loss": 0.6025,
+      "step": 834
+    },
+    {
+      "epoch": 1.4581105169340463,
+      "grad_norm": 2.104306221008301,
+      "learning_rate": 3.1109913750127175e-06,
+      "loss": 0.6286,
+      "step": 835
+    },
+    {
+      "epoch": 1.4598930481283423,
+      "grad_norm": 1.847529411315918,
+      "learning_rate": 3.0905404691595476e-06,
+      "loss": 0.6097,
+      "step": 836
+    },
+    {
+      "epoch": 1.4616755793226381,
+      "grad_norm": 1.9516671895980835,
+      "learning_rate": 3.0701447118387673e-06,
+      "loss": 0.6254,
+      "step": 837
+    },
+    {
+      "epoch": 1.463458110516934,
+      "grad_norm": 2.1860153675079346,
+      "learning_rate": 3.0498042658411276e-06,
+      "loss": 0.5936,
+      "step": 838
+    },
+    {
+      "epoch": 1.46524064171123,
+      "grad_norm": 2.162168502807617,
+      "learning_rate": 3.0295192935159e-06,
+      "loss": 0.6193,
+      "step": 839
+    },
+    {
+      "epoch": 1.4670231729055259,
+      "grad_norm": 1.9501457214355469,
+      "learning_rate": 3.009289956769593e-06,
+      "loss": 0.6031,
+      "step": 840
+    },
+    {
+      "epoch": 1.4688057040998217,
+      "grad_norm": 2.083054542541504,
+      "learning_rate": 2.989116417064655e-06,
+      "loss": 0.5807,
+      "step": 841
+    },
+    {
+      "epoch": 1.4705882352941178,
+      "grad_norm": 2.124634027481079,
+      "learning_rate": 2.9689988354181742e-06,
+      "loss": 0.6467,
+      "step": 842
+    },
+    {
+      "epoch": 1.4723707664884136,
+      "grad_norm": 2.0739710330963135,
+      "learning_rate": 2.9489373724006164e-06,
+      "loss": 0.6345,
+      "step": 843
+    },
+    {
+      "epoch": 1.4741532976827094,
+      "grad_norm": 1.7244806289672852,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.6698,
+      "step": 844
+    },
+    {
+      "epoch": 1.4759358288770055,
+      "grad_norm": 1.802803874015808,
+      "learning_rate": 2.908983442293253e-06,
+      "loss": 0.6248,
+      "step": 845
+    },
+    {
+      "epoch": 1.4777183600713013,
+      "grad_norm": 2.0008552074432373,
+      "learning_rate": 2.8890912940996784e-06,
+      "loss": 0.6227,
+      "step": 846
+    },
+    {
+      "epoch": 1.4777183600713013,
+      "eval_loss": 0.9287152290344238,
+      "eval_runtime": 329.6324,
+      "eval_samples_per_second": 8.925,
+      "eval_steps_per_second": 1.116,
+      "step": 846
+    },
+    {
+      "epoch": 1.4795008912655971,
+      "grad_norm": 2.082883358001709,
+      "learning_rate": 2.8692559023249457e-06,
+      "loss": 0.6235,
+      "step": 847
+    },
+    {
+      "epoch": 1.481283422459893,
+      "grad_norm": 2.0350584983825684,
+      "learning_rate": 2.8494774252871913e-06,
+      "loss": 0.603,
+      "step": 848
+    },
+    {
+      "epoch": 1.483065953654189,
+      "grad_norm": 2.247058868408203,
+      "learning_rate": 2.829756020850274e-06,
+      "loss": 0.6736,
+      "step": 849
+    },
+    {
+      "epoch": 1.4848484848484849,
+      "grad_norm": 2.168816328048706,
+      "learning_rate": 2.8100918464225304e-06,
+      "loss": 0.652,
+      "step": 850
+    },
+    {
+      "epoch": 1.4866310160427807,
+      "grad_norm": 2.2460670471191406,
+      "learning_rate": 2.7904850589555065e-06,
+      "loss": 0.6744,
+      "step": 851
+    },
+    {
+      "epoch": 1.4884135472370765,
+      "grad_norm": 1.918690800666809,
+      "learning_rate": 2.7709358149427114e-06,
+      "loss": 0.6582,
+      "step": 852
+    },
+    {
+      "epoch": 1.4901960784313726,
+      "grad_norm": 2.372283458709717,
+      "learning_rate": 2.751444270418361e-06,
+      "loss": 0.6133,
+      "step": 853
+    },
+    {
+      "epoch": 1.4919786096256684,
+      "grad_norm": 1.9066613912582397,
+      "learning_rate": 2.7320105809561415e-06,
+      "loss": 0.605,
+      "step": 854
+    },
+    {
+      "epoch": 1.4937611408199643,
+      "grad_norm": 2.066586494445801,
+      "learning_rate": 2.7126349016679587e-06,
+      "loss": 0.6999,
+      "step": 855
+    },
+    {
+      "epoch": 1.4955436720142603,
+      "grad_norm": 2.1307895183563232,
+      "learning_rate": 2.6933173872027096e-06,
+      "loss": 0.6343,
+      "step": 856
+    },
+    {
+      "epoch": 1.4973262032085561,
+      "grad_norm": 2.0337185859680176,
+      "learning_rate": 2.674058191745038e-06,
+      "loss": 0.6076,
+      "step": 857
+    },
+    {
+      "epoch": 1.499108734402852,
+      "grad_norm": 2.2143378257751465,
+      "learning_rate": 2.654857469014113e-06,
+      "loss": 0.6181,
+      "step": 858
+    },
+    {
+      "epoch": 1.500891265597148,
+      "grad_norm": 2.016242265701294,
+      "learning_rate": 2.6357153722623976e-06,
+      "loss": 0.66,
+      "step": 859
+    },
+    {
+      "epoch": 1.5026737967914439,
+      "grad_norm": 2.317138433456421,
+      "learning_rate": 2.6166320542744184e-06,
+      "loss": 0.6371,
+      "step": 860
+    },
+    {
+      "epoch": 1.5044563279857397,
+      "grad_norm": 1.8911103010177612,
+      "learning_rate": 2.5976076673655628e-06,
+      "loss": 0.6335,
+      "step": 861
+    },
+    {
+      "epoch": 1.5062388591800357,
+      "grad_norm": 2.6606857776641846,
+      "learning_rate": 2.5786423633808487e-06,
+      "loss": 0.6414,
+      "step": 862
+    },
+    {
+      "epoch": 1.5080213903743316,
+      "grad_norm": 1.818349003791809,
+      "learning_rate": 2.559736293693721e-06,
+      "loss": 0.6156,
+      "step": 863
+    },
+    {
+      "epoch": 1.5098039215686274,
+      "grad_norm": 1.7764534950256348,
+      "learning_rate": 2.5408896092048384e-06,
+      "loss": 0.6441,
+      "step": 864
+    },
+    {
+      "epoch": 1.5115864527629235,
+      "grad_norm": 2.2821059226989746,
+      "learning_rate": 2.522102460340874e-06,
+      "loss": 0.6253,
+      "step": 865
+    },
+    {
+      "epoch": 1.5133689839572193,
+      "grad_norm": 2.0185694694519043,
+      "learning_rate": 2.5033749970533015e-06,
+      "loss": 0.6091,
+      "step": 866
+    },
+    {
+      "epoch": 1.5151515151515151,
+      "grad_norm": 2.3071510791778564,
+      "learning_rate": 2.484707368817221e-06,
+      "loss": 0.6268,
+      "step": 867
+    },
+    {
+      "epoch": 1.5169340463458112,
+      "grad_norm": 2.0063817501068115,
+      "learning_rate": 2.4660997246301445e-06,
+      "loss": 0.6439,
+      "step": 868
+    },
+    {
+      "epoch": 1.5187165775401068,
+      "grad_norm": 2.095769166946411,
+      "learning_rate": 2.447552213010821e-06,
+      "loss": 0.6397,
+      "step": 869
+    },
+    {
+      "epoch": 1.5204991087344029,
+      "grad_norm": 2.276026487350464,
+      "learning_rate": 2.4290649819980404e-06,
+      "loss": 0.6755,
+      "step": 870
+    },
+    {
+      "epoch": 1.522281639928699,
+      "grad_norm": 2.0926215648651123,
+      "learning_rate": 2.410638179149465e-06,
+      "loss": 0.581,
+      "step": 871
+    },
+    {
+      "epoch": 1.5240641711229945,
+      "grad_norm": 1.8585364818572998,
+      "learning_rate": 2.3922719515404303e-06,
+      "loss": 0.6108,
+      "step": 872
+    },
+    {
+      "epoch": 1.5258467023172906,
+      "grad_norm": 2.319448709487915,
+      "learning_rate": 2.3739664457627974e-06,
+      "loss": 0.6521,
+      "step": 873
+    },
+    {
+      "epoch": 1.5276292335115864,
+      "grad_norm": 2.0976061820983887,
+      "learning_rate": 2.3557218079237608e-06,
+      "loss": 0.6235,
+      "step": 874
+    },
+    {
+      "epoch": 1.5294117647058822,
+      "grad_norm": 2.241476058959961,
+      "learning_rate": 2.337538183644702e-06,
+      "loss": 0.612,
+      "step": 875
+    },
+    {
+      "epoch": 1.5311942959001783,
+      "grad_norm": 2.2935774326324463,
+      "learning_rate": 2.319415718060011e-06,
+      "loss": 0.6827,
+      "step": 876
+    },
+    {
+      "epoch": 1.5329768270944741,
+      "grad_norm": 2.440673589706421,
+      "learning_rate": 2.301354555815928e-06,
+      "loss": 0.6346,
+      "step": 877
+    },
+    {
+      "epoch": 1.53475935828877,
+      "grad_norm": 2.0865209102630615,
+      "learning_rate": 2.283354841069403e-06,
+      "loss": 0.6396,
+      "step": 878
+    },
+    {
+      "epoch": 1.536541889483066,
+      "grad_norm": 2.1622796058654785,
+      "learning_rate": 2.2654167174869325e-06,
+      "loss": 0.6198,
+      "step": 879
+    },
+    {
+      "epoch": 1.5383244206773619,
+      "grad_norm": 2.0409114360809326,
+      "learning_rate": 2.2475403282434193e-06,
+      "loss": 0.6023,
+      "step": 880
+    },
+    {
+      "epoch": 1.5401069518716577,
+      "grad_norm": 2.097445249557495,
+      "learning_rate": 2.2297258160210255e-06,
+      "loss": 0.6092,
+      "step": 881
+    },
+    {
+      "epoch": 1.5418894830659537,
+      "grad_norm": 2.039428234100342,
+      "learning_rate": 2.211973323008041e-06,
+      "loss": 0.6371,
+      "step": 882
+    },
+    {
+      "epoch": 1.5436720142602496,
+      "grad_norm": 1.863448977470398,
+      "learning_rate": 2.1942829908977315e-06,
+      "loss": 0.6145,
+      "step": 883
+    },
+    {
+      "epoch": 1.5454545454545454,
+      "grad_norm": 1.8281656503677368,
+      "learning_rate": 2.176654960887231e-06,
+      "loss": 0.6095,
+      "step": 884
+    },
+    {
+      "epoch": 1.5472370766488415,
+      "grad_norm": 1.9864037036895752,
+      "learning_rate": 2.1590893736763974e-06,
+      "loss": 0.6077,
+      "step": 885
+    },
+    {
+      "epoch": 1.5490196078431373,
+      "grad_norm": 1.857020616531372,
+      "learning_rate": 2.1415863694666973e-06,
+      "loss": 0.602,
+      "step": 886
+    },
+    {
+      "epoch": 1.5508021390374331,
+      "grad_norm": 1.8649786710739136,
+      "learning_rate": 2.124146087960085e-06,
+      "loss": 0.6169,
+      "step": 887
+    },
+    {
+      "epoch": 1.5525846702317292,
+      "grad_norm": 1.9944473505020142,
+      "learning_rate": 2.106768668357888e-06,
+      "loss": 0.5788,
+      "step": 888
+    },
+    {
+      "epoch": 1.5543672014260248,
+      "grad_norm": 2.0956411361694336,
+      "learning_rate": 2.089454249359689e-06,
+      "loss": 0.6517,
+      "step": 889
+    },
+    {
+      "epoch": 1.5561497326203209,
+      "grad_norm": 2.172884702682495,
+      "learning_rate": 2.072202969162234e-06,
+      "loss": 0.6386,
+      "step": 890
+    },
+    {
+      "epoch": 1.557932263814617,
+      "grad_norm": 2.040494203567505,
+      "learning_rate": 2.055014965458314e-06,
+      "loss": 0.6114,
+      "step": 891
+    },
+    {
+      "epoch": 1.5597147950089125,
+      "grad_norm": 1.7873563766479492,
+      "learning_rate": 2.0378903754356772e-06,
+      "loss": 0.6609,
+      "step": 892
+    },
+    {
+      "epoch": 1.5614973262032086,
+      "grad_norm": 2.231397867202759,
+      "learning_rate": 2.0208293357759266e-06,
+      "loss": 0.6615,
+      "step": 893
+    },
+    {
+      "epoch": 1.5632798573975044,
+      "grad_norm": 1.6660290956497192,
+      "learning_rate": 2.0038319826534312e-06,
+      "loss": 0.5974,
+      "step": 894
+    },
+    {
+      "epoch": 1.5650623885918002,
+      "grad_norm": 2.1246089935302734,
+      "learning_rate": 1.986898451734243e-06,
+      "loss": 0.6234,
+      "step": 895
+    },
+    {
+      "epoch": 1.5668449197860963,
+      "grad_norm": 2.0004465579986572,
+      "learning_rate": 1.9700288781750043e-06,
+      "loss": 0.5632,
+      "step": 896
+    },
+    {
+      "epoch": 1.5686274509803921,
+      "grad_norm": 2.172769546508789,
+      "learning_rate": 1.953223396621884e-06,
+      "loss": 0.6343,
+      "step": 897
+    },
+    {
+      "epoch": 1.570409982174688,
+      "grad_norm": 2.3006234169006348,
+      "learning_rate": 1.936482141209486e-06,
+      "loss": 0.603,
+      "step": 898
+    },
+    {
+      "epoch": 1.572192513368984,
+      "grad_norm": 1.8977664709091187,
+      "learning_rate": 1.919805245559796e-06,
+      "loss": 0.6998,
+      "step": 899
+    },
+    {
+      "epoch": 1.5739750445632799,
+      "grad_norm": 1.7701358795166016,
+      "learning_rate": 1.903192842781094e-06,
+      "loss": 0.6292,
+      "step": 900
+    },
+    {
+      "epoch": 1.5757575757575757,
+      "grad_norm": 2.157860040664673,
+      "learning_rate": 1.8866450654669133e-06,
+      "loss": 0.6096,
+      "step": 901
+    },
+    {
+      "epoch": 1.5775401069518717,
+      "grad_norm": 1.5891040563583374,
+      "learning_rate": 1.870162045694971e-06,
+      "loss": 0.5585,
+      "step": 902
+    },
+    {
+      "epoch": 1.5793226381461676,
+      "grad_norm": 2.1544394493103027,
+      "learning_rate": 1.8537439150261126e-06,
+      "loss": 0.6436,
+      "step": 903
+    },
+    {
+      "epoch": 1.5811051693404634,
+      "grad_norm": 1.7956115007400513,
+      "learning_rate": 1.8373908045032685e-06,
+      "loss": 0.5886,
+      "step": 904
+    },
+    {
+      "epoch": 1.5828877005347595,
+      "grad_norm": 2.07564115524292,
+      "learning_rate": 1.821102844650403e-06,
+      "loss": 0.577,
+      "step": 905
+    },
+    {
+      "epoch": 1.5846702317290553,
+      "grad_norm": 2.2092342376708984,
+      "learning_rate": 1.8048801654714687e-06,
+      "loss": 0.6499,
+      "step": 906
+    },
+    {
+      "epoch": 1.5864527629233511,
+      "grad_norm": 2.127253532409668,
+      "learning_rate": 1.788722896449383e-06,
+      "loss": 0.5997,
+      "step": 907
+    },
+    {
+      "epoch": 1.5882352941176472,
+      "grad_norm": 2.1936185359954834,
+      "learning_rate": 1.772631166544979e-06,
+      "loss": 0.63,
+      "step": 908
+    },
+    {
+      "epoch": 1.5900178253119428,
+      "grad_norm": 1.8700515031814575,
+      "learning_rate": 1.7566051041959864e-06,
+      "loss": 0.6522,
+      "step": 909
+    },
+    {
+      "epoch": 1.5918003565062389,
+      "grad_norm": 2.1035735607147217,
+      "learning_rate": 1.7406448373160024e-06,
+      "loss": 0.6252,
+      "step": 910
+    },
+    {
+      "epoch": 1.593582887700535,
+      "grad_norm": 1.6864609718322754,
+      "learning_rate": 1.7247504932934688e-06,
+      "loss": 0.6087,
+      "step": 911
+    },
+    {
+      "epoch": 1.5953654188948305,
+      "grad_norm": 2.179560661315918,
+      "learning_rate": 1.7089221989906634e-06,
+      "loss": 0.6516,
+      "step": 912
+    },
+    {
+      "epoch": 1.5971479500891266,
+      "grad_norm": 1.7755348682403564,
+      "learning_rate": 1.693160080742673e-06,
+      "loss": 0.5788,
+      "step": 913
+    },
+    {
+      "epoch": 1.5989304812834224,
+      "grad_norm": 1.8449007272720337,
+      "learning_rate": 1.6774642643563955e-06,
+      "loss": 0.5863,
+      "step": 914
+    },
+    {
+      "epoch": 1.6007130124777182,
+      "grad_norm": 2.1267855167388916,
+      "learning_rate": 1.6618348751095448e-06,
+      "loss": 0.5896,
+      "step": 915
+    },
+    {
+      "epoch": 1.6024955436720143,
+      "grad_norm": 1.7636414766311646,
+      "learning_rate": 1.64627203774963e-06,
+      "loss": 0.6364,
+      "step": 916
+    },
+    {
+      "epoch": 1.6042780748663101,
+      "grad_norm": 1.9327129125595093,
+      "learning_rate": 1.630775876492967e-06,
+      "loss": 0.5962,
+      "step": 917
+    },
+    {
+      "epoch": 1.606060606060606,
+      "grad_norm": 2.0896499156951904,
+      "learning_rate": 1.615346515023698e-06,
+      "loss": 0.6945,
+      "step": 918
+    },
+    {
+      "epoch": 1.607843137254902,
+      "grad_norm": 1.82899808883667,
+      "learning_rate": 1.599984076492791e-06,
+      "loss": 0.5981,
+      "step": 919
+    },
+    {
+      "epoch": 1.6096256684491979,
+      "grad_norm": 1.8291279077529907,
+      "learning_rate": 1.5846886835170649e-06,
+      "loss": 0.6005,
+      "step": 920
+    },
+    {
+      "epoch": 1.6114081996434937,
+      "grad_norm": 2.2317283153533936,
+      "learning_rate": 1.5694604581782059e-06,
+      "loss": 0.6218,
+      "step": 921
+    },
+    {
+      "epoch": 1.6131907308377897,
+      "grad_norm": 2.006565809249878,
+      "learning_rate": 1.5542995220217961e-06,
+      "loss": 0.6008,
+      "step": 922
+    },
+    {
+      "epoch": 1.6149732620320856,
+      "grad_norm": 2.1319220066070557,
+      "learning_rate": 1.5392059960563444e-06,
+      "loss": 0.6309,
+      "step": 923
+    },
+    {
+      "epoch": 1.6167557932263814,
+      "grad_norm": 1.8702588081359863,
+      "learning_rate": 1.5241800007523123e-06,
+      "loss": 0.6081,
+      "step": 924
+    },
+    {
+      "epoch": 1.6185383244206775,
+      "grad_norm": 2.236720561981201,
+      "learning_rate": 1.5092216560411631e-06,
+      "loss": 0.6133,
+      "step": 925
+    },
+    {
+      "epoch": 1.6203208556149733,
+      "grad_norm": 2.221240282058716,
+      "learning_rate": 1.4943310813144006e-06,
+      "loss": 0.6624,
+      "step": 926
+    },
+    {
+      "epoch": 1.6221033868092691,
+      "grad_norm": 1.9202772378921509,
+      "learning_rate": 1.4795083954226153e-06,
+      "loss": 0.6069,
+      "step": 927
+    },
+    {
+      "epoch": 1.6238859180035652,
+      "grad_norm": 2.2788867950439453,
+      "learning_rate": 1.464753716674535e-06,
+      "loss": 0.6444,
+      "step": 928
+    },
+    {
+      "epoch": 1.6256684491978608,
+      "grad_norm": 1.9460335969924927,
+      "learning_rate": 1.4500671628360863e-06,
+      "loss": 0.541,
+      "step": 929
+    },
+    {
+      "epoch": 1.6274509803921569,
+      "grad_norm": 2.113252878189087,
+      "learning_rate": 1.4354488511294418e-06,
+      "loss": 0.6688,
+      "step": 930
+    },
+    {
+      "epoch": 1.629233511586453,
+      "grad_norm": 1.8469072580337524,
+      "learning_rate": 1.420898898232098e-06,
+      "loss": 0.6,
+      "step": 931
+    },
+    {
+      "epoch": 1.6310160427807485,
+      "grad_norm": 1.9188716411590576,
+      "learning_rate": 1.4064174202759405e-06,
+      "loss": 0.638,
+      "step": 932
+    },
+    {
+      "epoch": 1.6327985739750446,
+      "grad_norm": 2.0176281929016113,
+      "learning_rate": 1.3920045328463116e-06,
+      "loss": 0.5726,
+      "step": 933
+    },
+    {
+      "epoch": 1.6345811051693404,
+      "grad_norm": 2.013188362121582,
+      "learning_rate": 1.3776603509810938e-06,
+      "loss": 0.5888,
+      "step": 934
+    },
+    {
+      "epoch": 1.6363636363636362,
+      "grad_norm": 1.9507560729980469,
+      "learning_rate": 1.3633849891697882e-06,
+      "loss": 0.6518,
+      "step": 935
+    },
+    {
+      "epoch": 1.6381461675579323,
+      "grad_norm": 1.9589416980743408,
+      "learning_rate": 1.349178561352603e-06,
+      "loss": 0.604,
+      "step": 936
+    },
+    {
+      "epoch": 1.6399286987522281,
+      "grad_norm": 1.811722993850708,
+      "learning_rate": 1.335041180919543e-06,
+      "loss": 0.662,
+      "step": 937
+    },
+    {
+      "epoch": 1.641711229946524,
+      "grad_norm": 2.11613392829895,
+      "learning_rate": 1.3209729607095022e-06,
+      "loss": 0.6149,
+      "step": 938
+    },
+    {
+      "epoch": 1.64349376114082,
+      "grad_norm": 1.6558808088302612,
+      "learning_rate": 1.3069740130093678e-06,
+      "loss": 0.6301,
+      "step": 939
+    },
+    {
+      "epoch": 1.6452762923351159,
+      "grad_norm": 2.1352925300598145,
+      "learning_rate": 1.2930444495531237e-06,
+      "loss": 0.637,
+      "step": 940
+    },
+    {
+      "epoch": 1.6470588235294117,
+      "grad_norm": 2.76387357711792,
+      "learning_rate": 1.2791843815209494e-06,
+      "loss": 0.6461,
+      "step": 941
+    },
+    {
+      "epoch": 1.6488413547237077,
+      "grad_norm": 2.2259490489959717,
+      "learning_rate": 1.2653939195383448e-06,
+      "loss": 0.6129,
+      "step": 942
+    },
+    {
+      "epoch": 1.6506238859180036,
+      "grad_norm": 2.03933048248291,
+      "learning_rate": 1.251673173675243e-06,
+      "loss": 0.6154,
+      "step": 943
+    },
+    {
+      "epoch": 1.6524064171122994,
+      "grad_norm": 2.0698153972625732,
+      "learning_rate": 1.2380222534451302e-06,
+      "loss": 0.6206,
+      "step": 944
+    },
+    {
+      "epoch": 1.6541889483065955,
+      "grad_norm": 1.9460407495498657,
+      "learning_rate": 1.2244412678041707e-06,
+      "loss": 0.6721,
+      "step": 945
+    },
+    {
+      "epoch": 1.6559714795008913,
+      "grad_norm": 2.335196018218994,
+      "learning_rate": 1.2109303251503434e-06,
+      "loss": 0.6158,
+      "step": 946
+    },
+    {
+      "epoch": 1.6577540106951871,
+      "grad_norm": 2.0459136962890625,
+      "learning_rate": 1.1974895333225646e-06,
+      "loss": 0.6597,
+      "step": 947
+    },
+    {
+      "epoch": 1.6595365418894832,
+      "grad_norm": 2.3721280097961426,
+      "learning_rate": 1.1841189995998404e-06,
+      "loss": 0.6127,
+      "step": 948
+    },
+    {
+      "epoch": 1.661319073083779,
+      "grad_norm": 2.00299334526062,
+      "learning_rate": 1.1708188307004043e-06,
+      "loss": 0.564,
+      "step": 949
+    },
+    {
+      "epoch": 1.6631016042780749,
+      "grad_norm": 2.1417782306671143,
+      "learning_rate": 1.1575891327808664e-06,
+      "loss": 0.5406,
+      "step": 950
+    },
+    {
+      "epoch": 1.664884135472371,
+      "grad_norm": 2.224438428878784,
+      "learning_rate": 1.1444300114353647e-06,
+      "loss": 0.5907,
+      "step": 951
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.8061896562576294,
+      "learning_rate": 1.131341571694724e-06,
+      "loss": 0.6806,
+      "step": 952
+    },
+    {
+      "epoch": 1.6684491978609626,
+      "grad_norm": 1.9865063428878784,
+      "learning_rate": 1.1183239180256133e-06,
+      "loss": 0.5527,
+      "step": 953
+    },
+    {
+      "epoch": 1.6702317290552586,
+      "grad_norm": 1.6837379932403564,
+      "learning_rate": 1.1053771543297198e-06,
+      "loss": 0.6019,
+      "step": 954
+    },
+    {
+      "epoch": 1.6720142602495542,
+      "grad_norm": 2.2152931690216064,
+      "learning_rate": 1.0925013839429166e-06,
+      "loss": 0.6155,
+      "step": 955
+    },
+    {
+      "epoch": 1.6737967914438503,
+      "grad_norm": 1.8260977268218994,
+      "learning_rate": 1.0796967096344324e-06,
+      "loss": 0.6033,
+      "step": 956
+    },
+    {
+      "epoch": 1.6755793226381461,
+      "grad_norm": 2.218449592590332,
+      "learning_rate": 1.066963233606041e-06,
+      "loss": 0.6262,
+      "step": 957
+    },
+    {
+      "epoch": 1.677361853832442,
+      "grad_norm": 1.9861358404159546,
+      "learning_rate": 1.0543010574912305e-06,
+      "loss": 0.6212,
+      "step": 958
+    },
+    {
+      "epoch": 1.679144385026738,
+      "grad_norm": 2.2905385494232178,
+      "learning_rate": 1.0417102823544112e-06,
+      "loss": 0.5466,
+      "step": 959
+    },
+    {
+      "epoch": 1.6809269162210339,
+      "grad_norm": 2.123490571975708,
+      "learning_rate": 1.0291910086900968e-06,
+      "loss": 0.6122,
+      "step": 960
+    },
+    {
+      "epoch": 1.6827094474153297,
+      "grad_norm": 2.042818546295166,
+      "learning_rate": 1.0167433364221024e-06,
+      "loss": 0.6353,
+      "step": 961
+    },
+    {
+      "epoch": 1.6844919786096257,
+      "grad_norm": 1.9634802341461182,
+      "learning_rate": 1.0043673649027519e-06,
+      "loss": 0.6569,
+      "step": 962
+    },
+    {
+      "epoch": 1.6862745098039216,
+      "grad_norm": 2.4289674758911133,
+      "learning_rate": 9.92063192912085e-07,
+      "loss": 0.6073,
+      "step": 963
+    },
+    {
+      "epoch": 1.6880570409982174,
+      "grad_norm": 1.9238197803497314,
+      "learning_rate": 9.798309186570588e-07,
+      "loss": 0.6272,
+      "step": 964
+    },
+    {
+      "epoch": 1.6898395721925135,
+      "grad_norm": 2.2965667247772217,
+      "learning_rate": 9.676706397707769e-07,
+      "loss": 0.594,
+      "step": 965
+    },
+    {
+      "epoch": 1.6916221033868093,
+      "grad_norm": 2.0149905681610107,
+      "learning_rate": 9.555824533117064e-07,
+      "loss": 0.6251,
+      "step": 966
+    },
+    {
+      "epoch": 1.6934046345811051,
+      "grad_norm": 1.921858549118042,
+      "learning_rate": 9.435664557628976e-07,
+      "loss": 0.6106,
+      "step": 967
+    },
+    {
+      "epoch": 1.6951871657754012,
+      "grad_norm": 1.854486107826233,
+      "learning_rate": 9.316227430312196e-07,
+      "loss": 0.6605,
+      "step": 968
+    },
+    {
+      "epoch": 1.696969696969697,
+      "grad_norm": 2.0480058193206787,
+      "learning_rate": 9.197514104465955e-07,
+      "loss": 0.655,
+      "step": 969
+    },
+    {
+      "epoch": 1.6987522281639929,
+      "grad_norm": 2.107775926589966,
+      "learning_rate": 9.079525527612321e-07,
+      "loss": 0.5763,
+      "step": 970
+    },
+    {
+      "epoch": 1.700534759358289,
+      "grad_norm": 1.6507351398468018,
+      "learning_rate": 8.962262641488772e-07,
+      "loss": 0.5638,
+      "step": 971
+    },
+    {
+      "epoch": 1.7023172905525845,
+      "grad_norm": 1.974533200263977,
+      "learning_rate": 8.845726382040598e-07,
+      "loss": 0.6469,
+      "step": 972
+    },
+    {
+      "epoch": 1.7040998217468806,
+      "grad_norm": 2.504983901977539,
+      "learning_rate": 8.729917679413435e-07,
+      "loss": 0.5874,
+      "step": 973
+    },
+    {
+      "epoch": 1.7058823529411766,
+      "grad_norm": 2.3748109340667725,
+      "learning_rate": 8.614837457945868e-07,
+      "loss": 0.6213,
+      "step": 974
+    },
+    {
+      "epoch": 1.7076648841354722,
+      "grad_norm": 2.0784034729003906,
+      "learning_rate": 8.500486636162031e-07,
+      "loss": 0.6713,
+      "step": 975
+    },
+    {
+      "epoch": 1.7094474153297683,
+      "grad_norm": 2.2157113552093506,
+      "learning_rate": 8.38686612676427e-07,
+      "loss": 0.5642,
+      "step": 976
+    },
+    {
+      "epoch": 1.7112299465240641,
+      "grad_norm": 1.57169771194458,
+      "learning_rate": 8.273976836625897e-07,
+      "loss": 0.6263,
+      "step": 977
+    },
+    {
+      "epoch": 1.71301247771836,
+      "grad_norm": 2.015939474105835,
+      "learning_rate": 8.161819666783888e-07,
+      "loss": 0.5912,
+      "step": 978
+    },
+    {
+      "epoch": 1.714795008912656,
+      "grad_norm": 2.3326618671417236,
+      "learning_rate": 8.050395512431775e-07,
+      "loss": 0.5949,
+      "step": 979
+    },
+    {
+      "epoch": 1.7165775401069518,
+      "grad_norm": 2.1025454998016357,
+      "learning_rate": 7.939705262912423e-07,
+      "loss": 0.5597,
+      "step": 980
+    },
+    {
+      "epoch": 1.7183600713012477,
+      "grad_norm": 2.205465793609619,
+      "learning_rate": 7.829749801710961e-07,
+      "loss": 0.5117,
+      "step": 981
+    },
+    {
+      "epoch": 1.7201426024955437,
+      "grad_norm": 1.9855906963348389,
+      "learning_rate": 7.720530006447735e-07,
+      "loss": 0.6084,
+      "step": 982
+    },
+    {
+      "epoch": 1.7219251336898396,
+      "grad_norm": 2.0389535427093506,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.5774,
+      "step": 983
+    },
+    {
+      "epoch": 1.7237076648841354,
+      "grad_norm": 2.122084617614746,
+      "learning_rate": 7.504300894851546e-07,
+      "loss": 0.5661,
+      "step": 984
+    },
+    {
+      "epoch": 1.7254901960784315,
+      "grad_norm": 1.9116697311401367,
+      "learning_rate": 7.397293304372544e-07,
+      "loss": 0.5986,
+      "step": 985
+    },
+    {
+      "epoch": 1.7272727272727273,
+      "grad_norm": 1.990723967552185,
+      "learning_rate": 7.291024831525961e-07,
+      "loss": 0.5841,
+      "step": 986
+    },
+    {
+      "epoch": 1.7290552584670231,
+      "grad_norm": 2.1296963691711426,
+      "learning_rate": 7.185496324504093e-07,
+      "loss": 0.6501,
+      "step": 987
+    },
+    {
+      "epoch": 1.7290552584670231,
+      "eval_loss": 0.910054087638855,
+      "eval_runtime": 329.8702,
+      "eval_samples_per_second": 8.919,
+      "eval_steps_per_second": 1.116,
+      "step": 987
+    },
+    {
+      "epoch": 1.7308377896613192,
+      "grad_norm": 1.690833330154419,
+      "learning_rate": 7.080708625593103e-07,
+      "loss": 0.581,
+      "step": 988
+    },
+    {
+      "epoch": 1.732620320855615,
+      "grad_norm": 2.084317922592163,
+      "learning_rate": 6.976662571166348e-07,
+      "loss": 0.6071,
+      "step": 989
+    },
+    {
+      "epoch": 1.7344028520499108,
+      "grad_norm": 2.110623598098755,
+      "learning_rate": 6.87335899167767e-07,
+      "loss": 0.594,
+      "step": 990
+    },
+    {
+      "epoch": 1.736185383244207,
+      "grad_norm": 1.95883047580719,
+      "learning_rate": 6.77079871165478e-07,
+      "loss": 0.5884,
+      "step": 991
+    },
+    {
+      "epoch": 1.7379679144385025,
+      "grad_norm": 2.2620203495025635,
+      "learning_rate": 6.668982549692648e-07,
+      "loss": 0.569,
+      "step": 992
+    },
+    {
+      "epoch": 1.7397504456327986,
+      "grad_norm": 1.84707772731781,
+      "learning_rate": 6.567911318447018e-07,
+      "loss": 0.5643,
+      "step": 993
+    },
+    {
+      "epoch": 1.7415329768270946,
+      "grad_norm": 2.352470874786377,
+      "learning_rate": 6.467585824627886e-07,
+      "loss": 0.6341,
+      "step": 994
+    },
+    {
+      "epoch": 1.7433155080213902,
+      "grad_norm": 1.957634449005127,
+      "learning_rate": 6.368006868993071e-07,
+      "loss": 0.5805,
+      "step": 995
+    },
+    {
+      "epoch": 1.7450980392156863,
+      "grad_norm": 1.8500100374221802,
+      "learning_rate": 6.269175246341818e-07,
+      "loss": 0.6277,
+      "step": 996
+    },
+    {
+      "epoch": 1.7468805704099821,
+      "grad_norm": 2.1289892196655273,
+      "learning_rate": 6.171091745508484e-07,
+      "loss": 0.5556,
+      "step": 997
+    },
+    {
+      "epoch": 1.748663101604278,
+      "grad_norm": 2.2112717628479004,
+      "learning_rate": 6.073757149356185e-07,
+      "loss": 0.6379,
+      "step": 998
+    },
+    {
+      "epoch": 1.750445632798574,
+      "grad_norm": 1.6776251792907715,
+      "learning_rate": 5.977172234770589e-07,
+      "loss": 0.6103,
+      "step": 999
+    },
+    {
+      "epoch": 1.7522281639928698,
+      "grad_norm": 1.8559027910232544,
+      "learning_rate": 5.881337772653728e-07,
+      "loss": 0.6486,
+      "step": 1000
+    },
+    {
+      "epoch": 1.7540106951871657,
+      "grad_norm": 1.9796710014343262,
+      "learning_rate": 5.786254527917801e-07,
+      "loss": 0.5668,
+      "step": 1001
+    },
+    {
+      "epoch": 1.7557932263814617,
+      "grad_norm": 2.122915029525757,
+      "learning_rate": 5.691923259479093e-07,
+      "loss": 0.5901,
+      "step": 1002
+    },
+    {
+      "epoch": 1.7575757575757576,
+      "grad_norm": 1.7617076635360718,
+      "learning_rate": 5.598344720251935e-07,
+      "loss": 0.6103,
+      "step": 1003
+    },
+    {
+      "epoch": 1.7593582887700534,
+      "grad_norm": 2.4237117767333984,
+      "learning_rate": 5.505519657142677e-07,
+      "loss": 0.6668,
+      "step": 1004
+    },
+    {
+      "epoch": 1.7611408199643495,
+      "grad_norm": 2.0379695892333984,
+      "learning_rate": 5.413448811043676e-07,
+      "loss": 0.5855,
+      "step": 1005
+    },
+    {
+      "epoch": 1.7629233511586453,
+      "grad_norm": 2.066530704498291,
+      "learning_rate": 5.322132916827483e-07,
+      "loss": 0.606,
+      "step": 1006
+    },
+    {
+      "epoch": 1.7647058823529411,
+      "grad_norm": 2.2446866035461426,
+      "learning_rate": 5.231572703340904e-07,
+      "loss": 0.6629,
+      "step": 1007
+    },
+    {
+      "epoch": 1.7664884135472372,
+      "grad_norm": 2.5455820560455322,
+      "learning_rate": 5.141768893399224e-07,
+      "loss": 0.6387,
+      "step": 1008
+    },
+    {
+      "epoch": 1.768270944741533,
+      "grad_norm": 2.1626667976379395,
+      "learning_rate": 5.052722203780391e-07,
+      "loss": 0.6181,
+      "step": 1009
+    },
+    {
+      "epoch": 1.7700534759358288,
+      "grad_norm": 2.0961239337921143,
+      "learning_rate": 4.964433345219354e-07,
+      "loss": 0.6174,
+      "step": 1010
+    },
+    {
+      "epoch": 1.771836007130125,
+      "grad_norm": 2.0906381607055664,
+      "learning_rate": 4.876903022402324e-07,
+      "loss": 0.5807,
+      "step": 1011
+    },
+    {
+      "epoch": 1.7736185383244205,
+      "grad_norm": 1.672275185585022,
+      "learning_rate": 4.790131933961207e-07,
+      "loss": 0.5946,
+      "step": 1012
+    },
+    {
+      "epoch": 1.7754010695187166,
+      "grad_norm": 2.0058467388153076,
+      "learning_rate": 4.7041207724679904e-07,
+      "loss": 0.6311,
+      "step": 1013
+    },
+    {
+      "epoch": 1.7771836007130126,
+      "grad_norm": 1.8891334533691406,
+      "learning_rate": 4.6188702244292614e-07,
+      "loss": 0.6957,
+      "step": 1014
+    },
+    {
+      "epoch": 1.7789661319073082,
+      "grad_norm": 1.8195239305496216,
+      "learning_rate": 4.5343809702806606e-07,
+      "loss": 0.5564,
+      "step": 1015
+    },
+    {
+      "epoch": 1.7807486631016043,
+      "grad_norm": 1.9872279167175293,
+      "learning_rate": 4.4506536843815006e-07,
+      "loss": 0.6647,
+      "step": 1016
+    },
+    {
+      "epoch": 1.7825311942959001,
+      "grad_norm": 2.025951623916626,
+      "learning_rate": 4.367689035009359e-07,
+      "loss": 0.5382,
+      "step": 1017
+    },
+    {
+      "epoch": 1.784313725490196,
+      "grad_norm": 2.5888593196868896,
+      "learning_rate": 4.285487684354772e-07,
+      "loss": 0.6339,
+      "step": 1018
+    },
+    {
+      "epoch": 1.786096256684492,
+      "grad_norm": 1.9204456806182861,
+      "learning_rate": 4.2040502885159264e-07,
+      "loss": 0.6232,
+      "step": 1019
+    },
+    {
+      "epoch": 1.7878787878787878,
+      "grad_norm": 2.0572903156280518,
+      "learning_rate": 4.123377497493419e-07,
+      "loss": 0.5806,
+      "step": 1020
+    },
+    {
+      "epoch": 1.7896613190730837,
+      "grad_norm": 1.7260277271270752,
+      "learning_rate": 4.0434699551850973e-07,
+      "loss": 0.6254,
+      "step": 1021
+    },
+    {
+      "epoch": 1.7914438502673797,
+      "grad_norm": 1.9684414863586426,
+      "learning_rate": 3.96432829938086e-07,
+      "loss": 0.5893,
+      "step": 1022
+    },
+    {
+      "epoch": 1.7932263814616756,
+      "grad_norm": 2.3725926876068115,
+      "learning_rate": 3.88595316175765e-07,
+      "loss": 0.5836,
+      "step": 1023
+    },
+    {
+      "epoch": 1.7950089126559714,
+      "grad_norm": 1.6349846124649048,
+      "learning_rate": 3.808345167874361e-07,
+      "loss": 0.5887,
+      "step": 1024
+    },
+    {
+      "epoch": 1.7967914438502675,
+      "grad_norm": 2.258037567138672,
+      "learning_rate": 3.731504937166841e-07,
+      "loss": 0.609,
+      "step": 1025
+    },
+    {
+      "epoch": 1.7985739750445633,
+      "grad_norm": 2.426041603088379,
+      "learning_rate": 3.6554330829429716e-07,
+      "loss": 0.6012,
+      "step": 1026
+    },
+    {
+      "epoch": 1.8003565062388591,
+      "grad_norm": 2.0654661655426025,
+      "learning_rate": 3.5801302123777524e-07,
+      "loss": 0.6293,
+      "step": 1027
+    },
+    {
+      "epoch": 1.8021390374331552,
+      "grad_norm": 1.8019964694976807,
+      "learning_rate": 3.50559692650847e-07,
+      "loss": 0.6013,
+      "step": 1028
+    },
+    {
+      "epoch": 1.803921568627451,
+      "grad_norm": 2.0039703845977783,
+      "learning_rate": 3.4318338202298796e-07,
+      "loss": 0.6604,
+      "step": 1029
+    },
+    {
+      "epoch": 1.8057040998217468,
+      "grad_norm": 2.3911831378936768,
+      "learning_rate": 3.3588414822895097e-07,
+      "loss": 0.6017,
+      "step": 1030
+    },
+    {
+      "epoch": 1.807486631016043,
+      "grad_norm": 1.8550302982330322,
+      "learning_rate": 3.2866204952828773e-07,
+      "loss": 0.5959,
+      "step": 1031
+    },
+    {
+      "epoch": 1.8092691622103387,
+      "grad_norm": 2.110670328140259,
+      "learning_rate": 3.2151714356489225e-07,
+      "loss": 0.6788,
+      "step": 1032
+    },
+    {
+      "epoch": 1.8110516934046346,
+      "grad_norm": 2.0615427494049072,
+      "learning_rate": 3.1444948736653604e-07,
+      "loss": 0.6304,
+      "step": 1033
+    },
+    {
+      "epoch": 1.8128342245989306,
+      "grad_norm": 2.303654432296753,
+      "learning_rate": 3.0745913734441357e-07,
+      "loss": 0.5751,
+      "step": 1034
+    },
+    {
+      "epoch": 1.8146167557932262,
+      "grad_norm": 2.325732946395874,
+      "learning_rate": 3.005461492926931e-07,
+      "loss": 0.6593,
+      "step": 1035
+    },
+    {
+      "epoch": 1.8163992869875223,
+      "grad_norm": 2.306138515472412,
+      "learning_rate": 2.937105783880689e-07,
+      "loss": 0.6617,
+      "step": 1036
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 1.7654507160186768,
+      "learning_rate": 2.869524791893252e-07,
+      "loss": 0.6126,
+      "step": 1037
+    },
+    {
+      "epoch": 1.819964349376114,
+      "grad_norm": 1.7648675441741943,
+      "learning_rate": 2.8027190563689745e-07,
+      "loss": 0.6459,
+      "step": 1038
+    },
+    {
+      "epoch": 1.82174688057041,
+      "grad_norm": 2.3114426136016846,
+      "learning_rate": 2.736689110524404e-07,
+      "loss": 0.5672,
+      "step": 1039
+    },
+    {
+      "epoch": 1.8235294117647058,
+      "grad_norm": 2.0545334815979004,
+      "learning_rate": 2.671435481384066e-07,
+      "loss": 0.5506,
+      "step": 1040
+    },
+    {
+      "epoch": 1.8253119429590017,
+      "grad_norm": 1.8540985584259033,
+      "learning_rate": 2.60695868977624e-07,
+      "loss": 0.6197,
+      "step": 1041
+    },
+    {
+      "epoch": 1.8270944741532977,
+      "grad_norm": 1.8356308937072754,
+      "learning_rate": 2.5432592503288e-07,
+      "loss": 0.5953,
+      "step": 1042
+    },
+    {
+      "epoch": 1.8288770053475936,
+      "grad_norm": 2.0690879821777344,
+      "learning_rate": 2.480337671465083e-07,
+      "loss": 0.5921,
+      "step": 1043
+    },
+    {
+      "epoch": 1.8306595365418894,
+      "grad_norm": 1.937947392463684,
+      "learning_rate": 2.4181944553999026e-07,
+      "loss": 0.6009,
+      "step": 1044
+    },
+    {
+      "epoch": 1.8324420677361855,
+      "grad_norm": 2.166762351989746,
+      "learning_rate": 2.356830098135443e-07,
+      "loss": 0.6163,
+      "step": 1045
+    },
+    {
+      "epoch": 1.8342245989304813,
+      "grad_norm": 1.7820336818695068,
+      "learning_rate": 2.2962450894573606e-07,
+      "loss": 0.6497,
+      "step": 1046
+    },
+    {
+      "epoch": 1.8360071301247771,
+      "grad_norm": 1.9238687753677368,
+      "learning_rate": 2.236439912930899e-07,
+      "loss": 0.6168,
+      "step": 1047
+    },
+    {
+      "epoch": 1.8377896613190732,
+      "grad_norm": 2.0929696559906006,
+      "learning_rate": 2.1774150458969578e-07,
+      "loss": 0.5113,
+      "step": 1048
+    },
+    {
+      "epoch": 1.839572192513369,
+      "grad_norm": 2.0725159645080566,
+      "learning_rate": 2.1191709594683419e-07,
+      "loss": 0.528,
+      "step": 1049
+    },
+    {
+      "epoch": 1.8413547237076648,
+      "grad_norm": 2.1145312786102295,
+      "learning_rate": 2.0617081185259512e-07,
+      "loss": 0.588,
+      "step": 1050
+    },
+    {
+      "epoch": 1.843137254901961,
+      "grad_norm": 1.9781309366226196,
+      "learning_rate": 2.0050269817151413e-07,
+      "loss": 0.5931,
+      "step": 1051
+    },
+    {
+      "epoch": 1.8449197860962567,
+      "grad_norm": 2.162130832672119,
+      "learning_rate": 1.9491280014419689e-07,
+      "loss": 0.5989,
+      "step": 1052
+    },
+    {
+      "epoch": 1.8467023172905526,
+      "grad_norm": 1.7542837858200073,
+      "learning_rate": 1.8940116238696514e-07,
+      "loss": 0.6411,
+      "step": 1053
+    },
+    {
+      "epoch": 1.8484848484848486,
+      "grad_norm": 2.309709310531616,
+      "learning_rate": 1.8396782889150144e-07,
+      "loss": 0.6104,
+      "step": 1054
+    },
+    {
+      "epoch": 1.8502673796791442,
+      "grad_norm": 1.9894696474075317,
+      "learning_rate": 1.7861284302449267e-07,
+      "loss": 0.6426,
+      "step": 1055
+    },
+    {
+      "epoch": 1.8520499108734403,
+      "grad_norm": 2.1708126068115234,
+      "learning_rate": 1.7333624752728373e-07,
+      "loss": 0.5921,
+      "step": 1056
+    },
+    {
+      "epoch": 1.8538324420677363,
+      "grad_norm": 2.056997060775757,
+      "learning_rate": 1.6813808451554447e-07,
+      "loss": 0.6295,
+      "step": 1057
+    },
+    {
+      "epoch": 1.855614973262032,
+      "grad_norm": 2.0771663188934326,
+      "learning_rate": 1.630183954789233e-07,
+      "loss": 0.5634,
+      "step": 1058
+    },
+    {
+      "epoch": 1.857397504456328,
+      "grad_norm": 1.957069993019104,
+      "learning_rate": 1.5797722128072514e-07,
+      "loss": 0.6253,
+      "step": 1059
+    },
+    {
+      "epoch": 1.8591800356506238,
+      "grad_norm": 2.001437187194824,
+      "learning_rate": 1.5301460215757625e-07,
+      "loss": 0.6211,
+      "step": 1060
+    },
+    {
+      "epoch": 1.8609625668449197,
+      "grad_norm": 2.0400633811950684,
+      "learning_rate": 1.481305777191133e-07,
+      "loss": 0.5816,
+      "step": 1061
+    },
+    {
+      "epoch": 1.8627450980392157,
+      "grad_norm": 1.9419509172439575,
+      "learning_rate": 1.4332518694765708e-07,
+      "loss": 0.6496,
+      "step": 1062
+    },
+    {
+      "epoch": 1.8645276292335116,
+      "grad_norm": 2.0088133811950684,
+      "learning_rate": 1.385984681979069e-07,
+      "loss": 0.6479,
+      "step": 1063
+    },
+    {
+      "epoch": 1.8663101604278074,
+      "grad_norm": 2.402581214904785,
+      "learning_rate": 1.3395045919663674e-07,
+      "loss": 0.6767,
+      "step": 1064
+    },
+    {
+      "epoch": 1.8680926916221035,
+      "grad_norm": 2.071725845336914,
+      "learning_rate": 1.293811970423864e-07,
+      "loss": 0.5917,
+      "step": 1065
+    },
+    {
+      "epoch": 1.8698752228163993,
+      "grad_norm": 1.871656060218811,
+      "learning_rate": 1.2489071820517394e-07,
+      "loss": 0.5651,
+      "step": 1066
+    },
+    {
+      "epoch": 1.8716577540106951,
+      "grad_norm": 2.2825636863708496,
+      "learning_rate": 1.2047905852619834e-07,
+      "loss": 0.6595,
+      "step": 1067
+    },
+    {
+      "epoch": 1.8734402852049912,
+      "grad_norm": 2.1393330097198486,
+      "learning_rate": 1.1614625321755613e-07,
+      "loss": 0.6219,
+      "step": 1068
+    },
+    {
+      "epoch": 1.875222816399287,
+      "grad_norm": 2.1348400115966797,
+      "learning_rate": 1.1189233686195956e-07,
+      "loss": 0.5991,
+      "step": 1069
+    },
+    {
+      "epoch": 1.8770053475935828,
+      "grad_norm": 1.983951210975647,
+      "learning_rate": 1.0771734341246121e-07,
+      "loss": 0.565,
+      "step": 1070
+    },
+    {
+      "epoch": 1.878787878787879,
+      "grad_norm": 1.8638139963150024,
+      "learning_rate": 1.0362130619218424e-07,
+      "loss": 0.6242,
+      "step": 1071
+    },
+    {
+      "epoch": 1.8805704099821747,
+      "grad_norm": 2.2923965454101562,
+      "learning_rate": 9.96042578940526e-08,
+      "loss": 0.5726,
+      "step": 1072
+    },
+    {
+      "epoch": 1.8823529411764706,
+      "grad_norm": 2.1189980506896973,
+      "learning_rate": 9.566623058053337e-08,
+      "loss": 0.5842,
+      "step": 1073
+    },
+    {
+      "epoch": 1.8841354723707666,
+      "grad_norm": 1.616305947303772,
+      "learning_rate": 9.180725568338045e-08,
+      "loss": 0.6361,
+      "step": 1074
+    },
+    {
+      "epoch": 1.8859180035650622,
+      "grad_norm": 1.9860106706619263,
+      "learning_rate": 8.802736400338019e-08,
+      "loss": 0.6659,
+      "step": 1075
+    },
+    {
+      "epoch": 1.8877005347593583,
+      "grad_norm": 2.3532490730285645,
+      "learning_rate": 8.432658571011387e-08,
+      "loss": 0.6402,
+      "step": 1076
+    },
+    {
+      "epoch": 1.8894830659536543,
+      "grad_norm": 2.0499911308288574,
+      "learning_rate": 8.070495034170566e-08,
+      "loss": 0.6302,
+      "step": 1077
+    },
+    {
+      "epoch": 1.89126559714795,
+      "grad_norm": 2.055406332015991,
+      "learning_rate": 7.716248680459726e-08,
+      "loss": 0.5792,
+      "step": 1078
+    },
+    {
+      "epoch": 1.893048128342246,
+      "grad_norm": 1.9370893239974976,
+      "learning_rate": 7.369922337330914e-08,
+      "loss": 0.6076,
+      "step": 1079
+    },
+    {
+      "epoch": 1.8948306595365418,
+      "grad_norm": 2.0450425148010254,
+      "learning_rate": 7.031518769021972e-08,
+      "loss": 0.6294,
+      "step": 1080
+    },
+    {
+      "epoch": 1.8966131907308377,
+      "grad_norm": 1.9761147499084473,
+      "learning_rate": 6.701040676534432e-08,
+      "loss": 0.6191,
+      "step": 1081
+    },
+    {
+      "epoch": 1.8983957219251337,
+      "grad_norm": 2.1664581298828125,
+      "learning_rate": 6.378490697611761e-08,
+      "loss": 0.5854,
+      "step": 1082
+    },
+    {
+      "epoch": 1.9001782531194296,
+      "grad_norm": 2.0368752479553223,
+      "learning_rate": 6.063871406718381e-08,
+      "loss": 0.5508,
+      "step": 1083
+    },
+    {
+      "epoch": 1.9019607843137254,
+      "grad_norm": 2.349900007247925,
+      "learning_rate": 5.75718531501912e-08,
+      "loss": 0.6304,
+      "step": 1084
+    },
+    {
+      "epoch": 1.9037433155080214,
+      "grad_norm": 2.0724315643310547,
+      "learning_rate": 5.4584348703594634e-08,
+      "loss": 0.6062,
+      "step": 1085
+    },
+    {
+      "epoch": 1.9055258467023173,
+      "grad_norm": 2.279808282852173,
+      "learning_rate": 5.1676224572452246e-08,
+      "loss": 0.6588,
+      "step": 1086
+    },
+    {
+      "epoch": 1.9073083778966131,
+      "grad_norm": 1.9012436866760254,
+      "learning_rate": 4.884750396824567e-08,
+      "loss": 0.6473,
+      "step": 1087
+    },
+    {
+      "epoch": 1.9090909090909092,
+      "grad_norm": 1.8943966627120972,
+      "learning_rate": 4.609820946868682e-08,
+      "loss": 0.595,
+      "step": 1088
+    },
+    {
+      "epoch": 1.910873440285205,
+      "grad_norm": 2.4004600048065186,
+      "learning_rate": 4.3428363017540276e-08,
+      "loss": 0.6061,
+      "step": 1089
+    },
+    {
+      "epoch": 1.9126559714795008,
+      "grad_norm": 2.035853624343872,
+      "learning_rate": 4.083798592444899e-08,
+      "loss": 0.5977,
+      "step": 1090
+    },
+    {
+      "epoch": 1.914438502673797,
+      "grad_norm": 1.916986107826233,
+      "learning_rate": 3.832709886476438e-08,
+      "loss": 0.5385,
+      "step": 1091
+    },
+    {
+      "epoch": 1.9162210338680927,
+      "grad_norm": 1.8782941102981567,
+      "learning_rate": 3.589572187937651e-08,
+      "loss": 0.5656,
+      "step": 1092
+    },
+    {
+      "epoch": 1.9180035650623886,
+      "grad_norm": 2.144084930419922,
+      "learning_rate": 3.354387437456197e-08,
+      "loss": 0.5861,
+      "step": 1093
+    },
+    {
+      "epoch": 1.9197860962566846,
+      "grad_norm": 2.3000340461730957,
+      "learning_rate": 3.127157512182288e-08,
+      "loss": 0.5849,
+      "step": 1094
+    },
+    {
+      "epoch": 1.9215686274509802,
+      "grad_norm": 2.2352511882781982,
+      "learning_rate": 2.907884225774149e-08,
+      "loss": 0.6817,
+      "step": 1095
+    },
+    {
+      "epoch": 1.9233511586452763,
+      "grad_norm": 1.7509145736694336,
+      "learning_rate": 2.6965693283829143e-08,
+      "loss": 0.5748,
+      "step": 1096
+    },
+    {
+      "epoch": 1.9251336898395723,
+      "grad_norm": 2.180359363555908,
+      "learning_rate": 2.4932145066394186e-08,
+      "loss": 0.593,
+      "step": 1097
+    },
+    {
+      "epoch": 1.926916221033868,
+      "grad_norm": 2.239539623260498,
+      "learning_rate": 2.2978213836400974e-08,
+      "loss": 0.5085,
+      "step": 1098
+    },
+    {
+      "epoch": 1.928698752228164,
+      "grad_norm": 2.2124013900756836,
+      "learning_rate": 2.1103915189344403e-08,
+      "loss": 0.5729,
+      "step": 1099
+    },
+    {
+      "epoch": 1.93048128342246,
+      "grad_norm": 2.1283504962921143,
+      "learning_rate": 1.9309264085124458e-08,
+      "loss": 0.6068,
+      "step": 1100
+    },
+    {
+      "epoch": 1.9322638146167557,
+      "grad_norm": 2.4210431575775146,
+      "learning_rate": 1.7594274847926306e-08,
+      "loss": 0.6128,
+      "step": 1101
+    },
+    {
+      "epoch": 1.9340463458110517,
+      "grad_norm": 2.339913845062256,
+      "learning_rate": 1.5958961166104847e-08,
+      "loss": 0.5487,
+      "step": 1102
+    },
+    {
+      "epoch": 1.9358288770053476,
+      "grad_norm": 1.8890469074249268,
+      "learning_rate": 1.4403336092077002e-08,
+      "loss": 0.6371,
+      "step": 1103
+    },
+    {
+      "epoch": 1.9376114081996434,
+      "grad_norm": 2.159550666809082,
+      "learning_rate": 1.2927412042218479e-08,
+      "loss": 0.589,
+      "step": 1104
+    },
+    {
+      "epoch": 1.9393939393939394,
+      "grad_norm": 2.154710292816162,
+      "learning_rate": 1.1531200796762731e-08,
+      "loss": 0.607,
+      "step": 1105
+    },
+    {
+      "epoch": 1.9411764705882353,
+      "grad_norm": 2.176748275756836,
+      "learning_rate": 1.0214713499706596e-08,
+      "loss": 0.6424,
+      "step": 1106
+    },
+    {
+      "epoch": 1.9429590017825311,
+      "grad_norm": 2.3147072792053223,
+      "learning_rate": 8.977960658723694e-09,
+      "loss": 0.674,
+      "step": 1107
+    },
+    {
+      "epoch": 1.9447415329768272,
+      "grad_norm": 2.0270538330078125,
+      "learning_rate": 7.820952145078942e-09,
+      "loss": 0.6114,
+      "step": 1108
+    },
+    {
+      "epoch": 1.946524064171123,
+      "grad_norm": 1.8115181922912598,
+      "learning_rate": 6.743697193549725e-09,
+      "loss": 0.6498,
+      "step": 1109
+    },
+    {
+      "epoch": 1.9483065953654188,
+      "grad_norm": 1.941896677017212,
+      "learning_rate": 5.7462044023515186e-09,
+      "loss": 0.6068,
+      "step": 1110
+    },
+    {
+      "epoch": 1.950089126559715,
+      "grad_norm": 2.0113778114318848,
+      "learning_rate": 4.828481733073487e-09,
+      "loss": 0.5983,
+      "step": 1111
+    },
+    {
+      "epoch": 1.9518716577540107,
+      "grad_norm": 2.019991159439087,
+      "learning_rate": 3.9905365106085445e-09,
+      "loss": 0.5958,
+      "step": 1112
+    },
+    {
+      "epoch": 1.9536541889483066,
+      "grad_norm": 2.064058303833008,
+      "learning_rate": 3.2323754230989546e-09,
+      "loss": 0.6223,
+      "step": 1113
+    },
+    {
+      "epoch": 1.9554367201426026,
+      "grad_norm": 2.038113594055176,
+      "learning_rate": 2.5540045218819256e-09,
+      "loss": 0.6118,
+      "step": 1114
+    },
+    {
+      "epoch": 1.9572192513368984,
+      "grad_norm": 2.5111892223358154,
+      "learning_rate": 1.9554292214418734e-09,
+      "loss": 0.5805,
+      "step": 1115
+    },
+    {
+      "epoch": 1.9590017825311943,
+      "grad_norm": 2.1744213104248047,
+      "learning_rate": 1.436654299367124e-09,
+      "loss": 0.6322,
+      "step": 1116
+    },
+    {
+      "epoch": 1.9607843137254903,
+      "grad_norm": 1.8781476020812988,
+      "learning_rate": 9.976838963099421e-10,
+      "loss": 0.596,
+      "step": 1117
+    },
+    {
+      "epoch": 1.962566844919786,
+      "grad_norm": 2.229332447052002,
+      "learning_rate": 6.385215159565583e-10,
+      "loss": 0.6091,
+      "step": 1118
+    },
+    {
+      "epoch": 1.964349376114082,
+      "grad_norm": 2.3374135494232178,
+      "learning_rate": 3.5917002499719076e-10,
+      "loss": 0.5855,
+      "step": 1119
+    },
+    {
+      "epoch": 1.966131907308378,
+      "grad_norm": 1.931396722793579,
+      "learning_rate": 1.59631653102732e-10,
+      "loss": 0.6366,
+      "step": 1120
+    },
+    {
+      "epoch": 1.9679144385026737,
+      "grad_norm": 1.9510762691497803,
+      "learning_rate": 3.990799290809477e-11,
+      "loss": 0.638,
+      "step": 1121
+    },
+    {
+      "epoch": 1.9696969696969697,
+      "grad_norm": 2.2144649028778076,
+      "learning_rate": 0.0,
+      "loss": 0.554,
+      "step": 1122
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1122,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 281,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.428572157668229e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}