{
  "best_metric": 2.7514498233795166,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.9422850412249706,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004711425206124852,
      "grad_norm": 1.3946541547775269,
      "learning_rate": 1e-05,
      "loss": 4.1193,
      "step": 1
    },
    {
      "epoch": 0.004711425206124852,
      "eval_loss": 4.429855823516846,
      "eval_runtime": 5.7087,
      "eval_samples_per_second": 62.712,
      "eval_steps_per_second": 15.766,
      "step": 1
    },
    {
      "epoch": 0.009422850412249705,
      "grad_norm": 1.7446237802505493,
      "learning_rate": 2e-05,
      "loss": 3.5333,
      "step": 2
    },
    {
      "epoch": 0.014134275618374558,
      "grad_norm": 1.2714354991912842,
      "learning_rate": 3e-05,
      "loss": 3.8054,
      "step": 3
    },
    {
      "epoch": 0.01884570082449941,
      "grad_norm": 1.037392020225525,
      "learning_rate": 4e-05,
      "loss": 4.2302,
      "step": 4
    },
    {
      "epoch": 0.023557126030624265,
      "grad_norm": 1.5844902992248535,
      "learning_rate": 5e-05,
      "loss": 4.0444,
      "step": 5
    },
    {
      "epoch": 0.028268551236749116,
      "grad_norm": 1.3372470140457153,
      "learning_rate": 6e-05,
      "loss": 4.1766,
      "step": 6
    },
    {
      "epoch": 0.03297997644287397,
      "grad_norm": 1.3058269023895264,
      "learning_rate": 7e-05,
      "loss": 4.1232,
      "step": 7
    },
    {
      "epoch": 0.03769140164899882,
      "grad_norm": 1.0855653285980225,
      "learning_rate": 8e-05,
      "loss": 4.4008,
      "step": 8
    },
    {
      "epoch": 0.04240282685512368,
      "grad_norm": 1.2069677114486694,
      "learning_rate": 9e-05,
      "loss": 4.0673,
      "step": 9
    },
    {
      "epoch": 0.04711425206124853,
      "grad_norm": 1.1353275775909424,
      "learning_rate": 0.0001,
      "loss": 4.2198,
      "step": 10
    },
    {
      "epoch": 0.05182567726737338,
      "grad_norm": 0.9457094669342041,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.1115,
      "step": 11
    },
    {
      "epoch": 0.05653710247349823,
      "grad_norm": 1.022911787033081,
      "learning_rate": 9.997266286704631e-05,
      "loss": 4.0602,
      "step": 12
    },
    {
      "epoch": 0.061248527679623084,
      "grad_norm": 1.184678077697754,
      "learning_rate": 9.993849845741524e-05,
      "loss": 4.0178,
      "step": 13
    },
    {
      "epoch": 0.06595995288574794,
      "grad_norm": 1.3648606538772583,
      "learning_rate": 9.989068136093873e-05,
      "loss": 3.8917,
      "step": 14
    },
    {
      "epoch": 0.0706713780918728,
      "grad_norm": 1.3144981861114502,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.1258,
      "step": 15
    },
    {
      "epoch": 0.07538280329799764,
      "grad_norm": 1.241207480430603,
      "learning_rate": 9.975414512725057e-05,
      "loss": 3.8097,
      "step": 16
    },
    {
      "epoch": 0.0800942285041225,
      "grad_norm": 1.1597161293029785,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.9649,
      "step": 17
    },
    {
      "epoch": 0.08480565371024736,
      "grad_norm": 1.310815453529358,
      "learning_rate": 9.956320346634876e-05,
      "loss": 3.9916,
      "step": 18
    },
    {
      "epoch": 0.0895170789163722,
      "grad_norm": 1.0816928148269653,
      "learning_rate": 9.944739353007344e-05,
      "loss": 3.6277,
      "step": 19
    },
    {
      "epoch": 0.09422850412249706,
      "grad_norm": 1.219735026359558,
      "learning_rate": 9.931806517013612e-05,
      "loss": 3.6366,
      "step": 20
    },
    {
      "epoch": 0.0989399293286219,
      "grad_norm": 1.4704792499542236,
      "learning_rate": 9.917525374361912e-05,
      "loss": 3.4232,
      "step": 21
    },
    {
      "epoch": 0.10365135453474676,
      "grad_norm": 1.2693730592727661,
      "learning_rate": 9.901899829374047e-05,
      "loss": 3.9704,
      "step": 22
    },
    {
      "epoch": 0.10836277974087162,
      "grad_norm": 1.498766541481018,
      "learning_rate": 9.884934153917997e-05,
      "loss": 3.7581,
      "step": 23
    },
    {
      "epoch": 0.11307420494699646,
      "grad_norm": 1.684964656829834,
      "learning_rate": 9.86663298624003e-05,
      "loss": 3.9932,
      "step": 24
    },
    {
      "epoch": 0.11778563015312132,
      "grad_norm": 2.037986993789673,
      "learning_rate": 9.847001329696653e-05,
      "loss": 3.8869,
      "step": 25
    },
    {
      "epoch": 0.12249705535924617,
      "grad_norm": 2.0941431522369385,
      "learning_rate": 9.826044551386744e-05,
      "loss": 3.1425,
      "step": 26
    },
    {
      "epoch": 0.127208480565371,
      "grad_norm": 1.6686711311340332,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.9945,
      "step": 27
    },
    {
      "epoch": 0.13191990577149587,
      "grad_norm": 1.7616199254989624,
      "learning_rate": 9.780178907671789e-05,
      "loss": 3.4982,
      "step": 28
    },
    {
      "epoch": 0.13663133097762073,
      "grad_norm": 1.3488978147506714,
      "learning_rate": 9.755282581475769e-05,
      "loss": 3.2957,
      "step": 29
    },
    {
      "epoch": 0.1413427561837456,
      "grad_norm": 1.2035346031188965,
      "learning_rate": 9.729086208503174e-05,
      "loss": 3.0778,
      "step": 30
    },
    {
      "epoch": 0.14605418138987045,
      "grad_norm": 1.160496711730957,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.7507,
      "step": 31
    },
    {
      "epoch": 0.15076560659599528,
      "grad_norm": 1.103440284729004,
      "learning_rate": 9.672822322997305e-05,
      "loss": 3.2921,
      "step": 32
    },
    {
      "epoch": 0.15547703180212014,
      "grad_norm": 1.251758098602295,
      "learning_rate": 9.642770192448536e-05,
      "loss": 3.1819,
      "step": 33
    },
    {
      "epoch": 0.160188457008245,
      "grad_norm": 1.1060675382614136,
      "learning_rate": 9.611448774886924e-05,
      "loss": 3.3223,
      "step": 34
    },
    {
      "epoch": 0.16489988221436985,
      "grad_norm": 1.0669234991073608,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.9139,
      "step": 35
    },
    {
      "epoch": 0.1696113074204947,
      "grad_norm": 1.0646271705627441,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.9312,
      "step": 36
    },
    {
      "epoch": 0.17432273262661954,
      "grad_norm": 0.9158160090446472,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.7344,
      "step": 37
    },
    {
      "epoch": 0.1790341578327444,
      "grad_norm": 0.9490165114402771,
      "learning_rate": 9.473646649103818e-05,
      "loss": 3.3129,
      "step": 38
    },
    {
      "epoch": 0.18374558303886926,
      "grad_norm": 0.9325077533721924,
      "learning_rate": 9.43611409721806e-05,
      "loss": 3.0363,
      "step": 39
    },
    {
      "epoch": 0.18845700824499412,
      "grad_norm": 1.0368661880493164,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.7854,
      "step": 40
    },
    {
      "epoch": 0.19316843345111898,
      "grad_norm": 0.8712729215621948,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.7577,
      "step": 41
    },
    {
      "epoch": 0.1978798586572438,
      "grad_norm": 0.9019601345062256,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.995,
      "step": 42
    },
    {
      "epoch": 0.20259128386336867,
      "grad_norm": 0.9480728507041931,
      "learning_rate": 9.273963562927695e-05,
      "loss": 3.0736,
      "step": 43
    },
    {
      "epoch": 0.20730270906949352,
      "grad_norm": 0.8850538730621338,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.9836,
      "step": 44
    },
    {
      "epoch": 0.21201413427561838,
      "grad_norm": 0.8516348600387573,
      "learning_rate": 9.185832391312644e-05,
      "loss": 3.1142,
      "step": 45
    },
    {
      "epoch": 0.21672555948174324,
      "grad_norm": 0.9424264430999756,
      "learning_rate": 9.140044155740101e-05,
      "loss": 3.3857,
      "step": 46
    },
    {
      "epoch": 0.22143698468786807,
      "grad_norm": 0.9524626135826111,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.8493,
      "step": 47
    },
    {
      "epoch": 0.22614840989399293,
      "grad_norm": 0.9207716584205627,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.8681,
      "step": 48
    },
    {
      "epoch": 0.2308598351001178,
      "grad_norm": 0.8092623353004456,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.9606,
      "step": 49
    },
    {
      "epoch": 0.23557126030624265,
      "grad_norm": 0.915729284286499,
      "learning_rate": 8.945702546981969e-05,
      "loss": 3.1628,
      "step": 50
    },
    {
      "epoch": 0.23557126030624265,
      "eval_loss": 3.093031167984009,
      "eval_runtime": 5.6918,
      "eval_samples_per_second": 62.897,
      "eval_steps_per_second": 15.812,
      "step": 50
    },
    {
      "epoch": 0.24028268551236748,
      "grad_norm": 1.1849467754364014,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.2656,
      "step": 51
    },
    {
      "epoch": 0.24499411071849234,
      "grad_norm": 1.3945523500442505,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.2252,
      "step": 52
    },
    {
      "epoch": 0.2497055359246172,
      "grad_norm": 1.0486680269241333,
      "learning_rate": 8.788574348801675e-05,
      "loss": 3.4915,
      "step": 53
    },
    {
      "epoch": 0.254416961130742,
      "grad_norm": 0.9627955555915833,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.9888,
      "step": 54
    },
    {
      "epoch": 0.2591283863368669,
      "grad_norm": 1.5820244550704956,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.9965,
      "step": 55
    },
    {
      "epoch": 0.26383981154299174,
      "grad_norm": 0.8985310792922974,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.0081,
      "step": 56
    },
    {
      "epoch": 0.26855123674911663,
      "grad_norm": 0.9468170404434204,
      "learning_rate": 8.564642241456986e-05,
      "loss": 3.2599,
      "step": 57
    },
    {
      "epoch": 0.27326266195524146,
      "grad_norm": 0.8421167731285095,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.9656,
      "step": 58
    },
    {
      "epoch": 0.2779740871613663,
      "grad_norm": 0.9423841834068298,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.8243,
      "step": 59
    },
    {
      "epoch": 0.2826855123674912,
      "grad_norm": 0.9182611107826233,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.8801,
      "step": 60
    },
    {
      "epoch": 0.287396937573616,
      "grad_norm": 0.9046218991279602,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.0764,
      "step": 61
    },
    {
      "epoch": 0.2921083627797409,
      "grad_norm": 0.8965874314308167,
      "learning_rate": 8.262928807620843e-05,
      "loss": 3.1394,
      "step": 62
    },
    {
      "epoch": 0.2968197879858657,
      "grad_norm": 1.0667086839675903,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.688,
      "step": 63
    },
    {
      "epoch": 0.30153121319199055,
      "grad_norm": 0.8867380619049072,
      "learning_rate": 8.135881792367686e-05,
      "loss": 3.3729,
      "step": 64
    },
    {
      "epoch": 0.30624263839811544,
      "grad_norm": 0.8846852779388428,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.2973,
      "step": 65
    },
    {
      "epoch": 0.31095406360424027,
      "grad_norm": 0.8917059302330017,
      "learning_rate": 8.005405736415126e-05,
      "loss": 3.1064,
      "step": 66
    },
    {
      "epoch": 0.31566548881036516,
      "grad_norm": 0.9517539739608765,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.8648,
      "step": 67
    },
    {
      "epoch": 0.32037691401649,
      "grad_norm": 0.9675039649009705,
      "learning_rate": 7.871643313414718e-05,
      "loss": 3.2757,
      "step": 68
    },
    {
      "epoch": 0.3250883392226148,
      "grad_norm": 0.900853157043457,
      "learning_rate": 7.803575286758364e-05,
      "loss": 3.1051,
      "step": 69
    },
    {
      "epoch": 0.3297997644287397,
      "grad_norm": 1.0221682786941528,
      "learning_rate": 7.734740790612136e-05,
      "loss": 3.4485,
      "step": 70
    },
    {
      "epoch": 0.33451118963486454,
      "grad_norm": 0.9935013651847839,
      "learning_rate": 7.66515864363997e-05,
      "loss": 3.0571,
      "step": 71
    },
    {
      "epoch": 0.3392226148409894,
      "grad_norm": 1.0712761878967285,
      "learning_rate": 7.594847868906076e-05,
      "loss": 3.0314,
      "step": 72
    },
    {
      "epoch": 0.34393404004711425,
      "grad_norm": 1.1006646156311035,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.9187,
      "step": 73
    },
    {
      "epoch": 0.3486454652532391,
      "grad_norm": 1.3611395359039307,
      "learning_rate": 7.452117519152542e-05,
      "loss": 3.1346,
      "step": 74
    },
    {
      "epoch": 0.35335689045936397,
      "grad_norm": 1.2989857196807861,
      "learning_rate": 7.379736965185368e-05,
      "loss": 3.2491,
      "step": 75
    },
    {
      "epoch": 0.3580683156654888,
      "grad_norm": 1.1900298595428467,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.7104,
      "step": 76
    },
    {
      "epoch": 0.3627797408716137,
      "grad_norm": 1.2231336832046509,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.5591,
      "step": 77
    },
    {
      "epoch": 0.3674911660777385,
      "grad_norm": 1.140899658203125,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.8716,
      "step": 78
    },
    {
      "epoch": 0.37220259128386335,
      "grad_norm": 1.2476599216461182,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.6398,
      "step": 79
    },
    {
      "epoch": 0.37691401648998824,
      "grad_norm": 1.1672461032867432,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.7379,
      "step": 80
    },
    {
      "epoch": 0.38162544169611307,
      "grad_norm": 1.080532193183899,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.528,
      "step": 81
    },
    {
      "epoch": 0.38633686690223795,
      "grad_norm": 1.124290943145752,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.3585,
      "step": 82
    },
    {
      "epoch": 0.3910482921083628,
      "grad_norm": 0.9934729933738708,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.5504,
      "step": 83
    },
    {
      "epoch": 0.3957597173144876,
      "grad_norm": 1.2130358219146729,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.9468,
      "step": 84
    },
    {
      "epoch": 0.4004711425206125,
      "grad_norm": 1.0236927270889282,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.9596,
      "step": 85
    },
    {
      "epoch": 0.40518256772673733,
      "grad_norm": 0.9387683272361755,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.8189,
      "step": 86
    },
    {
      "epoch": 0.4098939929328622,
      "grad_norm": 0.9069825410842896,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.1935,
      "step": 87
    },
    {
      "epoch": 0.41460541813898705,
      "grad_norm": 0.9427089095115662,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.3151,
      "step": 88
    },
    {
      "epoch": 0.4193168433451119,
      "grad_norm": 0.8664374947547913,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.7139,
      "step": 89
    },
    {
      "epoch": 0.42402826855123676,
      "grad_norm": 0.9272505044937134,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.7058,
      "step": 90
    },
    {
      "epoch": 0.4287396937573616,
      "grad_norm": 0.7902560234069824,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.5351,
      "step": 91
    },
    {
      "epoch": 0.4334511189634865,
      "grad_norm": 0.7875714302062988,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.6184,
      "step": 92
    },
    {
      "epoch": 0.4381625441696113,
      "grad_norm": 0.9437817335128784,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.8083,
      "step": 93
    },
    {
      "epoch": 0.44287396937573614,
      "grad_norm": 0.8163449764251709,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.7748,
      "step": 94
    },
    {
      "epoch": 0.44758539458186103,
      "grad_norm": 0.9238842725753784,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.6347,
      "step": 95
    },
    {
      "epoch": 0.45229681978798586,
      "grad_norm": 0.8988198041915894,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.9642,
      "step": 96
    },
    {
      "epoch": 0.45700824499411075,
      "grad_norm": 0.8565858006477356,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 3.0546,
      "step": 97
    },
    {
      "epoch": 0.4617196702002356,
      "grad_norm": 0.9136887192726135,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.1682,
      "step": 98
    },
    {
      "epoch": 0.4664310954063604,
      "grad_norm": 0.8954570293426514,
      "learning_rate": 5.495227651252315e-05,
      "loss": 3.2578,
      "step": 99
    },
    {
      "epoch": 0.4711425206124853,
      "grad_norm": 0.8498767018318176,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.7291,
      "step": 100
    },
    {
      "epoch": 0.4711425206124853,
      "eval_loss": 2.874276876449585,
      "eval_runtime": 5.6805,
      "eval_samples_per_second": 63.023,
      "eval_steps_per_second": 15.844,
      "step": 100
    },
    {
      "epoch": 0.4758539458186101,
      "grad_norm": 1.2473000288009644,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.873,
      "step": 101
    },
    {
      "epoch": 0.48056537102473496,
      "grad_norm": 1.1726858615875244,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.0055,
      "step": 102
    },
    {
      "epoch": 0.48527679623085984,
      "grad_norm": 1.086951732635498,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.937,
      "step": 103
    },
    {
      "epoch": 0.48998822143698467,
      "grad_norm": 0.9103227853775024,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.101,
      "step": 104
    },
    {
      "epoch": 0.49469964664310956,
      "grad_norm": 0.8695147633552551,
      "learning_rate": 5e-05,
      "loss": 2.8832,
      "step": 105
    },
    {
      "epoch": 0.4994110718492344,
      "grad_norm": 0.8533949851989746,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.5325,
      "step": 106
    },
    {
      "epoch": 0.5041224970553593,
      "grad_norm": 0.9563857913017273,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.7812,
      "step": 107
    },
    {
      "epoch": 0.508833922261484,
      "grad_norm": 1.043393611907959,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.564,
      "step": 108
    },
    {
      "epoch": 0.5135453474676089,
      "grad_norm": 0.8984698057174683,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.928,
      "step": 109
    },
    {
      "epoch": 0.5182567726737338,
      "grad_norm": 0.9062252044677734,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.8584,
      "step": 110
    },
    {
      "epoch": 0.5229681978798587,
      "grad_norm": 0.8537540435791016,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.7673,
      "step": 111
    },
    {
      "epoch": 0.5276796230859835,
      "grad_norm": 0.9530758261680603,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.957,
      "step": 112
    },
    {
      "epoch": 0.5323910482921084,
      "grad_norm": 1.0364296436309814,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 3.2921,
      "step": 113
    },
    {
      "epoch": 0.5371024734982333,
      "grad_norm": 1.1131585836410522,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.0307,
      "step": 114
    },
    {
      "epoch": 0.541813898704358,
      "grad_norm": 1.1007980108261108,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.8602,
      "step": 115
    },
    {
      "epoch": 0.5465253239104829,
      "grad_norm": 0.9027460813522339,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.8229,
      "step": 116
    },
    {
      "epoch": 0.5512367491166078,
      "grad_norm": 1.0880588293075562,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.7676,
      "step": 117
    },
    {
      "epoch": 0.5559481743227326,
      "grad_norm": 1.158648133277893,
      "learning_rate": 3.933501846281267e-05,
      "loss": 3.0379,
      "step": 118
    },
    {
      "epoch": 0.5606595995288575,
      "grad_norm": 1.1515697240829468,
      "learning_rate": 3.852880399766243e-05,
      "loss": 3.2184,
      "step": 119
    },
    {
      "epoch": 0.5653710247349824,
      "grad_norm": 1.039744257926941,
      "learning_rate": 3.772572564296005e-05,
      "loss": 3.0673,
      "step": 120
    },
    {
      "epoch": 0.5700824499411072,
      "grad_norm": 1.1633607149124146,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 3.0795,
      "step": 121
    },
    {
      "epoch": 0.574793875147232,
      "grad_norm": 1.2394342422485352,
      "learning_rate": 3.612985456190778e-05,
      "loss": 3.1834,
      "step": 122
    },
    {
      "epoch": 0.5795053003533569,
      "grad_norm": 1.2689552307128906,
      "learning_rate": 3.533749813077677e-05,
      "loss": 3.3557,
      "step": 123
    },
    {
      "epoch": 0.5842167255594818,
      "grad_norm": 1.3590177297592163,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 3.0499,
      "step": 124
    },
    {
      "epoch": 0.5889281507656066,
      "grad_norm": 1.0699342489242554,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 3.1957,
      "step": 125
    },
    {
      "epoch": 0.5936395759717314,
      "grad_norm": 1.2349456548690796,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.2495,
      "step": 126
    },
    {
      "epoch": 0.5983510011778563,
      "grad_norm": 1.2268959283828735,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.3801,
      "step": 127
    },
    {
      "epoch": 0.6030624263839811,
      "grad_norm": 1.3209049701690674,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.8863,
      "step": 128
    },
    {
      "epoch": 0.607773851590106,
      "grad_norm": 1.1282166242599487,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.5438,
      "step": 129
    },
    {
      "epoch": 0.6124852767962309,
      "grad_norm": 1.3303471803665161,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.922,
      "step": 130
    },
    {
      "epoch": 0.6171967020023557,
      "grad_norm": 1.158387541770935,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.4659,
      "step": 131
    },
    {
      "epoch": 0.6219081272084805,
      "grad_norm": 1.2458516359329224,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.7004,
      "step": 132
    },
    {
      "epoch": 0.6266195524146054,
      "grad_norm": 1.0022237300872803,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.4997,
      "step": 133
    },
    {
      "epoch": 0.6313309776207303,
      "grad_norm": 1.0322611331939697,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.5472,
      "step": 134
    },
    {
      "epoch": 0.6360424028268551,
      "grad_norm": 1.1272794008255005,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.6108,
      "step": 135
    },
    {
      "epoch": 0.64075382803298,
      "grad_norm": 1.0378586053848267,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.1245,
      "step": 136
    },
    {
      "epoch": 0.6454652532391049,
      "grad_norm": 1.0337663888931274,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.5266,
      "step": 137
    },
    {
      "epoch": 0.6501766784452296,
      "grad_norm": 1.0917435884475708,
      "learning_rate": 2.405152131093926e-05,
      "loss": 3.0042,
      "step": 138
    },
    {
      "epoch": 0.6548881036513545,
      "grad_norm": 0.9059085249900818,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.3542,
      "step": 139
    },
    {
      "epoch": 0.6595995288574794,
      "grad_norm": 1.0263761281967163,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.9872,
      "step": 140
    },
    {
      "epoch": 0.6643109540636042,
      "grad_norm": 0.8678799271583557,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.7748,
      "step": 141
    },
    {
      "epoch": 0.6690223792697291,
      "grad_norm": 1.0620346069335938,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.5043,
      "step": 142
    },
    {
      "epoch": 0.673733804475854,
      "grad_norm": 0.9341846108436584,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.5321,
      "step": 143
    },
    {
      "epoch": 0.6784452296819788,
      "grad_norm": 0.9391326308250427,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.1601,
      "step": 144
    },
    {
      "epoch": 0.6831566548881036,
      "grad_norm": 0.9408818483352661,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.7244,
      "step": 145
    },
    {
      "epoch": 0.6878680800942285,
      "grad_norm": 0.9140599370002747,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.3333,
      "step": 146
    },
    {
      "epoch": 0.6925795053003534,
      "grad_norm": 0.9881057143211365,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.7188,
      "step": 147
    },
    {
      "epoch": 0.6972909305064782,
      "grad_norm": 0.8617043495178223,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.6069,
      "step": 148
    },
    {
      "epoch": 0.702002355712603,
      "grad_norm": 0.9651095867156982,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 3.0078,
      "step": 149
    },
    {
      "epoch": 0.7067137809187279,
      "grad_norm": 1.0257482528686523,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.7801,
      "step": 150
    },
    {
      "epoch": 0.7067137809187279,
      "eval_loss": 2.769987106323242,
      "eval_runtime": 5.6966,
      "eval_samples_per_second": 62.844,
      "eval_steps_per_second": 15.799,
      "step": 150
    },
    {
      "epoch": 0.7114252061248527,
      "grad_norm": 0.9879981279373169,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.7686,
      "step": 151
    },
    {
      "epoch": 0.7161366313309776,
      "grad_norm": 0.9102957844734192,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.4153,
      "step": 152
    },
    {
      "epoch": 0.7208480565371025,
      "grad_norm": 0.9467206001281738,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.9807,
      "step": 153
    },
    {
      "epoch": 0.7255594817432274,
      "grad_norm": 0.8312134742736816,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.4495,
      "step": 154
    },
    {
      "epoch": 0.7302709069493521,
      "grad_norm": 1.0083261728286743,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.9834,
      "step": 155
    },
    {
      "epoch": 0.734982332155477,
      "grad_norm": 0.926459014415741,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.8587,
      "step": 156
    },
    {
      "epoch": 0.7396937573616019,
      "grad_norm": 0.9118709564208984,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.8708,
      "step": 157
    },
    {
      "epoch": 0.7444051825677267,
      "grad_norm": 0.938479483127594,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.6886,
      "step": 158
    },
    {
      "epoch": 0.7491166077738516,
      "grad_norm": 1.0707321166992188,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 3.1004,
      "step": 159
    },
    {
      "epoch": 0.7538280329799765,
      "grad_norm": 0.9356095790863037,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.882,
      "step": 160
    },
    {
      "epoch": 0.7585394581861012,
      "grad_norm": 0.9403328895568848,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.8144,
      "step": 161
    },
    {
      "epoch": 0.7632508833922261,
      "grad_norm": 0.9856503009796143,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.6973,
      "step": 162
    },
    {
      "epoch": 0.767962308598351,
      "grad_norm": 0.9402594566345215,
      "learning_rate": 9.068759265665384e-06,
      "loss": 3.1375,
      "step": 163
    },
    {
      "epoch": 0.7726737338044759,
      "grad_norm": 1.0533093214035034,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.9059,
      "step": 164
    },
    {
      "epoch": 0.7773851590106007,
      "grad_norm": 1.0802216529846191,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.8909,
      "step": 165
    },
    {
      "epoch": 0.7820965842167256,
      "grad_norm": 1.1166588068008423,
      "learning_rate": 7.695237378953223e-06,
      "loss": 3.1018,
      "step": 166
    },
    {
      "epoch": 0.7868080094228505,
      "grad_norm": 1.0489593744277954,
      "learning_rate": 7.260364370723044e-06,
      "loss": 3.0847,
      "step": 167
    },
    {
      "epoch": 0.7915194346289752,
      "grad_norm": 1.2919156551361084,
      "learning_rate": 6.837175952121306e-06,
      "loss": 3.243,
      "step": 168
    },
    {
      "epoch": 0.7962308598351001,
      "grad_norm": 1.2311078310012817,
      "learning_rate": 6.425787818636131e-06,
      "loss": 3.344,
      "step": 169
    },
    {
      "epoch": 0.800942285041225,
      "grad_norm": 1.2963378429412842,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.9426,
      "step": 170
    },
    {
      "epoch": 0.8056537102473498,
      "grad_norm": 1.1835005283355713,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 3.004,
      "step": 171
    },
    {
      "epoch": 0.8103651354534747,
      "grad_norm": 1.3847728967666626,
      "learning_rate": 5.263533508961827e-06,
      "loss": 3.4973,
      "step": 172
    },
    {
      "epoch": 0.8150765606595995,
      "grad_norm": 1.259530782699585,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.9557,
      "step": 173
    },
    {
      "epoch": 0.8197879858657244,
      "grad_norm": 1.645922303199768,
      "learning_rate": 4.549673247541875e-06,
      "loss": 3.4497,
      "step": 174
    },
    {
      "epoch": 0.8244994110718492,
      "grad_norm": 1.6107027530670166,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.9002,
      "step": 175
    },
    {
      "epoch": 0.8292108362779741,
      "grad_norm": 0.8580393195152283,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.2651,
      "step": 176
    },
    {
      "epoch": 0.833922261484099,
      "grad_norm": 0.8875122666358948,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.1955,
      "step": 177
    },
    {
      "epoch": 0.8386336866902238,
      "grad_norm": 0.8691319227218628,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.5762,
      "step": 178
    },
    {
      "epoch": 0.8433451118963486,
      "grad_norm": 1.0163774490356445,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.6119,
      "step": 179
    },
    {
      "epoch": 0.8480565371024735,
      "grad_norm": 0.8920414447784424,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.5154,
      "step": 180
    },
    {
      "epoch": 0.8527679623085983,
      "grad_norm": 0.9631790518760681,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.5506,
      "step": 181
    },
    {
      "epoch": 0.8574793875147232,
      "grad_norm": 0.8805952072143555,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.679,
      "step": 182
    },
    {
      "epoch": 0.8621908127208481,
      "grad_norm": 0.9266687631607056,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.3374,
      "step": 183
    },
    {
      "epoch": 0.866902237926973,
      "grad_norm": 0.854870617389679,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.4187,
      "step": 184
    },
    {
      "epoch": 0.8716136631330977,
      "grad_norm": 0.9341062903404236,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.6638,
      "step": 185
    },
    {
      "epoch": 0.8763250883392226,
      "grad_norm": 0.9016166925430298,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.2794,
      "step": 186
    },
    {
      "epoch": 0.8810365135453475,
      "grad_norm": 1.0805704593658447,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.4947,
      "step": 187
    },
    {
      "epoch": 0.8857479387514723,
      "grad_norm": 0.8769288659095764,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.2467,
      "step": 188
    },
    {
      "epoch": 0.8904593639575972,
      "grad_norm": 0.9262085556983948,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.4642,
      "step": 189
    },
    {
      "epoch": 0.8951707891637221,
      "grad_norm": 0.8909001350402832,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.6498,
      "step": 190
    },
    {
      "epoch": 0.8998822143698468,
      "grad_norm": 0.9195206165313721,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.4863,
      "step": 191
    },
    {
      "epoch": 0.9045936395759717,
      "grad_norm": 0.8525665998458862,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.5905,
      "step": 192
    },
    {
      "epoch": 0.9093050647820966,
      "grad_norm": 0.8745928406715393,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.4172,
      "step": 193
    },
    {
      "epoch": 0.9140164899882215,
      "grad_norm": 0.9491853713989258,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.5457,
      "step": 194
    },
    {
      "epoch": 0.9187279151943463,
      "grad_norm": 0.9138966202735901,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.2656,
      "step": 195
    },
    {
      "epoch": 0.9234393404004712,
      "grad_norm": 0.8036891222000122,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.4428,
      "step": 196
    },
    {
      "epoch": 0.928150765606596,
      "grad_norm": 0.9975553154945374,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.3633,
      "step": 197
    },
    {
      "epoch": 0.9328621908127208,
      "grad_norm": 0.8577960133552551,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.2966,
      "step": 198
    },
    {
      "epoch": 0.9375736160188457,
      "grad_norm": 0.888018012046814,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.7802,
      "step": 199
    },
    {
      "epoch": 0.9422850412249706,
      "grad_norm": 0.9421773552894592,
      "learning_rate": 0.0,
      "loss": 2.8693,
      "step": 200
    },
    {
      "epoch": 0.9422850412249706,
      "eval_loss": 2.7514498233795166,
      "eval_runtime": 5.6843,
      "eval_samples_per_second": 62.98,
      "eval_steps_per_second": 15.833,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7441249896235008.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}