diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4767 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 3390,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 39.04010772705078,
+      "learning_rate": 2.5823527258633718e-06,
+      "loss": 0.3998,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.473583221435547,
+      "learning_rate": 3.694511509569653e-06,
+      "loss": 0.3997,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.636320114135742,
+      "learning_rate": 4.3450826928854785e-06,
+      "loss": 0.6072,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 13.199678421020508,
+      "learning_rate": 4.806670293275934e-06,
+      "loss": 0.3938,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.522007942199707,
+      "learning_rate": 5.1647054517267435e-06,
+      "loss": 0.398,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 26.822412490844727,
+      "learning_rate": 5.45724147659176e-06,
+      "loss": 0.3201,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.275376319885254,
+      "learning_rate": 5.704577161410798e-06,
+      "loss": 0.2569,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.292084693908691,
+      "learning_rate": 5.918829076982215e-06,
+      "loss": 0.2971,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.849486351013184,
+      "learning_rate": 6.107812659907586e-06,
+      "loss": 0.4501,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.202345848083496,
+      "learning_rate": 6.276864235433024e-06,
+      "loss": 0.3756,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.23438835144043,
+      "learning_rate": 6.429789987162081e-06,
+      "loss": 0.2825,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 6.466848373413086,
+      "learning_rate": 6.569400260298041e-06,
+      "loss": 0.3419,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.6180419921875,
+      "learning_rate": 6.697829261969583e-06,
+      "loss": 0.2617,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 17.76142120361328,
+      "learning_rate": 6.816735945117081e-06,
+      "loss": 0.3758,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 11.485393524169922,
+      "learning_rate": 6.92743541874885e-06,
+      "loss": 0.2856,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.610270977020264,
+      "learning_rate": 7.030987860688496e-06,
+      "loss": 0.2366,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.754855632781982,
+      "learning_rate": 7.12826042783297e-06,
+      "loss": 0.25,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.417834281921387,
+      "learning_rate": 7.219971443613867e-06,
+      "loss": 0.2623,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.7129106521606445,
+      "learning_rate": 7.306722622487238e-06,
+      "loss": 0.2763,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 15.627315521240234,
+      "learning_rate": 7.389023019139306e-06,
+      "loss": 0.2837,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 13.026738166809082,
+      "learning_rate": 7.467307128432906e-06,
+      "loss": 0.3366,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.662171363830566,
+      "learning_rate": 7.5419487708683615e-06,
+      "loss": 0.1562,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.789425373077393,
+      "learning_rate": 7.613271888931746e-06,
+      "loss": 0.21,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8265814781188965,
+      "learning_rate": 7.681559044004323e-06,
+      "loss": 0.2166,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 12.92542839050293,
+      "learning_rate": 7.747058177590116e-06,
+      "loss": 0.2111,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.322274684906006,
+      "learning_rate": 7.809988045675864e-06,
+      "loss": 0.3298,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.327513694763184,
+      "learning_rate": 7.870542626929693e-06,
+      "loss": 0.2813,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 6.368852615356445,
+      "learning_rate": 7.92889472882336e-06,
+      "loss": 0.1792,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 50.50636672973633,
+      "learning_rate": 7.985198960672682e-06,
+      "loss": 0.2303,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 7.011040210723877,
+      "learning_rate": 8.03959420245513e-06,
+      "loss": 0.2191,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 11.947832107543945,
+      "learning_rate": 8.092205668665385e-06,
+      "loss": 0.1825,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 27.491235733032227,
+      "learning_rate": 8.143146644394778e-06,
+      "loss": 0.1948,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.949389457702637,
+      "learning_rate": 8.192519954184187e-06,
+      "loss": 0.1383,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 23.224884033203125,
+      "learning_rate": 8.240419211539252e-06,
+      "loss": 0.2232,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.169686317443848,
+      "learning_rate": 8.286929887274171e-06,
+      "loss": 0.1737,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 11.372718811035156,
+      "learning_rate": 8.33213022732015e-06,
+      "loss": 0.2035,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 13.932453155517578,
+      "learning_rate": 8.376092044755856e-06,
+      "loss": 0.2031,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.361552715301514,
+      "learning_rate": 8.418881406193518e-06,
+      "loss": 0.2134,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.29440450668335,
+      "learning_rate": 8.46055922899169e-06,
+      "loss": 0.2055,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.917664527893066,
+      "learning_rate": 8.501181802845587e-06,
+      "loss": 0.1989,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.856716632843018,
+      "learning_rate": 8.540801246962568e-06,
+      "loss": 0.183,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 11.850785255432129,
+      "learning_rate": 8.579465912139187e-06,
+      "loss": 0.1972,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 9.407008171081543,
+      "learning_rate": 8.61722073552112e-06,
+      "loss": 0.2374,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 6.412520885467529,
+      "learning_rate": 8.654107554574644e-06,
+      "loss": 0.1816,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 8.370582580566406,
+      "learning_rate": 8.690165385770957e-06,
+      "loss": 0.2167,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 11.914291381835938,
+      "learning_rate": 8.725430672638028e-06,
+      "loss": 0.3155,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.829929828643799,
+      "learning_rate": 8.759937507133642e-06,
+      "loss": 0.1253,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 8.653116226196289,
+      "learning_rate": 8.793717827710604e-06,
+      "loss": 0.1955,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 7.200388431549072,
+      "learning_rate": 8.826801596958227e-06,
+      "loss": 0.2444,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 7.688900470733643,
+      "learning_rate": 8.859216961296396e-06,
+      "loss": 0.1624,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 9.976654052734375,
+      "learning_rate": 8.890990394855075e-06,
+      "loss": 0.1937,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 16.53554344177246,
+      "learning_rate": 8.922146829382146e-06,
+      "loss": 0.1787,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 41.63074493408203,
+      "learning_rate": 8.952709771776709e-06,
+      "loss": 0.2647,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 9.372986793518066,
+      "learning_rate": 8.982701410635975e-06,
+      "loss": 0.1872,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.8930535316467285,
+      "learning_rate": 9.012142713025453e-06,
+      "loss": 0.2721,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 5.028936386108398,
+      "learning_rate": 9.041053512529642e-06,
+      "loss": 0.1963,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 8.863526344299316,
+      "learning_rate": 9.069452589509345e-06,
+      "loss": 0.2346,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.268564224243164,
+      "learning_rate": 9.097357744378962e-06,
+      "loss": 0.1599,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.382570743560791,
+      "learning_rate": 9.124785864619847e-06,
+      "loss": 0.1709,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 15.576215744018555,
+      "learning_rate": 9.151752986161414e-06,
+      "loss": 0.3055,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 13.345590591430664,
+      "learning_rate": 9.17827434968874e-06,
+      "loss": 0.2155,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 6.941844940185547,
+      "learning_rate": 9.204364452371667e-06,
+      "loss": 0.1481,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 9.678308486938477,
+      "learning_rate": 9.230037095455012e-06,
+      "loss": 0.2374,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.632808208465576,
+      "learning_rate": 9.25530542810106e-06,
+      "loss": 0.1404,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 23.185867309570312,
+      "learning_rate": 9.280181987832955e-06,
+      "loss": 0.2102,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 14.097047805786133,
+      "learning_rate": 9.304678737890467e-06,
+      "loss": 0.1678,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.43781042098999,
+      "learning_rate": 9.328807101776708e-06,
+      "loss": 0.1666,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.9862120151519775,
+      "learning_rate": 9.352577995245534e-06,
+      "loss": 0.1145,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.2823405265808105,
+      "learning_rate": 9.376001855953853e-06,
+      "loss": 0.1804,
+      "step": 345
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 10.184122085571289,
+      "learning_rate": 9.399088670980451e-06,
+      "loss": 0.1874,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.782510280609131,
+      "learning_rate": 9.421848002392907e-06,
+      "loss": 0.1974,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.299713611602783,
+      "learning_rate": 9.44428901102643e-06,
+      "loss": 0.1949,
+      "step": 360
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 8.482523918151855,
+      "learning_rate": 9.466420478622641e-06,
+      "loss": 0.1424,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.908104419708252,
+      "learning_rate": 9.48825082846214e-06,
+      "loss": 0.2288,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 32.536109924316406,
+      "learning_rate": 9.509788144612221e-06,
+      "loss": 0.2137,
+      "step": 375
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.737333059310913,
+      "learning_rate": 9.531040189899801e-06,
+      "loss": 0.1574,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.724475860595703,
+      "learning_rate": 9.552014422709508e-06,
+      "loss": 0.1598,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 15.981143951416016,
+      "learning_rate": 9.57271801269797e-06,
+      "loss": 0.2177,
+      "step": 390
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.07174015045166,
+      "learning_rate": 9.59315785550709e-06,
+      "loss": 0.1467,
+      "step": 395
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.095094680786133,
+      "learning_rate": 9.613340586551869e-06,
+      "loss": 0.2482,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 15.071510314941406,
+      "learning_rate": 9.6332725939518e-06,
+      "loss": 0.1431,
+      "step": 405
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.946211099624634,
+      "learning_rate": 9.652960030668852e-06,
+      "loss": 0.1916,
+      "step": 410
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.153822898864746,
+      "learning_rate": 9.67240882590975e-06,
+      "loss": 0.2548,
+      "step": 415
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.152862071990967,
+      "learning_rate": 9.691624695845468e-06,
+      "loss": 0.3126,
+      "step": 420
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.666506767272949,
+      "learning_rate": 9.710613153696343e-06,
+      "loss": 0.2205,
+      "step": 425
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7740296125411987,
+      "learning_rate": 9.729379519227402e-06,
+      "loss": 0.1986,
+      "step": 430
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 10.573585510253906,
+      "learning_rate": 9.74792892769479e-06,
+      "loss": 0.1729,
+      "step": 435
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.632266521453857,
+      "learning_rate": 9.766266338280922e-06,
+      "loss": 0.2211,
+      "step": 440
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.620193958282471,
+      "learning_rate": 9.784396542053064e-06,
+      "loss": 0.1414,
+      "step": 445
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.095935344696045,
+      "learning_rate": 9.802324169477239e-06,
+      "loss": 0.1934,
+      "step": 450
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 7.872585296630859,
+      "learning_rate": 9.82005369751701e-06,
+      "loss": 0.2298,
+      "step": 455
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.002449989318848,
+      "learning_rate": 9.837589456344308e-06,
+      "loss": 0.1489,
+      "step": 460
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.583741188049316,
+      "learning_rate": 9.854935635687492e-06,
+      "loss": 0.2042,
+      "step": 465
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.471723556518555,
+      "learning_rate": 9.872096290839924e-06,
+      "loss": 0.2002,
+      "step": 470
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 24.396808624267578,
+      "learning_rate": 9.88907534835061e-06,
+      "loss": 0.3188,
+      "step": 475
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.080025672912598,
+      "learning_rate": 9.905876611416884e-06,
+      "loss": 0.1478,
+      "step": 480
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.60750675201416,
+      "learning_rate": 9.922503764997674e-06,
+      "loss": 0.242,
+      "step": 485
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.988901138305664,
+      "learning_rate": 9.938960380664506e-06,
+      "loss": 0.2102,
+      "step": 490
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 10.937402725219727,
+      "learning_rate": 9.955249921206294e-06,
+      "loss": 0.1608,
+      "step": 495
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.663753986358643,
+      "learning_rate": 9.971375745002678e-06,
+      "loss": 0.1964,
+      "step": 500
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.910350799560547,
+      "learning_rate": 9.987341110179803e-06,
+      "loss": 0.2184,
+      "step": 505
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.043577671051025,
+      "learning_rate": 1e-05,
+      "loss": 0.1667,
+      "step": 510
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.844428062438965,
+      "learning_rate": 9.994824552323776e-06,
+      "loss": 0.1524,
+      "step": 515
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 19.891033172607422,
+      "learning_rate": 9.989649104647554e-06,
+      "loss": 0.2023,
+      "step": 520
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 19.217378616333008,
+      "learning_rate": 9.984473656971329e-06,
+      "loss": 0.1474,
+      "step": 525
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 15.412911415100098,
+      "learning_rate": 9.979298209295105e-06,
+      "loss": 0.2625,
+      "step": 530
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 9.4769868850708,
+      "learning_rate": 9.97412276161888e-06,
+      "loss": 0.1401,
+      "step": 535
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 8.897192001342773,
+      "learning_rate": 9.968947313942657e-06,
+      "loss": 0.1465,
+      "step": 540
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 13.993078231811523,
+      "learning_rate": 9.963771866266433e-06,
+      "loss": 0.2059,
+      "step": 545
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 7.764007568359375,
+      "learning_rate": 9.958596418590208e-06,
+      "loss": 0.1501,
+      "step": 550
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.468727111816406,
+      "learning_rate": 9.953420970913986e-06,
+      "loss": 0.1989,
+      "step": 555
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.953731536865234,
+      "learning_rate": 9.948245523237761e-06,
+      "loss": 0.1465,
+      "step": 560
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.55965518951416,
+      "learning_rate": 9.943070075561537e-06,
+      "loss": 0.1702,
+      "step": 565
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 10.50852108001709,
+      "learning_rate": 9.937894627885312e-06,
+      "loss": 0.217,
+      "step": 570
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 17.3203125,
+      "learning_rate": 9.93271918020909e-06,
+      "loss": 0.1711,
+      "step": 575
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 8.020818710327148,
+      "learning_rate": 9.927543732532865e-06,
+      "loss": 0.1818,
+      "step": 580
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.031040191650391,
+      "learning_rate": 9.92236828485664e-06,
+      "loss": 0.2423,
+      "step": 585
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.64624547958374,
+      "learning_rate": 9.917192837180418e-06,
+      "loss": 0.1985,
+      "step": 590
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 12.593125343322754,
+      "learning_rate": 9.912017389504193e-06,
+      "loss": 0.1831,
+      "step": 595
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.998595714569092,
+      "learning_rate": 9.906841941827969e-06,
+      "loss": 0.2549,
+      "step": 600
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 13.419194221496582,
+      "learning_rate": 9.901666494151744e-06,
+      "loss": 0.1324,
+      "step": 605
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.427547454833984,
+      "learning_rate": 9.89649104647552e-06,
+      "loss": 0.1651,
+      "step": 610
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3035593032836914,
+      "learning_rate": 9.891315598799297e-06,
+      "loss": 0.2301,
+      "step": 615
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.742882251739502,
+      "learning_rate": 9.886140151123073e-06,
+      "loss": 0.1536,
+      "step": 620
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.335978984832764,
+      "learning_rate": 9.88096470344685e-06,
+      "loss": 0.134,
+      "step": 625
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.646374225616455,
+      "learning_rate": 9.875789255770626e-06,
+      "loss": 0.1301,
+      "step": 630
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 3.568737030029297,
+      "learning_rate": 9.870613808094401e-06,
+      "loss": 0.2048,
+      "step": 635
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.6915459632873535,
+      "learning_rate": 9.865438360418177e-06,
+      "loss": 0.0971,
+      "step": 640
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.331829071044922,
+      "learning_rate": 9.860262912741952e-06,
+      "loss": 0.1533,
+      "step": 645
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 9.648789405822754,
+      "learning_rate": 9.85508746506573e-06,
+      "loss": 0.1656,
+      "step": 650
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.055738925933838,
+      "learning_rate": 9.849912017389505e-06,
+      "loss": 0.1731,
+      "step": 655
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 18.190107345581055,
+      "learning_rate": 9.84473656971328e-06,
+      "loss": 0.2417,
+      "step": 660
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 10.200769424438477,
+      "learning_rate": 9.839561122037056e-06,
+      "loss": 0.1569,
+      "step": 665
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.0654609203338623,
+      "learning_rate": 9.834385674360833e-06,
+      "loss": 0.1962,
+      "step": 670
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.625911712646484,
+      "learning_rate": 9.829210226684609e-06,
+      "loss": 0.1562,
+      "step": 675
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 9.389242172241211,
+      "learning_rate": 9.824034779008384e-06,
+      "loss": 0.1203,
+      "step": 680
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.225521087646484,
+      "learning_rate": 9.818859331332162e-06,
+      "loss": 0.1874,
+      "step": 685
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.8926544189453125,
+      "learning_rate": 9.813683883655937e-06,
+      "loss": 0.0918,
+      "step": 690
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0372819900512695,
+      "learning_rate": 9.808508435979713e-06,
+      "loss": 0.1675,
+      "step": 695
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.38181209564209,
+      "learning_rate": 9.803332988303488e-06,
+      "loss": 0.1926,
+      "step": 700
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.62568998336792,
+      "learning_rate": 9.798157540627265e-06,
+      "loss": 0.1498,
+      "step": 705
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.986940383911133,
+      "learning_rate": 9.792982092951041e-06,
+      "loss": 0.2168,
+      "step": 710
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.664675235748291,
+      "learning_rate": 9.787806645274816e-06,
+      "loss": 0.1722,
+      "step": 715
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 11.244377136230469,
+      "learning_rate": 9.782631197598594e-06,
+      "loss": 0.179,
+      "step": 720
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.715181350708008,
+      "learning_rate": 9.77745574992237e-06,
+      "loss": 0.1302,
+      "step": 725
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.17584228515625,
+      "learning_rate": 9.772280302246145e-06,
+      "loss": 0.2597,
+      "step": 730
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.8422722816467285,
+      "learning_rate": 9.76710485456992e-06,
+      "loss": 0.1649,
+      "step": 735
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 13.395962715148926,
+      "learning_rate": 9.761929406893698e-06,
+      "loss": 0.1707,
+      "step": 740
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 8.246487617492676,
+      "learning_rate": 9.756753959217473e-06,
+      "loss": 0.1557,
+      "step": 745
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.046871185302734,
+      "learning_rate": 9.751578511541249e-06,
+      "loss": 0.1315,
+      "step": 750
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.337252616882324,
+      "learning_rate": 9.746403063865026e-06,
+      "loss": 0.2316,
+      "step": 755
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.383358001708984,
+      "learning_rate": 9.741227616188801e-06,
+      "loss": 0.2147,
+      "step": 760
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 8.423315048217773,
+      "learning_rate": 9.736052168512577e-06,
+      "loss": 0.2218,
+      "step": 765
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 19.063446044921875,
+      "learning_rate": 9.730876720836352e-06,
+      "loss": 0.1732,
+      "step": 770
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.8348522186279297,
+      "learning_rate": 9.72570127316013e-06,
+      "loss": 0.1946,
+      "step": 775
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 3.952862024307251,
+      "learning_rate": 9.720525825483905e-06,
+      "loss": 0.1215,
+      "step": 780
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.2658185958862305,
+      "learning_rate": 9.71535037780768e-06,
+      "loss": 0.1826,
+      "step": 785
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.492269992828369,
+      "learning_rate": 9.710174930131458e-06,
+      "loss": 0.1975,
+      "step": 790
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.331812381744385,
+      "learning_rate": 9.704999482455233e-06,
+      "loss": 0.1332,
+      "step": 795
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.971770763397217,
+      "learning_rate": 9.699824034779009e-06,
+      "loss": 0.1459,
+      "step": 800
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 8.650938034057617,
+      "learning_rate": 9.694648587102785e-06,
+      "loss": 0.1803,
+      "step": 805
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.830617904663086,
+      "learning_rate": 9.68947313942656e-06,
+      "loss": 0.1623,
+      "step": 810
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 17.95256996154785,
+      "learning_rate": 9.684297691750337e-06,
+      "loss": 0.1923,
+      "step": 815
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 14.101285934448242,
+      "learning_rate": 9.679122244074113e-06,
+      "loss": 0.1491,
+      "step": 820
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 12.107150077819824,
+      "learning_rate": 9.67394679639789e-06,
+      "loss": 0.1686,
+      "step": 825
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.124995708465576,
+      "learning_rate": 9.668771348721666e-06,
+      "loss": 0.1173,
+      "step": 830
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.4361252784729,
+      "learning_rate": 9.663595901045441e-06,
+      "loss": 0.1891,
+      "step": 835
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.013772010803223,
+      "learning_rate": 9.658420453369217e-06,
+      "loss": 0.1114,
+      "step": 840
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.897559642791748,
+      "learning_rate": 9.653245005692992e-06,
+      "loss": 0.1929,
+      "step": 845
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 11.063957214355469,
+      "learning_rate": 9.64806955801677e-06,
+      "loss": 0.2559,
+      "step": 850
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 9.405014991760254,
+      "learning_rate": 9.642894110340545e-06,
+      "loss": 0.2156,
+      "step": 855
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.925081729888916,
+      "learning_rate": 9.637718662664322e-06,
+      "loss": 0.1558,
+      "step": 860
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 9.78219223022461,
+      "learning_rate": 9.632543214988098e-06,
+      "loss": 0.197,
+      "step": 865
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.535521984100342,
+      "learning_rate": 9.627367767311873e-06,
+      "loss": 0.1499,
+      "step": 870
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.9827921390533447,
+      "learning_rate": 9.622192319635649e-06,
+      "loss": 0.1845,
+      "step": 875
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 7.4348368644714355,
+      "learning_rate": 9.617016871959424e-06,
+      "loss": 0.2157,
+      "step": 880
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.102104663848877,
+      "learning_rate": 9.611841424283202e-06,
+      "loss": 0.1661,
+      "step": 885
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.5984147787094116,
+      "learning_rate": 9.606665976606977e-06,
+      "loss": 0.242,
+      "step": 890
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.776930093765259,
+      "learning_rate": 9.601490528930754e-06,
+      "loss": 0.1212,
+      "step": 895
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.875964164733887,
+      "learning_rate": 9.59631508125453e-06,
+      "loss": 0.1273,
+      "step": 900
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 12.377617835998535,
+      "learning_rate": 9.591139633578305e-06,
+      "loss": 0.2036,
+      "step": 905
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.463315486907959,
+      "learning_rate": 9.585964185902081e-06,
+      "loss": 0.1353,
+      "step": 910
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.122995853424072,
+      "learning_rate": 9.580788738225857e-06,
+      "loss": 0.1307,
+      "step": 915
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.756727695465088,
+      "learning_rate": 9.575613290549634e-06,
+      "loss": 0.1825,
+      "step": 920
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.8547186851501465,
+      "learning_rate": 9.57043784287341e-06,
+      "loss": 0.1872,
+      "step": 925
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 4.761024475097656,
+      "learning_rate": 9.565262395197187e-06,
+      "loss": 0.1304,
+      "step": 930
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.3998799324035645,
+      "learning_rate": 9.560086947520962e-06,
+      "loss": 0.1979,
+      "step": 935
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 9.808250427246094,
+      "learning_rate": 9.554911499844738e-06,
+      "loss": 0.1894,
+      "step": 940
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.31424951553344727,
+      "learning_rate": 9.549736052168513e-06,
+      "loss": 0.0892,
+      "step": 945
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.293911933898926,
+      "learning_rate": 9.544560604492289e-06,
+      "loss": 0.1271,
+      "step": 950
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 14.079010009765625,
+      "learning_rate": 9.539385156816066e-06,
+      "loss": 0.2393,
+      "step": 955
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.396958827972412,
+      "learning_rate": 9.534209709139841e-06,
+      "loss": 0.2068,
+      "step": 960
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.725052356719971,
+      "learning_rate": 9.529034261463617e-06,
+      "loss": 0.1309,
+      "step": 965
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.885717868804932,
+      "learning_rate": 9.523858813787392e-06,
+      "loss": 0.1649,
+      "step": 970
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.779788970947266,
+      "learning_rate": 9.51868336611117e-06,
+      "loss": 0.1348,
+      "step": 975
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 9.494329452514648,
+      "learning_rate": 9.513507918434945e-06,
+      "loss": 0.1787,
+      "step": 980
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.396195888519287,
+      "learning_rate": 9.50833247075872e-06,
+      "loss": 0.1996,
+      "step": 985
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 11.899301528930664,
+      "learning_rate": 9.503157023082498e-06,
+      "loss": 0.1055,
+      "step": 990
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 13.169543266296387,
+      "learning_rate": 9.497981575406274e-06,
+      "loss": 0.2897,
+      "step": 995
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 9.189698219299316,
+      "learning_rate": 9.492806127730049e-06,
+      "loss": 0.1731,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.269948720932007,
+      "learning_rate": 9.487630680053825e-06,
+      "loss": 0.1266,
+      "step": 1005
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 7.035939693450928,
+      "learning_rate": 9.4824552323776e-06,
+      "loss": 0.2074,
+      "step": 1010
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3484182357788086,
+      "learning_rate": 9.477279784701377e-06,
+      "loss": 0.131,
+      "step": 1015
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.9915425777435303,
+      "learning_rate": 9.472104337025153e-06,
+      "loss": 0.1174,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.0831351280212402,
+      "learning_rate": 9.46692888934893e-06,
+      "loss": 0.1405,
+      "step": 1025
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.617198467254639,
+      "learning_rate": 9.461753441672706e-06,
+      "loss": 0.1713,
+      "step": 1030
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.402609348297119,
+      "learning_rate": 9.456577993996481e-06,
+      "loss": 0.1257,
+      "step": 1035
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9688963890075684,
+      "learning_rate": 9.451402546320257e-06,
+      "loss": 0.1317,
+      "step": 1040
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9347572326660156,
+      "learning_rate": 9.446227098644032e-06,
+      "loss": 0.1135,
+      "step": 1045
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.43341326713562,
+      "learning_rate": 9.44105165096781e-06,
+      "loss": 0.1259,
+      "step": 1050
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.623471736907959,
+      "learning_rate": 9.435876203291585e-06,
+      "loss": 0.1374,
+      "step": 1055
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.5787482261657715,
+      "learning_rate": 9.430700755615362e-06,
+      "loss": 0.1467,
+      "step": 1060
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.180797100067139,
+      "learning_rate": 9.425525307939138e-06,
+      "loss": 0.128,
+      "step": 1065
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 8.534868240356445,
+      "learning_rate": 9.420349860262913e-06,
+      "loss": 0.1406,
+      "step": 1070
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.018136978149414,
+      "learning_rate": 9.415174412586689e-06,
+      "loss": 0.1226,
+      "step": 1075
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.855718612670898,
+      "learning_rate": 9.409998964910464e-06,
+      "loss": 0.1391,
+      "step": 1080
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.620302677154541,
+      "learning_rate": 9.404823517234242e-06,
+      "loss": 0.164,
+      "step": 1085
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 14.70823860168457,
+      "learning_rate": 9.399648069558017e-06,
+      "loss": 0.1424,
+      "step": 1090
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.744147777557373,
+      "learning_rate": 9.394472621881794e-06,
+      "loss": 0.1077,
+      "step": 1095
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.587203025817871,
+      "learning_rate": 9.38929717420557e-06,
+      "loss": 0.102,
+      "step": 1100
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.23123025894165,
+      "learning_rate": 9.384121726529346e-06,
+      "loss": 0.1465,
+      "step": 1105
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.246310234069824,
+      "learning_rate": 9.378946278853121e-06,
+      "loss": 0.1028,
+      "step": 1110
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.397068500518799,
+      "learning_rate": 9.373770831176897e-06,
+      "loss": 0.1285,
+      "step": 1115
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3886682987213135,
+      "learning_rate": 9.368595383500674e-06,
+      "loss": 0.1158,
+      "step": 1120
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8311169147491455,
+      "learning_rate": 9.36341993582445e-06,
+      "loss": 0.1255,
+      "step": 1125
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.853452205657959,
+      "learning_rate": 9.358244488148227e-06,
+      "loss": 0.1404,
+      "step": 1130
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.4183835983276367,
+      "learning_rate": 9.353069040472002e-06,
+      "loss": 0.1393,
+      "step": 1135
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7233846187591553,
+      "learning_rate": 9.347893592795778e-06,
+      "loss": 0.0957,
+      "step": 1140
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.0931410789489746,
+      "learning_rate": 9.342718145119553e-06,
+      "loss": 0.0801,
+      "step": 1145
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.077810764312744,
+      "learning_rate": 9.337542697443329e-06,
+      "loss": 0.151,
+      "step": 1150
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.8209195137023926,
+      "learning_rate": 9.332367249767106e-06,
+      "loss": 0.1698,
+      "step": 1155
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.544396162033081,
+      "learning_rate": 9.327191802090881e-06,
+      "loss": 0.1098,
+      "step": 1160
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.884575366973877,
+      "learning_rate": 9.322016354414659e-06,
+      "loss": 0.1335,
+      "step": 1165
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 7.00750207901001,
+      "learning_rate": 9.316840906738434e-06,
+      "loss": 0.139,
+      "step": 1170
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9287322759628296,
+      "learning_rate": 9.31166545906221e-06,
+      "loss": 0.1356,
+      "step": 1175
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.5304596424102783,
+      "learning_rate": 9.306490011385985e-06,
+      "loss": 0.1421,
+      "step": 1180
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.7318196296691895,
+      "learning_rate": 9.301314563709761e-06,
+      "loss": 0.1368,
+      "step": 1185
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.634664297103882,
+      "learning_rate": 9.296139116033538e-06,
+      "loss": 0.134,
+      "step": 1190
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.424448013305664,
+      "learning_rate": 9.290963668357314e-06,
+      "loss": 0.0917,
+      "step": 1195
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.754941463470459,
+      "learning_rate": 9.285788220681091e-06,
+      "loss": 0.1162,
+      "step": 1200
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.112506866455078,
+      "learning_rate": 9.280612773004866e-06,
+      "loss": 0.1247,
+      "step": 1205
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.3786399364471436,
+      "learning_rate": 9.275437325328642e-06,
+      "loss": 0.1024,
+      "step": 1210
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.081037998199463,
+      "learning_rate": 9.270261877652417e-06,
+      "loss": 0.1543,
+      "step": 1215
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.124878883361816,
+      "learning_rate": 9.265086429976193e-06,
+      "loss": 0.1118,
+      "step": 1220
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 15.232734680175781,
+      "learning_rate": 9.25991098229997e-06,
+      "loss": 0.2026,
+      "step": 1225
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.690507411956787,
+      "learning_rate": 9.254735534623746e-06,
+      "loss": 0.1122,
+      "step": 1230
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.408130168914795,
+      "learning_rate": 9.249560086947523e-06,
+      "loss": 0.1255,
+      "step": 1235
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6263089179992676,
+      "learning_rate": 9.244384639271299e-06,
+      "loss": 0.1306,
+      "step": 1240
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4534480571746826,
+      "learning_rate": 9.239209191595074e-06,
+      "loss": 0.2347,
+      "step": 1245
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.317373752593994,
+      "learning_rate": 9.23403374391885e-06,
+      "loss": 0.12,
+      "step": 1250
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.0810723304748535,
+      "learning_rate": 9.228858296242625e-06,
+      "loss": 0.1477,
+      "step": 1255
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.283056735992432,
+      "learning_rate": 9.223682848566402e-06,
+      "loss": 0.0969,
+      "step": 1260
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.569967746734619,
+      "learning_rate": 9.218507400890178e-06,
+      "loss": 0.1203,
+      "step": 1265
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.6180992126464844,
+      "learning_rate": 9.213331953213953e-06,
+      "loss": 0.0803,
+      "step": 1270
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.16738224029541,
+      "learning_rate": 9.208156505537729e-06,
+      "loss": 0.1283,
+      "step": 1275
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 7.953784942626953,
+      "learning_rate": 9.202981057861505e-06,
+      "loss": 0.1768,
+      "step": 1280
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.1365580558776855,
+      "learning_rate": 9.197805610185282e-06,
+      "loss": 0.1379,
+      "step": 1285
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.42366099357605,
+      "learning_rate": 9.192630162509057e-06,
+      "loss": 0.152,
+      "step": 1290
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 3.656219959259033,
+      "learning_rate": 9.187454714832835e-06,
+      "loss": 0.1641,
+      "step": 1295
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.415257930755615,
+      "learning_rate": 9.18227926715661e-06,
+      "loss": 0.1181,
+      "step": 1300
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.553100109100342,
+      "learning_rate": 9.177103819480386e-06,
+      "loss": 0.1163,
+      "step": 1305
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 11.656464576721191,
+      "learning_rate": 9.171928371804161e-06,
+      "loss": 0.0927,
+      "step": 1310
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1004693508148193,
+      "learning_rate": 9.166752924127937e-06,
+      "loss": 0.1332,
+      "step": 1315
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.8113784790039062,
+      "learning_rate": 9.161577476451714e-06,
+      "loss": 0.1187,
+      "step": 1320
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.241288185119629,
+      "learning_rate": 9.15640202877549e-06,
+      "loss": 0.116,
+      "step": 1325
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.4864501953125,
+      "learning_rate": 9.151226581099267e-06,
+      "loss": 0.1342,
+      "step": 1330
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.652744770050049,
+      "learning_rate": 9.146051133423042e-06,
+      "loss": 0.1152,
+      "step": 1335
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.944018602371216,
+      "learning_rate": 9.140875685746818e-06,
+      "loss": 0.1134,
+      "step": 1340
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3197338581085205,
+      "learning_rate": 9.135700238070593e-06,
+      "loss": 0.1018,
+      "step": 1345
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.744292736053467,
+      "learning_rate": 9.130524790394369e-06,
+      "loss": 0.1008,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.1377034187316895,
+      "learning_rate": 9.125349342718146e-06,
+      "loss": 0.0839,
+      "step": 1355
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.71573543548584,
+      "learning_rate": 9.120173895041922e-06,
+      "loss": 0.1006,
+      "step": 1360
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.683949947357178,
+      "learning_rate": 9.114998447365699e-06,
+      "loss": 0.1325,
+      "step": 1365
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.3707456588745117,
+      "learning_rate": 9.109822999689474e-06,
+      "loss": 0.1107,
+      "step": 1370
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.6643354892730713,
+      "learning_rate": 9.10464755201325e-06,
+      "loss": 0.1424,
+      "step": 1375
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.6450676918029785,
+      "learning_rate": 9.099472104337025e-06,
+      "loss": 0.1209,
+      "step": 1380
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.182102680206299,
+      "learning_rate": 9.094296656660801e-06,
+      "loss": 0.1091,
+      "step": 1385
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.8936269283294678,
+      "learning_rate": 9.089121208984578e-06,
+      "loss": 0.1549,
+      "step": 1390
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 8.791295051574707,
+      "learning_rate": 9.083945761308354e-06,
+      "loss": 0.146,
+      "step": 1395
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 3.5521798133850098,
+      "learning_rate": 9.078770313632131e-06,
+      "loss": 0.1093,
+      "step": 1400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 6.907190799713135,
+      "learning_rate": 9.073594865955906e-06,
+      "loss": 0.1613,
+      "step": 1405
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 5.13857364654541,
+      "learning_rate": 9.068419418279682e-06,
+      "loss": 0.1152,
+      "step": 1410
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.6005055904388428,
+      "learning_rate": 9.063243970603458e-06,
+      "loss": 0.0915,
+      "step": 1415
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.14816951751709,
+      "learning_rate": 9.058068522927233e-06,
+      "loss": 0.1324,
+      "step": 1420
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.6514110565185547,
+      "learning_rate": 9.05289307525101e-06,
+      "loss": 0.1786,
+      "step": 1425
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.708466053009033,
+      "learning_rate": 9.047717627574786e-06,
+      "loss": 0.1084,
+      "step": 1430
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 3.2835967540740967,
+      "learning_rate": 9.042542179898563e-06,
+      "loss": 0.1523,
+      "step": 1435
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.636791467666626,
+      "learning_rate": 9.037366732222339e-06,
+      "loss": 0.1212,
+      "step": 1440
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 11.84289264678955,
+      "learning_rate": 9.032191284546114e-06,
+      "loss": 0.1382,
+      "step": 1445
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.9107322692871094,
+      "learning_rate": 9.02701583686989e-06,
+      "loss": 0.1423,
+      "step": 1450
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6146843433380127,
+      "learning_rate": 9.021840389193665e-06,
+      "loss": 0.0948,
+      "step": 1455
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.3949062824249268,
+      "learning_rate": 9.016664941517442e-06,
+      "loss": 0.0897,
+      "step": 1460
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.4755208492279053,
+      "learning_rate": 9.011489493841218e-06,
+      "loss": 0.1323,
+      "step": 1465
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 3.312089681625366,
+      "learning_rate": 9.006314046164995e-06,
+      "loss": 0.125,
+      "step": 1470
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.586395025253296,
+      "learning_rate": 9.00113859848877e-06,
+      "loss": 0.1028,
+      "step": 1475
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.178766965866089,
+      "learning_rate": 8.995963150812546e-06,
+      "loss": 0.1119,
+      "step": 1480
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.092270851135254,
+      "learning_rate": 8.990787703136322e-06,
+      "loss": 0.1102,
+      "step": 1485
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 3.5956788063049316,
+      "learning_rate": 8.985612255460097e-06,
+      "loss": 0.1169,
+      "step": 1490
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.364086627960205,
+      "learning_rate": 8.980436807783875e-06,
+      "loss": 0.1486,
+      "step": 1495
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.4916880130767822,
+      "learning_rate": 8.97526136010765e-06,
+      "loss": 0.0905,
+      "step": 1500
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.576471328735352,
+      "learning_rate": 8.970085912431427e-06,
+      "loss": 0.1147,
+      "step": 1505
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.7002522945404053,
+      "learning_rate": 8.964910464755203e-06,
+      "loss": 0.1215,
+      "step": 1510
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.3208682537078857,
+      "learning_rate": 8.959735017078978e-06,
+      "loss": 0.095,
+      "step": 1515
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.178728103637695,
+      "learning_rate": 8.954559569402754e-06,
+      "loss": 0.1216,
+      "step": 1520
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.2932112216949463,
+      "learning_rate": 8.94938412172653e-06,
+      "loss": 0.1066,
+      "step": 1525
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 9.658435821533203,
+      "learning_rate": 8.944208674050307e-06,
+      "loss": 0.1705,
+      "step": 1530
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.9766809940338135,
+      "learning_rate": 8.939033226374082e-06,
+      "loss": 0.1756,
+      "step": 1535
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.549077272415161,
+      "learning_rate": 8.933857778697858e-06,
+      "loss": 0.1021,
+      "step": 1540
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7569571733474731,
+      "learning_rate": 8.928682331021633e-06,
+      "loss": 0.1469,
+      "step": 1545
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 7.744059085845947,
+      "learning_rate": 8.923506883345409e-06,
+      "loss": 0.0861,
+      "step": 1550
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.042254686355591,
+      "learning_rate": 8.918331435669186e-06,
+      "loss": 0.1519,
+      "step": 1555
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.3460726737976074,
+      "learning_rate": 8.913155987992962e-06,
+      "loss": 0.159,
+      "step": 1560
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.934263229370117,
+      "learning_rate": 8.907980540316739e-06,
+      "loss": 0.1255,
+      "step": 1565
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 9.066960334777832,
+      "learning_rate": 8.902805092640514e-06,
+      "loss": 0.1466,
+      "step": 1570
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.1728413105010986,
+      "learning_rate": 8.89762964496429e-06,
+      "loss": 0.1256,
+      "step": 1575
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9333866834640503,
+      "learning_rate": 8.892454197288065e-06,
+      "loss": 0.106,
+      "step": 1580
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.112651824951172,
+      "learning_rate": 8.887278749611841e-06,
+      "loss": 0.1369,
+      "step": 1585
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.346231698989868,
+      "learning_rate": 8.882103301935618e-06,
+      "loss": 0.0928,
+      "step": 1590
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.798091888427734,
+      "learning_rate": 8.876927854259394e-06,
+      "loss": 0.1168,
+      "step": 1595
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.5871007442474365,
+      "learning_rate": 8.871752406583171e-06,
+      "loss": 0.1342,
+      "step": 1600
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.747342586517334,
+      "learning_rate": 8.866576958906947e-06,
+      "loss": 0.1094,
+      "step": 1605
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 4.007979393005371,
+      "learning_rate": 8.861401511230722e-06,
+      "loss": 0.0925,
+      "step": 1610
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.748072147369385,
+      "learning_rate": 8.856226063554498e-06,
+      "loss": 0.1125,
+      "step": 1615
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 11.549009323120117,
+      "learning_rate": 8.851050615878273e-06,
+      "loss": 0.1457,
+      "step": 1620
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.110932350158691,
+      "learning_rate": 8.84587516820205e-06,
+      "loss": 0.1361,
+      "step": 1625
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.154113292694092,
+      "learning_rate": 8.840699720525826e-06,
+      "loss": 0.1076,
+      "step": 1630
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.6666300296783447,
+      "learning_rate": 8.835524272849603e-06,
+      "loss": 0.1123,
+      "step": 1635
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.108492851257324,
+      "learning_rate": 8.830348825173379e-06,
+      "loss": 0.0934,
+      "step": 1640
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.6310579776763916,
+      "learning_rate": 8.825173377497154e-06,
+      "loss": 0.1053,
+      "step": 1645
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.3506951332092285,
+      "learning_rate": 8.81999792982093e-06,
+      "loss": 0.1281,
+      "step": 1650
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.073818683624268,
+      "learning_rate": 8.814822482144705e-06,
+      "loss": 0.1176,
+      "step": 1655
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.0923123359680176,
+      "learning_rate": 8.809647034468483e-06,
+      "loss": 0.1026,
+      "step": 1660
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.954655408859253,
+      "learning_rate": 8.804471586792258e-06,
+      "loss": 0.1143,
+      "step": 1665
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.5785014629364014,
+      "learning_rate": 8.799296139116035e-06,
+      "loss": 0.0953,
+      "step": 1670
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 3.692244529724121,
+      "learning_rate": 8.79412069143981e-06,
+      "loss": 0.1513,
+      "step": 1675
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 3.2623026371002197,
+      "learning_rate": 8.788945243763586e-06,
+      "loss": 0.0922,
+      "step": 1680
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.969079971313477,
+      "learning_rate": 8.783769796087362e-06,
+      "loss": 0.1724,
+      "step": 1685
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.769530296325684,
+      "learning_rate": 8.778594348411137e-06,
+      "loss": 0.1353,
+      "step": 1690
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.35957670211792,
+      "learning_rate": 8.773418900734915e-06,
+      "loss": 0.0852,
+      "step": 1695
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.807293891906738,
+      "learning_rate": 8.76824345305869e-06,
+      "loss": 0.1856,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.785705089569092,
+      "learning_rate": 8.763068005382467e-06,
+      "loss": 0.1055,
+      "step": 1705
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.076875686645508,
+      "learning_rate": 8.757892557706243e-06,
+      "loss": 0.1451,
+      "step": 1710
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.639549732208252,
+      "learning_rate": 8.752717110030019e-06,
+      "loss": 0.0881,
+      "step": 1715
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 7.621575355529785,
+      "learning_rate": 8.747541662353794e-06,
+      "loss": 0.1029,
+      "step": 1720
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.726953029632568,
+      "learning_rate": 8.74236621467757e-06,
+      "loss": 0.1517,
+      "step": 1725
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.103801727294922,
+      "learning_rate": 8.737190767001347e-06,
+      "loss": 0.13,
+      "step": 1730
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.969550609588623,
+      "learning_rate": 8.732015319325122e-06,
+      "loss": 0.1196,
+      "step": 1735
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.2284793853759766,
+      "learning_rate": 8.7268398716489e-06,
+      "loss": 0.0842,
+      "step": 1740
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 7.681682109832764,
+      "learning_rate": 8.721664423972675e-06,
+      "loss": 0.1357,
+      "step": 1745
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.924267292022705,
+      "learning_rate": 8.71648897629645e-06,
+      "loss": 0.1242,
+      "step": 1750
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.502907752990723,
+      "learning_rate": 8.711313528620226e-06,
+      "loss": 0.1036,
+      "step": 1755
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.329269886016846,
+      "learning_rate": 8.706138080944002e-06,
+      "loss": 0.136,
+      "step": 1760
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2087464332580566,
+      "learning_rate": 8.700962633267779e-06,
+      "loss": 0.1256,
+      "step": 1765
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 12.253616333007812,
+      "learning_rate": 8.695787185591554e-06,
+      "loss": 0.1627,
+      "step": 1770
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.6774630546569824,
+      "learning_rate": 8.69061173791533e-06,
+      "loss": 0.1309,
+      "step": 1775
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.795629858970642,
+      "learning_rate": 8.685436290239107e-06,
+      "loss": 0.0908,
+      "step": 1780
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.8215413093566895,
+      "learning_rate": 8.680260842562883e-06,
+      "loss": 0.1094,
+      "step": 1785
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.343129873275757,
+      "learning_rate": 8.675085394886658e-06,
+      "loss": 0.1093,
+      "step": 1790
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.326789140701294,
+      "learning_rate": 8.669909947210434e-06,
+      "loss": 0.0647,
+      "step": 1795
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 4.836015701293945,
+      "learning_rate": 8.664734499534211e-06,
+      "loss": 0.154,
+      "step": 1800
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.4941964149475098,
+      "learning_rate": 8.659559051857987e-06,
+      "loss": 0.11,
+      "step": 1805
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3634862899780273,
+      "learning_rate": 8.654383604181762e-06,
+      "loss": 0.0976,
+      "step": 1810
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.757242679595947,
+      "learning_rate": 8.64920815650554e-06,
+      "loss": 0.157,
+      "step": 1815
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.6024909019470215,
+      "learning_rate": 8.644032708829315e-06,
+      "loss": 0.1137,
+      "step": 1820
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.5130951404571533,
+      "learning_rate": 8.63885726115309e-06,
+      "loss": 0.1359,
+      "step": 1825
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.966015100479126,
+      "learning_rate": 8.633681813476866e-06,
+      "loss": 0.076,
+      "step": 1830
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 4.336612701416016,
+      "learning_rate": 8.628506365800643e-06,
+      "loss": 0.0859,
+      "step": 1835
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.2292232513427734,
+      "learning_rate": 8.623330918124419e-06,
+      "loss": 0.13,
+      "step": 1840
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.255389451980591,
+      "learning_rate": 8.618155470448194e-06,
+      "loss": 0.1597,
+      "step": 1845
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.719048023223877,
+      "learning_rate": 8.61298002277197e-06,
+      "loss": 0.1573,
+      "step": 1850
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.7208411693573,
+      "learning_rate": 8.607804575095745e-06,
+      "loss": 0.104,
+      "step": 1855
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8547576665878296,
+      "learning_rate": 8.602629127419523e-06,
+      "loss": 0.059,
+      "step": 1860
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.5389063358306885,
+      "learning_rate": 8.597453679743298e-06,
+      "loss": 0.1425,
+      "step": 1865
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.625467300415039,
+      "learning_rate": 8.592278232067075e-06,
+      "loss": 0.0647,
+      "step": 1870
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.22482967376709,
+      "learning_rate": 8.587102784390851e-06,
+      "loss": 0.1237,
+      "step": 1875
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.8151726722717285,
+      "learning_rate": 8.581927336714626e-06,
+      "loss": 0.1252,
+      "step": 1880
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.1997487545013428,
+      "learning_rate": 8.576751889038402e-06,
+      "loss": 0.0666,
+      "step": 1885
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.804878234863281,
+      "learning_rate": 8.571576441362178e-06,
+      "loss": 0.1488,
+      "step": 1890
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 12.413414001464844,
+      "learning_rate": 8.566400993685955e-06,
+      "loss": 0.1222,
+      "step": 1895
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.686859607696533,
+      "learning_rate": 8.56122554600973e-06,
+      "loss": 0.1067,
+      "step": 1900
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.566218376159668,
+      "learning_rate": 8.556050098333508e-06,
+      "loss": 0.08,
+      "step": 1905
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 8.065584182739258,
+      "learning_rate": 8.550874650657283e-06,
+      "loss": 0.154,
+      "step": 1910
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.6219561100006104,
+      "learning_rate": 8.545699202981059e-06,
+      "loss": 0.1673,
+      "step": 1915
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.5705161094665527,
+      "learning_rate": 8.540523755304834e-06,
+      "loss": 0.1166,
+      "step": 1920
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.346094846725464,
+      "learning_rate": 8.53534830762861e-06,
+      "loss": 0.0811,
+      "step": 1925
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.2743337154388428,
+      "learning_rate": 8.530172859952387e-06,
+      "loss": 0.1106,
+      "step": 1930
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.068463325500488,
+      "learning_rate": 8.524997412276162e-06,
+      "loss": 0.093,
+      "step": 1935
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.340872287750244,
+      "learning_rate": 8.519821964599938e-06,
+      "loss": 0.1139,
+      "step": 1940
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.84330153465271,
+      "learning_rate": 8.514646516923715e-06,
+      "loss": 0.1289,
+      "step": 1945
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.748030662536621,
+      "learning_rate": 8.50947106924749e-06,
+      "loss": 0.1314,
+      "step": 1950
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.315107822418213,
+      "learning_rate": 8.504295621571266e-06,
+      "loss": 0.1249,
+      "step": 1955
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0189619064331055,
+      "learning_rate": 8.499120173895042e-06,
+      "loss": 0.0938,
+      "step": 1960
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.526963710784912,
+      "learning_rate": 8.493944726218819e-06,
+      "loss": 0.124,
+      "step": 1965
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 9.746664047241211,
+      "learning_rate": 8.488769278542595e-06,
+      "loss": 0.1081,
+      "step": 1970
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.6642544269561768,
+      "learning_rate": 8.48359383086637e-06,
+      "loss": 0.1162,
+      "step": 1975
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.4622726440429688,
+      "learning_rate": 8.478418383190147e-06,
+      "loss": 0.1106,
+      "step": 1980
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.770214557647705,
+      "learning_rate": 8.473242935513923e-06,
+      "loss": 0.1001,
+      "step": 1985
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.70098352432251,
+      "learning_rate": 8.468067487837698e-06,
+      "loss": 0.1314,
+      "step": 1990
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.707994222640991,
+      "learning_rate": 8.462892040161474e-06,
+      "loss": 0.1319,
+      "step": 1995
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 9.308713912963867,
+      "learning_rate": 8.457716592485251e-06,
+      "loss": 0.0719,
+      "step": 2000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 4.883936405181885,
+      "learning_rate": 8.452541144809027e-06,
+      "loss": 0.1398,
+      "step": 2005
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9737305641174316,
+      "learning_rate": 8.447365697132802e-06,
+      "loss": 0.1068,
+      "step": 2010
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6990957260131836,
+      "learning_rate": 8.44219024945658e-06,
+      "loss": 0.1436,
+      "step": 2015
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 14.374921798706055,
+      "learning_rate": 8.437014801780355e-06,
+      "loss": 0.1821,
+      "step": 2020
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.804267168045044,
+      "learning_rate": 8.43183935410413e-06,
+      "loss": 0.1198,
+      "step": 2025
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.386476516723633,
+      "learning_rate": 8.426663906427906e-06,
+      "loss": 0.1082,
+      "step": 2030
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4496395587921143,
+      "learning_rate": 8.421488458751683e-06,
+      "loss": 0.0944,
+      "step": 2035
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.4833619594573975,
+      "learning_rate": 8.416313011075459e-06,
+      "loss": 0.0983,
+      "step": 2040
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.299015045166016,
+      "learning_rate": 8.411137563399234e-06,
+      "loss": 0.1113,
+      "step": 2045
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.734581470489502,
+      "learning_rate": 8.405962115723012e-06,
+      "loss": 0.0937,
+      "step": 2050
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.731231689453125,
+      "learning_rate": 8.400786668046787e-06,
+      "loss": 0.1026,
+      "step": 2055
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 8.851029396057129,
+      "learning_rate": 8.395611220370563e-06,
+      "loss": 0.135,
+      "step": 2060
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 8.13473129272461,
+      "learning_rate": 8.390435772694338e-06,
+      "loss": 0.1154,
+      "step": 2065
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.136192798614502,
+      "learning_rate": 8.385260325018115e-06,
+      "loss": 0.0957,
+      "step": 2070
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2493977546691895,
+      "learning_rate": 8.380084877341891e-06,
+      "loss": 0.1547,
+      "step": 2075
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 3.6070656776428223,
+      "learning_rate": 8.374909429665667e-06,
+      "loss": 0.1434,
+      "step": 2080
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.380437135696411,
+      "learning_rate": 8.369733981989444e-06,
+      "loss": 0.1333,
+      "step": 2085
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.745711326599121,
+      "learning_rate": 8.36455853431322e-06,
+      "loss": 0.1685,
+      "step": 2090
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2399375438690186,
+      "learning_rate": 8.359383086636995e-06,
+      "loss": 0.1315,
+      "step": 2095
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.057147741317749,
+      "learning_rate": 8.35420763896077e-06,
+      "loss": 0.1096,
+      "step": 2100
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.965878486633301,
+      "learning_rate": 8.349032191284548e-06,
+      "loss": 0.1168,
+      "step": 2105
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.02713680267334,
+      "learning_rate": 8.343856743608323e-06,
+      "loss": 0.1185,
+      "step": 2110
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 3.4443166255950928,
+      "learning_rate": 8.338681295932099e-06,
+      "loss": 0.0994,
+      "step": 2115
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.797217845916748,
+      "learning_rate": 8.333505848255874e-06,
+      "loss": 0.1487,
+      "step": 2120
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4136006832122803,
+      "learning_rate": 8.32833040057965e-06,
+      "loss": 0.1132,
+      "step": 2125
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.647338628768921,
+      "learning_rate": 8.323154952903427e-06,
+      "loss": 0.1052,
+      "step": 2130
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.742266654968262,
+      "learning_rate": 8.317979505227203e-06,
+      "loss": 0.1793,
+      "step": 2135
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 3.9803152084350586,
+      "learning_rate": 8.312804057550978e-06,
+      "loss": 0.0951,
+      "step": 2140
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.083047866821289,
+      "learning_rate": 8.307628609874755e-06,
+      "loss": 0.0876,
+      "step": 2145
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 7.0190110206604,
+      "learning_rate": 8.30245316219853e-06,
+      "loss": 0.1077,
+      "step": 2150
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.839928150177002,
+      "learning_rate": 8.297277714522306e-06,
+      "loss": 0.0978,
+      "step": 2155
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.520656943321228,
+      "learning_rate": 8.292102266846082e-06,
+      "loss": 0.0653,
+      "step": 2160
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.554316997528076,
+      "learning_rate": 8.286926819169859e-06,
+      "loss": 0.1021,
+      "step": 2165
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8181085586547852,
+      "learning_rate": 8.281751371493635e-06,
+      "loss": 0.0937,
+      "step": 2170
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.654700517654419,
+      "learning_rate": 8.27657592381741e-06,
+      "loss": 0.1007,
+      "step": 2175
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.064187526702881,
+      "learning_rate": 8.271400476141187e-06,
+      "loss": 0.1219,
+      "step": 2180
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.9234604835510254,
+      "learning_rate": 8.266225028464963e-06,
+      "loss": 0.111,
+      "step": 2185
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.7650468349456787,
+      "learning_rate": 8.261049580788738e-06,
+      "loss": 0.1052,
+      "step": 2190
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.4724655151367188,
+      "learning_rate": 8.255874133112514e-06,
+      "loss": 0.1064,
+      "step": 2195
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.5598931312561035,
+      "learning_rate": 8.250698685436291e-06,
+      "loss": 0.1032,
+      "step": 2200
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.593649387359619,
+      "learning_rate": 8.245523237760067e-06,
+      "loss": 0.1088,
+      "step": 2205
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.6620819568634033,
+      "learning_rate": 8.240347790083842e-06,
+      "loss": 0.0928,
+      "step": 2210
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 6.149742603302002,
+      "learning_rate": 8.23517234240762e-06,
+      "loss": 0.1639,
+      "step": 2215
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.050989866256714,
+      "learning_rate": 8.229996894731395e-06,
+      "loss": 0.166,
+      "step": 2220
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.9382867813110352,
+      "learning_rate": 8.22482144705517e-06,
+      "loss": 0.1032,
+      "step": 2225
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5945475101470947,
+      "learning_rate": 8.219645999378946e-06,
+      "loss": 0.1136,
+      "step": 2230
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.525472164154053,
+      "learning_rate": 8.214470551702723e-06,
+      "loss": 0.1659,
+      "step": 2235
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.980546236038208,
+      "learning_rate": 8.209295104026499e-06,
+      "loss": 0.1036,
+      "step": 2240
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.115708351135254,
+      "learning_rate": 8.204119656350274e-06,
+      "loss": 0.1247,
+      "step": 2245
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.5654399394989014,
+      "learning_rate": 8.198944208674052e-06,
+      "loss": 0.0956,
+      "step": 2250
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.081153154373169,
+      "learning_rate": 8.193768760997827e-06,
+      "loss": 0.1425,
+      "step": 2255
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.4958114624023438,
+      "learning_rate": 8.188593313321603e-06,
+      "loss": 0.1257,
+      "step": 2260
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.821100115776062,
+      "learning_rate": 8.183417865645378e-06,
+      "loss": 0.0976,
+      "step": 2265
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.486393928527832,
+      "learning_rate": 8.178242417969156e-06,
+      "loss": 0.1071,
+      "step": 2270
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.3406879901885986,
+      "learning_rate": 8.173066970292931e-06,
+      "loss": 0.1557,
+      "step": 2275
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.4600462913513184,
+      "learning_rate": 8.167891522616707e-06,
+      "loss": 0.0669,
+      "step": 2280
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.5978453159332275,
+      "learning_rate": 8.162716074940484e-06,
+      "loss": 0.0842,
+      "step": 2285
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.546857833862305,
+      "learning_rate": 8.15754062726426e-06,
+      "loss": 0.167,
+      "step": 2290
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.093022108078003,
+      "learning_rate": 8.152365179588035e-06,
+      "loss": 0.1355,
+      "step": 2295
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.2100273370742798,
+      "learning_rate": 8.14718973191181e-06,
+      "loss": 0.094,
+      "step": 2300
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.868842840194702,
+      "learning_rate": 8.142014284235588e-06,
+      "loss": 0.0752,
+      "step": 2305
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.411141395568848,
+      "learning_rate": 8.136838836559363e-06,
+      "loss": 0.1438,
+      "step": 2310
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.520047664642334,
+      "learning_rate": 8.131663388883139e-06,
+      "loss": 0.1071,
+      "step": 2315
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.8643572330474854,
+      "learning_rate": 8.126487941206916e-06,
+      "loss": 0.132,
+      "step": 2320
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 4.097289085388184,
+      "learning_rate": 8.121312493530692e-06,
+      "loss": 0.1181,
+      "step": 2325
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.704371213912964,
+      "learning_rate": 8.116137045854467e-06,
+      "loss": 0.0964,
+      "step": 2330
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.858574390411377,
+      "learning_rate": 8.110961598178243e-06,
+      "loss": 0.1035,
+      "step": 2335
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.610112190246582,
+      "learning_rate": 8.105786150502018e-06,
+      "loss": 0.1646,
+      "step": 2340
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0427603721618652,
+      "learning_rate": 8.100610702825795e-06,
+      "loss": 0.077,
+      "step": 2345
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.9987120628356934,
+      "learning_rate": 8.095435255149571e-06,
+      "loss": 0.1192,
+      "step": 2350
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.7207858562469482,
+      "learning_rate": 8.090259807473348e-06,
+      "loss": 0.0961,
+      "step": 2355
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.2569265365600586,
+      "learning_rate": 8.085084359797124e-06,
+      "loss": 0.1267,
+      "step": 2360
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 9.807589530944824,
+      "learning_rate": 8.0799089121209e-06,
+      "loss": 0.1377,
+      "step": 2365
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7769962549209595,
+      "learning_rate": 8.074733464444675e-06,
+      "loss": 0.1454,
+      "step": 2370
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.4914937019348145,
+      "learning_rate": 8.06955801676845e-06,
+      "loss": 0.1009,
+      "step": 2375
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1400229930877686,
+      "learning_rate": 8.064382569092227e-06,
+      "loss": 0.131,
+      "step": 2380
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.7774064540863037,
+      "learning_rate": 8.059207121416003e-06,
+      "loss": 0.108,
+      "step": 2385
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.377291202545166,
+      "learning_rate": 8.05403167373978e-06,
+      "loss": 0.102,
+      "step": 2390
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.7605440616607666,
+      "learning_rate": 8.048856226063556e-06,
+      "loss": 0.0857,
+      "step": 2395
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.146012783050537,
+      "learning_rate": 8.043680778387331e-06,
+      "loss": 0.1408,
+      "step": 2400
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.795672655105591,
+      "learning_rate": 8.038505330711107e-06,
+      "loss": 0.1385,
+      "step": 2405
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.02421760559082,
+      "learning_rate": 8.033329883034882e-06,
+      "loss": 0.1327,
+      "step": 2410
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7347333431243896,
+      "learning_rate": 8.02815443535866e-06,
+      "loss": 0.0791,
+      "step": 2415
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.233368873596191,
+      "learning_rate": 8.022978987682435e-06,
+      "loss": 0.1072,
+      "step": 2420
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6373865604400635,
+      "learning_rate": 8.01780354000621e-06,
+      "loss": 0.1166,
+      "step": 2425
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.997544527053833,
+      "learning_rate": 8.012628092329986e-06,
+      "loss": 0.1146,
+      "step": 2430
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4001853466033936,
+      "learning_rate": 8.007452644653763e-06,
+      "loss": 0.1853,
+      "step": 2435
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.30000376701355,
+      "learning_rate": 8.002277196977539e-06,
+      "loss": 0.134,
+      "step": 2440
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.6880128383636475,
+      "learning_rate": 7.997101749301315e-06,
+      "loss": 0.1246,
+      "step": 2445
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.9438652992248535,
+      "learning_rate": 7.991926301625092e-06,
+      "loss": 0.1245,
+      "step": 2450
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.15567684173584,
+      "learning_rate": 7.986750853948867e-06,
+      "loss": 0.1499,
+      "step": 2455
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 12.597527503967285,
+      "learning_rate": 7.981575406272643e-06,
+      "loss": 0.1426,
+      "step": 2460
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.984183311462402,
+      "learning_rate": 7.976399958596418e-06,
+      "loss": 0.1266,
+      "step": 2465
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3039536476135254,
+      "learning_rate": 7.971224510920196e-06,
+      "loss": 0.1156,
+      "step": 2470
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 8.08147144317627,
+      "learning_rate": 7.966049063243971e-06,
+      "loss": 0.1915,
+      "step": 2475
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 3.8845458030700684,
+      "learning_rate": 7.960873615567747e-06,
+      "loss": 0.1158,
+      "step": 2480
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.461251735687256,
+      "learning_rate": 7.955698167891524e-06,
+      "loss": 0.1054,
+      "step": 2485
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.557134628295898,
+      "learning_rate": 7.9505227202153e-06,
+      "loss": 0.1321,
+      "step": 2490
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.2354581356048584,
+      "learning_rate": 7.945347272539075e-06,
+      "loss": 0.1369,
+      "step": 2495
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.820423126220703,
+      "learning_rate": 7.94017182486285e-06,
+      "loss": 0.158,
+      "step": 2500
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.0720341205596924,
+      "learning_rate": 7.934996377186626e-06,
+      "loss": 0.1216,
+      "step": 2505
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.0163462162017822,
+      "learning_rate": 7.929820929510403e-06,
+      "loss": 0.0875,
+      "step": 2510
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.2006118297576904,
+      "learning_rate": 7.924645481834179e-06,
+      "loss": 0.0938,
+      "step": 2515
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1016287803649902,
+      "learning_rate": 7.919470034157956e-06,
+      "loss": 0.1077,
+      "step": 2520
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 3.5926766395568848,
+      "learning_rate": 7.914294586481732e-06,
+      "loss": 0.1345,
+      "step": 2525
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.409632205963135,
+      "learning_rate": 7.909119138805507e-06,
+      "loss": 0.1172,
+      "step": 2530
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9476617574691772,
+      "learning_rate": 7.903943691129283e-06,
+      "loss": 0.1744,
+      "step": 2535
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.9424796104431152,
+      "learning_rate": 7.898768243453058e-06,
+      "loss": 0.1969,
+      "step": 2540
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.397733211517334,
+      "learning_rate": 7.893592795776835e-06,
+      "loss": 0.1079,
+      "step": 2545
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.9149699211120605,
+      "learning_rate": 7.888417348100611e-06,
+      "loss": 0.1142,
+      "step": 2550
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.0017948150634766,
+      "learning_rate": 7.883241900424388e-06,
+      "loss": 0.0996,
+      "step": 2555
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.574118137359619,
+      "learning_rate": 7.878066452748164e-06,
+      "loss": 0.0866,
+      "step": 2560
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2571935653686523,
+      "learning_rate": 7.87289100507194e-06,
+      "loss": 0.0854,
+      "step": 2565
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.8886759281158447,
+      "learning_rate": 7.867715557395715e-06,
+      "loss": 0.0905,
+      "step": 2570
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.339482307434082,
+      "learning_rate": 7.86254010971949e-06,
+      "loss": 0.1458,
+      "step": 2575
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0696420669555664,
+      "learning_rate": 7.857364662043268e-06,
+      "loss": 0.148,
+      "step": 2580
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.3220736980438232,
+      "learning_rate": 7.852189214367043e-06,
+      "loss": 0.1233,
+      "step": 2585
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.099155426025391,
+      "learning_rate": 7.84701376669082e-06,
+      "loss": 0.0769,
+      "step": 2590
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7520205974578857,
+      "learning_rate": 7.841838319014596e-06,
+      "loss": 0.1055,
+      "step": 2595
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7609541416168213,
+      "learning_rate": 7.836662871338371e-06,
+      "loss": 0.114,
+      "step": 2600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.6962294578552246,
+      "learning_rate": 7.831487423662147e-06,
+      "loss": 0.1453,
+      "step": 2605
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.129779100418091,
+      "learning_rate": 7.826311975985922e-06,
+      "loss": 0.1092,
+      "step": 2610
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.9005229473114014,
+      "learning_rate": 7.8211365283097e-06,
+      "loss": 0.0843,
+      "step": 2615
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.7542483806610107,
+      "learning_rate": 7.815961080633475e-06,
+      "loss": 0.1022,
+      "step": 2620
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1400413513183594,
+      "learning_rate": 7.810785632957252e-06,
+      "loss": 0.0761,
+      "step": 2625
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4861457347869873,
+      "learning_rate": 7.805610185281028e-06,
+      "loss": 0.1202,
+      "step": 2630
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.827219009399414,
+      "learning_rate": 7.800434737604804e-06,
+      "loss": 0.0923,
+      "step": 2635
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.9459662437438965,
+      "learning_rate": 7.795259289928579e-06,
+      "loss": 0.1095,
+      "step": 2640
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.4152607917785645,
+      "learning_rate": 7.790083842252355e-06,
+      "loss": 0.0808,
+      "step": 2645
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1919751167297363,
+      "learning_rate": 7.784908394576132e-06,
+      "loss": 0.119,
+      "step": 2650
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 3.1704955101013184,
+      "learning_rate": 7.779732946899907e-06,
+      "loss": 0.0798,
+      "step": 2655
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2688515186309814,
+      "learning_rate": 7.774557499223685e-06,
+      "loss": 0.1138,
+      "step": 2660
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.7929760217666626,
+      "learning_rate": 7.76938205154746e-06,
+      "loss": 0.1114,
+      "step": 2665
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.473723411560059,
+      "learning_rate": 7.764206603871236e-06,
+      "loss": 0.1383,
+      "step": 2670
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.532488822937012,
+      "learning_rate": 7.759031156195011e-06,
+      "loss": 0.1072,
+      "step": 2675
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5835742950439453,
+      "learning_rate": 7.753855708518787e-06,
+      "loss": 0.0832,
+      "step": 2680
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.665186405181885,
+      "learning_rate": 7.748680260842564e-06,
+      "loss": 0.0949,
+      "step": 2685
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.426008462905884,
+      "learning_rate": 7.74350481316634e-06,
+      "loss": 0.072,
+      "step": 2690
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.6991891860961914,
+      "learning_rate": 7.738329365490117e-06,
+      "loss": 0.0696,
+      "step": 2695
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 11.929546356201172,
+      "learning_rate": 7.733153917813892e-06,
+      "loss": 0.1081,
+      "step": 2700
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.838849067687988,
+      "learning_rate": 7.727978470137668e-06,
+      "loss": 0.0934,
+      "step": 2705
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.855243682861328,
+      "learning_rate": 7.722803022461443e-06,
+      "loss": 0.1204,
+      "step": 2710
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.333618640899658,
+      "learning_rate": 7.717627574785219e-06,
+      "loss": 0.1097,
+      "step": 2715
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.9288158416748047,
+      "learning_rate": 7.712452127108996e-06,
+      "loss": 0.1217,
+      "step": 2720
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.649799823760986,
+      "learning_rate": 7.707276679432772e-06,
+      "loss": 0.1042,
+      "step": 2725
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.665895462036133,
+      "learning_rate": 7.702101231756547e-06,
+      "loss": 0.1248,
+      "step": 2730
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 4.786805152893066,
+      "learning_rate": 7.696925784080323e-06,
+      "loss": 0.1171,
+      "step": 2735
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.2108235359191895,
+      "learning_rate": 7.691750336404098e-06,
+      "loss": 0.0964,
+      "step": 2740
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.239229440689087,
+      "learning_rate": 7.686574888727876e-06,
+      "loss": 0.0924,
+      "step": 2745
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3956878185272217,
+      "learning_rate": 7.681399441051651e-06,
+      "loss": 0.146,
+      "step": 2750
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.120448589324951,
+      "learning_rate": 7.676223993375428e-06,
+      "loss": 0.1,
+      "step": 2755
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.652233839035034,
+      "learning_rate": 7.671048545699204e-06,
+      "loss": 0.0941,
+      "step": 2760
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.8857108354568481,
+      "learning_rate": 7.66587309802298e-06,
+      "loss": 0.113,
+      "step": 2765
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 9.084541320800781,
+      "learning_rate": 7.660697650346755e-06,
+      "loss": 0.1303,
+      "step": 2770
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.8233280181884766,
+      "learning_rate": 7.65552220267053e-06,
+      "loss": 0.1303,
+      "step": 2775
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.396145343780518,
+      "learning_rate": 7.650346754994308e-06,
+      "loss": 0.1187,
+      "step": 2780
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.240541934967041,
+      "learning_rate": 7.645171307318083e-06,
+      "loss": 0.0838,
+      "step": 2785
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.009382247924805,
+      "learning_rate": 7.63999585964186e-06,
+      "loss": 0.0773,
+      "step": 2790
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.281034469604492,
+      "learning_rate": 7.634820411965636e-06,
+      "loss": 0.1121,
+      "step": 2795
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.182842969894409,
+      "learning_rate": 7.629644964289411e-06,
+      "loss": 0.0628,
+      "step": 2800
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.3995907306671143,
+      "learning_rate": 7.624469516613188e-06,
+      "loss": 0.1446,
+      "step": 2805
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.0484015941619873,
+      "learning_rate": 7.619294068936963e-06,
+      "loss": 0.0951,
+      "step": 2810
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.213953971862793,
+      "learning_rate": 7.61411862126074e-06,
+      "loss": 0.116,
+      "step": 2815
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.406646251678467,
+      "learning_rate": 7.608943173584515e-06,
+      "loss": 0.1097,
+      "step": 2820
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.439655065536499,
+      "learning_rate": 7.603767725908292e-06,
+      "loss": 0.1631,
+      "step": 2825
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1134753227233887,
+      "learning_rate": 7.598592278232067e-06,
+      "loss": 0.1852,
+      "step": 2830
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3332927227020264,
+      "learning_rate": 7.5934168305558445e-06,
+      "loss": 0.1025,
+      "step": 2835
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 10.899185180664062,
+      "learning_rate": 7.58824138287962e-06,
+      "loss": 0.1129,
+      "step": 2840
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.261824607849121,
+      "learning_rate": 7.5830659352033956e-06,
+      "loss": 0.1177,
+      "step": 2845
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9256378412246704,
+      "learning_rate": 7.577890487527172e-06,
+      "loss": 0.0931,
+      "step": 2850
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.632373094558716,
+      "learning_rate": 7.5727150398509475e-06,
+      "loss": 0.0753,
+      "step": 2855
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.5675268173217773,
+      "learning_rate": 7.567539592174724e-06,
+      "loss": 0.084,
+      "step": 2860
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1067235469818115,
+      "learning_rate": 7.562364144498499e-06,
+      "loss": 0.083,
+      "step": 2865
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.160629987716675,
+      "learning_rate": 7.557188696822277e-06,
+      "loss": 0.0958,
+      "step": 2870
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.733227252960205,
+      "learning_rate": 7.552013249146052e-06,
+      "loss": 0.145,
+      "step": 2875
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.5898009538650513,
+      "learning_rate": 7.546837801469828e-06,
+      "loss": 0.1091,
+      "step": 2880
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 3.637584924697876,
+      "learning_rate": 7.541662353793604e-06,
+      "loss": 0.1023,
+      "step": 2885
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.8995022773742676,
+      "learning_rate": 7.53648690611738e-06,
+      "loss": 0.096,
+      "step": 2890
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1722774505615234,
+      "learning_rate": 7.531311458441156e-06,
+      "loss": 0.1172,
+      "step": 2895
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.399565696716309,
+      "learning_rate": 7.5261360107649315e-06,
+      "loss": 0.1025,
+      "step": 2900
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.6933958530426025,
+      "learning_rate": 7.520960563088707e-06,
+      "loss": 0.0847,
+      "step": 2905
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 7.4772210121154785,
+      "learning_rate": 7.515785115412484e-06,
+      "loss": 0.1071,
+      "step": 2910
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.8072834014892578,
+      "learning_rate": 7.51060966773626e-06,
+      "loss": 0.0759,
+      "step": 2915
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.314575672149658,
+      "learning_rate": 7.505434220060036e-06,
+      "loss": 0.0907,
+      "step": 2920
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.566117763519287,
+      "learning_rate": 7.500258772383812e-06,
+      "loss": 0.0719,
+      "step": 2925
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.5256733894348145,
+      "learning_rate": 7.495083324707588e-06,
+      "loss": 0.1027,
+      "step": 2930
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.8933703899383545,
+      "learning_rate": 7.489907877031364e-06,
+      "loss": 0.1151,
+      "step": 2935
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5803911685943604,
+      "learning_rate": 7.484732429355139e-06,
+      "loss": 0.0994,
+      "step": 2940
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 19.182605743408203,
+      "learning_rate": 7.479556981678916e-06,
+      "loss": 0.2395,
+      "step": 2945
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 14.523638725280762,
+      "learning_rate": 7.474381534002691e-06,
+      "loss": 0.0987,
+      "step": 2950
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5896825790405273,
+      "learning_rate": 7.469206086326468e-06,
+      "loss": 0.1155,
+      "step": 2955
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.121129989624023,
+      "learning_rate": 7.464030638650244e-06,
+      "loss": 0.0985,
+      "step": 2960
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 3.078057050704956,
+      "learning_rate": 7.45885519097402e-06,
+      "loss": 0.1142,
+      "step": 2965
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 13.759486198425293,
+      "learning_rate": 7.453679743297796e-06,
+      "loss": 0.1838,
+      "step": 2970
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.7202324867248535,
+      "learning_rate": 7.448504295621571e-06,
+      "loss": 0.1103,
+      "step": 2975
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5780439376831055,
+      "learning_rate": 7.443328847945348e-06,
+      "loss": 0.1002,
+      "step": 2980
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.044093132019043,
+      "learning_rate": 7.438153400269123e-06,
+      "loss": 0.097,
+      "step": 2985
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2831811904907227,
+      "learning_rate": 7.4329779525929005e-06,
+      "loss": 0.1004,
+      "step": 2990
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.22511625289917,
+      "learning_rate": 7.427802504916676e-06,
+      "loss": 0.1106,
+      "step": 2995
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5959653854370117,
+      "learning_rate": 7.422627057240452e-06,
+      "loss": 0.0746,
+      "step": 3000
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.123388767242432,
+      "learning_rate": 7.417451609564228e-06,
+      "loss": 0.1648,
+      "step": 3005
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.040205955505371,
+      "learning_rate": 7.4122761618880035e-06,
+      "loss": 0.1403,
+      "step": 3010
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9196488857269287,
+      "learning_rate": 7.40710071421178e-06,
+      "loss": 0.0945,
+      "step": 3015
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.9460086822509766,
+      "learning_rate": 7.401925266535555e-06,
+      "loss": 0.1193,
+      "step": 3020
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.386340856552124,
+      "learning_rate": 7.396749818859333e-06,
+      "loss": 0.0815,
+      "step": 3025
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 3.021364212036133,
+      "learning_rate": 7.391574371183108e-06,
+      "loss": 0.0989,
+      "step": 3030
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.4375131130218506,
+      "learning_rate": 7.3863989235068846e-06,
+      "loss": 0.083,
+      "step": 3035
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.7974884510040283,
+      "learning_rate": 7.38122347583066e-06,
+      "loss": 0.0969,
+      "step": 3040
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.3860740661621094,
+      "learning_rate": 7.376048028154436e-06,
+      "loss": 0.1077,
+      "step": 3045
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 8.257057189941406,
+      "learning_rate": 7.370872580478212e-06,
+      "loss": 0.2161,
+      "step": 3050
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 3.169840097427368,
+      "learning_rate": 7.3656971328019875e-06,
+      "loss": 0.0808,
+      "step": 3055
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.8728913068771362,
+      "learning_rate": 7.360521685125765e-06,
+      "loss": 0.102,
+      "step": 3060
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.9128949642181396,
+      "learning_rate": 7.35534623744954e-06,
+      "loss": 0.0893,
+      "step": 3065
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.6967766284942627,
+      "learning_rate": 7.350170789773316e-06,
+      "loss": 0.1134,
+      "step": 3070
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.915827751159668,
+      "learning_rate": 7.344995342097092e-06,
+      "loss": 0.1111,
+      "step": 3075
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 3.4506168365478516,
+      "learning_rate": 7.339819894420868e-06,
+      "loss": 0.1336,
+      "step": 3080
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.747591972351074,
+      "learning_rate": 7.334644446744644e-06,
+      "loss": 0.1,
+      "step": 3085
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.551347494125366,
+      "learning_rate": 7.32946899906842e-06,
+      "loss": 0.1001,
+      "step": 3090
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 11.74671745300293,
+      "learning_rate": 7.324293551392196e-06,
+      "loss": 0.08,
+      "step": 3095
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2714433670043945,
+      "learning_rate": 7.319118103715972e-06,
+      "loss": 0.0918,
+      "step": 3100
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.08971643447876,
+      "learning_rate": 7.313942656039747e-06,
+      "loss": 0.1043,
+      "step": 3105
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.7523419857025146,
+      "learning_rate": 7.308767208363524e-06,
+      "loss": 0.0934,
+      "step": 3110
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.065701961517334,
+      "learning_rate": 7.3035917606873e-06,
+      "loss": 0.1544,
+      "step": 3115
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.6904995441436768,
+      "learning_rate": 7.298416313011076e-06,
+      "loss": 0.12,
+      "step": 3120
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4377174377441406,
+      "learning_rate": 7.293240865334852e-06,
+      "loss": 0.0788,
+      "step": 3125
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.7418872117996216,
+      "learning_rate": 7.288065417658628e-06,
+      "loss": 0.0745,
+      "step": 3130
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9398059844970703,
+      "learning_rate": 7.282889969982404e-06,
+      "loss": 0.0994,
+      "step": 3135
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 8.430389404296875,
+      "learning_rate": 7.277714522306179e-06,
+      "loss": 0.1341,
+      "step": 3140
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4294533729553223,
+      "learning_rate": 7.2725390746299565e-06,
+      "loss": 0.0694,
+      "step": 3145
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 4.203413486480713,
+      "learning_rate": 7.267363626953732e-06,
+      "loss": 0.1104,
+      "step": 3150
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.8622099161148071,
+      "learning_rate": 7.2621881792775084e-06,
+      "loss": 0.1242,
+      "step": 3155
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 3.0647735595703125,
+      "learning_rate": 7.257012731601284e-06,
+      "loss": 0.1037,
+      "step": 3160
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.9662420749664307,
+      "learning_rate": 7.25183728392506e-06,
+      "loss": 0.0898,
+      "step": 3165
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.06348991394043,
+      "learning_rate": 7.246661836248836e-06,
+      "loss": 0.0696,
+      "step": 3170
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.7767813205718994,
+      "learning_rate": 7.2414863885726114e-06,
+      "loss": 0.1091,
+      "step": 3175
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 3.425811290740967,
+      "learning_rate": 7.236310940896389e-06,
+      "loss": 0.1329,
+      "step": 3180
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.831468105316162,
+      "learning_rate": 7.231135493220164e-06,
+      "loss": 0.1315,
+      "step": 3185
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2901949882507324,
+      "learning_rate": 7.2259600455439406e-06,
+      "loss": 0.0928,
+      "step": 3190
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.404942512512207,
+      "learning_rate": 7.220784597867716e-06,
+      "loss": 0.0882,
+      "step": 3195
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.917351007461548,
+      "learning_rate": 7.2156091501914925e-06,
+      "loss": 0.1244,
+      "step": 3200
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6030354499816895,
+      "learning_rate": 7.210433702515268e-06,
+      "loss": 0.0944,
+      "step": 3205
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2650647163391113,
+      "learning_rate": 7.2052582548390436e-06,
+      "loss": 0.1086,
+      "step": 3210
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 4.482391357421875,
+      "learning_rate": 7.200082807162821e-06,
+      "loss": 0.1283,
+      "step": 3215
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.1205341815948486,
+      "learning_rate": 7.194907359486596e-06,
+      "loss": 0.0955,
+      "step": 3220
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0902633666992188,
+      "learning_rate": 7.189731911810373e-06,
+      "loss": 0.0813,
+      "step": 3225
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.2730817794799805,
+      "learning_rate": 7.184556464134148e-06,
+      "loss": 0.1051,
+      "step": 3230
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.0574705600738525,
+      "learning_rate": 7.179381016457925e-06,
+      "loss": 0.1143,
+      "step": 3235
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 6.457324981689453,
+      "learning_rate": 7.1742055687817e-06,
+      "loss": 0.1419,
+      "step": 3240
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.399207353591919,
+      "learning_rate": 7.169030121105476e-06,
+      "loss": 0.0989,
+      "step": 3245
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 7.26680850982666,
+      "learning_rate": 7.163854673429252e-06,
+      "loss": 0.1087,
+      "step": 3250
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.541226863861084,
+      "learning_rate": 7.158679225753028e-06,
+      "loss": 0.121,
+      "step": 3255
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.7401907444000244,
+      "learning_rate": 7.153503778076805e-06,
+      "loss": 0.0746,
+      "step": 3260
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.075695991516113,
+      "learning_rate": 7.14832833040058e-06,
+      "loss": 0.1053,
+      "step": 3265
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.3619155883789062,
+      "learning_rate": 7.143152882724356e-06,
+      "loss": 0.0712,
+      "step": 3270
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 8.59334659576416,
+      "learning_rate": 7.137977435048132e-06,
+      "loss": 0.1189,
+      "step": 3275
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.417818069458008,
+      "learning_rate": 7.132801987371908e-06,
+      "loss": 0.1444,
+      "step": 3280
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 3.149252414703369,
+      "learning_rate": 7.127626539695684e-06,
+      "loss": 0.0966,
+      "step": 3285
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.921302318572998,
+      "learning_rate": 7.12245109201946e-06,
+      "loss": 0.0618,
+      "step": 3290
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4437150955200195,
+      "learning_rate": 7.117275644343237e-06,
+      "loss": 0.0901,
+      "step": 3295
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.587127208709717,
+      "learning_rate": 7.1121001966670125e-06,
+      "loss": 0.0792,
+      "step": 3300
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.467782020568848,
+      "learning_rate": 7.106924748990788e-06,
+      "loss": 0.0905,
+      "step": 3305
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.5796444416046143,
+      "learning_rate": 7.1017493013145645e-06,
+      "loss": 0.0863,
+      "step": 3310
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 4.151190757751465,
+      "learning_rate": 7.09657385363834e-06,
+      "loss": 0.1019,
+      "step": 3315
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.894544005393982,
+      "learning_rate": 7.091398405962116e-06,
+      "loss": 0.0898,
+      "step": 3320
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 11.871978759765625,
+      "learning_rate": 7.086222958285892e-06,
+      "loss": 0.1137,
+      "step": 3325
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.6164116859436035,
+      "learning_rate": 7.081047510609669e-06,
+      "loss": 0.0734,
+      "step": 3330
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.0714378356933594,
+      "learning_rate": 7.075872062933445e-06,
+      "loss": 0.0889,
+      "step": 3335
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.4325357675552368,
+      "learning_rate": 7.07069661525722e-06,
+      "loss": 0.0761,
+      "step": 3340
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.9930839538574219,
+      "learning_rate": 7.065521167580997e-06,
+      "loss": 0.0723,
+      "step": 3345
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.183719635009766,
+      "learning_rate": 7.060345719904772e-06,
+      "loss": 0.1258,
+      "step": 3350
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.605844259262085,
+      "learning_rate": 7.0551702722285485e-06,
+      "loss": 0.1743,
+      "step": 3355
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.9918246269226074,
+      "learning_rate": 7.049994824552324e-06,
+      "loss": 0.1155,
+      "step": 3360
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.7702308893203735,
+      "learning_rate": 7.0448193768761004e-06,
+      "loss": 0.1499,
+      "step": 3365
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.84514582157135,
+      "learning_rate": 7.039643929199876e-06,
+      "loss": 0.1302,
+      "step": 3370
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.0677855014801025,
+      "learning_rate": 7.0344684815236515e-06,
+      "loss": 0.1094,
+      "step": 3375
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.9211227893829346,
+      "learning_rate": 7.029293033847429e-06,
+      "loss": 0.0597,
+      "step": 3380
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.7878475189208984,
+      "learning_rate": 7.024117586171204e-06,
+      "loss": 0.0735,
+      "step": 3385
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.2111921310424805,
+      "learning_rate": 7.018942138494981e-06,
+      "loss": 0.0746,
+      "step": 3390
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 10170,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}