diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7043 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.3581420616596496,
+  "eval_steps": 500,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0013581420616596495,
+      "grad_norm": 15.13900089263916,
+      "learning_rate": 1.8e-06,
+      "loss": 1.4346,
+      "step": 10
+    },
+    {
+      "epoch": 0.002716284123319299,
+      "grad_norm": 11.375536918640137,
+      "learning_rate": 3.8e-06,
+      "loss": 1.1447,
+      "step": 20
+    },
+    {
+      "epoch": 0.004074426184978949,
+      "grad_norm": 5.97958517074585,
+      "learning_rate": 5.8e-06,
+      "loss": 0.5479,
+      "step": 30
+    },
+    {
+      "epoch": 0.005432568246638598,
+      "grad_norm": 5.709956645965576,
+      "learning_rate": 7.8e-06,
+      "loss": 0.2918,
+      "step": 40
+    },
+    {
+      "epoch": 0.006790710308298248,
+      "grad_norm": 4.446307182312012,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.2899,
+      "step": 50
+    },
+    {
+      "epoch": 0.008148852369957897,
+      "grad_norm": 1.92933988571167,
+      "learning_rate": 1.18e-05,
+      "loss": 0.1847,
+      "step": 60
+    },
+    {
+      "epoch": 0.009506994431617547,
+      "grad_norm": 3.599168539047241,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 0.215,
+      "step": 70
+    },
+    {
+      "epoch": 0.010865136493277196,
+      "grad_norm": 3.055508852005005,
+      "learning_rate": 1.58e-05,
+      "loss": 0.1934,
+      "step": 80
+    },
+    {
+      "epoch": 0.012223278554936847,
+      "grad_norm": 2.1978554725646973,
+      "learning_rate": 1.78e-05,
+      "loss": 0.1727,
+      "step": 90
+    },
+    {
+      "epoch": 0.013581420616596496,
+      "grad_norm": 2.0673773288726807,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 0.1812,
+      "step": 100
+    },
+    {
+      "epoch": 0.014939562678256146,
+      "grad_norm": 3.7078819274902344,
+      "learning_rate": 2.18e-05,
+      "loss": 0.1684,
+      "step": 110
+    },
+    {
+      "epoch": 0.016297704739915795,
+      "grad_norm": 2.541360855102539,
+      "learning_rate": 2.38e-05,
+      "loss": 0.1403,
+      "step": 120
+    },
+    {
+      "epoch": 0.017655846801575444,
+      "grad_norm": 3.4069161415100098,
+      "learning_rate": 2.58e-05,
+      "loss": 0.1601,
+      "step": 130
+    },
+    {
+      "epoch": 0.019013988863235094,
+      "grad_norm": 2.1283915042877197,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 0.1692,
+      "step": 140
+    },
+    {
+      "epoch": 0.020372130924894743,
+      "grad_norm": 1.4989871978759766,
+      "learning_rate": 2.98e-05,
+      "loss": 0.1541,
+      "step": 150
+    },
+    {
+      "epoch": 0.021730272986554392,
+      "grad_norm": 2.1739139556884766,
+      "learning_rate": 3.18e-05,
+      "loss": 0.1557,
+      "step": 160
+    },
+    {
+      "epoch": 0.023088415048214045,
+      "grad_norm": 1.9494563341140747,
+      "learning_rate": 3.38e-05,
+      "loss": 0.1368,
+      "step": 170
+    },
+    {
+      "epoch": 0.024446557109873694,
+      "grad_norm": 2.39599871635437,
+      "learning_rate": 3.58e-05,
+      "loss": 0.1452,
+      "step": 180
+    },
+    {
+      "epoch": 0.025804699171533343,
+      "grad_norm": 1.9096465110778809,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 0.1536,
+      "step": 190
+    },
+    {
+      "epoch": 0.027162841233192993,
+      "grad_norm": 2.842930793762207,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 0.1449,
+      "step": 200
+    },
+    {
+      "epoch": 0.028520983294852642,
+      "grad_norm": 2.820974826812744,
+      "learning_rate": 4.18e-05,
+      "loss": 0.1476,
+      "step": 210
+    },
+    {
+      "epoch": 0.02987912535651229,
+      "grad_norm": 1.5039032697677612,
+      "learning_rate": 4.38e-05,
+      "loss": 0.1888,
+      "step": 220
+    },
+    {
+      "epoch": 0.03123726741817194,
+      "grad_norm": 2.7140212059020996,
+      "learning_rate": 4.58e-05,
+      "loss": 0.1158,
+      "step": 230
+    },
+    {
+      "epoch": 0.03259540947983159,
+      "grad_norm": 1.7555482387542725,
+      "learning_rate": 4.78e-05,
+      "loss": 0.1713,
+      "step": 240
+    },
+    {
+      "epoch": 0.03395355154149124,
+      "grad_norm": 2.056089162826538,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 0.1211,
+      "step": 250
+    },
+    {
+      "epoch": 0.03531169360315089,
+      "grad_norm": 1.8953531980514526,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 0.1341,
+      "step": 260
+    },
+    {
+      "epoch": 0.03666983566481054,
+      "grad_norm": 3.864428997039795,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 0.1769,
+      "step": 270
+    },
+    {
+      "epoch": 0.03802797772647019,
+      "grad_norm": 1.2459381818771362,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 0.1678,
+      "step": 280
+    },
+    {
+      "epoch": 0.039386119788129836,
+      "grad_norm": 4.32213830947876,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 0.2122,
+      "step": 290
+    },
+    {
+      "epoch": 0.040744261849789486,
+      "grad_norm": 3.199658155441284,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 0.1574,
+      "step": 300
+    },
+    {
+      "epoch": 0.042102403911449135,
+      "grad_norm": 1.878484845161438,
+      "learning_rate": 6.18e-05,
+      "loss": 0.1789,
+      "step": 310
+    },
+    {
+      "epoch": 0.043460545973108784,
+      "grad_norm": 2.402588129043579,
+      "learning_rate": 6.38e-05,
+      "loss": 0.13,
+      "step": 320
+    },
+    {
+      "epoch": 0.044818688034768434,
+      "grad_norm": 1.7715047597885132,
+      "learning_rate": 6.58e-05,
+      "loss": 0.2056,
+      "step": 330
+    },
+    {
+      "epoch": 0.04617683009642809,
+      "grad_norm": 1.1538476943969727,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 0.1295,
+      "step": 340
+    },
+    {
+      "epoch": 0.04753497215808774,
+      "grad_norm": 1.3045623302459717,
+      "learning_rate": 6.98e-05,
+      "loss": 0.1385,
+      "step": 350
+    },
+    {
+      "epoch": 0.04889311421974739,
+      "grad_norm": 2.068037748336792,
+      "learning_rate": 7.18e-05,
+      "loss": 0.1133,
+      "step": 360
+    },
+    {
+      "epoch": 0.05025125628140704,
+      "grad_norm": 3.5157949924468994,
+      "learning_rate": 7.38e-05,
+      "loss": 0.1261,
+      "step": 370
+    },
+    {
+      "epoch": 0.05160939834306669,
+      "grad_norm": 1.8979589939117432,
+      "learning_rate": 7.58e-05,
+      "loss": 0.1502,
+      "step": 380
+    },
+    {
+      "epoch": 0.052967540404726336,
+      "grad_norm": 1.4680991172790527,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 0.1195,
+      "step": 390
+    },
+    {
+      "epoch": 0.054325682466385986,
+      "grad_norm": 1.8850711584091187,
+      "learning_rate": 7.98e-05,
+      "loss": 0.1298,
+      "step": 400
+    },
+    {
+      "epoch": 0.055683824528045635,
+      "grad_norm": 2.58443546295166,
+      "learning_rate": 8.18e-05,
+      "loss": 0.1294,
+      "step": 410
+    },
+    {
+      "epoch": 0.057041966589705284,
+      "grad_norm": 1.6674855947494507,
+      "learning_rate": 8.38e-05,
+      "loss": 0.1165,
+      "step": 420
+    },
+    {
+      "epoch": 0.05840010865136493,
+      "grad_norm": 1.5503572225570679,
+      "learning_rate": 8.58e-05,
+      "loss": 0.1758,
+      "step": 430
+    },
+    {
+      "epoch": 0.05975825071302458,
+      "grad_norm": 1.730098009109497,
+      "learning_rate": 8.78e-05,
+      "loss": 0.149,
+      "step": 440
+    },
+    {
+      "epoch": 0.06111639277468423,
+      "grad_norm": 2.6782047748565674,
+      "learning_rate": 8.98e-05,
+      "loss": 0.1668,
+      "step": 450
+    },
+    {
+      "epoch": 0.06247453483634388,
+      "grad_norm": 2.2021360397338867,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 0.1799,
+      "step": 460
+    },
+    {
+      "epoch": 0.06383267689800354,
+      "grad_norm": 3.8699429035186768,
+      "learning_rate": 9.38e-05,
+      "loss": 0.1666,
+      "step": 470
+    },
+    {
+      "epoch": 0.06519081895966318,
+      "grad_norm": 1.3565938472747803,
+      "learning_rate": 9.58e-05,
+      "loss": 0.2351,
+      "step": 480
+    },
+    {
+      "epoch": 0.06654896102132284,
+      "grad_norm": 1.031346321105957,
+      "learning_rate": 9.78e-05,
+      "loss": 0.1252,
+      "step": 490
+    },
+    {
+      "epoch": 0.06790710308298248,
+      "grad_norm": 1.745088815689087,
+      "learning_rate": 9.98e-05,
+      "loss": 0.1253,
+      "step": 500
+    },
+    {
+      "epoch": 0.06926524514464213,
+      "grad_norm": 1.7901244163513184,
+      "learning_rate": 9.9999778549206e-05,
+      "loss": 0.1296,
+      "step": 510
+    },
+    {
+      "epoch": 0.07062338720630178,
+      "grad_norm": 1.6305402517318726,
+      "learning_rate": 9.999901304280685e-05,
+      "loss": 0.1399,
+      "step": 520
+    },
+    {
+      "epoch": 0.07198152926796143,
+      "grad_norm": 2.305971145629883,
+      "learning_rate": 9.999770075521164e-05,
+      "loss": 0.1526,
+      "step": 530
+    },
+    {
+      "epoch": 0.07333967132962108,
+      "grad_norm": 3.5417497158050537,
+      "learning_rate": 9.99958417007713e-05,
+      "loss": 0.1571,
+      "step": 540
+    },
+    {
+      "epoch": 0.07469781339128073,
+      "grad_norm": 1.979875922203064,
+      "learning_rate": 9.999343589981615e-05,
+      "loss": 0.1422,
+      "step": 550
+    },
+    {
+      "epoch": 0.07605595545294037,
+      "grad_norm": 1.6565183401107788,
+      "learning_rate": 9.999048337865568e-05,
+      "loss": 0.1498,
+      "step": 560
+    },
+    {
+      "epoch": 0.07741409751460003,
+      "grad_norm": 1.6965197324752808,
+      "learning_rate": 9.998698416957815e-05,
+      "loss": 0.1463,
+      "step": 570
+    },
+    {
+      "epoch": 0.07877223957625967,
+      "grad_norm": 3.1182215213775635,
+      "learning_rate": 9.998293831085037e-05,
+      "loss": 0.1953,
+      "step": 580
+    },
+    {
+      "epoch": 0.08013038163791933,
+      "grad_norm": 3.3180224895477295,
+      "learning_rate": 9.997834584671719e-05,
+      "loss": 0.2329,
+      "step": 590
+    },
+    {
+      "epoch": 0.08148852369957897,
+      "grad_norm": 2.5061111450195312,
+      "learning_rate": 9.997320682740107e-05,
+      "loss": 0.15,
+      "step": 600
+    },
+    {
+      "epoch": 0.08284666576123863,
+      "grad_norm": 2.1334118843078613,
+      "learning_rate": 9.996752130910149e-05,
+      "loss": 0.1986,
+      "step": 610
+    },
+    {
+      "epoch": 0.08420480782289827,
+      "grad_norm": 1.7377362251281738,
+      "learning_rate": 9.99612893539944e-05,
+      "loss": 0.1345,
+      "step": 620
+    },
+    {
+      "epoch": 0.08556294988455793,
+      "grad_norm": 0.8570336103439331,
+      "learning_rate": 9.995451103023144e-05,
+      "loss": 0.1111,
+      "step": 630
+    },
+    {
+      "epoch": 0.08692109194621757,
+      "grad_norm": 1.112568736076355,
+      "learning_rate": 9.994718641193928e-05,
+      "loss": 0.0883,
+      "step": 640
+    },
+    {
+      "epoch": 0.08827923400787722,
+      "grad_norm": 1.2197680473327637,
+      "learning_rate": 9.993931557921874e-05,
+      "loss": 0.1344,
+      "step": 650
+    },
+    {
+      "epoch": 0.08963737606953687,
+      "grad_norm": 2.2512173652648926,
+      "learning_rate": 9.993089861814402e-05,
+      "loss": 0.1276,
+      "step": 660
+    },
+    {
+      "epoch": 0.09099551813119652,
+      "grad_norm": 1.4388880729675293,
+      "learning_rate": 9.992193562076166e-05,
+      "loss": 0.1289,
+      "step": 670
+    },
+    {
+      "epoch": 0.09235366019285618,
+      "grad_norm": 0.9686877727508545,
+      "learning_rate": 9.991242668508954e-05,
+      "loss": 0.1139,
+      "step": 680
+    },
+    {
+      "epoch": 0.09371180225451582,
+      "grad_norm": 2.7000038623809814,
+      "learning_rate": 9.990237191511587e-05,
+      "loss": 0.1157,
+      "step": 690
+    },
+    {
+      "epoch": 0.09506994431617548,
+      "grad_norm": 1.907443881034851,
+      "learning_rate": 9.989177142079802e-05,
+      "loss": 0.1187,
+      "step": 700
+    },
+    {
+      "epoch": 0.09642808637783512,
+      "grad_norm": 1.6974741220474243,
+      "learning_rate": 9.988062531806126e-05,
+      "loss": 0.131,
+      "step": 710
+    },
+    {
+      "epoch": 0.09778622843949478,
+      "grad_norm": 1.4445502758026123,
+      "learning_rate": 9.986893372879762e-05,
+      "loss": 0.0971,
+      "step": 720
+    },
+    {
+      "epoch": 0.09914437050115442,
+      "grad_norm": 1.9095450639724731,
+      "learning_rate": 9.985669678086443e-05,
+      "loss": 0.1616,
+      "step": 730
+    },
+    {
+      "epoch": 0.10050251256281408,
+      "grad_norm": 2.607119083404541,
+      "learning_rate": 9.984391460808298e-05,
+      "loss": 0.1446,
+      "step": 740
+    },
+    {
+      "epoch": 0.10186065462447372,
+      "grad_norm": 1.9929903745651245,
+      "learning_rate": 9.983058735023709e-05,
+      "loss": 0.1139,
+      "step": 750
+    },
+    {
+      "epoch": 0.10321879668613337,
+      "grad_norm": 1.4107511043548584,
+      "learning_rate": 9.98167151530715e-05,
+      "loss": 0.123,
+      "step": 760
+    },
+    {
+      "epoch": 0.10457693874779302,
+      "grad_norm": 3.6841492652893066,
+      "learning_rate": 9.980229816829034e-05,
+      "loss": 0.1392,
+      "step": 770
+    },
+    {
+      "epoch": 0.10593508080945267,
+      "grad_norm": 1.4337660074234009,
+      "learning_rate": 9.978733655355544e-05,
+      "loss": 0.177,
+      "step": 780
+    },
+    {
+      "epoch": 0.10729322287111231,
+      "grad_norm": 1.2946299314498901,
+      "learning_rate": 9.977183047248464e-05,
+      "loss": 0.1099,
+      "step": 790
+    },
+    {
+      "epoch": 0.10865136493277197,
+      "grad_norm": 1.617319107055664,
+      "learning_rate": 9.975578009464992e-05,
+      "loss": 0.1411,
+      "step": 800
+    },
+    {
+      "epoch": 0.11000950699443161,
+      "grad_norm": 1.680314302444458,
+      "learning_rate": 9.97391855955757e-05,
+      "loss": 0.1317,
+      "step": 810
+    },
+    {
+      "epoch": 0.11136764905609127,
+      "grad_norm": 1.4202923774719238,
+      "learning_rate": 9.972204715673669e-05,
+      "loss": 0.129,
+      "step": 820
+    },
+    {
+      "epoch": 0.11272579111775091,
+      "grad_norm": 1.5702614784240723,
+      "learning_rate": 9.970436496555617e-05,
+      "loss": 0.0859,
+      "step": 830
+    },
+    {
+      "epoch": 0.11408393317941057,
+      "grad_norm": 3.0650391578674316,
+      "learning_rate": 9.968613921540373e-05,
+      "loss": 0.113,
+      "step": 840
+    },
+    {
+      "epoch": 0.11544207524107021,
+      "grad_norm": 1.6806137561798096,
+      "learning_rate": 9.966737010559326e-05,
+      "loss": 0.1472,
+      "step": 850
+    },
+    {
+      "epoch": 0.11680021730272987,
+      "grad_norm": 1.7884191274642944,
+      "learning_rate": 9.964805784138072e-05,
+      "loss": 0.1225,
+      "step": 860
+    },
+    {
+      "epoch": 0.11815835936438951,
+      "grad_norm": 1.4735928773880005,
+      "learning_rate": 9.962820263396195e-05,
+      "loss": 0.1167,
+      "step": 870
+    },
+    {
+      "epoch": 0.11951650142604917,
+      "grad_norm": 1.0542923212051392,
+      "learning_rate": 9.960780470047033e-05,
+      "loss": 0.0953,
+      "step": 880
+    },
+    {
+      "epoch": 0.12087464348770881,
+      "grad_norm": 1.518430233001709,
+      "learning_rate": 9.958686426397437e-05,
+      "loss": 0.1114,
+      "step": 890
+    },
+    {
+      "epoch": 0.12223278554936846,
+      "grad_norm": 2.3001043796539307,
+      "learning_rate": 9.956538155347534e-05,
+      "loss": 0.12,
+      "step": 900
+    },
+    {
+      "epoch": 0.12359092761102812,
+      "grad_norm": 1.9716862440109253,
+      "learning_rate": 9.95433568039047e-05,
+      "loss": 0.0985,
+      "step": 910
+    },
+    {
+      "epoch": 0.12494906967268776,
+      "grad_norm": 2.6265206336975098,
+      "learning_rate": 9.952079025612162e-05,
+      "loss": 0.1405,
+      "step": 920
+    },
+    {
+      "epoch": 0.12630721173434742,
+      "grad_norm": 1.7564139366149902,
+      "learning_rate": 9.949768215691022e-05,
+      "loss": 0.102,
+      "step": 930
+    },
+    {
+      "epoch": 0.12766535379600708,
+      "grad_norm": 1.378955602645874,
+      "learning_rate": 9.9474032758977e-05,
+      "loss": 0.108,
+      "step": 940
+    },
+    {
+      "epoch": 0.1290234958576667,
+      "grad_norm": 1.4960349798202515,
+      "learning_rate": 9.944984232094794e-05,
+      "loss": 0.1314,
+      "step": 950
+    },
+    {
+      "epoch": 0.13038163791932636,
+      "grad_norm": 1.8896232843399048,
+      "learning_rate": 9.942511110736584e-05,
+      "loss": 0.1166,
+      "step": 960
+    },
+    {
+      "epoch": 0.13173977998098602,
+      "grad_norm": 0.845024824142456,
+      "learning_rate": 9.939983938868726e-05,
+      "loss": 0.1135,
+      "step": 970
+    },
+    {
+      "epoch": 0.13309792204264567,
+      "grad_norm": 1.8015919923782349,
+      "learning_rate": 9.93740274412797e-05,
+      "loss": 0.1059,
+      "step": 980
+    },
+    {
+      "epoch": 0.1344560641043053,
+      "grad_norm": 2.234103202819824,
+      "learning_rate": 9.934767554741846e-05,
+      "loss": 0.1144,
+      "step": 990
+    },
+    {
+      "epoch": 0.13581420616596496,
+      "grad_norm": 1.4727959632873535,
+      "learning_rate": 9.932078399528361e-05,
+      "loss": 0.12,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1371723482276246,
+      "grad_norm": 2.166718006134033,
+      "learning_rate": 9.929335307895689e-05,
+      "loss": 0.1076,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13853049028928427,
+      "grad_norm": 1.1835960149765015,
+      "learning_rate": 9.926538309841839e-05,
+      "loss": 0.1115,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1398886323509439,
+      "grad_norm": 1.4695322513580322,
+      "learning_rate": 9.923687435954334e-05,
+      "loss": 0.1066,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14124677441260355,
+      "grad_norm": 1.4083786010742188,
+      "learning_rate": 9.920782717409873e-05,
+      "loss": 0.0943,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1426049164742632,
+      "grad_norm": 1.679107904434204,
+      "learning_rate": 9.917824185973994e-05,
+      "loss": 0.1305,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14396305853592287,
+      "grad_norm": 1.5337637662887573,
+      "learning_rate": 9.914811874000723e-05,
+      "loss": 0.1169,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1453212005975825,
+      "grad_norm": 1.1444432735443115,
+      "learning_rate": 9.911745814432218e-05,
+      "loss": 0.0942,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14667934265924215,
+      "grad_norm": 1.2553508281707764,
+      "learning_rate": 9.90862604079842e-05,
+      "loss": 0.0696,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1480374847209018,
+      "grad_norm": 1.3284317255020142,
+      "learning_rate": 9.90545258721667e-05,
+      "loss": 0.1409,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14939562678256146,
+      "grad_norm": 1.4952031373977661,
+      "learning_rate": 9.90222548839135e-05,
+      "loss": 0.1033,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1507537688442211,
+      "grad_norm": 1.310085654258728,
+      "learning_rate": 9.898944779613495e-05,
+      "loss": 0.1145,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15211191090588075,
+      "grad_norm": 1.294023036956787,
+      "learning_rate": 9.89561049676041e-05,
+      "loss": 0.1213,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1534700529675404,
+      "grad_norm": 0.8721023797988892,
+      "learning_rate": 9.89222267629528e-05,
+      "loss": 0.118,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15482819502920006,
+      "grad_norm": 1.3580466508865356,
+      "learning_rate": 9.888781355266763e-05,
+      "loss": 0.1029,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15618633709085972,
+      "grad_norm": 0.880094051361084,
+      "learning_rate": 9.885286571308598e-05,
+      "loss": 0.081,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15754447915251935,
+      "grad_norm": 1.2899099588394165,
+      "learning_rate": 9.881738362639182e-05,
+      "loss": 0.1201,
+      "step": 1160
+    },
+    {
+      "epoch": 0.158902621214179,
+      "grad_norm": 1.4249815940856934,
+      "learning_rate": 9.878136768061154e-05,
+      "loss": 0.1245,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16026076327583866,
+      "grad_norm": 0.9232867956161499,
+      "learning_rate": 9.874481826960979e-05,
+      "loss": 0.091,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16161890533749831,
+      "grad_norm": 1.227941870689392,
+      "learning_rate": 9.870773579308503e-05,
+      "loss": 0.1387,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16297704739915794,
+      "grad_norm": 1.4979817867279053,
+      "learning_rate": 9.867012065656533e-05,
+      "loss": 0.1071,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1643351894608176,
+      "grad_norm": 1.4730525016784668,
+      "learning_rate": 9.863197327140376e-05,
+      "loss": 0.1096,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16569333152247726,
+      "grad_norm": 1.147582769393921,
+      "learning_rate": 9.859329405477403e-05,
+      "loss": 0.0888,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1670514735841369,
+      "grad_norm": 1.5011526346206665,
+      "learning_rate": 9.855408342966585e-05,
+      "loss": 0.1197,
+      "step": 1230
+    },
+    {
+      "epoch": 0.16840961564579654,
+      "grad_norm": 0.8365809917449951,
+      "learning_rate": 9.851434182488033e-05,
+      "loss": 0.1122,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1697677577074562,
+      "grad_norm": 0.9515891075134277,
+      "learning_rate": 9.84740696750253e-05,
+      "loss": 0.1164,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17112589976911585,
+      "grad_norm": 1.4293049573898315,
+      "learning_rate": 9.843326742051055e-05,
+      "loss": 0.078,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1724840418307755,
+      "grad_norm": 1.1116342544555664,
+      "learning_rate": 9.839193550754297e-05,
+      "loss": 0.09,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17384218389243514,
+      "grad_norm": 1.1830072402954102,
+      "learning_rate": 9.835007438812177e-05,
+      "loss": 0.0942,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1752003259540948,
+      "grad_norm": 2.254817485809326,
+      "learning_rate": 9.830768452003341e-05,
+      "loss": 0.0867,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17655846801575445,
+      "grad_norm": 2.1280407905578613,
+      "learning_rate": 9.826476636684671e-05,
+      "loss": 0.119,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1779166100774141,
+      "grad_norm": 1.9420000314712524,
+      "learning_rate": 9.822132039790773e-05,
+      "loss": 0.1166,
+      "step": 1310
+    },
+    {
+      "epoch": 0.17927475213907373,
+      "grad_norm": 1.0773438215255737,
+      "learning_rate": 9.817734708833461e-05,
+      "loss": 0.1294,
+      "step": 1320
+    },
+    {
+      "epoch": 0.1806328942007334,
+      "grad_norm": 1.073169231414795,
+      "learning_rate": 9.813284691901243e-05,
+      "loss": 0.0761,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18199103626239305,
+      "grad_norm": 1.6711218357086182,
+      "learning_rate": 9.808782037658792e-05,
+      "loss": 0.0946,
+      "step": 1340
+    },
+    {
+      "epoch": 0.1833491783240527,
+      "grad_norm": 2.1709039211273193,
+      "learning_rate": 9.804226795346411e-05,
+      "loss": 0.0793,
+      "step": 1350
+    },
+    {
+      "epoch": 0.18470732038571236,
+      "grad_norm": 1.3815443515777588,
+      "learning_rate": 9.799619014779503e-05,
+      "loss": 0.1333,
+      "step": 1360
+    },
+    {
+      "epoch": 0.186065462447372,
+      "grad_norm": 1.7657443284988403,
+      "learning_rate": 9.794958746348013e-05,
+      "loss": 0.0844,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18742360450903164,
+      "grad_norm": 0.7350079417228699,
+      "learning_rate": 9.790246041015896e-05,
+      "loss": 0.0859,
+      "step": 1380
+    },
+    {
+      "epoch": 0.1887817465706913,
+      "grad_norm": 1.9589179754257202,
+      "learning_rate": 9.785480950320538e-05,
+      "loss": 0.0879,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19013988863235096,
+      "grad_norm": 2.1242854595184326,
+      "learning_rate": 9.78066352637221e-05,
+      "loss": 0.0845,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19149803069401058,
+      "grad_norm": 0.807992160320282,
+      "learning_rate": 9.775793821853488e-05,
+      "loss": 0.0841,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19285617275567024,
+      "grad_norm": 1.382732629776001,
+      "learning_rate": 9.77087189001868e-05,
+      "loss": 0.1071,
+      "step": 1420
+    },
+    {
+      "epoch": 0.1942143148173299,
+      "grad_norm": 1.4420311450958252,
+      "learning_rate": 9.765897784693243e-05,
+      "loss": 0.094,
+      "step": 1430
+    },
+    {
+      "epoch": 0.19557245687898955,
+      "grad_norm": 1.5609487295150757,
+      "learning_rate": 9.760871560273197e-05,
+      "loss": 0.1056,
+      "step": 1440
+    },
+    {
+      "epoch": 0.19693059894064918,
+      "grad_norm": 1.3678183555603027,
+      "learning_rate": 9.755793271724526e-05,
+      "loss": 0.0813,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19828874100230884,
+      "grad_norm": 0.9879801273345947,
+      "learning_rate": 9.750662974582584e-05,
+      "loss": 0.0943,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1996468830639685,
+      "grad_norm": 1.612644076347351,
+      "learning_rate": 9.745480724951473e-05,
+      "loss": 0.0607,
+      "step": 1470
+    },
+    {
+      "epoch": 0.20100502512562815,
+      "grad_norm": 1.155566692352295,
+      "learning_rate": 9.740246579503447e-05,
+      "loss": 0.058,
+      "step": 1480
+    },
+    {
+      "epoch": 0.20236316718728778,
+      "grad_norm": 1.1730011701583862,
+      "learning_rate": 9.734960595478284e-05,
+      "loss": 0.0921,
+      "step": 1490
+    },
+    {
+      "epoch": 0.20372130924894744,
+      "grad_norm": 1.087755560874939,
+      "learning_rate": 9.729622830682657e-05,
+      "loss": 0.057,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2050794513106071,
+      "grad_norm": 0.874884843826294,
+      "learning_rate": 9.724233343489504e-05,
+      "loss": 0.0704,
+      "step": 1510
+    },
+    {
+      "epoch": 0.20643759337226675,
+      "grad_norm": 1.3692662715911865,
+      "learning_rate": 9.718792192837396e-05,
+      "loss": 0.0728,
+      "step": 1520
+    },
+    {
+      "epoch": 0.20779573543392638,
+      "grad_norm": 1.2251136302947998,
+      "learning_rate": 9.713299438229886e-05,
+      "loss": 0.0713,
+      "step": 1530
+    },
+    {
+      "epoch": 0.20915387749558603,
+      "grad_norm": 1.9464949369430542,
+      "learning_rate": 9.707755139734855e-05,
+      "loss": 0.1083,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2105120195572457,
+      "grad_norm": 0.6273180842399597,
+      "learning_rate": 9.702159357983866e-05,
+      "loss": 0.1153,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21187016161890534,
+      "grad_norm": 1.0268000364303589,
+      "learning_rate": 9.696512154171492e-05,
+      "loss": 0.1098,
+      "step": 1560
+    },
+    {
+      "epoch": 0.21322830368056497,
+      "grad_norm": 1.4694157838821411,
+      "learning_rate": 9.690813590054645e-05,
+      "loss": 0.0772,
+      "step": 1570
+    },
+    {
+      "epoch": 0.21458644574222463,
+      "grad_norm": 1.1054388284683228,
+      "learning_rate": 9.685063727951914e-05,
+      "loss": 0.089,
+      "step": 1580
+    },
+    {
+      "epoch": 0.21594458780388429,
+      "grad_norm": 1.0415211915969849,
+      "learning_rate": 9.679262630742865e-05,
+      "loss": 0.0775,
+      "step": 1590
+    },
+    {
+      "epoch": 0.21730272986554394,
+      "grad_norm": 1.5325464010238647,
+      "learning_rate": 9.673410361867373e-05,
+      "loss": 0.0865,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2186608719272036,
+      "grad_norm": 1.0572073459625244,
+      "learning_rate": 9.667506985324909e-05,
+      "loss": 0.1117,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22001901398886323,
+      "grad_norm": 1.6043555736541748,
+      "learning_rate": 9.661552565673855e-05,
+      "loss": 0.0928,
+      "step": 1620
+    },
+    {
+      "epoch": 0.22137715605052288,
+      "grad_norm": 1.0668857097625732,
+      "learning_rate": 9.655547168030789e-05,
+      "loss": 0.0897,
+      "step": 1630
+    },
+    {
+      "epoch": 0.22273529811218254,
+      "grad_norm": 1.8016750812530518,
+      "learning_rate": 9.649490858069777e-05,
+      "loss": 0.11,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2240934401738422,
+      "grad_norm": 1.0796738862991333,
+      "learning_rate": 9.643383702021658e-05,
+      "loss": 0.1175,
+      "step": 1650
+    },
+    {
+      "epoch": 0.22545158223550182,
+      "grad_norm": 0.7914559245109558,
+      "learning_rate": 9.637225766673307e-05,
+      "loss": 0.0788,
+      "step": 1660
+    },
+    {
+      "epoch": 0.22680972429716148,
+      "grad_norm": 1.3361483812332153,
+      "learning_rate": 9.631017119366922e-05,
+      "loss": 0.0865,
+      "step": 1670
+    },
+    {
+      "epoch": 0.22816786635882114,
+      "grad_norm": 0.6504874229431152,
+      "learning_rate": 9.624757827999273e-05,
+      "loss": 0.0873,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2295260084204808,
+      "grad_norm": 1.0988860130310059,
+      "learning_rate": 9.618447961020971e-05,
+      "loss": 0.075,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23088415048214042,
+      "grad_norm": 0.8990165591239929,
+      "learning_rate": 9.612087587435707e-05,
+      "loss": 0.1064,
+      "step": 1700
+    },
+    {
+      "epoch": 0.23224229254380008,
+      "grad_norm": 1.3480658531188965,
+      "learning_rate": 9.605676776799508e-05,
+      "loss": 0.086,
+      "step": 1710
+    },
+    {
+      "epoch": 0.23360043460545973,
+      "grad_norm": 1.031354546546936,
+      "learning_rate": 9.599215599219973e-05,
+      "loss": 0.115,
+      "step": 1720
+    },
+    {
+      "epoch": 0.2349585766671194,
+      "grad_norm": 1.1643404960632324,
+      "learning_rate": 9.592704125355505e-05,
+      "loss": 0.0986,
+      "step": 1730
+    },
+    {
+      "epoch": 0.23631671872877902,
+      "grad_norm": 1.3293263912200928,
+      "learning_rate": 9.586142426414538e-05,
+      "loss": 0.0839,
+      "step": 1740
+    },
+    {
+      "epoch": 0.23767486079043867,
+      "grad_norm": 2.1880764961242676,
+      "learning_rate": 9.57953057415476e-05,
+      "loss": 0.083,
+      "step": 1750
+    },
+    {
+      "epoch": 0.23903300285209833,
+      "grad_norm": 1.3033689260482788,
+      "learning_rate": 9.572868640882328e-05,
+      "loss": 0.1053,
+      "step": 1760
+    },
+    {
+      "epoch": 0.240391144913758,
+      "grad_norm": 1.0856456756591797,
+      "learning_rate": 9.56615669945108e-05,
+      "loss": 0.0989,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24174928697541762,
+      "grad_norm": 1.0210094451904297,
+      "learning_rate": 9.55939482326173e-05,
+      "loss": 0.0537,
+      "step": 1780
+    },
+    {
+      "epoch": 0.24310742903707727,
+      "grad_norm": 1.158023476600647,
+      "learning_rate": 9.552583086261069e-05,
+      "loss": 0.1145,
+      "step": 1790
+    },
+    {
+      "epoch": 0.24446557109873693,
+      "grad_norm": 0.9875665903091431,
+      "learning_rate": 9.545721562941168e-05,
+      "loss": 0.1044,
+      "step": 1800
+    },
+    {
+      "epoch": 0.24582371316039658,
+      "grad_norm": 1.575035810470581,
+      "learning_rate": 9.538810328338543e-05,
+      "loss": 0.0733,
+      "step": 1810
+    },
+    {
+      "epoch": 0.24718185522205624,
+      "grad_norm": 1.0316693782806396,
+      "learning_rate": 9.531849458033349e-05,
+      "loss": 0.1037,
+      "step": 1820
+    },
+    {
+      "epoch": 0.24853999728371587,
+      "grad_norm": 1.4554296731948853,
+      "learning_rate": 9.524839028148547e-05,
+      "loss": 0.0746,
+      "step": 1830
+    },
+    {
+      "epoch": 0.24989813934537553,
+      "grad_norm": 1.0667850971221924,
+      "learning_rate": 9.517779115349077e-05,
+      "loss": 0.0969,
+      "step": 1840
+    },
+    {
+      "epoch": 0.25125628140703515,
+      "grad_norm": 1.0878078937530518,
+      "learning_rate": 9.510669796841014e-05,
+      "loss": 0.0665,
+      "step": 1850
+    },
+    {
+      "epoch": 0.25261442346869484,
+      "grad_norm": 1.1045596599578857,
+      "learning_rate": 9.503511150370727e-05,
+      "loss": 0.0884,
+      "step": 1860
+    },
+    {
+      "epoch": 0.25397256553035447,
+      "grad_norm": 1.6121373176574707,
+      "learning_rate": 9.496303254224024e-05,
+      "loss": 0.1216,
+      "step": 1870
+    },
+    {
+      "epoch": 0.25533070759201415,
+      "grad_norm": 1.6281442642211914,
+      "learning_rate": 9.489046187225306e-05,
+      "loss": 0.0918,
+      "step": 1880
+    },
+    {
+      "epoch": 0.2566888496536738,
+      "grad_norm": 0.9419350028038025,
+      "learning_rate": 9.481740028736692e-05,
+      "loss": 0.0701,
+      "step": 1890
+    },
+    {
+      "epoch": 0.2580469917153334,
+      "grad_norm": 1.0548691749572754,
+      "learning_rate": 9.474384858657164e-05,
+      "loss": 0.0798,
+      "step": 1900
+    },
+    {
+      "epoch": 0.2594051337769931,
+      "grad_norm": 0.6551480889320374,
+      "learning_rate": 9.466980757421679e-05,
+      "loss": 0.081,
+      "step": 1910
+    },
+    {
+      "epoch": 0.2607632758386527,
+      "grad_norm": 0.9093461036682129,
+      "learning_rate": 9.459527806000305e-05,
+      "loss": 0.0611,
+      "step": 1920
+    },
+    {
+      "epoch": 0.26212141790031235,
+      "grad_norm": 2.1890456676483154,
+      "learning_rate": 9.452026085897325e-05,
+      "loss": 0.1553,
+      "step": 1930
+    },
+    {
+      "epoch": 0.26347955996197203,
+      "grad_norm": 1.161549687385559,
+      "learning_rate": 9.444475679150348e-05,
+      "loss": 0.0841,
+      "step": 1940
+    },
+    {
+      "epoch": 0.26483770202363166,
+      "grad_norm": 1.2468726634979248,
+      "learning_rate": 9.436876668329411e-05,
+      "loss": 0.0538,
+      "step": 1950
+    },
+    {
+      "epoch": 0.26619584408529134,
+      "grad_norm": 0.7669261693954468,
+      "learning_rate": 9.429229136536079e-05,
+      "loss": 0.0609,
+      "step": 1960
+    },
+    {
+      "epoch": 0.267553986146951,
+      "grad_norm": 2.21005916595459,
+      "learning_rate": 9.421533167402534e-05,
+      "loss": 0.0796,
+      "step": 1970
+    },
+    {
+      "epoch": 0.2689121282086106,
+      "grad_norm": 1.490836262702942,
+      "learning_rate": 9.413788845090666e-05,
+      "loss": 0.0707,
+      "step": 1980
+    },
+    {
+      "epoch": 0.2702702702702703,
+      "grad_norm": 0.7032498717308044,
+      "learning_rate": 9.405996254291136e-05,
+      "loss": 0.062,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2716284123319299,
+      "grad_norm": 0.9867807030677795,
+      "learning_rate": 9.398155480222474e-05,
+      "loss": 0.0589,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27298655439358954,
+      "grad_norm": 0.938808023929596,
+      "learning_rate": 9.390266608630128e-05,
+      "loss": 0.0536,
+      "step": 2010
+    },
+    {
+      "epoch": 0.2743446964552492,
+      "grad_norm": 0.9619332551956177,
+      "learning_rate": 9.38232972578553e-05,
+      "loss": 0.069,
+      "step": 2020
+    },
+    {
+      "epoch": 0.27570283851690885,
+      "grad_norm": 1.3301962614059448,
+      "learning_rate": 9.374344918485164e-05,
+      "loss": 0.0744,
+      "step": 2030
+    },
+    {
+      "epoch": 0.27706098057856854,
+      "grad_norm": 0.7898063063621521,
+      "learning_rate": 9.366312274049602e-05,
+      "loss": 0.0989,
+      "step": 2040
+    },
+    {
+      "epoch": 0.27841912264022817,
+      "grad_norm": 1.2995082139968872,
+      "learning_rate": 9.358231880322554e-05,
+      "loss": 0.0633,
+      "step": 2050
+    },
+    {
+      "epoch": 0.2797772647018878,
+      "grad_norm": 0.722120463848114,
+      "learning_rate": 9.350103825669916e-05,
+      "loss": 0.0708,
+      "step": 2060
+    },
+    {
+      "epoch": 0.2811354067635475,
+      "grad_norm": 0.7053869366645813,
+      "learning_rate": 9.341928198978787e-05,
+      "loss": 0.0665,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2824935488252071,
+      "grad_norm": 1.2359107732772827,
+      "learning_rate": 9.333705089656512e-05,
+      "loss": 0.0873,
+      "step": 2080
+    },
+    {
+      "epoch": 0.2838516908868668,
+      "grad_norm": 1.1944717168807983,
+      "learning_rate": 9.325434587629698e-05,
+      "loss": 0.0838,
+      "step": 2090
+    },
+    {
+      "epoch": 0.2852098329485264,
+      "grad_norm": 0.8943485617637634,
+      "learning_rate": 9.31711678334323e-05,
+      "loss": 0.0795,
+      "step": 2100
+    },
+    {
+      "epoch": 0.28656797501018605,
+      "grad_norm": 0.7796288132667542,
+      "learning_rate": 9.308751767759282e-05,
+      "loss": 0.0725,
+      "step": 2110
+    },
+    {
+      "epoch": 0.28792611707184573,
+      "grad_norm": 0.7403047680854797,
+      "learning_rate": 9.300339632356325e-05,
+      "loss": 0.0859,
+      "step": 2120
+    },
+    {
+      "epoch": 0.28928425913350536,
+      "grad_norm": 1.19515860080719,
+      "learning_rate": 9.291880469128124e-05,
+      "loss": 0.0593,
+      "step": 2130
+    },
+    {
+      "epoch": 0.290642401195165,
+      "grad_norm": 1.6060223579406738,
+      "learning_rate": 9.283374370582732e-05,
+      "loss": 0.082,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2920005432568247,
+      "grad_norm": 1.143458604812622,
+      "learning_rate": 9.274821429741482e-05,
+      "loss": 0.1073,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2933586853184843,
+      "grad_norm": 0.7964088916778564,
+      "learning_rate": 9.266221740137961e-05,
+      "loss": 0.0945,
+      "step": 2160
+    },
+    {
+      "epoch": 0.294716827380144,
+      "grad_norm": 0.8407586812973022,
+      "learning_rate": 9.257575395817001e-05,
+      "loss": 0.0575,
+      "step": 2170
+    },
+    {
+      "epoch": 0.2960749694418036,
+      "grad_norm": 1.1075048446655273,
+      "learning_rate": 9.248882491333637e-05,
+      "loss": 0.0949,
+      "step": 2180
+    },
+    {
+      "epoch": 0.29743311150346324,
+      "grad_norm": 1.127458930015564,
+      "learning_rate": 9.240143121752076e-05,
+      "loss": 0.0427,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2987912535651229,
+      "grad_norm": 1.554289698600769,
+      "learning_rate": 9.23135738264467e-05,
+      "loss": 0.0778,
+      "step": 2200
+    },
+    {
+      "epoch": 0.30014939562678256,
+      "grad_norm": 0.5087252259254456,
+      "learning_rate": 9.222525370090849e-05,
+      "loss": 0.0752,
+      "step": 2210
+    },
+    {
+      "epoch": 0.3015075376884422,
+      "grad_norm": 1.0896332263946533,
+      "learning_rate": 9.213647180676088e-05,
+      "loss": 0.1037,
+      "step": 2220
+    },
+    {
+      "epoch": 0.30286567975010187,
+      "grad_norm": 0.8647719621658325,
+      "learning_rate": 9.204722911490846e-05,
+      "loss": 0.0735,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3042238218117615,
+      "grad_norm": 1.0851221084594727,
+      "learning_rate": 9.1957526601295e-05,
+      "loss": 0.0729,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3055819638734212,
+      "grad_norm": 1.421687364578247,
+      "learning_rate": 9.186736524689281e-05,
+      "loss": 0.1028,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3069401059350808,
+      "grad_norm": 1.266473412513733,
+      "learning_rate": 9.177674603769204e-05,
+      "loss": 0.1133,
+      "step": 2260
+    },
+    {
+      "epoch": 0.30829824799674044,
+      "grad_norm": 1.2845805883407593,
+      "learning_rate": 9.168566996468983e-05,
+      "loss": 0.1187,
+      "step": 2270
+    },
+    {
+      "epoch": 0.3096563900584001,
+      "grad_norm": 1.4501416683197021,
+      "learning_rate": 9.159413802387951e-05,
+      "loss": 0.1038,
+      "step": 2280
+    },
+    {
+      "epoch": 0.31101453212005975,
+      "grad_norm": 1.2279877662658691,
+      "learning_rate": 9.150215121623974e-05,
+      "loss": 0.0976,
+      "step": 2290
+    },
+    {
+      "epoch": 0.31237267418171943,
+      "grad_norm": 1.1472997665405273,
+      "learning_rate": 9.140971054772349e-05,
+      "loss": 0.1106,
+      "step": 2300
+    },
+    {
+      "epoch": 0.31373081624337906,
+      "grad_norm": 1.1568089723587036,
+      "learning_rate": 9.131681702924713e-05,
+      "loss": 0.0661,
+      "step": 2310
+    },
+    {
+      "epoch": 0.3150889583050387,
+      "grad_norm": 0.8612898588180542,
+      "learning_rate": 9.122347167667926e-05,
+      "loss": 0.0936,
+      "step": 2320
+    },
+    {
+      "epoch": 0.3164471003666984,
+      "grad_norm": 1.5391892194747925,
+      "learning_rate": 9.112967551082973e-05,
+      "loss": 0.0909,
+      "step": 2330
+    },
+    {
+      "epoch": 0.317805242428358,
+      "grad_norm": 1.2783818244934082,
+      "learning_rate": 9.103542955743835e-05,
+      "loss": 0.1087,
+      "step": 2340
+    },
+    {
+      "epoch": 0.31916338449001763,
+      "grad_norm": 1.225325345993042,
+      "learning_rate": 9.094073484716381e-05,
+      "loss": 0.0801,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3205215265516773,
+      "grad_norm": 1.4914628267288208,
+      "learning_rate": 9.084559241557226e-05,
+      "loss": 0.0776,
+      "step": 2360
+    },
+    {
+      "epoch": 0.32187966861333694,
+      "grad_norm": 1.225519061088562,
+      "learning_rate": 9.075000330312608e-05,
+      "loss": 0.0842,
+      "step": 2370
+    },
+    {
+      "epoch": 0.32323781067499663,
+      "grad_norm": 1.0433491468429565,
+      "learning_rate": 9.065396855517253e-05,
+      "loss": 0.0755,
+      "step": 2380
+    },
+    {
+      "epoch": 0.32459595273665626,
+      "grad_norm": 1.4192088842391968,
+      "learning_rate": 9.055748922193219e-05,
+      "loss": 0.0748,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3259540947983159,
+      "grad_norm": 1.2184417247772217,
+      "learning_rate": 9.046056635848761e-05,
+      "loss": 0.1006,
+      "step": 2400
+    },
+    {
+      "epoch": 0.32731223685997557,
+      "grad_norm": 1.49359929561615,
+      "learning_rate": 9.036320102477169e-05,
+      "loss": 0.1139,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3286703789216352,
+      "grad_norm": 1.7678170204162598,
+      "learning_rate": 9.02653942855561e-05,
+      "loss": 0.0735,
+      "step": 2420
+    },
+    {
+      "epoch": 0.3300285209832948,
+      "grad_norm": 0.6798723340034485,
+      "learning_rate": 9.016714721043971e-05,
+      "loss": 0.0572,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3313866630449545,
+      "grad_norm": 0.9008175134658813,
+      "learning_rate": 9.006846087383675e-05,
+      "loss": 0.082,
+      "step": 2440
+    },
+    {
+      "epoch": 0.33274480510661414,
+      "grad_norm": 0.7593595385551453,
+      "learning_rate": 8.996933635496523e-05,
+      "loss": 0.0776,
+      "step": 2450
+    },
+    {
+      "epoch": 0.3341029471682738,
+      "grad_norm": 0.9014355540275574,
+      "learning_rate": 8.986977473783498e-05,
+      "loss": 0.0628,
+      "step": 2460
+    },
+    {
+      "epoch": 0.33546108922993345,
+      "grad_norm": 1.198567509651184,
+      "learning_rate": 8.97697771112359e-05,
+      "loss": 0.1024,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3368192312915931,
+      "grad_norm": 0.7001988291740417,
+      "learning_rate": 8.966934456872602e-05,
+      "loss": 0.0554,
+      "step": 2480
+    },
+    {
+      "epoch": 0.33817737335325276,
+      "grad_norm": 0.8445319533348083,
+      "learning_rate": 8.95684782086195e-05,
+      "loss": 0.064,
+      "step": 2490
+    },
+    {
+      "epoch": 0.3395355154149124,
+      "grad_norm": 1.2642648220062256,
+      "learning_rate": 8.946717913397476e-05,
+      "loss": 0.0671,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3408936574765721,
+      "grad_norm": 0.9681129455566406,
+      "learning_rate": 8.93654484525822e-05,
+      "loss": 0.0903,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3422517995382317,
+      "grad_norm": 0.8917770385742188,
+      "learning_rate": 8.926328727695226e-05,
+      "loss": 0.0961,
+      "step": 2520
+    },
+    {
+      "epoch": 0.34360994159989133,
+      "grad_norm": 1.0627943277359009,
+      "learning_rate": 8.916069672430319e-05,
+      "loss": 0.0655,
+      "step": 2530
+    },
+    {
+      "epoch": 0.344968083661551,
+      "grad_norm": 0.9439001679420471,
+      "learning_rate": 8.905767791654884e-05,
+      "loss": 0.0555,
+      "step": 2540
+    },
+    {
+      "epoch": 0.34632622572321065,
+      "grad_norm": 1.2661627531051636,
+      "learning_rate": 8.895423198028638e-05,
+      "loss": 0.0986,
+      "step": 2550
+    },
+    {
+      "epoch": 0.3476843677848703,
+      "grad_norm": 0.79547518491745,
+      "learning_rate": 8.885036004678402e-05,
+      "loss": 0.0711,
+      "step": 2560
+    },
+    {
+      "epoch": 0.34904250984652996,
+      "grad_norm": 0.7643102407455444,
+      "learning_rate": 8.874606325196857e-05,
+      "loss": 0.0747,
+      "step": 2570
+    },
+    {
+      "epoch": 0.3504006519081896,
+      "grad_norm": 0.559659481048584,
+      "learning_rate": 8.864134273641304e-05,
+      "loss": 0.0526,
+      "step": 2580
+    },
+    {
+      "epoch": 0.35175879396984927,
+      "grad_norm": 1.4627636671066284,
+      "learning_rate": 8.853619964532427e-05,
+      "loss": 0.0921,
+      "step": 2590
+    },
+    {
+      "epoch": 0.3531169360315089,
+      "grad_norm": 1.564150094985962,
+      "learning_rate": 8.843063512853019e-05,
+      "loss": 0.0742,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3544750780931685,
+      "grad_norm": 1.0845599174499512,
+      "learning_rate": 8.832465034046749e-05,
+      "loss": 0.0606,
+      "step": 2610
+    },
+    {
+      "epoch": 0.3558332201548282,
+      "grad_norm": 0.9278034567832947,
+      "learning_rate": 8.821824644016882e-05,
+      "loss": 0.0767,
+      "step": 2620
+    },
+    {
+      "epoch": 0.35719136221648784,
+      "grad_norm": 1.2881855964660645,
+      "learning_rate": 8.811142459125019e-05,
+      "loss": 0.0673,
+      "step": 2630
+    },
+    {
+      "epoch": 0.35854950427814747,
+      "grad_norm": 1.017382264137268,
+      "learning_rate": 8.800418596189822e-05,
+      "loss": 0.0846,
+      "step": 2640
+    },
+    {
+      "epoch": 0.35990764633980715,
+      "grad_norm": 0.4987911283969879,
+      "learning_rate": 8.789653172485737e-05,
+      "loss": 0.0475,
+      "step": 2650
+    },
+    {
+      "epoch": 0.3612657884014668,
+      "grad_norm": 1.4544811248779297,
+      "learning_rate": 8.778846305741715e-05,
+      "loss": 0.0485,
+      "step": 2660
+    },
+    {
+      "epoch": 0.36262393046312646,
+      "grad_norm": 1.5136889219284058,
+      "learning_rate": 8.767998114139918e-05,
+      "loss": 0.0537,
+      "step": 2670
+    },
+    {
+      "epoch": 0.3639820725247861,
+      "grad_norm": 1.4157406091690063,
+      "learning_rate": 8.757108716314429e-05,
+      "loss": 0.0738,
+      "step": 2680
+    },
+    {
+      "epoch": 0.3653402145864457,
+      "grad_norm": 0.7702262997627258,
+      "learning_rate": 8.746178231349962e-05,
+      "loss": 0.078,
+      "step": 2690
+    },
+    {
+      "epoch": 0.3666983566481054,
+      "grad_norm": 0.9803953766822815,
+      "learning_rate": 8.735206778780549e-05,
+      "loss": 0.0909,
+      "step": 2700
+    },
+    {
+      "epoch": 0.36805649870976503,
+      "grad_norm": 1.3174487352371216,
+      "learning_rate": 8.724194478588234e-05,
+      "loss": 0.0699,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3694146407714247,
+      "grad_norm": 0.705106258392334,
+      "learning_rate": 8.713141451201772e-05,
+      "loss": 0.0835,
+      "step": 2720
+    },
+    {
+      "epoch": 0.37077278283308435,
+      "grad_norm": 0.9962559342384338,
+      "learning_rate": 8.702047817495295e-05,
+      "loss": 0.0648,
+      "step": 2730
+    },
+    {
+      "epoch": 0.372130924894744,
+      "grad_norm": 1.1443382501602173,
+      "learning_rate": 8.69091369878701e-05,
+      "loss": 0.0807,
+      "step": 2740
+    },
+    {
+      "epoch": 0.37348906695640366,
+      "grad_norm": 1.3265937566757202,
+      "learning_rate": 8.679739216837849e-05,
+      "loss": 0.0703,
+      "step": 2750
+    },
+    {
+      "epoch": 0.3748472090180633,
+      "grad_norm": 0.8906628489494324,
+      "learning_rate": 8.66852449385016e-05,
+      "loss": 0.0547,
+      "step": 2760
+    },
+    {
+      "epoch": 0.3762053510797229,
+      "grad_norm": 1.0092452764511108,
+      "learning_rate": 8.657269652466356e-05,
+      "loss": 0.0602,
+      "step": 2770
+    },
+    {
+      "epoch": 0.3775634931413826,
+      "grad_norm": 0.6514250636100769,
+      "learning_rate": 8.645974815767577e-05,
+      "loss": 0.0504,
+      "step": 2780
+    },
+    {
+      "epoch": 0.37892163520304223,
+      "grad_norm": 1.0900055170059204,
+      "learning_rate": 8.634640107272351e-05,
+      "loss": 0.0683,
+      "step": 2790
+    },
+    {
+      "epoch": 0.3802797772647019,
+      "grad_norm": 0.9178528189659119,
+      "learning_rate": 8.623265650935234e-05,
+      "loss": 0.0546,
+      "step": 2800
+    },
+    {
+      "epoch": 0.38163791932636154,
+      "grad_norm": 1.3056002855300903,
+      "learning_rate": 8.611851571145456e-05,
+      "loss": 0.0778,
+      "step": 2810
+    },
+    {
+      "epoch": 0.38299606138802117,
+      "grad_norm": 1.3176368474960327,
+      "learning_rate": 8.600397992725566e-05,
+      "loss": 0.0918,
+      "step": 2820
+    },
+    {
+      "epoch": 0.38435420344968085,
+      "grad_norm": 1.3218613862991333,
+      "learning_rate": 8.588905040930061e-05,
+      "loss": 0.0716,
+      "step": 2830
+    },
+    {
+      "epoch": 0.3857123455113405,
+      "grad_norm": 1.2861921787261963,
+      "learning_rate": 8.577372841444022e-05,
+      "loss": 0.0886,
+      "step": 2840
+    },
+    {
+      "epoch": 0.3870704875730001,
+      "grad_norm": 1.6400715112686157,
+      "learning_rate": 8.565801520381736e-05,
+      "loss": 0.0738,
+      "step": 2850
+    },
+    {
+      "epoch": 0.3884286296346598,
+      "grad_norm": 1.4341719150543213,
+      "learning_rate": 8.554191204285313e-05,
+      "loss": 0.0719,
+      "step": 2860
+    },
+    {
+      "epoch": 0.3897867716963194,
+      "grad_norm": 0.750482976436615,
+      "learning_rate": 8.542542020123315e-05,
+      "loss": 0.0695,
+      "step": 2870
+    },
+    {
+      "epoch": 0.3911449137579791,
+      "grad_norm": 1.545171856880188,
+      "learning_rate": 8.530854095289347e-05,
+      "loss": 0.085,
+      "step": 2880
+    },
+    {
+      "epoch": 0.39250305581963874,
+      "grad_norm": 2.0415072441101074,
+      "learning_rate": 8.519127557600688e-05,
+      "loss": 0.09,
+      "step": 2890
+    },
+    {
+      "epoch": 0.39386119788129836,
+      "grad_norm": 1.1449663639068604,
+      "learning_rate": 8.507362535296871e-05,
+      "loss": 0.0544,
+      "step": 2900
+    },
+    {
+      "epoch": 0.39521933994295805,
+      "grad_norm": 0.5266981720924377,
+      "learning_rate": 8.495559157038299e-05,
+      "loss": 0.085,
+      "step": 2910
+    },
+    {
+      "epoch": 0.3965774820046177,
+      "grad_norm": 0.6705682277679443,
+      "learning_rate": 8.483717551904823e-05,
+      "loss": 0.0512,
+      "step": 2920
+    },
+    {
+      "epoch": 0.39793562406627736,
+      "grad_norm": 0.8800350427627563,
+      "learning_rate": 8.47183784939434e-05,
+      "loss": 0.0848,
+      "step": 2930
+    },
+    {
+      "epoch": 0.399293766127937,
+      "grad_norm": 0.7678025364875793,
+      "learning_rate": 8.459920179421374e-05,
+      "loss": 0.0721,
+      "step": 2940
+    },
+    {
+      "epoch": 0.4006519081895966,
+      "grad_norm": 0.8100093007087708,
+      "learning_rate": 8.447964672315656e-05,
+      "loss": 0.0557,
+      "step": 2950
+    },
+    {
+      "epoch": 0.4020100502512563,
+      "grad_norm": 0.9654191136360168,
+      "learning_rate": 8.435971458820692e-05,
+      "loss": 0.0462,
+      "step": 2960
+    },
+    {
+      "epoch": 0.40336819231291593,
+      "grad_norm": 0.6833099126815796,
+      "learning_rate": 8.423940670092345e-05,
+      "loss": 0.0664,
+      "step": 2970
+    },
+    {
+      "epoch": 0.40472633437457556,
+      "grad_norm": 0.7810555696487427,
+      "learning_rate": 8.411872437697394e-05,
+      "loss": 0.0517,
+      "step": 2980
+    },
+    {
+      "epoch": 0.40608447643623524,
+      "grad_norm": 1.0646721124649048,
+      "learning_rate": 8.399766893612096e-05,
+      "loss": 0.0713,
+      "step": 2990
+    },
+    {
+      "epoch": 0.40744261849789487,
+      "grad_norm": 0.8468383550643921,
+      "learning_rate": 8.38762417022074e-05,
+      "loss": 0.0597,
+      "step": 3000
+    },
+    {
+      "epoch": 0.40880076055955455,
+      "grad_norm": 0.5564700365066528,
+      "learning_rate": 8.375444400314204e-05,
+      "loss": 0.0544,
+      "step": 3010
+    },
+    {
+      "epoch": 0.4101589026212142,
+      "grad_norm": 0.9086338877677917,
+      "learning_rate": 8.3632277170885e-05,
+      "loss": 0.0588,
+      "step": 3020
+    },
+    {
+      "epoch": 0.4115170446828738,
+      "grad_norm": 1.0646495819091797,
+      "learning_rate": 8.350974254143318e-05,
+      "loss": 0.0609,
+      "step": 3030
+    },
+    {
+      "epoch": 0.4128751867445335,
+      "grad_norm": 0.8048272728919983,
+      "learning_rate": 8.338684145480566e-05,
+      "loss": 0.0733,
+      "step": 3040
+    },
+    {
+      "epoch": 0.4142333288061931,
+      "grad_norm": 0.9169145226478577,
+      "learning_rate": 8.326357525502904e-05,
+      "loss": 0.0471,
+      "step": 3050
+    },
+    {
+      "epoch": 0.41559147086785275,
+      "grad_norm": 0.6720271706581116,
+      "learning_rate": 8.313994529012273e-05,
+      "loss": 0.0866,
+      "step": 3060
+    },
+    {
+      "epoch": 0.41694961292951244,
+      "grad_norm": 0.9976987838745117,
+      "learning_rate": 8.301595291208422e-05,
+      "loss": 0.0664,
+      "step": 3070
+    },
+    {
+      "epoch": 0.41830775499117206,
+      "grad_norm": 0.7122252583503723,
+      "learning_rate": 8.289159947687427e-05,
+      "loss": 0.0473,
+      "step": 3080
+    },
+    {
+      "epoch": 0.41966589705283175,
+      "grad_norm": 1.3988673686981201,
+      "learning_rate": 8.276688634440216e-05,
+      "loss": 0.0639,
+      "step": 3090
+    },
+    {
+      "epoch": 0.4210240391144914,
+      "grad_norm": 0.8595008850097656,
+      "learning_rate": 8.26418148785107e-05,
+      "loss": 0.07,
+      "step": 3100
+    },
+    {
+      "epoch": 0.422382181176151,
+      "grad_norm": 0.749464213848114,
+      "learning_rate": 8.251638644696141e-05,
+      "loss": 0.0843,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4237403232378107,
+      "grad_norm": 0.6274281740188599,
+      "learning_rate": 8.23906024214195e-05,
+      "loss": 0.0623,
+      "step": 3120
+    },
+    {
+      "epoch": 0.4250984652994703,
+      "grad_norm": 0.8080748319625854,
+      "learning_rate": 8.226446417743897e-05,
+      "loss": 0.0392,
+      "step": 3130
+    },
+    {
+      "epoch": 0.42645660736112995,
+      "grad_norm": 1.1160558462142944,
+      "learning_rate": 8.213797309444742e-05,
+      "loss": 0.0479,
+      "step": 3140
+    },
+    {
+      "epoch": 0.42781474942278963,
+      "grad_norm": 0.8156780004501343,
+      "learning_rate": 8.201113055573105e-05,
+      "loss": 0.0527,
+      "step": 3150
+    },
+    {
+      "epoch": 0.42917289148444926,
+      "grad_norm": 0.8623396158218384,
+      "learning_rate": 8.188393794841958e-05,
+      "loss": 0.0472,
+      "step": 3160
+    },
+    {
+      "epoch": 0.43053103354610894,
+      "grad_norm": 1.4742708206176758,
+      "learning_rate": 8.175639666347094e-05,
+      "loss": 0.064,
+      "step": 3170
+    },
+    {
+      "epoch": 0.43188917560776857,
+      "grad_norm": 1.2746484279632568,
+      "learning_rate": 8.162850809565623e-05,
+      "loss": 0.0866,
+      "step": 3180
+    },
+    {
+      "epoch": 0.4332473176694282,
+      "grad_norm": 0.600433886051178,
+      "learning_rate": 8.150027364354431e-05,
+      "loss": 0.0691,
+      "step": 3190
+    },
+    {
+      "epoch": 0.4346054597310879,
+      "grad_norm": 0.5896084308624268,
+      "learning_rate": 8.137169470948662e-05,
+      "loss": 0.0545,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4359636017927475,
+      "grad_norm": 0.9488576650619507,
+      "learning_rate": 8.124277269960179e-05,
+      "loss": 0.0471,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4373217438544072,
+      "grad_norm": 0.9625917077064514,
+      "learning_rate": 8.111350902376023e-05,
+      "loss": 0.0551,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4386798859160668,
+      "grad_norm": 1.6023871898651123,
+      "learning_rate": 8.098390509556883e-05,
+      "loss": 0.0608,
+      "step": 3230
+    },
+    {
+      "epoch": 0.44003802797772645,
+      "grad_norm": 0.7293844223022461,
+      "learning_rate": 8.085396233235536e-05,
+      "loss": 0.087,
+      "step": 3240
+    },
+    {
+      "epoch": 0.44139617003938614,
+      "grad_norm": 0.8570722937583923,
+      "learning_rate": 8.072368215515306e-05,
+      "loss": 0.0668,
+      "step": 3250
+    },
+    {
+      "epoch": 0.44275431210104577,
+      "grad_norm": 0.6232367157936096,
+      "learning_rate": 8.059306598868506e-05,
+      "loss": 0.1004,
+      "step": 3260
+    },
+    {
+      "epoch": 0.4441124541627054,
+      "grad_norm": 0.9708699584007263,
+      "learning_rate": 8.046211526134888e-05,
+      "loss": 0.0684,
+      "step": 3270
+    },
+    {
+      "epoch": 0.4454705962243651,
+      "grad_norm": 1.0352977514266968,
+      "learning_rate": 8.033083140520065e-05,
+      "loss": 0.0991,
+      "step": 3280
+    },
+    {
+      "epoch": 0.4468287382860247,
+      "grad_norm": 0.8723713159561157,
+      "learning_rate": 8.019921585593962e-05,
+      "loss": 0.0698,
+      "step": 3290
+    },
+    {
+      "epoch": 0.4481868803476844,
+      "grad_norm": 0.8299247026443481,
+      "learning_rate": 8.006727005289232e-05,
+      "loss": 0.078,
+      "step": 3300
+    },
+    {
+      "epoch": 0.449545022409344,
+      "grad_norm": 1.4721529483795166,
+      "learning_rate": 7.993499543899692e-05,
+      "loss": 0.0601,
+      "step": 3310
+    },
+    {
+      "epoch": 0.45090316447100365,
+      "grad_norm": 0.5306909084320068,
+      "learning_rate": 7.980239346078742e-05,
+      "loss": 0.0696,
+      "step": 3320
+    },
+    {
+      "epoch": 0.45226130653266333,
+      "grad_norm": 1.502251148223877,
+      "learning_rate": 7.966946556837778e-05,
+      "loss": 0.0491,
+      "step": 3330
+    },
+    {
+      "epoch": 0.45361944859432296,
+      "grad_norm": 0.7173418402671814,
+      "learning_rate": 7.953621321544616e-05,
+      "loss": 0.0647,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4549775906559826,
+      "grad_norm": 1.245213270187378,
+      "learning_rate": 7.940263785921896e-05,
+      "loss": 0.0628,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4563357327176423,
+      "grad_norm": 0.7357654571533203,
+      "learning_rate": 7.926874096045482e-05,
+      "loss": 0.0624,
+      "step": 3360
+    },
+    {
+      "epoch": 0.4576938747793019,
+      "grad_norm": 1.215211033821106,
+      "learning_rate": 7.913452398342881e-05,
+      "loss": 0.0558,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4590520168409616,
+      "grad_norm": 0.8133662343025208,
+      "learning_rate": 7.89999883959163e-05,
+      "loss": 0.0556,
+      "step": 3380
+    },
+    {
+      "epoch": 0.4604101589026212,
+      "grad_norm": 0.7595717310905457,
+      "learning_rate": 7.886513566917687e-05,
+      "loss": 0.0561,
+      "step": 3390
+    },
+    {
+      "epoch": 0.46176830096428084,
+      "grad_norm": 0.6180480122566223,
+      "learning_rate": 7.872996727793838e-05,
+      "loss": 0.058,
+      "step": 3400
+    },
+    {
+      "epoch": 0.4631264430259405,
+      "grad_norm": 1.3587250709533691,
+      "learning_rate": 7.859448470038069e-05,
+      "loss": 0.0661,
+      "step": 3410
+    },
+    {
+      "epoch": 0.46448458508760015,
+      "grad_norm": 1.241623044013977,
+      "learning_rate": 7.845868941811956e-05,
+      "loss": 0.0862,
+      "step": 3420
+    },
+    {
+      "epoch": 0.46584272714925984,
+      "grad_norm": 1.2678231000900269,
+      "learning_rate": 7.832258291619043e-05,
+      "loss": 0.073,
+      "step": 3430
+    },
+    {
+      "epoch": 0.46720086921091947,
+      "grad_norm": 0.6042308211326599,
+      "learning_rate": 7.81861666830322e-05,
+      "loss": 0.0454,
+      "step": 3440
+    },
+    {
+      "epoch": 0.4685590112725791,
+      "grad_norm": 0.6415855884552002,
+      "learning_rate": 7.804944221047097e-05,
+      "loss": 0.0594,
+      "step": 3450
+    },
+    {
+      "epoch": 0.4699171533342388,
+      "grad_norm": 0.8118975162506104,
+      "learning_rate": 7.791241099370364e-05,
+      "loss": 0.0418,
+      "step": 3460
+    },
+    {
+      "epoch": 0.4712752953958984,
+      "grad_norm": 0.9041648507118225,
+      "learning_rate": 7.777507453128163e-05,
+      "loss": 0.0745,
+      "step": 3470
+    },
+    {
+      "epoch": 0.47263343745755804,
+      "grad_norm": 0.933545708656311,
+      "learning_rate": 7.763743432509451e-05,
+      "loss": 0.0595,
+      "step": 3480
+    },
+    {
+      "epoch": 0.4739915795192177,
+      "grad_norm": 0.8928757309913635,
+      "learning_rate": 7.749949188035353e-05,
+      "loss": 0.0569,
+      "step": 3490
+    },
+    {
+      "epoch": 0.47534972158087735,
+      "grad_norm": 0.9314897060394287,
+      "learning_rate": 7.736124870557516e-05,
+      "loss": 0.0618,
+      "step": 3500
+    },
+    {
+      "epoch": 0.47670786364253703,
+      "grad_norm": 0.9116345643997192,
+      "learning_rate": 7.722270631256459e-05,
+      "loss": 0.0597,
+      "step": 3510
+    },
+    {
+      "epoch": 0.47806600570419666,
+      "grad_norm": 1.1705352067947388,
+      "learning_rate": 7.708386621639925e-05,
+      "loss": 0.0626,
+      "step": 3520
+    },
+    {
+      "epoch": 0.4794241477658563,
+      "grad_norm": 0.37668225169181824,
+      "learning_rate": 7.694472993541219e-05,
+      "loss": 0.0501,
+      "step": 3530
+    },
+    {
+      "epoch": 0.480782289827516,
+      "grad_norm": 0.8684849739074707,
+      "learning_rate": 7.680529899117547e-05,
+      "loss": 0.0617,
+      "step": 3540
+    },
+    {
+      "epoch": 0.4821404318891756,
+      "grad_norm": 1.0204918384552002,
+      "learning_rate": 7.666557490848358e-05,
+      "loss": 0.0686,
+      "step": 3550
+    },
+    {
+      "epoch": 0.48349857395083523,
+      "grad_norm": 1.3644721508026123,
+      "learning_rate": 7.65255592153367e-05,
+      "loss": 0.0545,
+      "step": 3560
+    },
+    {
+      "epoch": 0.4848567160124949,
+      "grad_norm": 2.151400566101074,
+      "learning_rate": 7.638525344292402e-05,
+      "loss": 0.0544,
+      "step": 3570
+    },
+    {
+      "epoch": 0.48621485807415454,
+      "grad_norm": 0.5766331553459167,
+      "learning_rate": 7.624465912560697e-05,
+      "loss": 0.0719,
+      "step": 3580
+    },
+    {
+      "epoch": 0.4875730001358142,
+      "grad_norm": 1.2946696281433105,
+      "learning_rate": 7.610377780090249e-05,
+      "loss": 0.0555,
+      "step": 3590
+    },
+    {
+      "epoch": 0.48893114219747386,
+      "grad_norm": 0.7257918119430542,
+      "learning_rate": 7.596261100946618e-05,
+      "loss": 0.069,
+      "step": 3600
+    },
+    {
+      "epoch": 0.4902892842591335,
+      "grad_norm": 0.6641822457313538,
+      "learning_rate": 7.582116029507542e-05,
+      "loss": 0.0903,
+      "step": 3610
+    },
+    {
+      "epoch": 0.49164742632079317,
+      "grad_norm": 0.41628071665763855,
+      "learning_rate": 7.56794272046126e-05,
+      "loss": 0.0527,
+      "step": 3620
+    },
+    {
+      "epoch": 0.4930055683824528,
+      "grad_norm": 0.6902533769607544,
+      "learning_rate": 7.55374132880481e-05,
+      "loss": 0.0406,
+      "step": 3630
+    },
+    {
+      "epoch": 0.4943637104441125,
+      "grad_norm": 1.3925668001174927,
+      "learning_rate": 7.539512009842333e-05,
+      "loss": 0.0877,
+      "step": 3640
+    },
+    {
+      "epoch": 0.4957218525057721,
+      "grad_norm": 1.115399956703186,
+      "learning_rate": 7.525254919183382e-05,
+      "loss": 0.0476,
+      "step": 3650
+    },
+    {
+      "epoch": 0.49707999456743174,
+      "grad_norm": 1.3451963663101196,
+      "learning_rate": 7.510970212741215e-05,
+      "loss": 0.0779,
+      "step": 3660
+    },
+    {
+      "epoch": 0.4984381366290914,
+      "grad_norm": 0.7136859893798828,
+      "learning_rate": 7.496658046731096e-05,
+      "loss": 0.068,
+      "step": 3670
+    },
+    {
+      "epoch": 0.49979627869075105,
+      "grad_norm": 0.7799562215805054,
+      "learning_rate": 7.482318577668578e-05,
+      "loss": 0.0619,
+      "step": 3680
+    },
+    {
+      "epoch": 0.5011544207524107,
+      "grad_norm": 0.7798644304275513,
+      "learning_rate": 7.467951962367796e-05,
+      "loss": 0.0894,
+      "step": 3690
+    },
+    {
+      "epoch": 0.5025125628140703,
+      "grad_norm": 0.9806691408157349,
+      "learning_rate": 7.453558357939755e-05,
+      "loss": 0.0717,
+      "step": 3700
+    },
+    {
+      "epoch": 0.50387070487573,
+      "grad_norm": 1.134883999824524,
+      "learning_rate": 7.439137921790606e-05,
+      "loss": 0.0461,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5052288469373897,
+      "grad_norm": 0.7467165589332581,
+      "learning_rate": 7.42469081161993e-05,
+      "loss": 0.0832,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5065869889990493,
+      "grad_norm": 1.4849263429641724,
+      "learning_rate": 7.410217185419006e-05,
+      "loss": 0.0605,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5079451310607089,
+      "grad_norm": 0.6116840243339539,
+      "learning_rate": 7.395717201469095e-05,
+      "loss": 0.0791,
+      "step": 3740
+    },
+    {
+      "epoch": 0.5093032731223686,
+      "grad_norm": 0.3688634932041168,
+      "learning_rate": 7.381191018339696e-05,
+      "loss": 0.0513,
+      "step": 3750
+    },
+    {
+      "epoch": 0.5106614151840283,
+      "grad_norm": 0.9582529067993164,
+      "learning_rate": 7.36663879488682e-05,
+      "loss": 0.0541,
+      "step": 3760
+    },
+    {
+      "epoch": 0.5120195572456879,
+      "grad_norm": 0.6000323295593262,
+      "learning_rate": 7.352060690251254e-05,
+      "loss": 0.0621,
+      "step": 3770
+    },
+    {
+      "epoch": 0.5133776993073476,
+      "grad_norm": 0.6939341425895691,
+      "learning_rate": 7.337456863856811e-05,
+      "loss": 0.0484,
+      "step": 3780
+    },
+    {
+      "epoch": 0.5147358413690072,
+      "grad_norm": 0.7146658897399902,
+      "learning_rate": 7.3228274754086e-05,
+      "loss": 0.0557,
+      "step": 3790
+    },
+    {
+      "epoch": 0.5160939834306668,
+      "grad_norm": 0.9297583699226379,
+      "learning_rate": 7.308172684891267e-05,
+      "loss": 0.044,
+      "step": 3800
+    },
+    {
+      "epoch": 0.5174521254923264,
+      "grad_norm": 0.8100739121437073,
+      "learning_rate": 7.293492652567255e-05,
+      "loss": 0.0612,
+      "step": 3810
+    },
+    {
+      "epoch": 0.5188102675539862,
+      "grad_norm": 0.4675349295139313,
+      "learning_rate": 7.278787538975043e-05,
+      "loss": 0.0788,
+      "step": 3820
+    },
+    {
+      "epoch": 0.5201684096156458,
+      "grad_norm": 0.9917894005775452,
+      "learning_rate": 7.2640575049274e-05,
+      "loss": 0.0729,
+      "step": 3830
+    },
+    {
+      "epoch": 0.5215265516773054,
+      "grad_norm": 0.7156324982643127,
+      "learning_rate": 7.249302711509616e-05,
+      "loss": 0.0731,
+      "step": 3840
+    },
+    {
+      "epoch": 0.5228846937389651,
+      "grad_norm": 1.0420186519622803,
+      "learning_rate": 7.23452332007775e-05,
+      "loss": 0.0519,
+      "step": 3850
+    },
+    {
+      "epoch": 0.5242428358006247,
+      "grad_norm": 0.43901902437210083,
+      "learning_rate": 7.219719492256858e-05,
+      "loss": 0.0451,
+      "step": 3860
+    },
+    {
+      "epoch": 0.5256009778622844,
+      "grad_norm": 0.6867021322250366,
+      "learning_rate": 7.20489138993923e-05,
+      "loss": 0.0686,
+      "step": 3870
+    },
+    {
+      "epoch": 0.5269591199239441,
+      "grad_norm": 1.6295194625854492,
+      "learning_rate": 7.190039175282614e-05,
+      "loss": 0.0686,
+      "step": 3880
+    },
+    {
+      "epoch": 0.5283172619856037,
+      "grad_norm": 0.8770433068275452,
+      "learning_rate": 7.175163010708455e-05,
+      "loss": 0.057,
+      "step": 3890
+    },
+    {
+      "epoch": 0.5296754040472633,
+      "grad_norm": 0.9314376711845398,
+      "learning_rate": 7.1602630589001e-05,
+      "loss": 0.0426,
+      "step": 3900
+    },
+    {
+      "epoch": 0.531033546108923,
+      "grad_norm": 0.6964649558067322,
+      "learning_rate": 7.14533948280104e-05,
+      "loss": 0.0607,
+      "step": 3910
+    },
+    {
+      "epoch": 0.5323916881705827,
+      "grad_norm": 0.7150620222091675,
+      "learning_rate": 7.130392445613109e-05,
+      "loss": 0.0521,
+      "step": 3920
+    },
+    {
+      "epoch": 0.5337498302322423,
+      "grad_norm": 0.6022738218307495,
+      "learning_rate": 7.115422110794711e-05,
+      "loss": 0.0569,
+      "step": 3930
+    },
+    {
+      "epoch": 0.535107972293902,
+      "grad_norm": 0.8522818088531494,
+      "learning_rate": 7.100428642059033e-05,
+      "loss": 0.061,
+      "step": 3940
+    },
+    {
+      "epoch": 0.5364661143555616,
+      "grad_norm": 0.6194635033607483,
+      "learning_rate": 7.08541220337224e-05,
+      "loss": 0.0611,
+      "step": 3950
+    },
+    {
+      "epoch": 0.5378242564172212,
+      "grad_norm": 0.6505541205406189,
+      "learning_rate": 7.070372958951706e-05,
+      "loss": 0.0753,
+      "step": 3960
+    },
+    {
+      "epoch": 0.5391823984788809,
+      "grad_norm": 0.5155308842658997,
+      "learning_rate": 7.055311073264194e-05,
+      "loss": 0.064,
+      "step": 3970
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 1.2939543724060059,
+      "learning_rate": 7.040226711024077e-05,
+      "loss": 0.0365,
+      "step": 3980
+    },
+    {
+      "epoch": 0.5418986826022002,
+      "grad_norm": 1.216658592224121,
+      "learning_rate": 7.02512003719152e-05,
+      "loss": 0.0912,
+      "step": 3990
+    },
+    {
+      "epoch": 0.5432568246638598,
+      "grad_norm": 1.5436028242111206,
+      "learning_rate": 7.00999121697069e-05,
+      "loss": 0.0564,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5446149667255195,
+      "grad_norm": 0.7875412702560425,
+      "learning_rate": 6.99484041580794e-05,
+      "loss": 0.0714,
+      "step": 4010
+    },
+    {
+      "epoch": 0.5459731087871791,
+      "grad_norm": 1.1042916774749756,
+      "learning_rate": 6.979667799390004e-05,
+      "loss": 0.089,
+      "step": 4020
+    },
+    {
+      "epoch": 0.5473312508488388,
+      "grad_norm": 0.7690709233283997,
+      "learning_rate": 6.964473533642185e-05,
+      "loss": 0.0624,
+      "step": 4030
+    },
+    {
+      "epoch": 0.5486893929104985,
+      "grad_norm": 0.4494988024234772,
+      "learning_rate": 6.949257784726539e-05,
+      "loss": 0.0688,
+      "step": 4040
+    },
+    {
+      "epoch": 0.5500475349721581,
+      "grad_norm": 0.8128408789634705,
+      "learning_rate": 6.934020719040056e-05,
+      "loss": 0.0373,
+      "step": 4050
+    },
+    {
+      "epoch": 0.5514056770338177,
+      "grad_norm": 0.4633880853652954,
+      "learning_rate": 6.918762503212848e-05,
+      "loss": 0.0498,
+      "step": 4060
+    },
+    {
+      "epoch": 0.5527638190954773,
+      "grad_norm": 0.6356334686279297,
+      "learning_rate": 6.903483304106319e-05,
+      "loss": 0.0532,
+      "step": 4070
+    },
+    {
+      "epoch": 0.5541219611571371,
+      "grad_norm": 0.5694980025291443,
+      "learning_rate": 6.888183288811341e-05,
+      "loss": 0.0826,
+      "step": 4080
+    },
+    {
+      "epoch": 0.5554801032187967,
+      "grad_norm": 0.7099330425262451,
+      "learning_rate": 6.87286262464643e-05,
+      "loss": 0.0462,
+      "step": 4090
+    },
+    {
+      "epoch": 0.5568382452804563,
+      "grad_norm": 2.591799736022949,
+      "learning_rate": 6.857521479155915e-05,
+      "loss": 0.0725,
+      "step": 4100
+    },
+    {
+      "epoch": 0.558196387342116,
+      "grad_norm": 0.5880663394927979,
+      "learning_rate": 6.842160020108104e-05,
+      "loss": 0.058,
+      "step": 4110
+    },
+    {
+      "epoch": 0.5595545294037756,
+      "grad_norm": 1.146309733390808,
+      "learning_rate": 6.826778415493455e-05,
+      "loss": 0.0397,
+      "step": 4120
+    },
+    {
+      "epoch": 0.5609126714654353,
+      "grad_norm": 0.7618089914321899,
+      "learning_rate": 6.811376833522729e-05,
+      "loss": 0.0492,
+      "step": 4130
+    },
+    {
+      "epoch": 0.562270813527095,
+      "grad_norm": 1.4201722145080566,
+      "learning_rate": 6.795955442625159e-05,
+      "loss": 0.0524,
+      "step": 4140
+    },
+    {
+      "epoch": 0.5636289555887546,
+      "grad_norm": 1.026930809020996,
+      "learning_rate": 6.780514411446608e-05,
+      "loss": 0.0463,
+      "step": 4150
+    },
+    {
+      "epoch": 0.5649870976504142,
+      "grad_norm": 1.0731076002120972,
+      "learning_rate": 6.765053908847716e-05,
+      "loss": 0.0632,
+      "step": 4160
+    },
+    {
+      "epoch": 0.5663452397120738,
+      "grad_norm": 0.7732483744621277,
+      "learning_rate": 6.749574103902064e-05,
+      "loss": 0.0516,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5677033817737336,
+      "grad_norm": 0.9263116717338562,
+      "learning_rate": 6.734075165894317e-05,
+      "loss": 0.0807,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5690615238353932,
+      "grad_norm": 1.1348145008087158,
+      "learning_rate": 6.71855726431838e-05,
+      "loss": 0.0526,
+      "step": 4190
+    },
+    {
+      "epoch": 0.5704196658970528,
+      "grad_norm": 0.9622387886047363,
+      "learning_rate": 6.703020568875538e-05,
+      "loss": 0.0557,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5717778079587125,
+      "grad_norm": 0.9638462066650391,
+      "learning_rate": 6.687465249472603e-05,
+      "loss": 0.0651,
+      "step": 4210
+    },
+    {
+      "epoch": 0.5731359500203721,
+      "grad_norm": 0.6449846625328064,
+      "learning_rate": 6.671891476220055e-05,
+      "loss": 0.0638,
+      "step": 4220
+    },
+    {
+      "epoch": 0.5744940920820317,
+      "grad_norm": 0.9194169044494629,
+      "learning_rate": 6.656299419430183e-05,
+      "loss": 0.0541,
+      "step": 4230
+    },
+    {
+      "epoch": 0.5758522341436915,
+      "grad_norm": 0.9876245260238647,
+      "learning_rate": 6.640689249615223e-05,
+      "loss": 0.0579,
+      "step": 4240
+    },
+    {
+      "epoch": 0.5772103762053511,
+      "grad_norm": 0.7524198889732361,
+      "learning_rate": 6.625061137485491e-05,
+      "loss": 0.0408,
+      "step": 4250
+    },
+    {
+      "epoch": 0.5785685182670107,
+      "grad_norm": 0.6558626890182495,
+      "learning_rate": 6.609415253947517e-05,
+      "loss": 0.0504,
+      "step": 4260
+    },
+    {
+      "epoch": 0.5799266603286704,
+      "grad_norm": 1.788628339767456,
+      "learning_rate": 6.593751770102178e-05,
+      "loss": 0.0577,
+      "step": 4270
+    },
+    {
+      "epoch": 0.58128480239033,
+      "grad_norm": 0.5676286816596985,
+      "learning_rate": 6.578070857242823e-05,
+      "loss": 0.047,
+      "step": 4280
+    },
+    {
+      "epoch": 0.5826429444519897,
+      "grad_norm": 0.9517272710800171,
+      "learning_rate": 6.562372686853402e-05,
+      "loss": 0.0631,
+      "step": 4290
+    },
+    {
+      "epoch": 0.5840010865136493,
+      "grad_norm": 0.9986273050308228,
+      "learning_rate": 6.546657430606593e-05,
+      "loss": 0.0478,
+      "step": 4300
+    },
+    {
+      "epoch": 0.585359228575309,
+      "grad_norm": 0.9061160087585449,
+      "learning_rate": 6.530925260361918e-05,
+      "loss": 0.069,
+      "step": 4310
+    },
+    {
+      "epoch": 0.5867173706369686,
+      "grad_norm": 0.9209710955619812,
+      "learning_rate": 6.515176348163871e-05,
+      "loss": 0.0622,
+      "step": 4320
+    },
+    {
+      "epoch": 0.5880755126986282,
+      "grad_norm": 1.1966285705566406,
+      "learning_rate": 6.499410866240032e-05,
+      "loss": 0.0513,
+      "step": 4330
+    },
+    {
+      "epoch": 0.589433654760288,
+      "grad_norm": 1.3566375970840454,
+      "learning_rate": 6.48362898699919e-05,
+      "loss": 0.0452,
+      "step": 4340
+    },
+    {
+      "epoch": 0.5907917968219476,
+      "grad_norm": 1.0169988870620728,
+      "learning_rate": 6.467830883029443e-05,
+      "loss": 0.0818,
+      "step": 4350
+    },
+    {
+      "epoch": 0.5921499388836072,
+      "grad_norm": 0.78096604347229,
+      "learning_rate": 6.452016727096326e-05,
+      "loss": 0.0621,
+      "step": 4360
+    },
+    {
+      "epoch": 0.5935080809452669,
+      "grad_norm": 0.9473659992218018,
+      "learning_rate": 6.436186692140916e-05,
+      "loss": 0.0521,
+      "step": 4370
+    },
+    {
+      "epoch": 0.5948662230069265,
+      "grad_norm": 1.2153605222702026,
+      "learning_rate": 6.420340951277938e-05,
+      "loss": 0.0669,
+      "step": 4380
+    },
+    {
+      "epoch": 0.5962243650685862,
+      "grad_norm": 0.7906447052955627,
+      "learning_rate": 6.404479677793874e-05,
+      "loss": 0.0589,
+      "step": 4390
+    },
+    {
+      "epoch": 0.5975825071302459,
+      "grad_norm": 0.6616839170455933,
+      "learning_rate": 6.388603045145075e-05,
+      "loss": 0.0379,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5989406491919055,
+      "grad_norm": 0.7233092188835144,
+      "learning_rate": 6.372711226955843e-05,
+      "loss": 0.054,
+      "step": 4410
+    },
+    {
+      "epoch": 0.6002987912535651,
+      "grad_norm": 0.4105149209499359,
+      "learning_rate": 6.356804397016564e-05,
+      "loss": 0.0604,
+      "step": 4420
+    },
+    {
+      "epoch": 0.6016569333152247,
+      "grad_norm": 1.0449178218841553,
+      "learning_rate": 6.340882729281779e-05,
+      "loss": 0.0458,
+      "step": 4430
+    },
+    {
+      "epoch": 0.6030150753768844,
+      "grad_norm": 0.44361162185668945,
+      "learning_rate": 6.324946397868294e-05,
+      "loss": 0.0364,
+      "step": 4440
+    },
+    {
+      "epoch": 0.6043732174385441,
+      "grad_norm": 0.46793973445892334,
+      "learning_rate": 6.308995577053276e-05,
+      "loss": 0.0468,
+      "step": 4450
+    },
+    {
+      "epoch": 0.6057313595002037,
+      "grad_norm": 0.4983607232570648,
+      "learning_rate": 6.293030441272347e-05,
+      "loss": 0.0325,
+      "step": 4460
+    },
+    {
+      "epoch": 0.6070895015618634,
+      "grad_norm": 0.7550007104873657,
+      "learning_rate": 6.277051165117677e-05,
+      "loss": 0.0549,
+      "step": 4470
+    },
+    {
+      "epoch": 0.608447643623523,
+      "grad_norm": 1.0227235555648804,
+      "learning_rate": 6.261057923336064e-05,
+      "loss": 0.044,
+      "step": 4480
+    },
+    {
+      "epoch": 0.6098057856851826,
+      "grad_norm": 0.8695225715637207,
+      "learning_rate": 6.245050890827042e-05,
+      "loss": 0.0529,
+      "step": 4490
+    },
+    {
+      "epoch": 0.6111639277468424,
+      "grad_norm": 0.6711282730102539,
+      "learning_rate": 6.229030242640952e-05,
+      "loss": 0.0628,
+      "step": 4500
+    },
+    {
+      "epoch": 0.612522069808502,
+      "grad_norm": 0.9320592284202576,
+      "learning_rate": 6.212996153977037e-05,
+      "loss": 0.0616,
+      "step": 4510
+    },
+    {
+      "epoch": 0.6138802118701616,
+      "grad_norm": 0.9257829189300537,
+      "learning_rate": 6.196948800181523e-05,
+      "loss": 0.0435,
+      "step": 4520
+    },
+    {
+      "epoch": 0.6152383539318212,
+      "grad_norm": 1.133319616317749,
+      "learning_rate": 6.180888356745695e-05,
+      "loss": 0.0484,
+      "step": 4530
+    },
+    {
+      "epoch": 0.6165964959934809,
+      "grad_norm": 0.7320436239242554,
+      "learning_rate": 6.164814999303995e-05,
+      "loss": 0.0444,
+      "step": 4540
+    },
+    {
+      "epoch": 0.6179546380551406,
+      "grad_norm": 0.8527611494064331,
+      "learning_rate": 6.148728903632081e-05,
+      "loss": 0.0414,
+      "step": 4550
+    },
+    {
+      "epoch": 0.6193127801168002,
+      "grad_norm": 1.1138105392456055,
+      "learning_rate": 6.132630245644921e-05,
+      "loss": 0.0472,
+      "step": 4560
+    },
+    {
+      "epoch": 0.6206709221784599,
+      "grad_norm": 1.063890814781189,
+      "learning_rate": 6.116519201394857e-05,
+      "loss": 0.0754,
+      "step": 4570
+    },
+    {
+      "epoch": 0.6220290642401195,
+      "grad_norm": 0.8675426244735718,
+      "learning_rate": 6.10039594706969e-05,
+      "loss": 0.0605,
+      "step": 4580
+    },
+    {
+      "epoch": 0.6233872063017791,
+      "grad_norm": 0.6923426985740662,
+      "learning_rate": 6.084260658990744e-05,
+      "loss": 0.0439,
+      "step": 4590
+    },
+    {
+      "epoch": 0.6247453483634389,
+      "grad_norm": 0.759075403213501,
+      "learning_rate": 6.068113513610943e-05,
+      "loss": 0.0414,
+      "step": 4600
+    },
+    {
+      "epoch": 0.6261034904250985,
+      "grad_norm": 0.497366726398468,
+      "learning_rate": 6.0519546875128876e-05,
+      "loss": 0.0539,
+      "step": 4610
+    },
+    {
+      "epoch": 0.6274616324867581,
+      "grad_norm": 0.5042897462844849,
+      "learning_rate": 6.035784357406906e-05,
+      "loss": 0.0462,
+      "step": 4620
+    },
+    {
+      "epoch": 0.6288197745484178,
+      "grad_norm": 0.7236697673797607,
+      "learning_rate": 6.01960270012914e-05,
+      "loss": 0.0326,
+      "step": 4630
+    },
+    {
+      "epoch": 0.6301779166100774,
+      "grad_norm": 0.5361550450325012,
+      "learning_rate": 6.003409892639599e-05,
+      "loss": 0.0233,
+      "step": 4640
+    },
+    {
+      "epoch": 0.631536058671737,
+      "grad_norm": 1.1324068307876587,
+      "learning_rate": 5.9872061120202336e-05,
+      "loss": 0.0752,
+      "step": 4650
+    },
+    {
+      "epoch": 0.6328942007333968,
+      "grad_norm": 0.7560549974441528,
+      "learning_rate": 5.9709915354729914e-05,
+      "loss": 0.0596,
+      "step": 4660
+    },
+    {
+      "epoch": 0.6342523427950564,
+      "grad_norm": 0.9377226829528809,
+      "learning_rate": 5.9547663403178824e-05,
+      "loss": 0.0618,
+      "step": 4670
+    },
+    {
+      "epoch": 0.635610484856716,
+      "grad_norm": 0.7963283061981201,
+      "learning_rate": 5.9385307039910445e-05,
+      "loss": 0.0408,
+      "step": 4680
+    },
+    {
+      "epoch": 0.6369686269183756,
+      "grad_norm": 0.5524522066116333,
+      "learning_rate": 5.922284804042792e-05,
+      "loss": 0.0401,
+      "step": 4690
+    },
+    {
+      "epoch": 0.6383267689800353,
+      "grad_norm": 0.6885665059089661,
+      "learning_rate": 5.906028818135687e-05,
+      "loss": 0.0422,
+      "step": 4700
+    },
+    {
+      "epoch": 0.639684911041695,
+      "grad_norm": 1.0520758628845215,
+      "learning_rate": 5.889762924042585e-05,
+      "loss": 0.05,
+      "step": 4710
+    },
+    {
+      "epoch": 0.6410430531033546,
+      "grad_norm": 1.1661704778671265,
+      "learning_rate": 5.873487299644699e-05,
+      "loss": 0.0575,
+      "step": 4720
+    },
+    {
+      "epoch": 0.6424011951650143,
+      "grad_norm": 0.5852257013320923,
+      "learning_rate": 5.857202122929649e-05,
+      "loss": 0.0643,
+      "step": 4730
+    },
+    {
+      "epoch": 0.6437593372266739,
+      "grad_norm": 0.5681099891662598,
+      "learning_rate": 5.840907571989518e-05,
+      "loss": 0.0451,
+      "step": 4740
+    },
+    {
+      "epoch": 0.6451174792883335,
+      "grad_norm": 0.6470234394073486,
+      "learning_rate": 5.824603825018904e-05,
+      "loss": 0.0414,
+      "step": 4750
+    },
+    {
+      "epoch": 0.6464756213499933,
+      "grad_norm": 0.7073606252670288,
+      "learning_rate": 5.808291060312975e-05,
+      "loss": 0.0617,
+      "step": 4760
+    },
+    {
+      "epoch": 0.6478337634116529,
+      "grad_norm": 0.4726971983909607,
+      "learning_rate": 5.7919694562655083e-05,
+      "loss": 0.0742,
+      "step": 4770
+    },
+    {
+      "epoch": 0.6491919054733125,
+      "grad_norm": 0.6359230279922485,
+      "learning_rate": 5.775639191366954e-05,
+      "loss": 0.063,
+      "step": 4780
+    },
+    {
+      "epoch": 0.6505500475349721,
+      "grad_norm": 1.0476025342941284,
+      "learning_rate": 5.75930044420247e-05,
+      "loss": 0.0587,
+      "step": 4790
+    },
+    {
+      "epoch": 0.6519081895966318,
+      "grad_norm": 0.8771790862083435,
+      "learning_rate": 5.74295339344998e-05,
+      "loss": 0.0475,
+      "step": 4800
+    },
+    {
+      "epoch": 0.6532663316582915,
+      "grad_norm": 0.7339069247245789,
+      "learning_rate": 5.726598217878211e-05,
+      "loss": 0.0402,
+      "step": 4810
+    },
+    {
+      "epoch": 0.6546244737199511,
+      "grad_norm": 0.8165606260299683,
+      "learning_rate": 5.71023509634474e-05,
+      "loss": 0.0469,
+      "step": 4820
+    },
+    {
+      "epoch": 0.6559826157816108,
+      "grad_norm": 0.8948976397514343,
+      "learning_rate": 5.693864207794049e-05,
+      "loss": 0.0726,
+      "step": 4830
+    },
+    {
+      "epoch": 0.6573407578432704,
+      "grad_norm": 0.7649142742156982,
+      "learning_rate": 5.677485731255545e-05,
+      "loss": 0.0483,
+      "step": 4840
+    },
+    {
+      "epoch": 0.65869889990493,
+      "grad_norm": 1.0208269357681274,
+      "learning_rate": 5.6610998458416296e-05,
+      "loss": 0.0447,
+      "step": 4850
+    },
+    {
+      "epoch": 0.6600570419665897,
+      "grad_norm": 0.8817300200462341,
+      "learning_rate": 5.644706730745716e-05,
+      "loss": 0.0523,
+      "step": 4860
+    },
+    {
+      "epoch": 0.6614151840282494,
+      "grad_norm": 0.9240930676460266,
+      "learning_rate": 5.628306565240287e-05,
+      "loss": 0.0516,
+      "step": 4870
+    },
+    {
+      "epoch": 0.662773326089909,
+      "grad_norm": 1.2007157802581787,
+      "learning_rate": 5.611899528674923e-05,
+      "loss": 0.0397,
+      "step": 4880
+    },
+    {
+      "epoch": 0.6641314681515686,
+      "grad_norm": 0.4762249290943146,
+      "learning_rate": 5.595485800474349e-05,
+      "loss": 0.0416,
+      "step": 4890
+    },
+    {
+      "epoch": 0.6654896102132283,
+      "grad_norm": 1.1518303155899048,
+      "learning_rate": 5.579065560136467e-05,
+      "loss": 0.057,
+      "step": 4900
+    },
+    {
+      "epoch": 0.6668477522748879,
+      "grad_norm": 0.5936382412910461,
+      "learning_rate": 5.562638987230392e-05,
+      "loss": 0.0463,
+      "step": 4910
+    },
+    {
+      "epoch": 0.6682058943365476,
+      "grad_norm": 1.4759844541549683,
+      "learning_rate": 5.546206261394498e-05,
+      "loss": 0.0377,
+      "step": 4920
+    },
+    {
+      "epoch": 0.6695640363982073,
+      "grad_norm": 0.6873993873596191,
+      "learning_rate": 5.529767562334437e-05,
+      "loss": 0.0674,
+      "step": 4930
+    },
+    {
+      "epoch": 0.6709221784598669,
+      "grad_norm": 0.45289376378059387,
+      "learning_rate": 5.5133230698211926e-05,
+      "loss": 0.0413,
+      "step": 4940
+    },
+    {
+      "epoch": 0.6722803205215265,
+      "grad_norm": 0.46766775846481323,
+      "learning_rate": 5.496872963689096e-05,
+      "loss": 0.0839,
+      "step": 4950
+    },
+    {
+      "epoch": 0.6736384625831862,
+      "grad_norm": 0.6315058469772339,
+      "learning_rate": 5.4804174238338756e-05,
+      "loss": 0.0478,
+      "step": 4960
+    },
+    {
+      "epoch": 0.6749966046448459,
+      "grad_norm": 0.7202677726745605,
+      "learning_rate": 5.463956630210678e-05,
+      "loss": 0.0654,
+      "step": 4970
+    },
+    {
+      "epoch": 0.6763547467065055,
+      "grad_norm": 0.6324037909507751,
+      "learning_rate": 5.4474907628321046e-05,
+      "loss": 0.0388,
+      "step": 4980
+    },
+    {
+      "epoch": 0.6777128887681652,
+      "grad_norm": 0.6857421398162842,
+      "learning_rate": 5.431020001766244e-05,
+      "loss": 0.0541,
+      "step": 4990
+    },
+    {
+      "epoch": 0.6790710308298248,
+      "grad_norm": 0.8715236783027649,
+      "learning_rate": 5.4145445271346986e-05,
+      "loss": 0.0312,
+      "step": 5000
+    },
+    {
+      "epoch": 0.6804291728914844,
+      "grad_norm": 0.5386717915534973,
+      "learning_rate": 5.398064519110622e-05,
+      "loss": 0.0397,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6817873149531442,
+      "grad_norm": 0.6696141362190247,
+      "learning_rate": 5.3815801579167394e-05,
+      "loss": 0.0335,
+      "step": 5020
+    },
+    {
+      "epoch": 0.6831454570148038,
+      "grad_norm": 0.8183315396308899,
+      "learning_rate": 5.365091623823382e-05,
+      "loss": 0.0359,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6845035990764634,
+      "grad_norm": 1.2229604721069336,
+      "learning_rate": 5.348599097146521e-05,
+      "loss": 0.0368,
+      "step": 5040
+    },
+    {
+      "epoch": 0.685861741138123,
+      "grad_norm": 0.6841098666191101,
+      "learning_rate": 5.3321027582457836e-05,
+      "loss": 0.0386,
+      "step": 5050
+    },
+    {
+      "epoch": 0.6872198831997827,
+      "grad_norm": 0.592751681804657,
+      "learning_rate": 5.315602787522491e-05,
+      "loss": 0.0339,
+      "step": 5060
+    },
+    {
+      "epoch": 0.6885780252614423,
+      "grad_norm": 1.0147932767868042,
+      "learning_rate": 5.299099365417678e-05,
+      "loss": 0.0439,
+      "step": 5070
+    },
+    {
+      "epoch": 0.689936167323102,
+      "grad_norm": 0.46367761492729187,
+      "learning_rate": 5.2825926724101236e-05,
+      "loss": 0.0384,
+      "step": 5080
+    },
+    {
+      "epoch": 0.6912943093847617,
+      "grad_norm": 0.6336700916290283,
+      "learning_rate": 5.26608288901438e-05,
+      "loss": 0.0486,
+      "step": 5090
+    },
+    {
+      "epoch": 0.6926524514464213,
+      "grad_norm": 0.48966488242149353,
+      "learning_rate": 5.24957019577879e-05,
+      "loss": 0.0306,
+      "step": 5100
+    },
+    {
+      "epoch": 0.6940105935080809,
+      "grad_norm": 0.9978145360946655,
+      "learning_rate": 5.2330547732835266e-05,
+      "loss": 0.0571,
+      "step": 5110
+    },
+    {
+      "epoch": 0.6953687355697405,
+      "grad_norm": 1.1210969686508179,
+      "learning_rate": 5.2165368021385996e-05,
+      "loss": 0.0558,
+      "step": 5120
+    },
+    {
+      "epoch": 0.6967268776314003,
+      "grad_norm": 1.4904037714004517,
+      "learning_rate": 5.200016462981897e-05,
+      "loss": 0.041,
+      "step": 5130
+    },
+    {
+      "epoch": 0.6980850196930599,
+      "grad_norm": 1.9081432819366455,
+      "learning_rate": 5.1834939364772015e-05,
+      "loss": 0.0623,
+      "step": 5140
+    },
+    {
+      "epoch": 0.6994431617547195,
+      "grad_norm": 0.6883187890052795,
+      "learning_rate": 5.166969403312214e-05,
+      "loss": 0.0607,
+      "step": 5150
+    },
+    {
+      "epoch": 0.7008013038163792,
+      "grad_norm": 1.1940135955810547,
+      "learning_rate": 5.1504430441965844e-05,
+      "loss": 0.0556,
+      "step": 5160
+    },
+    {
+      "epoch": 0.7021594458780388,
+      "grad_norm": 1.034138798713684,
+      "learning_rate": 5.133915039859923e-05,
+      "loss": 0.052,
+      "step": 5170
+    },
+    {
+      "epoch": 0.7035175879396985,
+      "grad_norm": 1.4998527765274048,
+      "learning_rate": 5.1173855710498444e-05,
+      "loss": 0.0431,
+      "step": 5180
+    },
+    {
+      "epoch": 0.7048757300013582,
+      "grad_norm": 0.9458922147750854,
+      "learning_rate": 5.100854818529967e-05,
+      "loss": 0.0612,
+      "step": 5190
+    },
+    {
+      "epoch": 0.7062338720630178,
+      "grad_norm": 1.2267506122589111,
+      "learning_rate": 5.084322963077951e-05,
+      "loss": 0.0351,
+      "step": 5200
+    },
+    {
+      "epoch": 0.7075920141246774,
+      "grad_norm": 0.5052388906478882,
+      "learning_rate": 5.067790185483522e-05,
+      "loss": 0.0432,
+      "step": 5210
+    },
+    {
+      "epoch": 0.708950156186337,
+      "grad_norm": 0.758340060710907,
+      "learning_rate": 5.0512566665464844e-05,
+      "loss": 0.0505,
+      "step": 5220
+    },
+    {
+      "epoch": 0.7103082982479968,
+      "grad_norm": 0.7887386083602905,
+      "learning_rate": 5.034722587074755e-05,
+      "loss": 0.0367,
+      "step": 5230
+    },
+    {
+      "epoch": 0.7116664403096564,
+      "grad_norm": 0.6212475299835205,
+      "learning_rate": 5.018188127882375e-05,
+      "loss": 0.0331,
+      "step": 5240
+    },
+    {
+      "epoch": 0.713024582371316,
+      "grad_norm": 0.5125031471252441,
+      "learning_rate": 5.0016534697875417e-05,
+      "loss": 0.0247,
+      "step": 5250
+    },
+    {
+      "epoch": 0.7143827244329757,
+      "grad_norm": 0.6737174987792969,
+      "learning_rate": 4.9851187936106294e-05,
+      "loss": 0.0399,
+      "step": 5260
+    },
+    {
+      "epoch": 0.7157408664946353,
+      "grad_norm": 0.6427016854286194,
+      "learning_rate": 4.968584280172206e-05,
+      "loss": 0.0579,
+      "step": 5270
+    },
+    {
+      "epoch": 0.7170990085562949,
+      "grad_norm": 0.829805850982666,
+      "learning_rate": 4.95205011029106e-05,
+      "loss": 0.0616,
+      "step": 5280
+    },
+    {
+      "epoch": 0.7184571506179547,
+      "grad_norm": 1.0251703262329102,
+      "learning_rate": 4.935516464782227e-05,
+      "loss": 0.0385,
+      "step": 5290
+    },
+    {
+      "epoch": 0.7198152926796143,
+      "grad_norm": 0.7390406131744385,
+      "learning_rate": 4.918983524455003e-05,
+      "loss": 0.041,
+      "step": 5300
+    },
+    {
+      "epoch": 0.7211734347412739,
+      "grad_norm": 0.5993945002555847,
+      "learning_rate": 4.9024514701109766e-05,
+      "loss": 0.05,
+      "step": 5310
+    },
+    {
+      "epoch": 0.7225315768029336,
+      "grad_norm": 1.0277793407440186,
+      "learning_rate": 4.885920482542043e-05,
+      "loss": 0.0536,
+      "step": 5320
+    },
+    {
+      "epoch": 0.7238897188645932,
+      "grad_norm": 1.0288736820220947,
+      "learning_rate": 4.869390742528438e-05,
+      "loss": 0.065,
+      "step": 5330
+    },
+    {
+      "epoch": 0.7252478609262529,
+      "grad_norm": 0.7350568771362305,
+      "learning_rate": 4.852862430836744e-05,
+      "loss": 0.062,
+      "step": 5340
+    },
+    {
+      "epoch": 0.7266060029879126,
+      "grad_norm": 1.1282691955566406,
+      "learning_rate": 4.836335728217933e-05,
+      "loss": 0.0567,
+      "step": 5350
+    },
+    {
+      "epoch": 0.7279641450495722,
+      "grad_norm": 0.8829731345176697,
+      "learning_rate": 4.819810815405379e-05,
+      "loss": 0.074,
+      "step": 5360
+    },
+    {
+      "epoch": 0.7293222871112318,
+      "grad_norm": 0.8731906414031982,
+      "learning_rate": 4.803287873112877e-05,
+      "loss": 0.0531,
+      "step": 5370
+    },
+    {
+      "epoch": 0.7306804291728914,
+      "grad_norm": 0.48923322558403015,
+      "learning_rate": 4.786767082032681e-05,
+      "loss": 0.0288,
+      "step": 5380
+    },
+    {
+      "epoch": 0.7320385712345512,
+      "grad_norm": 0.9502625465393066,
+      "learning_rate": 4.77024862283351e-05,
+      "loss": 0.0397,
+      "step": 5390
+    },
+    {
+      "epoch": 0.7333967132962108,
+      "grad_norm": 0.7046083807945251,
+      "learning_rate": 4.753732676158593e-05,
+      "loss": 0.0387,
+      "step": 5400
+    },
+    {
+      "epoch": 0.7347548553578704,
+      "grad_norm": 0.7452012896537781,
+      "learning_rate": 4.737219422623672e-05,
+      "loss": 0.041,
+      "step": 5410
+    },
+    {
+      "epoch": 0.7361129974195301,
+      "grad_norm": 0.7140015363693237,
+      "learning_rate": 4.720709042815044e-05,
+      "loss": 0.0353,
+      "step": 5420
+    },
+    {
+      "epoch": 0.7374711394811897,
+      "grad_norm": 0.9651252627372742,
+      "learning_rate": 4.704201717287578e-05,
+      "loss": 0.0438,
+      "step": 5430
+    },
+    {
+      "epoch": 0.7388292815428494,
+      "grad_norm": 0.6355652213096619,
+      "learning_rate": 4.6876976265627404e-05,
+      "loss": 0.0755,
+      "step": 5440
+    },
+    {
+      "epoch": 0.7401874236045091,
+      "grad_norm": 0.9455762505531311,
+      "learning_rate": 4.671196951126626e-05,
+      "loss": 0.0423,
+      "step": 5450
+    },
+    {
+      "epoch": 0.7415455656661687,
+      "grad_norm": 0.8203608393669128,
+      "learning_rate": 4.654699871427971e-05,
+      "loss": 0.045,
+      "step": 5460
+    },
+    {
+      "epoch": 0.7429037077278283,
+      "grad_norm": 0.6200745105743408,
+      "learning_rate": 4.6382065678762034e-05,
+      "loss": 0.0361,
+      "step": 5470
+    },
+    {
+      "epoch": 0.744261849789488,
+      "grad_norm": 1.0056647062301636,
+      "learning_rate": 4.6217172208394424e-05,
+      "loss": 0.0579,
+      "step": 5480
+    },
+    {
+      "epoch": 0.7456199918511476,
+      "grad_norm": 0.7042694687843323,
+      "learning_rate": 4.605232010642549e-05,
+      "loss": 0.0408,
+      "step": 5490
+    },
+    {
+      "epoch": 0.7469781339128073,
+      "grad_norm": 1.097289800643921,
+      "learning_rate": 4.588751117565142e-05,
+      "loss": 0.0477,
+      "step": 5500
+    },
+    {
+      "epoch": 0.748336275974467,
+      "grad_norm": 0.8351175785064697,
+      "learning_rate": 4.5722747218396214e-05,
+      "loss": 0.0376,
+      "step": 5510
+    },
+    {
+      "epoch": 0.7496944180361266,
+      "grad_norm": 0.7338027358055115,
+      "learning_rate": 4.5558030036492194e-05,
+      "loss": 0.0577,
+      "step": 5520
+    },
+    {
+      "epoch": 0.7510525600977862,
+      "grad_norm": 0.32611626386642456,
+      "learning_rate": 4.539336143125999e-05,
+      "loss": 0.0397,
+      "step": 5530
+    },
+    {
+      "epoch": 0.7524107021594458,
+      "grad_norm": 1.1413090229034424,
+      "learning_rate": 4.522874320348916e-05,
+      "loss": 0.0515,
+      "step": 5540
+    },
+    {
+      "epoch": 0.7537688442211056,
+      "grad_norm": 1.0175182819366455,
+      "learning_rate": 4.506417715341821e-05,
+      "loss": 0.0296,
+      "step": 5550
+    },
+    {
+      "epoch": 0.7551269862827652,
+      "grad_norm": 0.6192682385444641,
+      "learning_rate": 4.489966508071511e-05,
+      "loss": 0.0596,
+      "step": 5560
+    },
+    {
+      "epoch": 0.7564851283444248,
+      "grad_norm": 0.8577454686164856,
+      "learning_rate": 4.4735208784457575e-05,
+      "loss": 0.0533,
+      "step": 5570
+    },
+    {
+      "epoch": 0.7578432704060845,
+      "grad_norm": 1.3309111595153809,
+      "learning_rate": 4.457081006311325e-05,
+      "loss": 0.0587,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7592014124677441,
+      "grad_norm": 0.9071065783500671,
+      "learning_rate": 4.440647071452027e-05,
+      "loss": 0.0365,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7605595545294038,
+      "grad_norm": 0.8144100308418274,
+      "learning_rate": 4.424219253586737e-05,
+      "loss": 0.03,
+      "step": 5600
+    },
+    {
+      "epoch": 0.7619176965910635,
+      "grad_norm": 0.8778440952301025,
+      "learning_rate": 4.407797732367443e-05,
+      "loss": 0.0411,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7632758386527231,
+      "grad_norm": 0.8001221418380737,
+      "learning_rate": 4.391382687377268e-05,
+      "loss": 0.0754,
+      "step": 5620
+    },
+    {
+      "epoch": 0.7646339807143827,
+      "grad_norm": 0.5695180892944336,
+      "learning_rate": 4.374974298128512e-05,
+      "loss": 0.0588,
+      "step": 5630
+    },
+    {
+      "epoch": 0.7659921227760423,
+      "grad_norm": 1.0194611549377441,
+      "learning_rate": 4.358572744060699e-05,
+      "loss": 0.0594,
+      "step": 5640
+    },
+    {
+      "epoch": 0.7673502648377021,
+      "grad_norm": 1.3697640895843506,
+      "learning_rate": 4.342178204538588e-05,
+      "loss": 0.0522,
+      "step": 5650
+    },
+    {
+      "epoch": 0.7687084068993617,
+      "grad_norm": 1.3831275701522827,
+      "learning_rate": 4.325790858850241e-05,
+      "loss": 0.0464,
+      "step": 5660
+    },
+    {
+      "epoch": 0.7700665489610213,
+      "grad_norm": 1.1755640506744385,
+      "learning_rate": 4.309410886205043e-05,
+      "loss": 0.054,
+      "step": 5670
+    },
+    {
+      "epoch": 0.771424691022681,
+      "grad_norm": 0.6365404725074768,
+      "learning_rate": 4.293038465731752e-05,
+      "loss": 0.0568,
+      "step": 5680
+    },
+    {
+      "epoch": 0.7727828330843406,
+      "grad_norm": 0.8792571425437927,
+      "learning_rate": 4.276673776476533e-05,
+      "loss": 0.0664,
+      "step": 5690
+    },
+    {
+      "epoch": 0.7741409751460002,
+      "grad_norm": 0.7772958874702454,
+      "learning_rate": 4.260316997401007e-05,
+      "loss": 0.0402,
+      "step": 5700
+    },
+    {
+      "epoch": 0.77549911720766,
+      "grad_norm": 0.7393807172775269,
+      "learning_rate": 4.243968307380293e-05,
+      "loss": 0.0344,
+      "step": 5710
+    },
+    {
+      "epoch": 0.7768572592693196,
+      "grad_norm": 0.8659506440162659,
+      "learning_rate": 4.22762788520104e-05,
+      "loss": 0.0542,
+      "step": 5720
+    },
+    {
+      "epoch": 0.7782154013309792,
+      "grad_norm": 1.5391348600387573,
+      "learning_rate": 4.211295909559491e-05,
+      "loss": 0.0462,
+      "step": 5730
+    },
+    {
+      "epoch": 0.7795735433926388,
+      "grad_norm": 0.7516659498214722,
+      "learning_rate": 4.194972559059511e-05,
+      "loss": 0.0436,
+      "step": 5740
+    },
+    {
+      "epoch": 0.7809316854542985,
+      "grad_norm": 0.9339722990989685,
+      "learning_rate": 4.178658012210651e-05,
+      "loss": 0.0395,
+      "step": 5750
+    },
+    {
+      "epoch": 0.7822898275159582,
+      "grad_norm": 0.9163076281547546,
+      "learning_rate": 4.162352447426177e-05,
+      "loss": 0.0326,
+      "step": 5760
+    },
+    {
+      "epoch": 0.7836479695776178,
+      "grad_norm": 0.9076511859893799,
+      "learning_rate": 4.146056043021135e-05,
+      "loss": 0.0439,
+      "step": 5770
+    },
+    {
+      "epoch": 0.7850061116392775,
+      "grad_norm": 0.4349239468574524,
+      "learning_rate": 4.1297689772103944e-05,
+      "loss": 0.0388,
+      "step": 5780
+    },
+    {
+      "epoch": 0.7863642537009371,
+      "grad_norm": 0.5277696251869202,
+      "learning_rate": 4.113491428106694e-05,
+      "loss": 0.0365,
+      "step": 5790
+    },
+    {
+      "epoch": 0.7877223957625967,
+      "grad_norm": 0.6830906271934509,
+      "learning_rate": 4.0972235737187055e-05,
+      "loss": 0.0618,
+      "step": 5800
+    },
+    {
+      "epoch": 0.7890805378242565,
+      "grad_norm": 0.6524956822395325,
+      "learning_rate": 4.080965591949076e-05,
+      "loss": 0.0477,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7904386798859161,
+      "grad_norm": 2.3854546546936035,
+      "learning_rate": 4.0647176605924924e-05,
+      "loss": 0.0486,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7917968219475757,
+      "grad_norm": 0.35581135749816895,
+      "learning_rate": 4.0484799573337255e-05,
+      "loss": 0.0512,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7931549640092354,
+      "grad_norm": 0.7135029435157776,
+      "learning_rate": 4.032252659745699e-05,
+      "loss": 0.0454,
+      "step": 5840
+    },
+    {
+      "epoch": 0.794513106070895,
+      "grad_norm": 0.4069244861602783,
+      "learning_rate": 4.016035945287539e-05,
+      "loss": 0.048,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7958712481325547,
+      "grad_norm": 0.7290052771568298,
+      "learning_rate": 3.999829991302635e-05,
+      "loss": 0.0441,
+      "step": 5860
+    },
+    {
+      "epoch": 0.7972293901942143,
+      "grad_norm": 0.9676533937454224,
+      "learning_rate": 3.983634975016707e-05,
+      "loss": 0.0311,
+      "step": 5870
+    },
+    {
+      "epoch": 0.798587532255874,
+      "grad_norm": 0.6402451395988464,
+      "learning_rate": 3.967451073535854e-05,
+      "loss": 0.0471,
+      "step": 5880
+    },
+    {
+      "epoch": 0.7999456743175336,
+      "grad_norm": 0.9151315093040466,
+      "learning_rate": 3.951278463844633e-05,
+      "loss": 0.0365,
+      "step": 5890
+    },
+    {
+      "epoch": 0.8013038163791932,
+      "grad_norm": 0.8081924319267273,
+      "learning_rate": 3.935117322804111e-05,
+      "loss": 0.0425,
+      "step": 5900
+    },
+    {
+      "epoch": 0.8026619584408529,
+      "grad_norm": 0.5337335467338562,
+      "learning_rate": 3.918967827149938e-05,
+      "loss": 0.0574,
+      "step": 5910
+    },
+    {
+      "epoch": 0.8040201005025126,
+      "grad_norm": 0.7247133255004883,
+      "learning_rate": 3.9028301534904094e-05,
+      "loss": 0.0326,
+      "step": 5920
+    },
+    {
+      "epoch": 0.8053782425641722,
+      "grad_norm": 0.6503797769546509,
+      "learning_rate": 3.88670447830454e-05,
+      "loss": 0.028,
+      "step": 5930
+    },
+    {
+      "epoch": 0.8067363846258319,
+      "grad_norm": 0.6459348201751709,
+      "learning_rate": 3.870590977940132e-05,
+      "loss": 0.0367,
+      "step": 5940
+    },
+    {
+      "epoch": 0.8080945266874915,
+      "grad_norm": 0.5170432925224304,
+      "learning_rate": 3.8544898286118404e-05,
+      "loss": 0.0413,
+      "step": 5950
+    },
+    {
+      "epoch": 0.8094526687491511,
+      "grad_norm": 0.6468885540962219,
+      "learning_rate": 3.838401206399257e-05,
+      "loss": 0.044,
+      "step": 5960
+    },
+    {
+      "epoch": 0.8108108108108109,
+      "grad_norm": 0.7732337713241577,
+      "learning_rate": 3.822325287244975e-05,
+      "loss": 0.0362,
+      "step": 5970
+    },
+    {
+      "epoch": 0.8121689528724705,
+      "grad_norm": 1.1559529304504395,
+      "learning_rate": 3.8062622469526725e-05,
+      "loss": 0.0704,
+      "step": 5980
+    },
+    {
+      "epoch": 0.8135270949341301,
+      "grad_norm": 0.877435028553009,
+      "learning_rate": 3.790212261185183e-05,
+      "loss": 0.0634,
+      "step": 5990
+    },
+    {
+      "epoch": 0.8148852369957897,
+      "grad_norm": 0.6502079367637634,
+      "learning_rate": 3.7741755054625794e-05,
+      "loss": 0.0279,
+      "step": 6000
+    },
+    {
+      "epoch": 0.8162433790574494,
+      "grad_norm": 0.43969351053237915,
+      "learning_rate": 3.758152155160255e-05,
+      "loss": 0.0493,
+      "step": 6010
+    },
+    {
+      "epoch": 0.8176015211191091,
+      "grad_norm": 0.5557388663291931,
+      "learning_rate": 3.742142385506999e-05,
+      "loss": 0.0293,
+      "step": 6020
+    },
+    {
+      "epoch": 0.8189596631807687,
+      "grad_norm": 0.5442578792572021,
+      "learning_rate": 3.72614637158309e-05,
+      "loss": 0.0388,
+      "step": 6030
+    },
+    {
+      "epoch": 0.8203178052424284,
+      "grad_norm": 0.6480793952941895,
+      "learning_rate": 3.710164288318371e-05,
+      "loss": 0.0183,
+      "step": 6040
+    },
+    {
+      "epoch": 0.821675947304088,
+      "grad_norm": 0.647675633430481,
+      "learning_rate": 3.694196310490345e-05,
+      "loss": 0.0428,
+      "step": 6050
+    },
+    {
+      "epoch": 0.8230340893657476,
+      "grad_norm": 0.7709089517593384,
+      "learning_rate": 3.678242612722259e-05,
+      "loss": 0.0283,
+      "step": 6060
+    },
+    {
+      "epoch": 0.8243922314274073,
+      "grad_norm": 0.6709498763084412,
+      "learning_rate": 3.6623033694811953e-05,
+      "loss": 0.0334,
+      "step": 6070
+    },
+    {
+      "epoch": 0.825750373489067,
+      "grad_norm": 0.8446496725082397,
+      "learning_rate": 3.6463787550761665e-05,
+      "loss": 0.0494,
+      "step": 6080
+    },
+    {
+      "epoch": 0.8271085155507266,
+      "grad_norm": 0.8736713528633118,
+      "learning_rate": 3.630468943656202e-05,
+      "loss": 0.0346,
+      "step": 6090
+    },
+    {
+      "epoch": 0.8284666576123862,
+      "grad_norm": 0.7283543348312378,
+      "learning_rate": 3.6145741092084523e-05,
+      "loss": 0.045,
+      "step": 6100
+    },
+    {
+      "epoch": 0.8298247996740459,
+      "grad_norm": 0.940607488155365,
+      "learning_rate": 3.598694425556278e-05,
+      "loss": 0.0468,
+      "step": 6110
+    },
+    {
+      "epoch": 0.8311829417357055,
+      "grad_norm": 0.8215283155441284,
+      "learning_rate": 3.58283006635736e-05,
+      "loss": 0.0379,
+      "step": 6120
+    },
+    {
+      "epoch": 0.8325410837973652,
+      "grad_norm": 1.0448235273361206,
+      "learning_rate": 3.566981205101781e-05,
+      "loss": 0.0306,
+      "step": 6130
+    },
+    {
+      "epoch": 0.8338992258590249,
+      "grad_norm": 0.9097661972045898,
+      "learning_rate": 3.5511480151101556e-05,
+      "loss": 0.0369,
+      "step": 6140
+    },
+    {
+      "epoch": 0.8352573679206845,
+      "grad_norm": 0.5590543746948242,
+      "learning_rate": 3.5353306695317104e-05,
+      "loss": 0.0326,
+      "step": 6150
+    },
+    {
+      "epoch": 0.8366155099823441,
+      "grad_norm": 0.3810138702392578,
+      "learning_rate": 3.519529341342402e-05,
+      "loss": 0.0547,
+      "step": 6160
+    },
+    {
+      "epoch": 0.8379736520440038,
+      "grad_norm": 0.8633129000663757,
+      "learning_rate": 3.503744203343026e-05,
+      "loss": 0.0422,
+      "step": 6170
+    },
+    {
+      "epoch": 0.8393317941056635,
+      "grad_norm": 0.6816096901893616,
+      "learning_rate": 3.487975428157318e-05,
+      "loss": 0.042,
+      "step": 6180
+    },
+    {
+      "epoch": 0.8406899361673231,
+      "grad_norm": 0.44773173332214355,
+      "learning_rate": 3.472223188230083e-05,
+      "loss": 0.0372,
+      "step": 6190
+    },
+    {
+      "epoch": 0.8420480782289828,
+      "grad_norm": 0.8800703883171082,
+      "learning_rate": 3.4564876558252866e-05,
+      "loss": 0.0371,
+      "step": 6200
+    },
+    {
+      "epoch": 0.8434062202906424,
+      "grad_norm": 0.37842467427253723,
+      "learning_rate": 3.440769003024195e-05,
+      "loss": 0.0376,
+      "step": 6210
+    },
+    {
+      "epoch": 0.844764362352302,
+      "grad_norm": 0.33254295587539673,
+      "learning_rate": 3.425067401723477e-05,
+      "loss": 0.0233,
+      "step": 6220
+    },
+    {
+      "epoch": 0.8461225044139618,
+      "grad_norm": 0.6304585337638855,
+      "learning_rate": 3.409383023633325e-05,
+      "loss": 0.0358,
+      "step": 6230
+    },
+    {
+      "epoch": 0.8474806464756214,
+      "grad_norm": 0.3811233639717102,
+      "learning_rate": 3.3937160402755894e-05,
+      "loss": 0.0294,
+      "step": 6240
+    },
+    {
+      "epoch": 0.848838788537281,
+      "grad_norm": 0.9511606097221375,
+      "learning_rate": 3.378066622981885e-05,
+      "loss": 0.0281,
+      "step": 6250
+    },
+    {
+      "epoch": 0.8501969305989406,
+      "grad_norm": 0.8596386313438416,
+      "learning_rate": 3.362434942891738e-05,
+      "loss": 0.0476,
+      "step": 6260
+    },
+    {
+      "epoch": 0.8515550726606003,
+      "grad_norm": 1.0428494215011597,
+      "learning_rate": 3.346821170950693e-05,
+      "loss": 0.0368,
+      "step": 6270
+    },
+    {
+      "epoch": 0.8529132147222599,
+      "grad_norm": 0.7779479622840881,
+      "learning_rate": 3.3312254779084585e-05,
+      "loss": 0.043,
+      "step": 6280
+    },
+    {
+      "epoch": 0.8542713567839196,
+      "grad_norm": 0.8371151089668274,
+      "learning_rate": 3.315648034317039e-05,
+      "loss": 0.0427,
+      "step": 6290
+    },
+    {
+      "epoch": 0.8556294988455793,
+      "grad_norm": 0.844372034072876,
+      "learning_rate": 3.3000890105288564e-05,
+      "loss": 0.0452,
+      "step": 6300
+    },
+    {
+      "epoch": 0.8569876409072389,
+      "grad_norm": 2.701012372970581,
+      "learning_rate": 3.284548576694908e-05,
+      "loss": 0.0579,
+      "step": 6310
+    },
+    {
+      "epoch": 0.8583457829688985,
+      "grad_norm": 0.5503039956092834,
+      "learning_rate": 3.2690269027628815e-05,
+      "loss": 0.0374,
+      "step": 6320
+    },
+    {
+      "epoch": 0.8597039250305581,
+      "grad_norm": 1.1068071126937866,
+      "learning_rate": 3.253524158475324e-05,
+      "loss": 0.0329,
+      "step": 6330
+    },
+    {
+      "epoch": 0.8610620670922179,
+      "grad_norm": 0.30357858538627625,
+      "learning_rate": 3.238040513367757e-05,
+      "loss": 0.0265,
+      "step": 6340
+    },
+    {
+      "epoch": 0.8624202091538775,
+      "grad_norm": 0.5525499582290649,
+      "learning_rate": 3.222576136766843e-05,
+      "loss": 0.0518,
+      "step": 6350
+    },
+    {
+      "epoch": 0.8637783512155371,
+      "grad_norm": 0.7420355081558228,
+      "learning_rate": 3.2071311977885324e-05,
+      "loss": 0.0385,
+      "step": 6360
+    },
+    {
+      "epoch": 0.8651364932771968,
+      "grad_norm": 0.591984212398529,
+      "learning_rate": 3.191705865336197e-05,
+      "loss": 0.0377,
+      "step": 6370
+    },
+    {
+      "epoch": 0.8664946353388564,
+      "grad_norm": 0.718184232711792,
+      "learning_rate": 3.1763003080988075e-05,
+      "loss": 0.0283,
+      "step": 6380
+    },
+    {
+      "epoch": 0.8678527774005161,
+      "grad_norm": 0.6680257320404053,
+      "learning_rate": 3.160914694549063e-05,
+      "loss": 0.0339,
+      "step": 6390
+    },
+    {
+      "epoch": 0.8692109194621758,
+      "grad_norm": 0.4477512240409851,
+      "learning_rate": 3.145549192941573e-05,
+      "loss": 0.0325,
+      "step": 6400
+    },
+    {
+      "epoch": 0.8705690615238354,
+      "grad_norm": 1.1212290525436401,
+      "learning_rate": 3.130203971310999e-05,
+      "loss": 0.0361,
+      "step": 6410
+    },
+    {
+      "epoch": 0.871927203585495,
+      "grad_norm": 0.7342495918273926,
+      "learning_rate": 3.114879197470225e-05,
+      "loss": 0.0251,
+      "step": 6420
+    },
+    {
+      "epoch": 0.8732853456471547,
+      "grad_norm": 0.8969946503639221,
+      "learning_rate": 3.0995750390085285e-05,
+      "loss": 0.0521,
+      "step": 6430
+    },
+    {
+      "epoch": 0.8746434877088144,
+      "grad_norm": 0.46025049686431885,
+      "learning_rate": 3.084291663289728e-05,
+      "loss": 0.0485,
+      "step": 6440
+    },
+    {
+      "epoch": 0.876001629770474,
+      "grad_norm": 0.5148270726203918,
+      "learning_rate": 3.069029237450375e-05,
+      "loss": 0.0602,
+      "step": 6450
+    },
+    {
+      "epoch": 0.8773597718321337,
+      "grad_norm": 0.5505500435829163,
+      "learning_rate": 3.053787928397911e-05,
+      "loss": 0.0379,
+      "step": 6460
+    },
+    {
+      "epoch": 0.8787179138937933,
+      "grad_norm": 1.1663702726364136,
+      "learning_rate": 3.0385679028088526e-05,
+      "loss": 0.0365,
+      "step": 6470
+    },
+    {
+      "epoch": 0.8800760559554529,
+      "grad_norm": 0.3477430045604706,
+      "learning_rate": 3.023369327126959e-05,
+      "loss": 0.0123,
+      "step": 6480
+    },
+    {
+      "epoch": 0.8814341980171125,
+      "grad_norm": 0.8028808832168579,
+      "learning_rate": 3.0081923675614198e-05,
+      "loss": 0.0326,
+      "step": 6490
+    },
+    {
+      "epoch": 0.8827923400787723,
+      "grad_norm": 0.27157557010650635,
+      "learning_rate": 2.993037190085034e-05,
+      "loss": 0.064,
+      "step": 6500
+    },
+    {
+      "epoch": 0.8841504821404319,
+      "grad_norm": 0.6656714677810669,
+      "learning_rate": 2.977903960432392e-05,
+      "loss": 0.0348,
+      "step": 6510
+    },
+    {
+      "epoch": 0.8855086242020915,
+      "grad_norm": 1.1999881267547607,
+      "learning_rate": 2.9627928440980722e-05,
+      "loss": 0.0399,
+      "step": 6520
+    },
+    {
+      "epoch": 0.8868667662637512,
+      "grad_norm": 0.9073282480239868,
+      "learning_rate": 2.9477040063348183e-05,
+      "loss": 0.0362,
+      "step": 6530
+    },
+    {
+      "epoch": 0.8882249083254108,
+      "grad_norm": 0.9923334717750549,
+      "learning_rate": 2.9326376121517456e-05,
+      "loss": 0.033,
+      "step": 6540
+    },
+    {
+      "epoch": 0.8895830503870705,
+      "grad_norm": 0.7126595377922058,
+      "learning_rate": 2.9175938263125236e-05,
+      "loss": 0.0271,
+      "step": 6550
+    },
+    {
+      "epoch": 0.8909411924487302,
+      "grad_norm": 0.8900442123413086,
+      "learning_rate": 2.9025728133335873e-05,
+      "loss": 0.0344,
+      "step": 6560
+    },
+    {
+      "epoch": 0.8922993345103898,
+      "grad_norm": 0.9497235417366028,
+      "learning_rate": 2.8875747374823288e-05,
+      "loss": 0.0276,
+      "step": 6570
+    },
+    {
+      "epoch": 0.8936574765720494,
+      "grad_norm": 0.397804856300354,
+      "learning_rate": 2.872599762775298e-05,
+      "loss": 0.0362,
+      "step": 6580
+    },
+    {
+      "epoch": 0.895015618633709,
+      "grad_norm": 0.35318711400032043,
+      "learning_rate": 2.857648052976425e-05,
+      "loss": 0.0267,
+      "step": 6590
+    },
+    {
+      "epoch": 0.8963737606953688,
+      "grad_norm": 0.30047762393951416,
+      "learning_rate": 2.8427197715952047e-05,
+      "loss": 0.0188,
+      "step": 6600
+    },
+    {
+      "epoch": 0.8977319027570284,
+      "grad_norm": 0.40104982256889343,
+      "learning_rate": 2.8278150818849393e-05,
+      "loss": 0.0553,
+      "step": 6610
+    },
+    {
+      "epoch": 0.899090044818688,
+      "grad_norm": 0.43015170097351074,
+      "learning_rate": 2.812934146840922e-05,
+      "loss": 0.0332,
+      "step": 6620
+    },
+    {
+      "epoch": 0.9004481868803477,
+      "grad_norm": 0.30686715245246887,
+      "learning_rate": 2.7980771291986764e-05,
+      "loss": 0.0386,
+      "step": 6630
+    },
+    {
+      "epoch": 0.9018063289420073,
+      "grad_norm": 0.42802298069000244,
+      "learning_rate": 2.783244191432167e-05,
+      "loss": 0.0351,
+      "step": 6640
+    },
+    {
+      "epoch": 0.903164471003667,
+      "grad_norm": 1.033292531967163,
+      "learning_rate": 2.768435495752022e-05,
+      "loss": 0.0468,
+      "step": 6650
+    },
+    {
+      "epoch": 0.9045226130653267,
+      "grad_norm": 0.8372592926025391,
+      "learning_rate": 2.753651204103771e-05,
+      "loss": 0.0307,
+      "step": 6660
+    },
+    {
+      "epoch": 0.9058807551269863,
+      "grad_norm": 0.6943889856338501,
+      "learning_rate": 2.7388914781660523e-05,
+      "loss": 0.0411,
+      "step": 6670
+    },
+    {
+      "epoch": 0.9072388971886459,
+      "grad_norm": 1.1943504810333252,
+      "learning_rate": 2.7241564793488693e-05,
+      "loss": 0.039,
+      "step": 6680
+    },
+    {
+      "epoch": 0.9085970392503055,
+      "grad_norm": 0.513438880443573,
+      "learning_rate": 2.7094463687918037e-05,
+      "loss": 0.0339,
+      "step": 6690
+    },
+    {
+      "epoch": 0.9099551813119652,
+      "grad_norm": 0.4000544250011444,
+      "learning_rate": 2.694761307362268e-05,
+      "loss": 0.0277,
+      "step": 6700
+    },
+    {
+      "epoch": 0.9113133233736249,
+      "grad_norm": 0.6161653995513916,
+      "learning_rate": 2.6801014556537467e-05,
+      "loss": 0.022,
+      "step": 6710
+    },
+    {
+      "epoch": 0.9126714654352845,
+      "grad_norm": 0.5668504238128662,
+      "learning_rate": 2.6654669739840243e-05,
+      "loss": 0.0345,
+      "step": 6720
+    },
+    {
+      "epoch": 0.9140296074969442,
+      "grad_norm": 0.6454136967658997,
+      "learning_rate": 2.650858022393451e-05,
+      "loss": 0.0317,
+      "step": 6730
+    },
+    {
+      "epoch": 0.9153877495586038,
+      "grad_norm": 0.41939571499824524,
+      "learning_rate": 2.6362747606431747e-05,
+      "loss": 0.0449,
+      "step": 6740
+    },
+    {
+      "epoch": 0.9167458916202634,
+      "grad_norm": 0.913457453250885,
+      "learning_rate": 2.6217173482134172e-05,
+      "loss": 0.0478,
+      "step": 6750
+    },
+    {
+      "epoch": 0.9181040336819232,
+      "grad_norm": 0.32109588384628296,
+      "learning_rate": 2.6071859443017044e-05,
+      "loss": 0.0557,
+      "step": 6760
+    },
+    {
+      "epoch": 0.9194621757435828,
+      "grad_norm": 0.4987392723560333,
+      "learning_rate": 2.5926807078211414e-05,
+      "loss": 0.0387,
+      "step": 6770
+    },
+    {
+      "epoch": 0.9208203178052424,
+      "grad_norm": 0.7392056584358215,
+      "learning_rate": 2.5782017973986728e-05,
+      "loss": 0.0242,
+      "step": 6780
+    },
+    {
+      "epoch": 0.9221784598669021,
+      "grad_norm": 0.49856331944465637,
+      "learning_rate": 2.5637493713733374e-05,
+      "loss": 0.044,
+      "step": 6790
+    },
+    {
+      "epoch": 0.9235366019285617,
+      "grad_norm": 0.6359718441963196,
+      "learning_rate": 2.549323587794559e-05,
+      "loss": 0.0378,
+      "step": 6800
+    },
+    {
+      "epoch": 0.9248947439902214,
+      "grad_norm": 0.5166847705841064,
+      "learning_rate": 2.5349246044203895e-05,
+      "loss": 0.0444,
+      "step": 6810
+    },
+    {
+      "epoch": 0.926252886051881,
+      "grad_norm": 0.7145246863365173,
+      "learning_rate": 2.520552578715808e-05,
+      "loss": 0.027,
+      "step": 6820
+    },
+    {
+      "epoch": 0.9276110281135407,
+      "grad_norm": 0.5690740942955017,
+      "learning_rate": 2.506207667850981e-05,
+      "loss": 0.039,
+      "step": 6830
+    },
+    {
+      "epoch": 0.9289691701752003,
+      "grad_norm": 1.0720932483673096,
+      "learning_rate": 2.4918900286995555e-05,
+      "loss": 0.0305,
+      "step": 6840
+    },
+    {
+      "epoch": 0.9303273122368599,
+      "grad_norm": 0.5707837343215942,
+      "learning_rate": 2.4775998178369458e-05,
+      "loss": 0.0391,
+      "step": 6850
+    },
+    {
+      "epoch": 0.9316854542985197,
+      "grad_norm": 0.8946691751480103,
+      "learning_rate": 2.4633371915386017e-05,
+      "loss": 0.0376,
+      "step": 6860
+    },
+    {
+      "epoch": 0.9330435963601793,
+      "grad_norm": 0.5458039045333862,
+      "learning_rate": 2.4491023057783235e-05,
+      "loss": 0.0444,
+      "step": 6870
+    },
+    {
+      "epoch": 0.9344017384218389,
+      "grad_norm": 0.6177228689193726,
+      "learning_rate": 2.4348953162265375e-05,
+      "loss": 0.0364,
+      "step": 6880
+    },
+    {
+      "epoch": 0.9357598804834986,
+      "grad_norm": 0.44035449624061584,
+      "learning_rate": 2.420716378248607e-05,
+      "loss": 0.0417,
+      "step": 6890
+    },
+    {
+      "epoch": 0.9371180225451582,
+      "grad_norm": 1.3996977806091309,
+      "learning_rate": 2.4065656469031266e-05,
+      "loss": 0.05,
+      "step": 6900
+    },
+    {
+      "epoch": 0.9384761646068178,
+      "grad_norm": 0.4726139307022095,
+      "learning_rate": 2.3924432769402268e-05,
+      "loss": 0.0329,
+      "step": 6910
+    },
+    {
+      "epoch": 0.9398343066684776,
+      "grad_norm": 0.7002353072166443,
+      "learning_rate": 2.3783494227998844e-05,
+      "loss": 0.0349,
+      "step": 6920
+    },
+    {
+      "epoch": 0.9411924487301372,
+      "grad_norm": 0.4759741723537445,
+      "learning_rate": 2.3642842386102264e-05,
+      "loss": 0.0538,
+      "step": 6930
+    },
+    {
+      "epoch": 0.9425505907917968,
+      "grad_norm": 0.696948766708374,
+      "learning_rate": 2.3502478781858567e-05,
+      "loss": 0.0333,
+      "step": 6940
+    },
+    {
+      "epoch": 0.9439087328534564,
+      "grad_norm": 0.7952139973640442,
+      "learning_rate": 2.3362404950261628e-05,
+      "loss": 0.0383,
+      "step": 6950
+    },
+    {
+      "epoch": 0.9452668749151161,
+      "grad_norm": 0.6942967176437378,
+      "learning_rate": 2.3222622423136458e-05,
+      "loss": 0.0616,
+      "step": 6960
+    },
+    {
+      "epoch": 0.9466250169767758,
+      "grad_norm": 0.6597992181777954,
+      "learning_rate": 2.3083132729122332e-05,
+      "loss": 0.0265,
+      "step": 6970
+    },
+    {
+      "epoch": 0.9479831590384354,
+      "grad_norm": 0.8483303785324097,
+      "learning_rate": 2.294393739365621e-05,
+      "loss": 0.037,
+      "step": 6980
+    },
+    {
+      "epoch": 0.9493413011000951,
+      "grad_norm": 0.7203598022460938,
+      "learning_rate": 2.2805037938956e-05,
+      "loss": 0.0415,
+      "step": 6990
+    },
+    {
+      "epoch": 0.9506994431617547,
+      "grad_norm": 1.0278226137161255,
+      "learning_rate": 2.266643588400386e-05,
+      "loss": 0.027,
+      "step": 7000
+    },
+    {
+      "epoch": 0.9520575852234143,
+      "grad_norm": 0.9354854822158813,
+      "learning_rate": 2.252813274452969e-05,
+      "loss": 0.0335,
+      "step": 7010
+    },
+    {
+      "epoch": 0.9534157272850741,
+      "grad_norm": 0.5468366146087646,
+      "learning_rate": 2.2390130032994427e-05,
+      "loss": 0.0291,
+      "step": 7020
+    },
+    {
+      "epoch": 0.9547738693467337,
+      "grad_norm": 0.7657195925712585,
+      "learning_rate": 2.2252429258573633e-05,
+      "loss": 0.0205,
+      "step": 7030
+    },
+    {
+      "epoch": 0.9561320114083933,
+      "grad_norm": 0.5784329771995544,
+      "learning_rate": 2.2115031927140904e-05,
+      "loss": 0.047,
+      "step": 7040
+    },
+    {
+      "epoch": 0.957490153470053,
+      "grad_norm": 0.5740514993667603,
+      "learning_rate": 2.1977939541251463e-05,
+      "loss": 0.0387,
+      "step": 7050
+    },
+    {
+      "epoch": 0.9588482955317126,
+      "grad_norm": 0.6613061428070068,
+      "learning_rate": 2.1841153600125684e-05,
+      "loss": 0.0355,
+      "step": 7060
+    },
+    {
+      "epoch": 0.9602064375933723,
+      "grad_norm": 0.5087347626686096,
+      "learning_rate": 2.170467559963267e-05,
+      "loss": 0.0404,
+      "step": 7070
+    },
+    {
+      "epoch": 0.961564579655032,
+      "grad_norm": 0.8660780787467957,
+      "learning_rate": 2.1568507032273982e-05,
+      "loss": 0.0396,
+      "step": 7080
+    },
+    {
+      "epoch": 0.9629227217166916,
+      "grad_norm": 0.613242506980896,
+      "learning_rate": 2.1432649387167264e-05,
+      "loss": 0.028,
+      "step": 7090
+    },
+    {
+      "epoch": 0.9642808637783512,
+      "grad_norm": 0.541179358959198,
+      "learning_rate": 2.1297104150029973e-05,
+      "loss": 0.0324,
+      "step": 7100
+    },
+    {
+      "epoch": 0.9656390058400108,
+      "grad_norm": 0.8271092176437378,
+      "learning_rate": 2.116187280316307e-05,
+      "loss": 0.0155,
+      "step": 7110
+    },
+    {
+      "epoch": 0.9669971479016705,
+      "grad_norm": 0.25554633140563965,
+      "learning_rate": 2.1026956825434908e-05,
+      "loss": 0.0265,
+      "step": 7120
+    },
+    {
+      "epoch": 0.9683552899633302,
+      "grad_norm": 0.7277560234069824,
+      "learning_rate": 2.0892357692265017e-05,
+      "loss": 0.0334,
+      "step": 7130
+    },
+    {
+      "epoch": 0.9697134320249898,
+      "grad_norm": 0.6272832155227661,
+      "learning_rate": 2.0758076875607947e-05,
+      "loss": 0.0437,
+      "step": 7140
+    },
+    {
+      "epoch": 0.9710715740866495,
+      "grad_norm": 0.5331401228904724,
+      "learning_rate": 2.0624115843937207e-05,
+      "loss": 0.0698,
+      "step": 7150
+    },
+    {
+      "epoch": 0.9724297161483091,
+      "grad_norm": 0.8539391756057739,
+      "learning_rate": 2.0490476062229157e-05,
+      "loss": 0.0491,
+      "step": 7160
+    },
+    {
+      "epoch": 0.9737878582099687,
+      "grad_norm": 0.7826774716377258,
+      "learning_rate": 2.035715899194704e-05,
+      "loss": 0.0579,
+      "step": 7170
+    },
+    {
+      "epoch": 0.9751460002716285,
+      "grad_norm": 0.7377849221229553,
+      "learning_rate": 2.022416609102499e-05,
+      "loss": 0.0248,
+      "step": 7180
+    },
+    {
+      "epoch": 0.9765041423332881,
+      "grad_norm": 0.6041225790977478,
+      "learning_rate": 2.009149881385205e-05,
+      "loss": 0.0503,
+      "step": 7190
+    },
+    {
+      "epoch": 0.9778622843949477,
+      "grad_norm": 1.1008758544921875,
+      "learning_rate": 1.995915861125634e-05,
+      "loss": 0.0515,
+      "step": 7200
+    },
+    {
+      "epoch": 0.9792204264566073,
+      "grad_norm": 0.4391670227050781,
+      "learning_rate": 1.9827146930489065e-05,
+      "loss": 0.0312,
+      "step": 7210
+    },
+    {
+      "epoch": 0.980578568518267,
+      "grad_norm": 0.4211924076080322,
+      "learning_rate": 1.9695465215208848e-05,
+      "loss": 0.0429,
+      "step": 7220
+    },
+    {
+      "epoch": 0.9819367105799267,
+      "grad_norm": 1.0390186309814453,
+      "learning_rate": 1.9564114905465813e-05,
+      "loss": 0.0282,
+      "step": 7230
+    },
+    {
+      "epoch": 0.9832948526415863,
+      "grad_norm": 1.0719592571258545,
+      "learning_rate": 1.9433097437685936e-05,
+      "loss": 0.0285,
+      "step": 7240
+    },
+    {
+      "epoch": 0.984652994703246,
+      "grad_norm": 0.4206024408340454,
+      "learning_rate": 1.930241424465521e-05,
+      "loss": 0.0235,
+      "step": 7250
+    },
+    {
+      "epoch": 0.9860111367649056,
+      "grad_norm": 0.35964375734329224,
+      "learning_rate": 1.9172066755504115e-05,
+      "loss": 0.0241,
+      "step": 7260
+    },
+    {
+      "epoch": 0.9873692788265652,
+      "grad_norm": 0.24809835851192474,
+      "learning_rate": 1.9042056395691914e-05,
+      "loss": 0.0199,
+      "step": 7270
+    },
+    {
+      "epoch": 0.988727420888225,
+      "grad_norm": 0.4773974120616913,
+      "learning_rate": 1.8912384586991066e-05,
+      "loss": 0.0268,
+      "step": 7280
+    },
+    {
+      "epoch": 0.9900855629498846,
+      "grad_norm": 0.5127553343772888,
+      "learning_rate": 1.8783052747471717e-05,
+      "loss": 0.0298,
+      "step": 7290
+    },
+    {
+      "epoch": 0.9914437050115442,
+      "grad_norm": 0.5278441309928894,
+      "learning_rate": 1.865406229148611e-05,
+      "loss": 0.0286,
+      "step": 7300
+    },
+    {
+      "epoch": 0.9928018470732038,
+      "grad_norm": 0.6264014840126038,
+      "learning_rate": 1.8525414629653233e-05,
+      "loss": 0.0377,
+      "step": 7310
+    },
+    {
+      "epoch": 0.9941599891348635,
+      "grad_norm": 1.1351443529129028,
+      "learning_rate": 1.8397111168843255e-05,
+      "loss": 0.0371,
+      "step": 7320
+    },
+    {
+      "epoch": 0.9955181311965231,
+      "grad_norm": 0.5309675931930542,
+      "learning_rate": 1.8269153312162323e-05,
+      "loss": 0.0418,
+      "step": 7330
+    },
+    {
+      "epoch": 0.9968762732581828,
+      "grad_norm": 0.9954323172569275,
+      "learning_rate": 1.8141542458937054e-05,
+      "loss": 0.0364,
+      "step": 7340
+    },
+    {
+      "epoch": 0.9982344153198425,
+      "grad_norm": 0.42543384432792664,
+      "learning_rate": 1.8014280004699268e-05,
+      "loss": 0.027,
+      "step": 7350
+    },
+    {
+      "epoch": 0.9995925573815021,
+      "grad_norm": 0.487441748380661,
+      "learning_rate": 1.788736734117078e-05,
+      "loss": 0.0273,
+      "step": 7360
+    },
+    {
+      "epoch": 1.0009506994431618,
+      "grad_norm": 0.43553733825683594,
+      "learning_rate": 1.7760805856248152e-05,
+      "loss": 0.0303,
+      "step": 7370
+    },
+    {
+      "epoch": 1.0023088415048214,
+      "grad_norm": 0.6573531627655029,
+      "learning_rate": 1.7634596933987518e-05,
+      "loss": 0.035,
+      "step": 7380
+    },
+    {
+      "epoch": 1.003666983566481,
+      "grad_norm": 0.7375355958938599,
+      "learning_rate": 1.7508741954589404e-05,
+      "loss": 0.0301,
+      "step": 7390
+    },
+    {
+      "epoch": 1.0050251256281406,
+      "grad_norm": 0.4842686951160431,
+      "learning_rate": 1.7383242294383717e-05,
+      "loss": 0.0194,
+      "step": 7400
+    },
+    {
+      "epoch": 1.0063832676898004,
+      "grad_norm": 0.9063589572906494,
+      "learning_rate": 1.7258099325814632e-05,
+      "loss": 0.0325,
+      "step": 7410
+    },
+    {
+      "epoch": 1.00774140975146,
+      "grad_norm": 0.5210592150688171,
+      "learning_rate": 1.7133314417425594e-05,
+      "loss": 0.0192,
+      "step": 7420
+    },
+    {
+      "epoch": 1.0090995518131196,
+      "grad_norm": 0.5844367146492004,
+      "learning_rate": 1.7008888933844408e-05,
+      "loss": 0.0363,
+      "step": 7430
+    },
+    {
+      "epoch": 1.0104576938747794,
+      "grad_norm": 0.8706324696540833,
+      "learning_rate": 1.6884824235768172e-05,
+      "loss": 0.0316,
+      "step": 7440
+    },
+    {
+      "epoch": 1.0118158359364389,
+      "grad_norm": 0.721836268901825,
+      "learning_rate": 1.6761121679948592e-05,
+      "loss": 0.0391,
+      "step": 7450
+    },
+    {
+      "epoch": 1.0131739779980986,
+      "grad_norm": 0.428141713142395,
+      "learning_rate": 1.663778261917695e-05,
+      "loss": 0.0182,
+      "step": 7460
+    },
+    {
+      "epoch": 1.0145321200597583,
+      "grad_norm": 0.9551045298576355,
+      "learning_rate": 1.651480840226952e-05,
+      "loss": 0.0399,
+      "step": 7470
+    },
+    {
+      "epoch": 1.0158902621214179,
+      "grad_norm": 0.4620324373245239,
+      "learning_rate": 1.639220037405258e-05,
+      "loss": 0.0202,
+      "step": 7480
+    },
+    {
+      "epoch": 1.0172484041830776,
+      "grad_norm": 0.3737846314907074,
+      "learning_rate": 1.6269959875347906e-05,
+      "loss": 0.021,
+      "step": 7490
+    },
+    {
+      "epoch": 1.0186065462447371,
+      "grad_norm": 0.4276089072227478,
+      "learning_rate": 1.614808824295802e-05,
+      "loss": 0.0427,
+      "step": 7500
+    },
+    {
+      "epoch": 1.0199646883063969,
+      "grad_norm": 0.7729580402374268,
+      "learning_rate": 1.602658680965152e-05,
+      "loss": 0.0204,
+      "step": 7510
+    },
+    {
+      "epoch": 1.0213228303680566,
+      "grad_norm": 0.21997807919979095,
+      "learning_rate": 1.5905456904148686e-05,
+      "loss": 0.0217,
+      "step": 7520
+    },
+    {
+      "epoch": 1.0226809724297161,
+      "grad_norm": 0.4168502688407898,
+      "learning_rate": 1.57846998511067e-05,
+      "loss": 0.0382,
+      "step": 7530
+    },
+    {
+      "epoch": 1.0240391144913759,
+      "grad_norm": 0.8020449280738831,
+      "learning_rate": 1.566431697110538e-05,
+      "loss": 0.0281,
+      "step": 7540
+    },
+    {
+      "epoch": 1.0253972565530354,
+      "grad_norm": 0.8150598406791687,
+      "learning_rate": 1.554430958063259e-05,
+      "loss": 0.0391,
+      "step": 7550
+    },
+    {
+      "epoch": 1.0267553986146951,
+      "grad_norm": 0.4424729645252228,
+      "learning_rate": 1.5424678992069912e-05,
+      "loss": 0.0251,
+      "step": 7560
+    },
+    {
+      "epoch": 1.0281135406763549,
+      "grad_norm": 0.7714619040489197,
+      "learning_rate": 1.5305426513678362e-05,
+      "loss": 0.0348,
+      "step": 7570
+    },
+    {
+      "epoch": 1.0294716827380144,
+      "grad_norm": 0.48755723237991333,
+      "learning_rate": 1.518655344958388e-05,
+      "loss": 0.0208,
+      "step": 7580
+    },
+    {
+      "epoch": 1.030829824799674,
+      "grad_norm": 0.7660195827484131,
+      "learning_rate": 1.5068061099763275e-05,
+      "loss": 0.0412,
+      "step": 7590
+    },
+    {
+      "epoch": 1.0321879668613336,
+      "grad_norm": 0.8019341230392456,
+      "learning_rate": 1.494995076002988e-05,
+      "loss": 0.0596,
+      "step": 7600
+    },
+    {
+      "epoch": 1.0335461089229934,
+      "grad_norm": 0.7356421947479248,
+      "learning_rate": 1.4832223722019456e-05,
+      "loss": 0.0358,
+      "step": 7610
+    },
+    {
+      "epoch": 1.034904250984653,
+      "grad_norm": 1.0923794507980347,
+      "learning_rate": 1.4714881273176035e-05,
+      "loss": 0.0298,
+      "step": 7620
+    },
+    {
+      "epoch": 1.0362623930463126,
+      "grad_norm": 0.6139436364173889,
+      "learning_rate": 1.4597924696737835e-05,
+      "loss": 0.0296,
+      "step": 7630
+    },
+    {
+      "epoch": 1.0376205351079724,
+      "grad_norm": 0.6711189150810242,
+      "learning_rate": 1.4481355271723252e-05,
+      "loss": 0.0347,
+      "step": 7640
+    },
+    {
+      "epoch": 1.0389786771696319,
+      "grad_norm": 0.9619874358177185,
+      "learning_rate": 1.4365174272916809e-05,
+      "loss": 0.0358,
+      "step": 7650
+    },
+    {
+      "epoch": 1.0403368192312916,
+      "grad_norm": 0.7057786583900452,
+      "learning_rate": 1.4249382970855319e-05,
+      "loss": 0.0286,
+      "step": 7660
+    },
+    {
+      "epoch": 1.0416949612929511,
+      "grad_norm": 0.8339405655860901,
+      "learning_rate": 1.4133982631813903e-05,
+      "loss": 0.0266,
+      "step": 7670
+    },
+    {
+      "epoch": 1.0430531033546109,
+      "grad_norm": 0.23086152970790863,
+      "learning_rate": 1.4018974517792194e-05,
+      "loss": 0.0233,
+      "step": 7680
+    },
+    {
+      "epoch": 1.0444112454162706,
+      "grad_norm": 0.4098237156867981,
+      "learning_rate": 1.390435988650048e-05,
+      "loss": 0.0427,
+      "step": 7690
+    },
+    {
+      "epoch": 1.0457693874779301,
+      "grad_norm": 0.6336156129837036,
+      "learning_rate": 1.3790139991346006e-05,
+      "loss": 0.0212,
+      "step": 7700
+    },
+    {
+      "epoch": 1.0471275295395899,
+      "grad_norm": 0.5233216881752014,
+      "learning_rate": 1.367631608141926e-05,
+      "loss": 0.0231,
+      "step": 7710
+    },
+    {
+      "epoch": 1.0484856716012494,
+      "grad_norm": 0.7864448428153992,
+      "learning_rate": 1.3562889401480278e-05,
+      "loss": 0.0246,
+      "step": 7720
+    },
+    {
+      "epoch": 1.0498438136629091,
+      "grad_norm": 0.6876363754272461,
+      "learning_rate": 1.3449861191945074e-05,
+      "loss": 0.0325,
+      "step": 7730
+    },
+    {
+      "epoch": 1.0512019557245689,
+      "grad_norm": 0.6086934208869934,
+      "learning_rate": 1.3337232688872009e-05,
+      "loss": 0.0215,
+      "step": 7740
+    },
+    {
+      "epoch": 1.0525600977862284,
+      "grad_norm": 0.2169996201992035,
+      "learning_rate": 1.3225005123948364e-05,
+      "loss": 0.0263,
+      "step": 7750
+    },
+    {
+      "epoch": 1.0539182398478881,
+      "grad_norm": 0.5260003805160522,
+      "learning_rate": 1.311317972447681e-05,
+      "loss": 0.0169,
+      "step": 7760
+    },
+    {
+      "epoch": 1.0552763819095476,
+      "grad_norm": 0.30321717262268066,
+      "learning_rate": 1.3001757713361996e-05,
+      "loss": 0.0343,
+      "step": 7770
+    },
+    {
+      "epoch": 1.0566345239712074,
+      "grad_norm": 0.14077503979206085,
+      "learning_rate": 1.2890740309097204e-05,
+      "loss": 0.0381,
+      "step": 7780
+    },
+    {
+      "epoch": 1.0579926660328671,
+      "grad_norm": 0.2967330813407898,
+      "learning_rate": 1.2780128725750944e-05,
+      "loss": 0.028,
+      "step": 7790
+    },
+    {
+      "epoch": 1.0593508080945266,
+      "grad_norm": 0.6631761193275452,
+      "learning_rate": 1.266992417295379e-05,
+      "loss": 0.0381,
+      "step": 7800
+    },
+    {
+      "epoch": 1.0607089501561864,
+      "grad_norm": 0.38740065693855286,
+      "learning_rate": 1.2560127855885073e-05,
+      "loss": 0.0325,
+      "step": 7810
+    },
+    {
+      "epoch": 1.062067092217846,
+      "grad_norm": 0.5752584338188171,
+      "learning_rate": 1.2450740975259745e-05,
+      "loss": 0.035,
+      "step": 7820
+    },
+    {
+      "epoch": 1.0634252342795056,
+      "grad_norm": 0.6170110106468201,
+      "learning_rate": 1.234176472731517e-05,
+      "loss": 0.0311,
+      "step": 7830
+    },
+    {
+      "epoch": 1.0647833763411654,
+      "grad_norm": 0.5072628259658813,
+      "learning_rate": 1.2233200303798158e-05,
+      "loss": 0.0311,
+      "step": 7840
+    },
+    {
+      "epoch": 1.066141518402825,
+      "grad_norm": 0.8664295077323914,
+      "learning_rate": 1.2125048891951846e-05,
+      "loss": 0.0459,
+      "step": 7850
+    },
+    {
+      "epoch": 1.0674996604644846,
+      "grad_norm": 0.2910117506980896,
+      "learning_rate": 1.2017311674502745e-05,
+      "loss": 0.0163,
+      "step": 7860
+    },
+    {
+      "epoch": 1.0688578025261442,
+      "grad_norm": 0.6841345429420471,
+      "learning_rate": 1.1909989829647822e-05,
+      "loss": 0.0297,
+      "step": 7870
+    },
+    {
+      "epoch": 1.070215944587804,
+      "grad_norm": 0.45628541707992554,
+      "learning_rate": 1.1803084531041553e-05,
+      "loss": 0.0206,
+      "step": 7880
+    },
+    {
+      "epoch": 1.0715740866494636,
+      "grad_norm": 0.9766668081283569,
+      "learning_rate": 1.1696596947783162e-05,
+      "loss": 0.0328,
+      "step": 7890
+    },
+    {
+      "epoch": 1.0729322287111231,
+      "grad_norm": 0.6276238560676575,
+      "learning_rate": 1.1590528244403803e-05,
+      "loss": 0.0257,
+      "step": 7900
+    },
+    {
+      "epoch": 1.0742903707727829,
+      "grad_norm": 0.30070820450782776,
+      "learning_rate": 1.148487958085382e-05,
+      "loss": 0.0628,
+      "step": 7910
+    },
+    {
+      "epoch": 1.0756485128344424,
+      "grad_norm": 0.6255337595939636,
+      "learning_rate": 1.1379652112490086e-05,
+      "loss": 0.0228,
+      "step": 7920
+    },
+    {
+      "epoch": 1.0770066548961021,
+      "grad_norm": 0.37872710824012756,
+      "learning_rate": 1.1274846990063315e-05,
+      "loss": 0.0219,
+      "step": 7930
+    },
+    {
+      "epoch": 1.0783647969577619,
+      "grad_norm": 0.8163110613822937,
+      "learning_rate": 1.117046535970554e-05,
+      "loss": 0.0275,
+      "step": 7940
+    },
+    {
+      "epoch": 1.0797229390194214,
+      "grad_norm": 0.3901977241039276,
+      "learning_rate": 1.106650836291755e-05,
+      "loss": 0.0472,
+      "step": 7950
+    },
+    {
+      "epoch": 1.0810810810810811,
+      "grad_norm": 0.31426215171813965,
+      "learning_rate": 1.0962977136556418e-05,
+      "loss": 0.032,
+      "step": 7960
+    },
+    {
+      "epoch": 1.0824392231427407,
+      "grad_norm": 0.7033277153968811,
+      "learning_rate": 1.0859872812823024e-05,
+      "loss": 0.0307,
+      "step": 7970
+    },
+    {
+      "epoch": 1.0837973652044004,
+      "grad_norm": 0.7162697911262512,
+      "learning_rate": 1.0757196519249747e-05,
+      "loss": 0.0259,
+      "step": 7980
+    },
+    {
+      "epoch": 1.0851555072660601,
+      "grad_norm": 0.5055510997772217,
+      "learning_rate": 1.0654949378688077e-05,
+      "loss": 0.0248,
+      "step": 7990
+    },
+    {
+      "epoch": 1.0865136493277197,
+      "grad_norm": 0.5954899191856384,
+      "learning_rate": 1.0553132509296376e-05,
+      "loss": 0.035,
+      "step": 8000
+    },
+    {
+      "epoch": 1.0878717913893794,
+      "grad_norm": 0.6421067118644714,
+      "learning_rate": 1.0451747024527613e-05,
+      "loss": 0.0145,
+      "step": 8010
+    },
+    {
+      "epoch": 1.089229933451039,
+      "grad_norm": 0.5341675281524658,
+      "learning_rate": 1.0350794033117189e-05,
+      "loss": 0.0262,
+      "step": 8020
+    },
+    {
+      "epoch": 1.0905880755126987,
+      "grad_norm": 0.7875378131866455,
+      "learning_rate": 1.0250274639070856e-05,
+      "loss": 0.0339,
+      "step": 8030
+    },
+    {
+      "epoch": 1.0919462175743582,
+      "grad_norm": 0.3432125747203827,
+      "learning_rate": 1.0150189941652599e-05,
+      "loss": 0.0247,
+      "step": 8040
+    },
+    {
+      "epoch": 1.093304359636018,
+      "grad_norm": 0.30299338698387146,
+      "learning_rate": 1.0050541035372635e-05,
+      "loss": 0.0324,
+      "step": 8050
+    },
+    {
+      "epoch": 1.0946625016976776,
+      "grad_norm": 0.5707754492759705,
+      "learning_rate": 9.951329009975458e-06,
+      "loss": 0.0341,
+      "step": 8060
+    },
+    {
+      "epoch": 1.0960206437593372,
+      "grad_norm": 0.6863912343978882,
+      "learning_rate": 9.852554950427845e-06,
+      "loss": 0.032,
+      "step": 8070
+    },
+    {
+      "epoch": 1.097378785820997,
+      "grad_norm": 0.5907803177833557,
+      "learning_rate": 9.754219936907105e-06,
+      "loss": 0.0242,
+      "step": 8080
+    },
+    {
+      "epoch": 1.0987369278826564,
+      "grad_norm": 0.2799040675163269,
+      "learning_rate": 9.656325044789194e-06,
+      "loss": 0.0185,
+      "step": 8090
+    },
+    {
+      "epoch": 1.1000950699443162,
+      "grad_norm": 0.6462242603302002,
+      "learning_rate": 9.55887134463697e-06,
+      "loss": 0.0233,
+      "step": 8100
+    },
+    {
+      "epoch": 1.101453212005976,
+      "grad_norm": 0.5629631280899048,
+      "learning_rate": 9.461859902188475e-06,
+      "loss": 0.0203,
+      "step": 8110
+    },
+    {
+      "epoch": 1.1028113540676354,
+      "grad_norm": 0.4194990396499634,
+      "learning_rate": 9.365291778345303e-06,
+      "loss": 0.0332,
+      "step": 8120
+    },
+    {
+      "epoch": 1.1041694961292952,
+      "grad_norm": 0.4346142113208771,
+      "learning_rate": 9.269168029160991e-06,
+      "loss": 0.0186,
+      "step": 8130
+    },
+    {
+      "epoch": 1.1055276381909547,
+      "grad_norm": 0.6989235281944275,
+      "learning_rate": 9.173489705829447e-06,
+      "loss": 0.0203,
+      "step": 8140
+    },
+    {
+      "epoch": 1.1068857802526144,
+      "grad_norm": 0.7293084263801575,
+      "learning_rate": 9.078257854673516e-06,
+      "loss": 0.0483,
+      "step": 8150
+    },
+    {
+      "epoch": 1.1082439223142742,
+      "grad_norm": 0.5558478236198425,
+      "learning_rate": 8.983473517133429e-06,
+      "loss": 0.0555,
+      "step": 8160
+    },
+    {
+      "epoch": 1.1096020643759337,
+      "grad_norm": 0.4024063050746918,
+      "learning_rate": 8.889137729755537e-06,
+      "loss": 0.0339,
+      "step": 8170
+    },
+    {
+      "epoch": 1.1109602064375934,
+      "grad_norm": 0.9031903743743896,
+      "learning_rate": 8.79525152418087e-06,
+      "loss": 0.0288,
+      "step": 8180
+    },
+    {
+      "epoch": 1.112318348499253,
+      "grad_norm": 0.86002516746521,
+      "learning_rate": 8.701815927133961e-06,
+      "loss": 0.0201,
+      "step": 8190
+    },
+    {
+      "epoch": 1.1136764905609127,
+      "grad_norm": 0.8369676470756531,
+      "learning_rate": 8.608831960411534e-06,
+      "loss": 0.0333,
+      "step": 8200
+    },
+    {
+      "epoch": 1.1150346326225724,
+      "grad_norm": 0.5633930563926697,
+      "learning_rate": 8.516300640871321e-06,
+      "loss": 0.0222,
+      "step": 8210
+    },
+    {
+      "epoch": 1.116392774684232,
+      "grad_norm": 0.5516213178634644,
+      "learning_rate": 8.424222980421038e-06,
+      "loss": 0.0401,
+      "step": 8220
+    },
+    {
+      "epoch": 1.1177509167458917,
+      "grad_norm": 1.0462805032730103,
+      "learning_rate": 8.332599986007184e-06,
+      "loss": 0.042,
+      "step": 8230
+    },
+    {
+      "epoch": 1.1191090588075512,
+      "grad_norm": 1.9067347049713135,
+      "learning_rate": 8.241432659604203e-06,
+      "loss": 0.0301,
+      "step": 8240
+    },
+    {
+      "epoch": 1.120467200869211,
+      "grad_norm": 0.3867623805999756,
+      "learning_rate": 8.150721998203331e-06,
+      "loss": 0.0325,
+      "step": 8250
+    },
+    {
+      "epoch": 1.1218253429308707,
+      "grad_norm": 0.6784263849258423,
+      "learning_rate": 8.06046899380184e-06,
+      "loss": 0.0365,
+      "step": 8260
+    },
+    {
+      "epoch": 1.1231834849925302,
+      "grad_norm": 0.5138591527938843,
+      "learning_rate": 7.970674633392133e-06,
+      "loss": 0.0258,
+      "step": 8270
+    },
+    {
+      "epoch": 1.12454162705419,
+      "grad_norm": 0.20440071821212769,
+      "learning_rate": 7.881339898950924e-06,
+      "loss": 0.0385,
+      "step": 8280
+    },
+    {
+      "epoch": 1.1258997691158494,
+      "grad_norm": 0.7685686349868774,
+      "learning_rate": 7.792465767428597e-06,
+      "loss": 0.0222,
+      "step": 8290
+    },
+    {
+      "epoch": 1.1272579111775092,
+      "grad_norm": 0.5112767219543457,
+      "learning_rate": 7.704053210738376e-06,
+      "loss": 0.0491,
+      "step": 8300
+    },
+    {
+      "epoch": 1.1286160532391687,
+      "grad_norm": 0.8115878701210022,
+      "learning_rate": 7.6161031957458494e-06,
+      "loss": 0.0322,
+      "step": 8310
+    },
+    {
+      "epoch": 1.1299741953008284,
+      "grad_norm": 0.4085262715816498,
+      "learning_rate": 7.5286166842582605e-06,
+      "loss": 0.0189,
+      "step": 8320
+    },
+    {
+      "epoch": 1.1313323373624882,
+      "grad_norm": 0.246077761054039,
+      "learning_rate": 7.4415946330140814e-06,
+      "loss": 0.0273,
+      "step": 8330
+    },
+    {
+      "epoch": 1.1326904794241477,
+      "grad_norm": 0.7145042419433594,
+      "learning_rate": 7.3550379936725644e-06,
+      "loss": 0.0324,
+      "step": 8340
+    },
+    {
+      "epoch": 1.1340486214858074,
+      "grad_norm": 0.30695274472236633,
+      "learning_rate": 7.2689477128032035e-06,
+      "loss": 0.0246,
+      "step": 8350
+    },
+    {
+      "epoch": 1.1354067635474672,
+      "grad_norm": 0.2302057445049286,
+      "learning_rate": 7.183324731875551e-06,
+      "loss": 0.0134,
+      "step": 8360
+    },
+    {
+      "epoch": 1.1367649056091267,
+      "grad_norm": 0.40955549478530884,
+      "learning_rate": 7.098169987248782e-06,
+      "loss": 0.0218,
+      "step": 8370
+    },
+    {
+      "epoch": 1.1381230476707864,
+      "grad_norm": 0.48199403285980225,
+      "learning_rate": 7.013484410161553e-06,
+      "loss": 0.0271,
+      "step": 8380
+    },
+    {
+      "epoch": 1.139481189732446,
+      "grad_norm": 0.5325714945793152,
+      "learning_rate": 6.92926892672176e-06,
+      "loss": 0.0211,
+      "step": 8390
+    },
+    {
+      "epoch": 1.1408393317941057,
+      "grad_norm": 0.487539678812027,
+      "learning_rate": 6.845524457896446e-06,
+      "loss": 0.0306,
+      "step": 8400
+    },
+    {
+      "epoch": 1.1421974738557652,
+      "grad_norm": 0.5802561044692993,
+      "learning_rate": 6.7622519195017165e-06,
+      "loss": 0.028,
+      "step": 8410
+    },
+    {
+      "epoch": 1.143555615917425,
+      "grad_norm": 0.1833496242761612,
+      "learning_rate": 6.679452222192684e-06,
+      "loss": 0.0306,
+      "step": 8420
+    },
+    {
+      "epoch": 1.1449137579790847,
+      "grad_norm": 0.5052464008331299,
+      "learning_rate": 6.597126271453579e-06,
+      "loss": 0.0272,
+      "step": 8430
+    },
+    {
+      "epoch": 1.1462719000407442,
+      "grad_norm": 0.6907228231430054,
+      "learning_rate": 6.51527496758782e-06,
+      "loss": 0.0273,
+      "step": 8440
+    },
+    {
+      "epoch": 1.147630042102404,
+      "grad_norm": 0.42919477820396423,
+      "learning_rate": 6.433899205708155e-06,
+      "loss": 0.0403,
+      "step": 8450
+    },
+    {
+      "epoch": 1.1489881841640637,
+      "grad_norm": 0.8584298491477966,
+      "learning_rate": 6.352999875726856e-06,
+      "loss": 0.0551,
+      "step": 8460
+    },
+    {
+      "epoch": 1.1503463262257232,
+      "grad_norm": 0.35242265462875366,
+      "learning_rate": 6.272577862346052e-06,
+      "loss": 0.0402,
+      "step": 8470
+    },
+    {
+      "epoch": 1.151704468287383,
+      "grad_norm": 0.6699735522270203,
+      "learning_rate": 6.192634045047996e-06,
+      "loss": 0.028,
+      "step": 8480
+    },
+    {
+      "epoch": 1.1530626103490424,
+      "grad_norm": 0.6612886786460876,
+      "learning_rate": 6.113169298085458e-06,
+      "loss": 0.0421,
+      "step": 8490
+    },
+    {
+      "epoch": 1.1544207524107022,
+      "grad_norm": 0.8134211897850037,
+      "learning_rate": 6.034184490472195e-06,
+      "loss": 0.0218,
+      "step": 8500
+    },
+    {
+      "epoch": 1.1557788944723617,
+      "grad_norm": 0.5644611716270447,
+      "learning_rate": 5.955680485973386e-06,
+      "loss": 0.0353,
+      "step": 8510
+    },
+    {
+      "epoch": 1.1571370365340214,
+      "grad_norm": 0.2927319407463074,
+      "learning_rate": 5.877658143096265e-06,
+      "loss": 0.0269,
+      "step": 8520
+    },
+    {
+      "epoch": 1.1584951785956812,
+      "grad_norm": 0.7623069882392883,
+      "learning_rate": 5.800118315080661e-06,
+      "loss": 0.0448,
+      "step": 8530
+    },
+    {
+      "epoch": 1.1598533206573407,
+      "grad_norm": 0.4477306306362152,
+      "learning_rate": 5.723061849889716e-06,
+      "loss": 0.0417,
+      "step": 8540
+    },
+    {
+      "epoch": 1.1612114627190004,
+      "grad_norm": 0.5128278136253357,
+      "learning_rate": 5.646489590200604e-06,
+      "loss": 0.0284,
+      "step": 8550
+    },
+    {
+      "epoch": 1.1625696047806602,
+      "grad_norm": 0.44137388467788696,
+      "learning_rate": 5.570402373395256e-06,
+      "loss": 0.0236,
+      "step": 8560
+    },
+    {
+      "epoch": 1.1639277468423197,
+      "grad_norm": 0.4181285798549652,
+      "learning_rate": 5.494801031551305e-06,
+      "loss": 0.0465,
+      "step": 8570
+    },
+    {
+      "epoch": 1.1652858889039794,
+      "grad_norm": 0.4736121594905853,
+      "learning_rate": 5.41968639143291e-06,
+      "loss": 0.0235,
+      "step": 8580
+    },
+    {
+      "epoch": 1.166644030965639,
+      "grad_norm": 0.6000255942344666,
+      "learning_rate": 5.345059274481751e-06,
+      "loss": 0.0294,
+      "step": 8590
+    },
+    {
+      "epoch": 1.1680021730272987,
+      "grad_norm": 0.4622021019458771,
+      "learning_rate": 5.270920496808002e-06,
+      "loss": 0.0334,
+      "step": 8600
+    },
+    {
+      "epoch": 1.1693603150889582,
+      "grad_norm": 0.42154496908187866,
+      "learning_rate": 5.1972708691814695e-06,
+      "loss": 0.0347,
+      "step": 8610
+    },
+    {
+      "epoch": 1.170718457150618,
+      "grad_norm": 0.4720594882965088,
+      "learning_rate": 5.124111197022674e-06,
+      "loss": 0.0298,
+      "step": 8620
+    },
+    {
+      "epoch": 1.1720765992122777,
+      "grad_norm": 0.7988888621330261,
+      "learning_rate": 5.051442280394081e-06,
+      "loss": 0.0538,
+      "step": 8630
+    },
+    {
+      "epoch": 1.1734347412739372,
+      "grad_norm": 0.3680790066719055,
+      "learning_rate": 4.979264913991322e-06,
+      "loss": 0.0226,
+      "step": 8640
+    },
+    {
+      "epoch": 1.174792883335597,
+      "grad_norm": 0.4571763873100281,
+      "learning_rate": 4.907579887134489e-06,
+      "loss": 0.0181,
+      "step": 8650
+    },
+    {
+      "epoch": 1.1761510253972565,
+      "grad_norm": 0.25367143750190735,
+      "learning_rate": 4.836387983759572e-06,
+      "loss": 0.0258,
+      "step": 8660
+    },
+    {
+      "epoch": 1.1775091674589162,
+      "grad_norm": 0.30869343876838684,
+      "learning_rate": 4.765689982409816e-06,
+      "loss": 0.021,
+      "step": 8670
+    },
+    {
+      "epoch": 1.178867309520576,
+      "grad_norm": 0.534633457660675,
+      "learning_rate": 4.695486656227233e-06,
+      "loss": 0.0178,
+      "step": 8680
+    },
+    {
+      "epoch": 1.1802254515822355,
+      "grad_norm": 0.26899537444114685,
+      "learning_rate": 4.625778772944156e-06,
+      "loss": 0.0189,
+      "step": 8690
+    },
+    {
+      "epoch": 1.1815835936438952,
+      "grad_norm": 0.2262059450149536,
+      "learning_rate": 4.556567094874825e-06,
+      "loss": 0.0143,
+      "step": 8700
+    },
+    {
+      "epoch": 1.1829417357055547,
+      "grad_norm": 0.4114440083503723,
+      "learning_rate": 4.487852378907059e-06,
+      "loss": 0.0224,
+      "step": 8710
+    },
+    {
+      "epoch": 1.1842998777672145,
+      "grad_norm": 0.33333584666252136,
+      "learning_rate": 4.419635376493986e-06,
+      "loss": 0.0143,
+      "step": 8720
+    },
+    {
+      "epoch": 1.1856580198288742,
+      "grad_norm": 0.5655884742736816,
+      "learning_rate": 4.351916833645825e-06,
+      "loss": 0.0254,
+      "step": 8730
+    },
+    {
+      "epoch": 1.1870161618905337,
+      "grad_norm": 0.7168158292770386,
+      "learning_rate": 4.284697490921691e-06,
+      "loss": 0.0279,
+      "step": 8740
+    },
+    {
+      "epoch": 1.1883743039521935,
+      "grad_norm": 0.48739928007125854,
+      "learning_rate": 4.2179780834215585e-06,
+      "loss": 0.017,
+      "step": 8750
+    },
+    {
+      "epoch": 1.189732446013853,
+      "grad_norm": 0.3397294282913208,
+      "learning_rate": 4.151759340778178e-06,
+      "loss": 0.0267,
+      "step": 8760
+    },
+    {
+      "epoch": 1.1910905880755127,
+      "grad_norm": 0.5716491937637329,
+      "learning_rate": 4.086041987149109e-06,
+      "loss": 0.0308,
+      "step": 8770
+    },
+    {
+      "epoch": 1.1924487301371722,
+      "grad_norm": 0.6204937100410461,
+      "learning_rate": 4.020826741208811e-06,
+      "loss": 0.0218,
+      "step": 8780
+    },
+    {
+      "epoch": 1.193806872198832,
+      "grad_norm": 0.6282355785369873,
+      "learning_rate": 3.956114316140746e-06,
+      "loss": 0.0493,
+      "step": 8790
+    },
+    {
+      "epoch": 1.1951650142604917,
+      "grad_norm": 0.825742781162262,
+      "learning_rate": 3.891905419629643e-06,
+      "loss": 0.0266,
+      "step": 8800
+    },
+    {
+      "epoch": 1.1965231563221512,
+      "grad_norm": 0.35509225726127625,
+      "learning_rate": 3.8282007538536946e-06,
+      "loss": 0.0395,
+      "step": 8810
+    },
+    {
+      "epoch": 1.197881298383811,
+      "grad_norm": 0.3297845423221588,
+      "learning_rate": 3.7650010154769265e-06,
+      "loss": 0.0167,
+      "step": 8820
+    },
+    {
+      "epoch": 1.1992394404454707,
+      "grad_norm": 0.598496675491333,
+      "learning_rate": 3.7023068956415608e-06,
+      "loss": 0.0423,
+      "step": 8830
+    },
+    {
+      "epoch": 1.2005975825071302,
+      "grad_norm": 0.48857778310775757,
+      "learning_rate": 3.6401190799604303e-06,
+      "loss": 0.0291,
+      "step": 8840
+    },
+    {
+      "epoch": 1.20195572456879,
+      "grad_norm": 0.593111515045166,
+      "learning_rate": 3.578438248509536e-06,
+      "loss": 0.0348,
+      "step": 8850
+    },
+    {
+      "epoch": 1.2033138666304495,
+      "grad_norm": 0.32693949341773987,
+      "learning_rate": 3.5172650758205583e-06,
+      "loss": 0.0182,
+      "step": 8860
+    },
+    {
+      "epoch": 1.2046720086921092,
+      "grad_norm": 0.5082590579986572,
+      "learning_rate": 3.45660023087353e-06,
+      "loss": 0.0255,
+      "step": 8870
+    },
+    {
+      "epoch": 1.2060301507537687,
+      "grad_norm": 0.8863648176193237,
+      "learning_rate": 3.3964443770894528e-06,
+      "loss": 0.0222,
+      "step": 8880
+    },
+    {
+      "epoch": 1.2073882928154285,
+      "grad_norm": 0.4241459369659424,
+      "learning_rate": 3.3367981723231245e-06,
+      "loss": 0.0348,
+      "step": 8890
+    },
+    {
+      "epoch": 1.2087464348770882,
+      "grad_norm": 0.2838013768196106,
+      "learning_rate": 3.2776622688558746e-06,
+      "loss": 0.0352,
+      "step": 8900
+    },
+    {
+      "epoch": 1.2101045769387477,
+      "grad_norm": 0.4814279079437256,
+      "learning_rate": 3.2190373133884677e-06,
+      "loss": 0.0291,
+      "step": 8910
+    },
+    {
+      "epoch": 1.2114627190004075,
+      "grad_norm": 0.1825248748064041,
+      "learning_rate": 3.1609239470340446e-06,
+      "loss": 0.0204,
+      "step": 8920
+    },
+    {
+      "epoch": 1.2128208610620672,
+      "grad_norm": 0.5995619297027588,
+      "learning_rate": 3.1033228053110373e-06,
+      "loss": 0.0305,
+      "step": 8930
+    },
+    {
+      "epoch": 1.2141790031237267,
+      "grad_norm": 0.3867414891719818,
+      "learning_rate": 3.0462345181363314e-06,
+      "loss": 0.0161,
+      "step": 8940
+    },
+    {
+      "epoch": 1.2155371451853865,
+      "grad_norm": 0.4504413604736328,
+      "learning_rate": 2.9896597098182654e-06,
+      "loss": 0.0208,
+      "step": 8950
+    },
+    {
+      "epoch": 1.216895287247046,
+      "grad_norm": 1.0150460004806519,
+      "learning_rate": 2.933598999049891e-06,
+      "loss": 0.0419,
+      "step": 8960
+    },
+    {
+      "epoch": 1.2182534293087057,
+      "grad_norm": 0.384197860956192,
+      "learning_rate": 2.8780529989021697e-06,
+      "loss": 0.0279,
+      "step": 8970
+    },
+    {
+      "epoch": 1.2196115713703652,
+      "grad_norm": 0.7119029760360718,
+      "learning_rate": 2.823022316817242e-06,
+      "loss": 0.032,
+      "step": 8980
+    },
+    {
+      "epoch": 1.220969713432025,
+      "grad_norm": 0.5933554768562317,
+      "learning_rate": 2.7685075546018456e-06,
+      "loss": 0.0176,
+      "step": 8990
+    },
+    {
+      "epoch": 1.2223278554936847,
+      "grad_norm": 0.6779228448867798,
+      "learning_rate": 2.7145093084206598e-06,
+      "loss": 0.0236,
+      "step": 9000
+    },
+    {
+      "epoch": 1.2236859975553442,
+      "grad_norm": 0.8965504765510559,
+      "learning_rate": 2.661028168789892e-06,
+      "loss": 0.0385,
+      "step": 9010
+    },
+    {
+      "epoch": 1.225044139617004,
+      "grad_norm": 0.31116700172424316,
+      "learning_rate": 2.6080647205706855e-06,
+      "loss": 0.0601,
+      "step": 9020
+    },
+    {
+      "epoch": 1.2264022816786635,
+      "grad_norm": 0.3436891436576843,
+      "learning_rate": 2.555619542962834e-06,
+      "loss": 0.0173,
+      "step": 9030
+    },
+    {
+      "epoch": 1.2277604237403232,
+      "grad_norm": 0.3875465989112854,
+      "learning_rate": 2.503693209498409e-06,
+      "loss": 0.0323,
+      "step": 9040
+    },
+    {
+      "epoch": 1.229118565801983,
+      "grad_norm": 0.5696704983711243,
+      "learning_rate": 2.452286288035449e-06,
+      "loss": 0.0147,
+      "step": 9050
+    },
+    {
+      "epoch": 1.2304767078636425,
+      "grad_norm": 0.42824694514274597,
+      "learning_rate": 2.4013993407518363e-06,
+      "loss": 0.0327,
+      "step": 9060
+    },
+    {
+      "epoch": 1.2318348499253022,
+      "grad_norm": 0.7306038737297058,
+      "learning_rate": 2.351032924139063e-06,
+      "loss": 0.0429,
+      "step": 9070
+    },
+    {
+      "epoch": 1.2331929919869618,
+      "grad_norm": 0.6035571694374084,
+      "learning_rate": 2.30118758899619e-06,
+      "loss": 0.0258,
+      "step": 9080
+    },
+    {
+      "epoch": 1.2345511340486215,
+      "grad_norm": 0.274285227060318,
+      "learning_rate": 2.2518638804238157e-06,
+      "loss": 0.0753,
+      "step": 9090
+    },
+    {
+      "epoch": 1.2359092761102812,
+      "grad_norm": 0.2084880769252777,
+      "learning_rate": 2.203062337818118e-06,
+      "loss": 0.0202,
+      "step": 9100
+    },
+    {
+      "epoch": 1.2372674181719407,
+      "grad_norm": 0.39178210496902466,
+      "learning_rate": 2.1547834948649483e-06,
+      "loss": 0.0148,
+      "step": 9110
+    },
+    {
+      "epoch": 1.2386255602336005,
+      "grad_norm": 0.4211605191230774,
+      "learning_rate": 2.1070278795340017e-06,
+      "loss": 0.0225,
+      "step": 9120
+    },
+    {
+      "epoch": 1.23998370229526,
+      "grad_norm": 0.45165959000587463,
+      "learning_rate": 2.059796014073029e-06,
+      "loss": 0.0272,
+      "step": 9130
+    },
+    {
+      "epoch": 1.2413418443569197,
+      "grad_norm": 0.3119717240333557,
+      "learning_rate": 2.01308841500214e-06,
+      "loss": 0.0404,
+      "step": 9140
+    },
+    {
+      "epoch": 1.2426999864185793,
+      "grad_norm": 0.5744400024414062,
+      "learning_rate": 1.9669055931081704e-06,
+      "loss": 0.0232,
+      "step": 9150
+    },
+    {
+      "epoch": 1.244058128480239,
+      "grad_norm": 0.6806374192237854,
+      "learning_rate": 1.9212480534390507e-06,
+      "loss": 0.0205,
+      "step": 9160
+    },
+    {
+      "epoch": 1.2454162705418987,
+      "grad_norm": 0.5028724074363708,
+      "learning_rate": 1.8761162952983246e-06,
+      "loss": 0.0227,
+      "step": 9170
+    },
+    {
+      "epoch": 1.2467744126035583,
+      "grad_norm": 0.6477677822113037,
+      "learning_rate": 1.8315108122396618e-06,
+      "loss": 0.0313,
+      "step": 9180
+    },
+    {
+      "epoch": 1.248132554665218,
+      "grad_norm": 0.29398593306541443,
+      "learning_rate": 1.787432092061475e-06,
+      "loss": 0.0268,
+      "step": 9190
+    },
+    {
+      "epoch": 1.2494906967268777,
+      "grad_norm": 0.5440655946731567,
+      "learning_rate": 1.743880616801602e-06,
+      "loss": 0.0215,
+      "step": 9200
+    },
+    {
+      "epoch": 1.2508488387885373,
+      "grad_norm": 0.41061124205589294,
+      "learning_rate": 1.7008568627319865e-06,
+      "loss": 0.0183,
+      "step": 9210
+    },
+    {
+      "epoch": 1.252206980850197,
+      "grad_norm": 0.24148662388324738,
+      "learning_rate": 1.6583613003535226e-06,
+      "loss": 0.0218,
+      "step": 9220
+    },
+    {
+      "epoch": 1.2535651229118565,
+      "grad_norm": 0.627099335193634,
+      "learning_rate": 1.6163943943908522e-06,
+      "loss": 0.0227,
+      "step": 9230
+    },
+    {
+      "epoch": 1.2549232649735163,
+      "grad_norm": 0.4970042109489441,
+      "learning_rate": 1.5749566037873476e-06,
+      "loss": 0.0288,
+      "step": 9240
+    },
+    {
+      "epoch": 1.2562814070351758,
+      "grad_norm": 0.6812019944190979,
+      "learning_rate": 1.5340483817000428e-06,
+      "loss": 0.0216,
+      "step": 9250
+    },
+    {
+      "epoch": 1.2576395490968355,
+      "grad_norm": 0.5812270045280457,
+      "learning_rate": 1.4936701754947101e-06,
+      "loss": 0.0245,
+      "step": 9260
+    },
+    {
+      "epoch": 1.2589976911584952,
+      "grad_norm": 0.618738055229187,
+      "learning_rate": 1.4538224267409361e-06,
+      "loss": 0.0293,
+      "step": 9270
+    },
+    {
+      "epoch": 1.2603558332201548,
+      "grad_norm": 0.784549355506897,
+      "learning_rate": 1.414505571207314e-06,
+      "loss": 0.028,
+      "step": 9280
+    },
+    {
+      "epoch": 1.2617139752818145,
+      "grad_norm": 0.6772966980934143,
+      "learning_rate": 1.3757200388566816e-06,
+      "loss": 0.029,
+      "step": 9290
+    },
+    {
+      "epoch": 1.2630721173434742,
+      "grad_norm": 0.37525832653045654,
+      "learning_rate": 1.3374662538414074e-06,
+      "loss": 0.0235,
+      "step": 9300
+    },
+    {
+      "epoch": 1.2644302594051338,
+      "grad_norm": 0.2933613061904907,
+      "learning_rate": 1.2997446344987617e-06,
+      "loss": 0.0268,
+      "step": 9310
+    },
+    {
+      "epoch": 1.2657884014667935,
+      "grad_norm": 0.7963993549346924,
+      "learning_rate": 1.262555593346315e-06,
+      "loss": 0.0268,
+      "step": 9320
+    },
+    {
+      "epoch": 1.267146543528453,
+      "grad_norm": 0.458940714597702,
+      "learning_rate": 1.2258995370774685e-06,
+      "loss": 0.0265,
+      "step": 9330
+    },
+    {
+      "epoch": 1.2685046855901128,
+      "grad_norm": 0.8191608786582947,
+      "learning_rate": 1.1897768665569798e-06,
+      "loss": 0.0161,
+      "step": 9340
+    },
+    {
+      "epoch": 1.2698628276517723,
+      "grad_norm": 0.18732419610023499,
+      "learning_rate": 1.1541879768165954e-06,
+      "loss": 0.033,
+      "step": 9350
+    },
+    {
+      "epoch": 1.271220969713432,
+      "grad_norm": 0.3815140128135681,
+      "learning_rate": 1.1191332570507085e-06,
+      "loss": 0.0227,
+      "step": 9360
+    },
+    {
+      "epoch": 1.2725791117750918,
+      "grad_norm": 0.6988071799278259,
+      "learning_rate": 1.0846130906121132e-06,
+      "loss": 0.036,
+      "step": 9370
+    },
+    {
+      "epoch": 1.2739372538367513,
+      "grad_norm": 0.4975379705429077,
+      "learning_rate": 1.0506278550078131e-06,
+      "loss": 0.0252,
+      "step": 9380
+    },
+    {
+      "epoch": 1.275295395898411,
+      "grad_norm": 0.8296396136283875,
+      "learning_rate": 1.0171779218949185e-06,
+      "loss": 0.0315,
+      "step": 9390
+    },
+    {
+      "epoch": 1.2766535379600708,
+      "grad_norm": 0.9181087613105774,
+      "learning_rate": 9.842636570765174e-07,
+      "loss": 0.0284,
+      "step": 9400
+    },
+    {
+      "epoch": 1.2780116800217303,
+      "grad_norm": 0.3783533573150635,
+      "learning_rate": 9.518854204977612e-07,
+      "loss": 0.0174,
+      "step": 9410
+    },
+    {
+      "epoch": 1.2793698220833898,
+      "grad_norm": 0.8029483556747437,
+      "learning_rate": 9.200435662418349e-07,
+      "loss": 0.0472,
+      "step": 9420
+    },
+    {
+      "epoch": 1.2807279641450495,
+      "grad_norm": 0.5075349807739258,
+      "learning_rate": 8.887384425261658e-07,
+      "loss": 0.0226,
+      "step": 9430
+    },
+    {
+      "epoch": 1.2820861062067093,
+      "grad_norm": 0.5316408276557922,
+      "learning_rate": 8.579703916985648e-07,
+      "loss": 0.0222,
+      "step": 9440
+    },
+    {
+      "epoch": 1.2834442482683688,
+      "grad_norm": 0.30262675881385803,
+      "learning_rate": 8.277397502335194e-07,
+      "loss": 0.0221,
+      "step": 9450
+    },
+    {
+      "epoch": 1.2848023903300285,
+      "grad_norm": 0.2954307496547699,
+      "learning_rate": 7.980468487284675e-07,
+      "loss": 0.0472,
+      "step": 9460
+    },
+    {
+      "epoch": 1.2861605323916883,
+      "grad_norm": 0.4719064235687256,
+      "learning_rate": 7.688920119002297e-07,
+      "loss": 0.0226,
+      "step": 9470
+    },
+    {
+      "epoch": 1.2875186744533478,
+      "grad_norm": 0.31883105635643005,
+      "learning_rate": 7.402755585814269e-07,
+      "loss": 0.0189,
+      "step": 9480
+    },
+    {
+      "epoch": 1.2888768165150075,
+      "grad_norm": 0.3226218819618225,
+      "learning_rate": 7.121978017170073e-07,
+      "loss": 0.0168,
+      "step": 9490
+    },
+    {
+      "epoch": 1.2902349585766673,
+      "grad_norm": 0.3375738859176636,
+      "learning_rate": 6.846590483608306e-07,
+      "loss": 0.0184,
+      "step": 9500
+    },
+    {
+      "epoch": 1.2915931006383268,
+      "grad_norm": 0.3047260344028473,
+      "learning_rate": 6.576595996722834e-07,
+      "loss": 0.0205,
+      "step": 9510
+    },
+    {
+      "epoch": 1.2929512426999863,
+      "grad_norm": 0.4995407462120056,
+      "learning_rate": 6.311997509130141e-07,
+      "loss": 0.0361,
+      "step": 9520
+    },
+    {
+      "epoch": 1.294309384761646,
+      "grad_norm": 0.6594052910804749,
+      "learning_rate": 6.052797914436803e-07,
+      "loss": 0.0416,
+      "step": 9530
+    },
+    {
+      "epoch": 1.2956675268233058,
+      "grad_norm": 0.897195041179657,
+      "learning_rate": 5.799000047208181e-07,
+      "loss": 0.0213,
+      "step": 9540
+    },
+    {
+      "epoch": 1.2970256688849653,
+      "grad_norm": 0.4758073091506958,
+      "learning_rate": 5.550606682937054e-07,
+      "loss": 0.0342,
+      "step": 9550
+    },
+    {
+      "epoch": 1.298383810946625,
+      "grad_norm": 0.8537939190864563,
+      "learning_rate": 5.307620538013481e-07,
+      "loss": 0.0207,
+      "step": 9560
+    },
+    {
+      "epoch": 1.2997419530082848,
+      "grad_norm": 0.3283558487892151,
+      "learning_rate": 5.070044269694874e-07,
+      "loss": 0.0198,
+      "step": 9570
+    },
+    {
+      "epoch": 1.3011000950699443,
+      "grad_norm": 0.270744651556015,
+      "learning_rate": 4.837880476077417e-07,
+      "loss": 0.031,
+      "step": 9580
+    },
+    {
+      "epoch": 1.302458237131604,
+      "grad_norm": 0.6325442790985107,
+      "learning_rate": 4.6111316960670835e-07,
+      "loss": 0.0308,
+      "step": 9590
+    },
+    {
+      "epoch": 1.3038163791932635,
+      "grad_norm": 0.7390625476837158,
+      "learning_rate": 4.389800409352218e-07,
+      "loss": 0.0282,
+      "step": 9600
+    },
+    {
+      "epoch": 1.3051745212549233,
+      "grad_norm": 0.31492725014686584,
+      "learning_rate": 4.173889036376277e-07,
+      "loss": 0.0326,
+      "step": 9610
+    },
+    {
+      "epoch": 1.3065326633165828,
+      "grad_norm": 0.43414315581321716,
+      "learning_rate": 3.963399938311463e-07,
+      "loss": 0.0239,
+      "step": 9620
+    },
+    {
+      "epoch": 1.3078908053782425,
+      "grad_norm": 0.5755314230918884,
+      "learning_rate": 3.7583354170328545e-07,
+      "loss": 0.0236,
+      "step": 9630
+    },
+    {
+      "epoch": 1.3092489474399023,
+      "grad_norm": 0.6316108703613281,
+      "learning_rate": 3.558697715093207e-07,
+      "loss": 0.0474,
+      "step": 9640
+    },
+    {
+      "epoch": 1.3106070895015618,
+      "grad_norm": 0.9187036752700806,
+      "learning_rate": 3.3644890156983576e-07,
+      "loss": 0.0212,
+      "step": 9650
+    },
+    {
+      "epoch": 1.3119652315632215,
+      "grad_norm": 0.22826598584651947,
+      "learning_rate": 3.175711442683638e-07,
+      "loss": 0.0292,
+      "step": 9660
+    },
+    {
+      "epoch": 1.3133233736248813,
+      "grad_norm": 0.7134349346160889,
+      "learning_rate": 2.9923670604902197e-07,
+      "loss": 0.0386,
+      "step": 9670
+    },
+    {
+      "epoch": 1.3146815156865408,
+      "grad_norm": 0.8570911884307861,
+      "learning_rate": 2.814457874143028e-07,
+      "loss": 0.0239,
+      "step": 9680
+    },
+    {
+      "epoch": 1.3160396577482005,
+      "grad_norm": 0.08484259992837906,
+      "learning_rate": 2.641985829228366e-07,
+      "loss": 0.0284,
+      "step": 9690
+    },
+    {
+      "epoch": 1.31739779980986,
+      "grad_norm": 0.43286362290382385,
+      "learning_rate": 2.474952811872877e-07,
+      "loss": 0.0196,
+      "step": 9700
+    },
+    {
+      "epoch": 1.3187559418715198,
+      "grad_norm": 0.6312990784645081,
+      "learning_rate": 2.3133606487228397e-07,
+      "loss": 0.028,
+      "step": 9710
+    },
+    {
+      "epoch": 1.3201140839331793,
+      "grad_norm": 0.4311346709728241,
+      "learning_rate": 2.157211106924295e-07,
+      "loss": 0.0167,
+      "step": 9720
+    },
+    {
+      "epoch": 1.321472225994839,
+      "grad_norm": 0.5412164330482483,
+      "learning_rate": 2.006505894103672e-07,
+      "loss": 0.03,
+      "step": 9730
+    },
+    {
+      "epoch": 1.3228303680564988,
+      "grad_norm": 0.723097026348114,
+      "learning_rate": 1.8612466583489696e-07,
+      "loss": 0.0186,
+      "step": 9740
+    },
+    {
+      "epoch": 1.3241885101181583,
+      "grad_norm": 0.4908280670642853,
+      "learning_rate": 1.7214349881918834e-07,
+      "loss": 0.0156,
+      "step": 9750
+    },
+    {
+      "epoch": 1.325546652179818,
+      "grad_norm": 0.3422550857067108,
+      "learning_rate": 1.5870724125904845e-07,
+      "loss": 0.0282,
+      "step": 9760
+    },
+    {
+      "epoch": 1.3269047942414778,
+      "grad_norm": 0.36814767122268677,
+      "learning_rate": 1.4581604009124006e-07,
+      "loss": 0.0467,
+      "step": 9770
+    },
+    {
+      "epoch": 1.3282629363031373,
+      "grad_norm": 0.22741085290908813,
+      "learning_rate": 1.334700362918717e-07,
+      "loss": 0.0291,
+      "step": 9780
+    },
+    {
+      "epoch": 1.329621078364797,
+      "grad_norm": 0.6424657106399536,
+      "learning_rate": 1.2166936487486015e-07,
+      "loss": 0.0312,
+      "step": 9790
+    },
+    {
+      "epoch": 1.3309792204264566,
+      "grad_norm": 0.300128698348999,
+      "learning_rate": 1.1041415489045914e-07,
+      "loss": 0.0212,
+      "step": 9800
+    },
+    {
+      "epoch": 1.3323373624881163,
+      "grad_norm": 0.4029289484024048,
+      "learning_rate": 9.970452942384412e-08,
+      "loss": 0.023,
+      "step": 9810
+    },
+    {
+      "epoch": 1.3336955045497758,
+      "grad_norm": 0.780843198299408,
+      "learning_rate": 8.954060559375754e-08,
+      "loss": 0.0289,
+      "step": 9820
+    },
+    {
+      "epoch": 1.3350536466114356,
+      "grad_norm": 1.1147595643997192,
+      "learning_rate": 7.99224945512489e-08,
+      "loss": 0.0389,
+      "step": 9830
+    },
+    {
+      "epoch": 1.3364117886730953,
+      "grad_norm": 0.14290760457515717,
+      "learning_rate": 7.085030147843675e-08,
+      "loss": 0.0244,
+      "step": 9840
+    },
+    {
+      "epoch": 1.3377699307347548,
+      "grad_norm": 0.5255113244056702,
+      "learning_rate": 6.232412558736523e-08,
+      "loss": 0.0265,
+      "step": 9850
+    },
+    {
+      "epoch": 1.3391280727964145,
+      "grad_norm": 0.6413520574569702,
+      "learning_rate": 5.434406011893822e-08,
+      "loss": 0.0239,
+      "step": 9860
+    },
+    {
+      "epoch": 1.3404862148580743,
+      "grad_norm": 0.4884924590587616,
+      "learning_rate": 4.6910192341864664e-08,
+      "loss": 0.0253,
+      "step": 9870
+    },
+    {
+      "epoch": 1.3418443569197338,
+      "grad_norm": 0.48118856549263,
+      "learning_rate": 4.0022603551737035e-08,
+      "loss": 0.0411,
+      "step": 9880
+    },
+    {
+      "epoch": 1.3432024989813933,
+      "grad_norm": 0.3284280300140381,
+      "learning_rate": 3.3681369070120985e-08,
+      "loss": 0.029,
+      "step": 9890
+    },
+    {
+      "epoch": 1.344560641043053,
+      "grad_norm": 0.4529123604297638,
+      "learning_rate": 2.7886558243744866e-08,
+      "loss": 0.0253,
+      "step": 9900
+    },
+    {
+      "epoch": 1.3459187831047128,
+      "grad_norm": 0.4403287470340729,
+      "learning_rate": 2.2638234443722596e-08,
+      "loss": 0.0203,
+      "step": 9910
+    },
+    {
+      "epoch": 1.3472769251663723,
+      "grad_norm": 0.9165987968444824,
+      "learning_rate": 1.7936455064887504e-08,
+      "loss": 0.0227,
+      "step": 9920
+    },
+    {
+      "epoch": 1.348635067228032,
+      "grad_norm": 0.7782343029975891,
+      "learning_rate": 1.378127152514841e-08,
+      "loss": 0.0349,
+      "step": 9930
+    },
+    {
+      "epoch": 1.3499932092896918,
+      "grad_norm": 0.3763273358345032,
+      "learning_rate": 1.0172729264917857e-08,
+      "loss": 0.0332,
+      "step": 9940
+    },
+    {
+      "epoch": 1.3513513513513513,
+      "grad_norm": 0.2052953541278839,
+      "learning_rate": 7.1108677466458215e-09,
+      "loss": 0.0315,
+      "step": 9950
+    },
+    {
+      "epoch": 1.352709493413011,
+      "grad_norm": 0.7022479176521301,
+      "learning_rate": 4.595720454353414e-09,
+      "loss": 0.0244,
+      "step": 9960
+    },
+    {
+      "epoch": 1.3540676354746706,
+      "grad_norm": 0.2550441324710846,
+      "learning_rate": 2.627314893294264e-09,
+      "loss": 0.0182,
+      "step": 9970
+    },
+    {
+      "epoch": 1.3554257775363303,
+      "grad_norm": 0.5177278518676758,
+      "learning_rate": 1.2056725896270048e-09,
+      "loss": 0.0346,
+      "step": 9980
+    },
+    {
+      "epoch": 1.3567839195979898,
+      "grad_norm": 0.2763776481151581,
+      "learning_rate": 3.308090902098826e-10,
+      "loss": 0.0251,
+      "step": 9990
+    },
+    {
+      "epoch": 1.3581420616596496,
+      "grad_norm": 0.6695942878723145,
+      "learning_rate": 2.7339624120159555e-12,
+      "loss": 0.0422,
+      "step": 10000
+    },
+    {
+      "epoch": 1.3581420616596496,
+      "step": 10000,
+      "total_flos": 0.0,
+      "train_loss": 0.06336962224394083,
+      "train_runtime": 4217.7139,
+      "train_samples_per_second": 9.484,
+      "train_steps_per_second": 2.371
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}