diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5278 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 3728,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.000536480686695279,
+      "grad_norm": 2.0863339521447086,
+      "learning_rate": 2.6809651474530834e-08,
+      "loss": 0.2851,
+      "step": 1
+    },
+    {
+      "epoch": 0.002682403433476395,
+      "grad_norm": 2.019867688346717,
+      "learning_rate": 1.3404825737265417e-07,
+      "loss": 0.281,
+      "step": 5
+    },
+    {
+      "epoch": 0.00536480686695279,
+      "grad_norm": 2.043241454451721,
+      "learning_rate": 2.6809651474530835e-07,
+      "loss": 0.272,
+      "step": 10
+    },
+    {
+      "epoch": 0.008047210300429184,
+      "grad_norm": 1.8858706299088983,
+      "learning_rate": 4.021447721179625e-07,
+      "loss": 0.2711,
+      "step": 15
+    },
+    {
+      "epoch": 0.01072961373390558,
+      "grad_norm": 1.7673503995687727,
+      "learning_rate": 5.361930294906167e-07,
+      "loss": 0.2735,
+      "step": 20
+    },
+    {
+      "epoch": 0.013412017167381975,
+      "grad_norm": 2.323464675632955,
+      "learning_rate": 6.702412868632709e-07,
+      "loss": 0.2725,
+      "step": 25
+    },
+    {
+      "epoch": 0.016094420600858368,
+      "grad_norm": 2.5673838750685647,
+      "learning_rate": 8.04289544235925e-07,
+      "loss": 0.2744,
+      "step": 30
+    },
+    {
+      "epoch": 0.018776824034334765,
+      "grad_norm": 1.9187222847932,
+      "learning_rate": 9.383378016085791e-07,
+      "loss": 0.2761,
+      "step": 35
+    },
+    {
+      "epoch": 0.02145922746781116,
+      "grad_norm": 1.8984966074581404,
+      "learning_rate": 1.0723860589812334e-06,
+      "loss": 0.2655,
+      "step": 40
+    },
+    {
+      "epoch": 0.024141630901287552,
+      "grad_norm": 1.8659220642427434,
+      "learning_rate": 1.2064343163538874e-06,
+      "loss": 0.2793,
+      "step": 45
+    },
+    {
+      "epoch": 0.02682403433476395,
+      "grad_norm": 1.9238255304447283,
+      "learning_rate": 1.3404825737265418e-06,
+      "loss": 0.2669,
+      "step": 50
+    },
+    {
+      "epoch": 0.029506437768240343,
+      "grad_norm": 1.8501661181906428,
+      "learning_rate": 1.4745308310991958e-06,
+      "loss": 0.2711,
+      "step": 55
+    },
+    {
+      "epoch": 0.032188841201716736,
+      "grad_norm": 1.9878940051253793,
+      "learning_rate": 1.60857908847185e-06,
+      "loss": 0.26,
+      "step": 60
+    },
+    {
+      "epoch": 0.03487124463519313,
+      "grad_norm": 2.094483621764117,
+      "learning_rate": 1.7426273458445042e-06,
+      "loss": 0.2694,
+      "step": 65
+    },
+    {
+      "epoch": 0.03755364806866953,
+      "grad_norm": 2.058467307343078,
+      "learning_rate": 1.8766756032171582e-06,
+      "loss": 0.266,
+      "step": 70
+    },
+    {
+      "epoch": 0.040236051502145924,
+      "grad_norm": 2.008096065466058,
+      "learning_rate": 2.0107238605898126e-06,
+      "loss": 0.2623,
+      "step": 75
+    },
+    {
+      "epoch": 0.04291845493562232,
+      "grad_norm": 1.849039686452229,
+      "learning_rate": 2.1447721179624668e-06,
+      "loss": 0.2677,
+      "step": 80
+    },
+    {
+      "epoch": 0.04560085836909871,
+      "grad_norm": 2.092389670059048,
+      "learning_rate": 2.278820375335121e-06,
+      "loss": 0.2618,
+      "step": 85
+    },
+    {
+      "epoch": 0.048283261802575105,
+      "grad_norm": 1.8886682449515093,
+      "learning_rate": 2.4128686327077747e-06,
+      "loss": 0.2657,
+      "step": 90
+    },
+    {
+      "epoch": 0.050965665236051505,
+      "grad_norm": 2.2114135594284305,
+      "learning_rate": 2.5469168900804294e-06,
+      "loss": 0.2676,
+      "step": 95
+    },
+    {
+      "epoch": 0.0536480686695279,
+      "grad_norm": 1.9846013216504284,
+      "learning_rate": 2.6809651474530836e-06,
+      "loss": 0.269,
+      "step": 100
+    },
+    {
+      "epoch": 0.05633047210300429,
+      "grad_norm": 1.9945348953325985,
+      "learning_rate": 2.8150134048257378e-06,
+      "loss": 0.2681,
+      "step": 105
+    },
+    {
+      "epoch": 0.059012875536480686,
+      "grad_norm": 1.8180659313873635,
+      "learning_rate": 2.9490616621983915e-06,
+      "loss": 0.2704,
+      "step": 110
+    },
+    {
+      "epoch": 0.06169527896995708,
+      "grad_norm": 2.225519442017157,
+      "learning_rate": 3.0831099195710457e-06,
+      "loss": 0.2731,
+      "step": 115
+    },
+    {
+      "epoch": 0.06437768240343347,
+      "grad_norm": 2.2160771277529117,
+      "learning_rate": 3.2171581769437e-06,
+      "loss": 0.2686,
+      "step": 120
+    },
+    {
+      "epoch": 0.06706008583690987,
+      "grad_norm": 1.9249390642006592,
+      "learning_rate": 3.351206434316354e-06,
+      "loss": 0.2761,
+      "step": 125
+    },
+    {
+      "epoch": 0.06974248927038626,
+      "grad_norm": 1.9328682746531594,
+      "learning_rate": 3.4852546916890083e-06,
+      "loss": 0.2714,
+      "step": 130
+    },
+    {
+      "epoch": 0.07242489270386267,
+      "grad_norm": 2.1186813690148214,
+      "learning_rate": 3.6193029490616625e-06,
+      "loss": 0.2701,
+      "step": 135
+    },
+    {
+      "epoch": 0.07510729613733906,
+      "grad_norm": 2.1933333909107287,
+      "learning_rate": 3.7533512064343163e-06,
+      "loss": 0.2791,
+      "step": 140
+    },
+    {
+      "epoch": 0.07778969957081545,
+      "grad_norm": 2.129819677044279,
+      "learning_rate": 3.8873994638069705e-06,
+      "loss": 0.2797,
+      "step": 145
+    },
+    {
+      "epoch": 0.08047210300429185,
+      "grad_norm": 2.067811024124003,
+      "learning_rate": 4.021447721179625e-06,
+      "loss": 0.2762,
+      "step": 150
+    },
+    {
+      "epoch": 0.08315450643776824,
+      "grad_norm": 2.1707210015708083,
+      "learning_rate": 4.155495978552279e-06,
+      "loss": 0.279,
+      "step": 155
+    },
+    {
+      "epoch": 0.08583690987124463,
+      "grad_norm": 2.2019940721149007,
+      "learning_rate": 4.2895442359249335e-06,
+      "loss": 0.2738,
+      "step": 160
+    },
+    {
+      "epoch": 0.08851931330472103,
+      "grad_norm": 2.2099864687326662,
+      "learning_rate": 4.423592493297587e-06,
+      "loss": 0.268,
+      "step": 165
+    },
+    {
+      "epoch": 0.09120171673819742,
+      "grad_norm": 2.22468718531686,
+      "learning_rate": 4.557640750670242e-06,
+      "loss": 0.2829,
+      "step": 170
+    },
+    {
+      "epoch": 0.09388412017167382,
+      "grad_norm": 2.1800134691918767,
+      "learning_rate": 4.691689008042896e-06,
+      "loss": 0.2856,
+      "step": 175
+    },
+    {
+      "epoch": 0.09656652360515021,
+      "grad_norm": 2.103884261534838,
+      "learning_rate": 4.8257372654155495e-06,
+      "loss": 0.2791,
+      "step": 180
+    },
+    {
+      "epoch": 0.0992489270386266,
+      "grad_norm": 2.0714593934541408,
+      "learning_rate": 4.959785522788204e-06,
+      "loss": 0.273,
+      "step": 185
+    },
+    {
+      "epoch": 0.10193133047210301,
+      "grad_norm": 2.058846978716926,
+      "learning_rate": 5.093833780160859e-06,
+      "loss": 0.284,
+      "step": 190
+    },
+    {
+      "epoch": 0.1046137339055794,
+      "grad_norm": 2.076409950200697,
+      "learning_rate": 5.2278820375335125e-06,
+      "loss": 0.2792,
+      "step": 195
+    },
+    {
+      "epoch": 0.1072961373390558,
+      "grad_norm": 2.1243909119008766,
+      "learning_rate": 5.361930294906167e-06,
+      "loss": 0.282,
+      "step": 200
+    },
+    {
+      "epoch": 0.10997854077253219,
+      "grad_norm": 2.3388393818726585,
+      "learning_rate": 5.495978552278821e-06,
+      "loss": 0.2925,
+      "step": 205
+    },
+    {
+      "epoch": 0.11266094420600858,
+      "grad_norm": 2.451045371521958,
+      "learning_rate": 5.6300268096514755e-06,
+      "loss": 0.2994,
+      "step": 210
+    },
+    {
+      "epoch": 0.11534334763948498,
+      "grad_norm": 2.1487473131618255,
+      "learning_rate": 5.764075067024129e-06,
+      "loss": 0.2848,
+      "step": 215
+    },
+    {
+      "epoch": 0.11802575107296137,
+      "grad_norm": 2.4814960698366604,
+      "learning_rate": 5.898123324396783e-06,
+      "loss": 0.2774,
+      "step": 220
+    },
+    {
+      "epoch": 0.12070815450643776,
+      "grad_norm": 2.410915656859707,
+      "learning_rate": 6.032171581769437e-06,
+      "loss": 0.2809,
+      "step": 225
+    },
+    {
+      "epoch": 0.12339055793991416,
+      "grad_norm": 2.396133050866378,
+      "learning_rate": 6.1662198391420915e-06,
+      "loss": 0.2849,
+      "step": 230
+    },
+    {
+      "epoch": 0.12607296137339055,
+      "grad_norm": 2.2801377751578578,
+      "learning_rate": 6.300268096514745e-06,
+      "loss": 0.2918,
+      "step": 235
+    },
+    {
+      "epoch": 0.12875536480686695,
+      "grad_norm": 2.34490728705815,
+      "learning_rate": 6.4343163538874e-06,
+      "loss": 0.2987,
+      "step": 240
+    },
+    {
+      "epoch": 0.13143776824034334,
+      "grad_norm": 2.3642964302216214,
+      "learning_rate": 6.5683646112600545e-06,
+      "loss": 0.2889,
+      "step": 245
+    },
+    {
+      "epoch": 0.13412017167381973,
+      "grad_norm": 2.380305337416477,
+      "learning_rate": 6.702412868632708e-06,
+      "loss": 0.2938,
+      "step": 250
+    },
+    {
+      "epoch": 0.13680257510729613,
+      "grad_norm": 2.2797615102374853,
+      "learning_rate": 6.836461126005363e-06,
+      "loss": 0.2839,
+      "step": 255
+    },
+    {
+      "epoch": 0.13948497854077252,
+      "grad_norm": 2.3413285353478455,
+      "learning_rate": 6.970509383378017e-06,
+      "loss": 0.2919,
+      "step": 260
+    },
+    {
+      "epoch": 0.14216738197424894,
+      "grad_norm": 2.3862345681550874,
+      "learning_rate": 7.104557640750671e-06,
+      "loss": 0.2937,
+      "step": 265
+    },
+    {
+      "epoch": 0.14484978540772533,
+      "grad_norm": 2.3320150218284303,
+      "learning_rate": 7.238605898123325e-06,
+      "loss": 0.2985,
+      "step": 270
+    },
+    {
+      "epoch": 0.14753218884120173,
+      "grad_norm": 2.35670634905087,
+      "learning_rate": 7.37265415549598e-06,
+      "loss": 0.2989,
+      "step": 275
+    },
+    {
+      "epoch": 0.15021459227467812,
+      "grad_norm": 2.416487625597099,
+      "learning_rate": 7.506702412868633e-06,
+      "loss": 0.3064,
+      "step": 280
+    },
+    {
+      "epoch": 0.15289699570815452,
+      "grad_norm": 2.433303207584211,
+      "learning_rate": 7.640750670241287e-06,
+      "loss": 0.2919,
+      "step": 285
+    },
+    {
+      "epoch": 0.1555793991416309,
+      "grad_norm": 2.3339942787804766,
+      "learning_rate": 7.774798927613941e-06,
+      "loss": 0.2992,
+      "step": 290
+    },
+    {
+      "epoch": 0.1582618025751073,
+      "grad_norm": 2.359252347466936,
+      "learning_rate": 7.908847184986595e-06,
+      "loss": 0.3022,
+      "step": 295
+    },
+    {
+      "epoch": 0.1609442060085837,
+      "grad_norm": 2.584700516100613,
+      "learning_rate": 8.04289544235925e-06,
+      "loss": 0.2936,
+      "step": 300
+    },
+    {
+      "epoch": 0.1636266094420601,
+      "grad_norm": 2.6035628076524473,
+      "learning_rate": 8.176943699731904e-06,
+      "loss": 0.3063,
+      "step": 305
+    },
+    {
+      "epoch": 0.16630901287553648,
+      "grad_norm": 2.344260798418792,
+      "learning_rate": 8.310991957104558e-06,
+      "loss": 0.3025,
+      "step": 310
+    },
+    {
+      "epoch": 0.16899141630901288,
+      "grad_norm": 2.3092835960344074,
+      "learning_rate": 8.445040214477213e-06,
+      "loss": 0.3088,
+      "step": 315
+    },
+    {
+      "epoch": 0.17167381974248927,
+      "grad_norm": 2.4791987185928663,
+      "learning_rate": 8.579088471849867e-06,
+      "loss": 0.3066,
+      "step": 320
+    },
+    {
+      "epoch": 0.17435622317596566,
+      "grad_norm": 2.4438632774093674,
+      "learning_rate": 8.71313672922252e-06,
+      "loss": 0.3023,
+      "step": 325
+    },
+    {
+      "epoch": 0.17703862660944206,
+      "grad_norm": 2.6299682555734147,
+      "learning_rate": 8.847184986595175e-06,
+      "loss": 0.3067,
+      "step": 330
+    },
+    {
+      "epoch": 0.17972103004291845,
+      "grad_norm": 2.3929731274938097,
+      "learning_rate": 8.98123324396783e-06,
+      "loss": 0.2967,
+      "step": 335
+    },
+    {
+      "epoch": 0.18240343347639484,
+      "grad_norm": 2.489569539616038,
+      "learning_rate": 9.115281501340484e-06,
+      "loss": 0.3028,
+      "step": 340
+    },
+    {
+      "epoch": 0.18508583690987124,
+      "grad_norm": 2.4774372151229436,
+      "learning_rate": 9.249329758713138e-06,
+      "loss": 0.3184,
+      "step": 345
+    },
+    {
+      "epoch": 0.18776824034334763,
+      "grad_norm": 2.6390058785683244,
+      "learning_rate": 9.383378016085791e-06,
+      "loss": 0.3129,
+      "step": 350
+    },
+    {
+      "epoch": 0.19045064377682402,
+      "grad_norm": 2.559704702422249,
+      "learning_rate": 9.517426273458445e-06,
+      "loss": 0.3189,
+      "step": 355
+    },
+    {
+      "epoch": 0.19313304721030042,
+      "grad_norm": 2.5923004925849202,
+      "learning_rate": 9.651474530831099e-06,
+      "loss": 0.3179,
+      "step": 360
+    },
+    {
+      "epoch": 0.1958154506437768,
+      "grad_norm": 2.5402736760924305,
+      "learning_rate": 9.785522788203754e-06,
+      "loss": 0.3149,
+      "step": 365
+    },
+    {
+      "epoch": 0.1984978540772532,
+      "grad_norm": 2.425562826716764,
+      "learning_rate": 9.919571045576408e-06,
+      "loss": 0.3199,
+      "step": 370
+    },
+    {
+      "epoch": 0.20118025751072963,
+      "grad_norm": 2.5852834491116146,
+      "learning_rate": 9.999991231716779e-06,
+      "loss": 0.3229,
+      "step": 375
+    },
+    {
+      "epoch": 0.20386266094420602,
+      "grad_norm": 2.5995942631655193,
+      "learning_rate": 9.999892588883699e-06,
+      "loss": 0.318,
+      "step": 380
+    },
+    {
+      "epoch": 0.2065450643776824,
+      "grad_norm": 2.7426449146592917,
+      "learning_rate": 9.99968434503304e-06,
+      "loss": 0.3207,
+      "step": 385
+    },
+    {
+      "epoch": 0.2092274678111588,
+      "grad_norm": 2.589479755333145,
+      "learning_rate": 9.999366504729645e-06,
+      "loss": 0.3261,
+      "step": 390
+    },
+    {
+      "epoch": 0.2119098712446352,
+      "grad_norm": 2.5197646152950184,
+      "learning_rate": 9.998939074940788e-06,
+      "loss": 0.3156,
+      "step": 395
+    },
+    {
+      "epoch": 0.2145922746781116,
+      "grad_norm": 2.652204652785218,
+      "learning_rate": 9.998402065036018e-06,
+      "loss": 0.3287,
+      "step": 400
+    },
+    {
+      "epoch": 0.217274678111588,
+      "grad_norm": 2.941963697553996,
+      "learning_rate": 9.997755486786954e-06,
+      "loss": 0.3152,
+      "step": 405
+    },
+    {
+      "epoch": 0.21995708154506438,
+      "grad_norm": 2.514230531663027,
+      "learning_rate": 9.996999354367028e-06,
+      "loss": 0.3224,
+      "step": 410
+    },
+    {
+      "epoch": 0.22263948497854077,
+      "grad_norm": 2.613820892247126,
+      "learning_rate": 9.996133684351172e-06,
+      "loss": 0.3223,
+      "step": 415
+    },
+    {
+      "epoch": 0.22532188841201717,
+      "grad_norm": 2.60586168357607,
+      "learning_rate": 9.995158495715459e-06,
+      "loss": 0.3304,
+      "step": 420
+    },
+    {
+      "epoch": 0.22800429184549356,
+      "grad_norm": 2.73014784630188,
+      "learning_rate": 9.994073809836677e-06,
+      "loss": 0.3233,
+      "step": 425
+    },
+    {
+      "epoch": 0.23068669527896996,
+      "grad_norm": 2.808246799344738,
+      "learning_rate": 9.992879650491877e-06,
+      "loss": 0.3287,
+      "step": 430
+    },
+    {
+      "epoch": 0.23336909871244635,
+      "grad_norm": 2.4427023808762733,
+      "learning_rate": 9.991576043857833e-06,
+      "loss": 0.3246,
+      "step": 435
+    },
+    {
+      "epoch": 0.23605150214592274,
+      "grad_norm": 2.8001081105122734,
+      "learning_rate": 9.990163018510484e-06,
+      "loss": 0.3254,
+      "step": 440
+    },
+    {
+      "epoch": 0.23873390557939914,
+      "grad_norm": 2.5756753850013236,
+      "learning_rate": 9.988640605424298e-06,
+      "loss": 0.3303,
+      "step": 445
+    },
+    {
+      "epoch": 0.24141630901287553,
+      "grad_norm": 2.6382834663452464,
+      "learning_rate": 9.987008837971595e-06,
+      "loss": 0.3259,
+      "step": 450
+    },
+    {
+      "epoch": 0.24409871244635192,
+      "grad_norm": 2.874308167284836,
+      "learning_rate": 9.98526775192182e-06,
+      "loss": 0.3287,
+      "step": 455
+    },
+    {
+      "epoch": 0.24678111587982832,
+      "grad_norm": 2.577949722207028,
+      "learning_rate": 9.983417385440755e-06,
+      "loss": 0.3285,
+      "step": 460
+    },
+    {
+      "epoch": 0.2494635193133047,
+      "grad_norm": 2.605209899345705,
+      "learning_rate": 9.981457779089678e-06,
+      "loss": 0.3393,
+      "step": 465
+    },
+    {
+      "epoch": 0.2521459227467811,
+      "grad_norm": 2.669405429442268,
+      "learning_rate": 9.979388975824485e-06,
+      "loss": 0.3297,
+      "step": 470
+    },
+    {
+      "epoch": 0.2548283261802575,
+      "grad_norm": 2.6050099070733737,
+      "learning_rate": 9.977211020994735e-06,
+      "loss": 0.3267,
+      "step": 475
+    },
+    {
+      "epoch": 0.2575107296137339,
+      "grad_norm": 2.5738484109153066,
+      "learning_rate": 9.97492396234267e-06,
+      "loss": 0.3241,
+      "step": 480
+    },
+    {
+      "epoch": 0.2601931330472103,
+      "grad_norm": 2.6751920468972887,
+      "learning_rate": 9.972527850002154e-06,
+      "loss": 0.3211,
+      "step": 485
+    },
+    {
+      "epoch": 0.2628755364806867,
+      "grad_norm": 2.6515899868690687,
+      "learning_rate": 9.970022736497588e-06,
+      "loss": 0.3343,
+      "step": 490
+    },
+    {
+      "epoch": 0.2655579399141631,
+      "grad_norm": 2.833435887671815,
+      "learning_rate": 9.96740867674275e-06,
+      "loss": 0.3213,
+      "step": 495
+    },
+    {
+      "epoch": 0.26824034334763946,
+      "grad_norm": 2.5723877096036127,
+      "learning_rate": 9.964685728039596e-06,
+      "loss": 0.3273,
+      "step": 500
+    },
+    {
+      "epoch": 0.2709227467811159,
+      "grad_norm": 2.421773456878221,
+      "learning_rate": 9.961853950076992e-06,
+      "loss": 0.325,
+      "step": 505
+    },
+    {
+      "epoch": 0.27360515021459225,
+      "grad_norm": 2.72433255261523,
+      "learning_rate": 9.958913404929423e-06,
+      "loss": 0.3259,
+      "step": 510
+    },
+    {
+      "epoch": 0.2762875536480687,
+      "grad_norm": 2.511740357955884,
+      "learning_rate": 9.955864157055623e-06,
+      "loss": 0.3335,
+      "step": 515
+    },
+    {
+      "epoch": 0.27896995708154504,
+      "grad_norm": 2.570325380578064,
+      "learning_rate": 9.95270627329716e-06,
+      "loss": 0.3238,
+      "step": 520
+    },
+    {
+      "epoch": 0.28165236051502146,
+      "grad_norm": 2.807983699039292,
+      "learning_rate": 9.949439822876975e-06,
+      "loss": 0.3338,
+      "step": 525
+    },
+    {
+      "epoch": 0.2843347639484979,
+      "grad_norm": 2.66736822736115,
+      "learning_rate": 9.94606487739787e-06,
+      "loss": 0.3377,
+      "step": 530
+    },
+    {
+      "epoch": 0.28701716738197425,
+      "grad_norm": 2.603191440095429,
+      "learning_rate": 9.942581510840919e-06,
+      "loss": 0.3348,
+      "step": 535
+    },
+    {
+      "epoch": 0.28969957081545067,
+      "grad_norm": 2.5473646746035414,
+      "learning_rate": 9.93898979956387e-06,
+      "loss": 0.3261,
+      "step": 540
+    },
+    {
+      "epoch": 0.29238197424892703,
+      "grad_norm": 2.599401887659365,
+      "learning_rate": 9.935289822299456e-06,
+      "loss": 0.33,
+      "step": 545
+    },
+    {
+      "epoch": 0.29506437768240346,
+      "grad_norm": 2.40975591177543,
+      "learning_rate": 9.931481660153672e-06,
+      "loss": 0.3286,
+      "step": 550
+    },
+    {
+      "epoch": 0.2977467811158798,
+      "grad_norm": 2.4894490507567197,
+      "learning_rate": 9.927565396604001e-06,
+      "loss": 0.3291,
+      "step": 555
+    },
+    {
+      "epoch": 0.30042918454935624,
+      "grad_norm": 2.5722016173007125,
+      "learning_rate": 9.923541117497586e-06,
+      "loss": 0.337,
+      "step": 560
+    },
+    {
+      "epoch": 0.3031115879828326,
+      "grad_norm": 2.5550725925072877,
+      "learning_rate": 9.919408911049333e-06,
+      "loss": 0.3346,
+      "step": 565
+    },
+    {
+      "epoch": 0.30579399141630903,
+      "grad_norm": 2.5433106872337072,
+      "learning_rate": 9.915168867839997e-06,
+      "loss": 0.3364,
+      "step": 570
+    },
+    {
+      "epoch": 0.3084763948497854,
+      "grad_norm": 2.450943392872261,
+      "learning_rate": 9.910821080814184e-06,
+      "loss": 0.3354,
+      "step": 575
+    },
+    {
+      "epoch": 0.3111587982832618,
+      "grad_norm": 2.546643411521105,
+      "learning_rate": 9.90636564527832e-06,
+      "loss": 0.3325,
+      "step": 580
+    },
+    {
+      "epoch": 0.3138412017167382,
+      "grad_norm": 2.4957026778211526,
+      "learning_rate": 9.901802658898552e-06,
+      "loss": 0.3279,
+      "step": 585
+    },
+    {
+      "epoch": 0.3165236051502146,
+      "grad_norm": 2.6557291209936413,
+      "learning_rate": 9.897132221698624e-06,
+      "loss": 0.3401,
+      "step": 590
+    },
+    {
+      "epoch": 0.31920600858369097,
+      "grad_norm": 2.6636098661539283,
+      "learning_rate": 9.892354436057665e-06,
+      "loss": 0.3265,
+      "step": 595
+    },
+    {
+      "epoch": 0.3218884120171674,
+      "grad_norm": 2.4076735348215395,
+      "learning_rate": 9.887469406707962e-06,
+      "loss": 0.3384,
+      "step": 600
+    },
+    {
+      "epoch": 0.32457081545064376,
+      "grad_norm": 2.6766617663933006,
+      "learning_rate": 9.882477240732652e-06,
+      "loss": 0.3364,
+      "step": 605
+    },
+    {
+      "epoch": 0.3272532188841202,
+      "grad_norm": 2.3629027511324665,
+      "learning_rate": 9.877378047563378e-06,
+      "loss": 0.3377,
+      "step": 610
+    },
+    {
+      "epoch": 0.32993562231759654,
+      "grad_norm": 2.454480581820658,
+      "learning_rate": 9.872171938977895e-06,
+      "loss": 0.3472,
+      "step": 615
+    },
+    {
+      "epoch": 0.33261802575107297,
+      "grad_norm": 2.467294504215762,
+      "learning_rate": 9.866859029097613e-06,
+      "loss": 0.3394,
+      "step": 620
+    },
+    {
+      "epoch": 0.33530042918454933,
+      "grad_norm": 2.434454219419277,
+      "learning_rate": 9.8614394343851e-06,
+      "loss": 0.3384,
+      "step": 625
+    },
+    {
+      "epoch": 0.33798283261802575,
+      "grad_norm": 2.588554106379298,
+      "learning_rate": 9.855913273641531e-06,
+      "loss": 0.3486,
+      "step": 630
+    },
+    {
+      "epoch": 0.3406652360515021,
+      "grad_norm": 2.559726676405475,
+      "learning_rate": 9.850280668004072e-06,
+      "loss": 0.3368,
+      "step": 635
+    },
+    {
+      "epoch": 0.34334763948497854,
+      "grad_norm": 2.5431858822060085,
+      "learning_rate": 9.844541740943239e-06,
+      "loss": 0.3212,
+      "step": 640
+    },
+    {
+      "epoch": 0.34603004291845496,
+      "grad_norm": 2.5877373233512446,
+      "learning_rate": 9.838696618260182e-06,
+      "loss": 0.3326,
+      "step": 645
+    },
+    {
+      "epoch": 0.3487124463519313,
+      "grad_norm": 2.5309813288979917,
+      "learning_rate": 9.832745428083934e-06,
+      "loss": 0.3331,
+      "step": 650
+    },
+    {
+      "epoch": 0.35139484978540775,
+      "grad_norm": 2.4354927332842364,
+      "learning_rate": 9.826688300868597e-06,
+      "loss": 0.3413,
+      "step": 655
+    },
+    {
+      "epoch": 0.3540772532188841,
+      "grad_norm": 2.5015034823966893,
+      "learning_rate": 9.820525369390486e-06,
+      "loss": 0.3409,
+      "step": 660
+    },
+    {
+      "epoch": 0.35675965665236054,
+      "grad_norm": 2.5950385006418766,
+      "learning_rate": 9.814256768745212e-06,
+      "loss": 0.333,
+      "step": 665
+    },
+    {
+      "epoch": 0.3594420600858369,
+      "grad_norm": 2.661302238907792,
+      "learning_rate": 9.80788263634473e-06,
+      "loss": 0.3332,
+      "step": 670
+    },
+    {
+      "epoch": 0.3621244635193133,
+      "grad_norm": 2.684828328695924,
+      "learning_rate": 9.801403111914324e-06,
+      "loss": 0.3358,
+      "step": 675
+    },
+    {
+      "epoch": 0.3648068669527897,
+      "grad_norm": 2.4226357461522245,
+      "learning_rate": 9.794818337489535e-06,
+      "loss": 0.3395,
+      "step": 680
+    },
+    {
+      "epoch": 0.3674892703862661,
+      "grad_norm": 2.633280418543926,
+      "learning_rate": 9.788128457413064e-06,
+      "loss": 0.3415,
+      "step": 685
+    },
+    {
+      "epoch": 0.3701716738197425,
+      "grad_norm": 2.7709279052969924,
+      "learning_rate": 9.78133361833159e-06,
+      "loss": 0.3396,
+      "step": 690
+    },
+    {
+      "epoch": 0.3728540772532189,
+      "grad_norm": 2.4891302152361683,
+      "learning_rate": 9.774433969192569e-06,
+      "loss": 0.3413,
+      "step": 695
+    },
+    {
+      "epoch": 0.37553648068669526,
+      "grad_norm": 2.448140021927178,
+      "learning_rate": 9.767429661240966e-06,
+      "loss": 0.337,
+      "step": 700
+    },
+    {
+      "epoch": 0.3782188841201717,
+      "grad_norm": 2.4979263971763292,
+      "learning_rate": 9.760320848015932e-06,
+      "loss": 0.3356,
+      "step": 705
+    },
+    {
+      "epoch": 0.38090128755364805,
+      "grad_norm": 2.5647155012401575,
+      "learning_rate": 9.75310768534745e-06,
+      "loss": 0.3276,
+      "step": 710
+    },
+    {
+      "epoch": 0.38358369098712447,
+      "grad_norm": 2.678305589560622,
+      "learning_rate": 9.745790331352907e-06,
+      "loss": 0.3332,
+      "step": 715
+    },
+    {
+      "epoch": 0.38626609442060084,
+      "grad_norm": 2.444142509679277,
+      "learning_rate": 9.73836894643364e-06,
+      "loss": 0.3411,
+      "step": 720
+    },
+    {
+      "epoch": 0.38894849785407726,
+      "grad_norm": 2.5901281336103605,
+      "learning_rate": 9.730843693271413e-06,
+      "loss": 0.3433,
+      "step": 725
+    },
+    {
+      "epoch": 0.3916309012875536,
+      "grad_norm": 2.6069745191548277,
+      "learning_rate": 9.723214736824847e-06,
+      "loss": 0.3269,
+      "step": 730
+    },
+    {
+      "epoch": 0.39431330472103004,
+      "grad_norm": 2.466120600187352,
+      "learning_rate": 9.715482244325816e-06,
+      "loss": 0.3438,
+      "step": 735
+    },
+    {
+      "epoch": 0.3969957081545064,
+      "grad_norm": 2.704267296891685,
+      "learning_rate": 9.707646385275766e-06,
+      "loss": 0.3408,
+      "step": 740
+    },
+    {
+      "epoch": 0.39967811158798283,
+      "grad_norm": 2.4389005799413037,
+      "learning_rate": 9.699707331442016e-06,
+      "loss": 0.3438,
+      "step": 745
+    },
+    {
+      "epoch": 0.40236051502145925,
+      "grad_norm": 2.5076690789214364,
+      "learning_rate": 9.691665256853978e-06,
+      "loss": 0.3329,
+      "step": 750
+    },
+    {
+      "epoch": 0.4050429184549356,
+      "grad_norm": 2.4796685866213717,
+      "learning_rate": 9.683520337799353e-06,
+      "loss": 0.3379,
+      "step": 755
+    },
+    {
+      "epoch": 0.40772532188841204,
+      "grad_norm": 4.153485428842475,
+      "learning_rate": 9.675272752820258e-06,
+      "loss": 0.3319,
+      "step": 760
+    },
+    {
+      "epoch": 0.4104077253218884,
+      "grad_norm": 2.5641045685817727,
+      "learning_rate": 9.666922682709317e-06,
+      "loss": 0.3434,
+      "step": 765
+    },
+    {
+      "epoch": 0.4130901287553648,
+      "grad_norm": 2.5739549498843477,
+      "learning_rate": 9.6584703105057e-06,
+      "loss": 0.3367,
+      "step": 770
+    },
+    {
+      "epoch": 0.4157725321888412,
+      "grad_norm": 2.506896854676805,
+      "learning_rate": 9.649915821491107e-06,
+      "loss": 0.3362,
+      "step": 775
+    },
+    {
+      "epoch": 0.4184549356223176,
+      "grad_norm": 2.4070337960809294,
+      "learning_rate": 9.641259403185706e-06,
+      "loss": 0.3352,
+      "step": 780
+    },
+    {
+      "epoch": 0.421137339055794,
+      "grad_norm": 2.4353089647919783,
+      "learning_rate": 9.632501245344024e-06,
+      "loss": 0.3397,
+      "step": 785
+    },
+    {
+      "epoch": 0.4238197424892704,
+      "grad_norm": 2.643735395039157,
+      "learning_rate": 9.623641539950787e-06,
+      "loss": 0.3495,
+      "step": 790
+    },
+    {
+      "epoch": 0.42650214592274677,
+      "grad_norm": 2.478560964786462,
+      "learning_rate": 9.614680481216712e-06,
+      "loss": 0.3432,
+      "step": 795
+    },
+    {
+      "epoch": 0.4291845493562232,
+      "grad_norm": 2.577614990446464,
+      "learning_rate": 9.60561826557425e-06,
+      "loss": 0.3383,
+      "step": 800
+    },
+    {
+      "epoch": 0.43186695278969955,
+      "grad_norm": 2.5172456862717825,
+      "learning_rate": 9.596455091673282e-06,
+      "loss": 0.3373,
+      "step": 805
+    },
+    {
+      "epoch": 0.434549356223176,
+      "grad_norm": 2.4359928535441138,
+      "learning_rate": 9.587191160376758e-06,
+      "loss": 0.3407,
+      "step": 810
+    },
+    {
+      "epoch": 0.43723175965665234,
+      "grad_norm": 2.465539807171554,
+      "learning_rate": 9.577826674756301e-06,
+      "loss": 0.3344,
+      "step": 815
+    },
+    {
+      "epoch": 0.43991416309012876,
+      "grad_norm": 2.2960811499386153,
+      "learning_rate": 9.56836184008775e-06,
+      "loss": 0.3382,
+      "step": 820
+    },
+    {
+      "epoch": 0.44259656652360513,
+      "grad_norm": 2.372331098589979,
+      "learning_rate": 9.558796863846663e-06,
+      "loss": 0.3339,
+      "step": 825
+    },
+    {
+      "epoch": 0.44527896995708155,
+      "grad_norm": 2.581631110852197,
+      "learning_rate": 9.549131955703772e-06,
+      "loss": 0.323,
+      "step": 830
+    },
+    {
+      "epoch": 0.4479613733905579,
+      "grad_norm": 2.3697193776417707,
+      "learning_rate": 9.539367327520382e-06,
+      "loss": 0.3458,
+      "step": 835
+    },
+    {
+      "epoch": 0.45064377682403434,
+      "grad_norm": 2.5130569622507135,
+      "learning_rate": 9.529503193343726e-06,
+      "loss": 0.3418,
+      "step": 840
+    },
+    {
+      "epoch": 0.4533261802575107,
+      "grad_norm": 2.4362641659919806,
+      "learning_rate": 9.519539769402282e-06,
+      "loss": 0.3454,
+      "step": 845
+    },
+    {
+      "epoch": 0.4560085836909871,
+      "grad_norm": 2.6421344257402106,
+      "learning_rate": 9.509477274101019e-06,
+      "loss": 0.3373,
+      "step": 850
+    },
+    {
+      "epoch": 0.45869098712446355,
+      "grad_norm": 2.38501065072516,
+      "learning_rate": 9.499315928016619e-06,
+      "loss": 0.3282,
+      "step": 855
+    },
+    {
+      "epoch": 0.4613733905579399,
+      "grad_norm": 2.414855309500928,
+      "learning_rate": 9.489055953892644e-06,
+      "loss": 0.3254,
+      "step": 860
+    },
+    {
+      "epoch": 0.46405579399141633,
+      "grad_norm": 2.5052929819879077,
+      "learning_rate": 9.478697576634646e-06,
+      "loss": 0.3326,
+      "step": 865
+    },
+    {
+      "epoch": 0.4667381974248927,
+      "grad_norm": 2.489365469431507,
+      "learning_rate": 9.46824102330524e-06,
+      "loss": 0.3393,
+      "step": 870
+    },
+    {
+      "epoch": 0.4694206008583691,
+      "grad_norm": 2.4291904681349568,
+      "learning_rate": 9.457686523119128e-06,
+      "loss": 0.335,
+      "step": 875
+    },
+    {
+      "epoch": 0.4721030042918455,
+      "grad_norm": 2.4057320722711757,
+      "learning_rate": 9.447034307438068e-06,
+      "loss": 0.3269,
+      "step": 880
+    },
+    {
+      "epoch": 0.4747854077253219,
+      "grad_norm": 2.5244213882736237,
+      "learning_rate": 9.436284609765818e-06,
+      "loss": 0.335,
+      "step": 885
+    },
+    {
+      "epoch": 0.47746781115879827,
+      "grad_norm": 2.429979427144697,
+      "learning_rate": 9.425437665742998e-06,
+      "loss": 0.3376,
+      "step": 890
+    },
+    {
+      "epoch": 0.4801502145922747,
+      "grad_norm": 2.50403465078303,
+      "learning_rate": 9.414493713141936e-06,
+      "loss": 0.3317,
+      "step": 895
+    },
+    {
+      "epoch": 0.48283261802575106,
+      "grad_norm": 2.3406910645072374,
+      "learning_rate": 9.403452991861452e-06,
+      "loss": 0.3323,
+      "step": 900
+    },
+    {
+      "epoch": 0.4855150214592275,
+      "grad_norm": 2.4499576375372776,
+      "learning_rate": 9.392315743921606e-06,
+      "loss": 0.3434,
+      "step": 905
+    },
+    {
+      "epoch": 0.48819742489270385,
+      "grad_norm": 2.3837396316239925,
+      "learning_rate": 9.381082213458384e-06,
+      "loss": 0.3379,
+      "step": 910
+    },
+    {
+      "epoch": 0.49087982832618027,
+      "grad_norm": 2.283083285021325,
+      "learning_rate": 9.36975264671835e-06,
+      "loss": 0.3416,
+      "step": 915
+    },
+    {
+      "epoch": 0.49356223175965663,
+      "grad_norm": 2.453493510513127,
+      "learning_rate": 9.358327292053244e-06,
+      "loss": 0.3332,
+      "step": 920
+    },
+    {
+      "epoch": 0.49624463519313305,
+      "grad_norm": 2.497590155715813,
+      "learning_rate": 9.346806399914547e-06,
+      "loss": 0.332,
+      "step": 925
+    },
+    {
+      "epoch": 0.4989270386266094,
+      "grad_norm": 2.4578743205162854,
+      "learning_rate": 9.335190222847988e-06,
+      "loss": 0.3221,
+      "step": 930
+    },
+    {
+      "epoch": 0.5016094420600858,
+      "grad_norm": 2.441472272597816,
+      "learning_rate": 9.323479015488e-06,
+      "loss": 0.3268,
+      "step": 935
+    },
+    {
+      "epoch": 0.5042918454935622,
+      "grad_norm": 2.319230879720979,
+      "learning_rate": 9.311673034552146e-06,
+      "loss": 0.3289,
+      "step": 940
+    },
+    {
+      "epoch": 0.5069742489270386,
+      "grad_norm": 2.4121543980436524,
+      "learning_rate": 9.299772538835492e-06,
+      "loss": 0.3343,
+      "step": 945
+    },
+    {
+      "epoch": 0.509656652360515,
+      "grad_norm": 2.394868024721506,
+      "learning_rate": 9.28777778920493e-06,
+      "loss": 0.3271,
+      "step": 950
+    },
+    {
+      "epoch": 0.5123390557939914,
+      "grad_norm": 2.4436982704679786,
+      "learning_rate": 9.27568904859346e-06,
+      "loss": 0.3363,
+      "step": 955
+    },
+    {
+      "epoch": 0.5150214592274678,
+      "grad_norm": 2.3832231268645057,
+      "learning_rate": 9.26350658199443e-06,
+      "loss": 0.3362,
+      "step": 960
+    },
+    {
+      "epoch": 0.5177038626609443,
+      "grad_norm": 2.327822759382941,
+      "learning_rate": 9.251230656455722e-06,
+      "loss": 0.3398,
+      "step": 965
+    },
+    {
+      "epoch": 0.5203862660944206,
+      "grad_norm": 2.439342840620097,
+      "learning_rate": 9.238861541073909e-06,
+      "loss": 0.3416,
+      "step": 970
+    },
+    {
+      "epoch": 0.523068669527897,
+      "grad_norm": 2.385481519390706,
+      "learning_rate": 9.226399506988336e-06,
+      "loss": 0.3307,
+      "step": 975
+    },
+    {
+      "epoch": 0.5257510729613734,
+      "grad_norm": 2.4266226741554187,
+      "learning_rate": 9.213844827375196e-06,
+      "loss": 0.3389,
+      "step": 980
+    },
+    {
+      "epoch": 0.5284334763948498,
+      "grad_norm": 2.466403363273194,
+      "learning_rate": 9.201197777441533e-06,
+      "loss": 0.3266,
+      "step": 985
+    },
+    {
+      "epoch": 0.5311158798283262,
+      "grad_norm": 2.356016335190441,
+      "learning_rate": 9.188458634419213e-06,
+      "loss": 0.333,
+      "step": 990
+    },
+    {
+      "epoch": 0.5337982832618026,
+      "grad_norm": 2.434189564629439,
+      "learning_rate": 9.175627677558842e-06,
+      "loss": 0.3231,
+      "step": 995
+    },
+    {
+      "epoch": 0.5364806866952789,
+      "grad_norm": 2.335622267331965,
+      "learning_rate": 9.162705188123647e-06,
+      "loss": 0.319,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5391630901287554,
+      "grad_norm": 2.3022256652508886,
+      "learning_rate": 9.149691449383313e-06,
+      "loss": 0.3329,
+      "step": 1005
+    },
+    {
+      "epoch": 0.5418454935622318,
+      "grad_norm": 2.352723118113408,
+      "learning_rate": 9.136586746607767e-06,
+      "loss": 0.3243,
+      "step": 1010
+    },
+    {
+      "epoch": 0.5445278969957081,
+      "grad_norm": 2.3738093381466197,
+      "learning_rate": 9.123391367060937e-06,
+      "loss": 0.3403,
+      "step": 1015
+    },
+    {
+      "epoch": 0.5472103004291845,
+      "grad_norm": 2.486621636388831,
+      "learning_rate": 9.110105599994436e-06,
+      "loss": 0.3416,
+      "step": 1020
+    },
+    {
+      "epoch": 0.549892703862661,
+      "grad_norm": 2.183874840676628,
+      "learning_rate": 9.096729736641242e-06,
+      "loss": 0.3334,
+      "step": 1025
+    },
+    {
+      "epoch": 0.5525751072961373,
+      "grad_norm": 2.350951978916979,
+      "learning_rate": 9.0832640702093e-06,
+      "loss": 0.3379,
+      "step": 1030
+    },
+    {
+      "epoch": 0.5552575107296137,
+      "grad_norm": 2.4066282347575676,
+      "learning_rate": 9.0697088958751e-06,
+      "loss": 0.3354,
+      "step": 1035
+    },
+    {
+      "epoch": 0.5579399141630901,
+      "grad_norm": 2.1575463339610113,
+      "learning_rate": 9.056064510777204e-06,
+      "loss": 0.3248,
+      "step": 1040
+    },
+    {
+      "epoch": 0.5606223175965666,
+      "grad_norm": 2.616297568647876,
+      "learning_rate": 9.042331214009736e-06,
+      "loss": 0.3309,
+      "step": 1045
+    },
+    {
+      "epoch": 0.5633047210300429,
+      "grad_norm": 2.327723872267469,
+      "learning_rate": 9.028509306615825e-06,
+      "loss": 0.3301,
+      "step": 1050
+    },
+    {
+      "epoch": 0.5659871244635193,
+      "grad_norm": 2.3750699885317426,
+      "learning_rate": 9.014599091581e-06,
+      "loss": 0.3305,
+      "step": 1055
+    },
+    {
+      "epoch": 0.5686695278969958,
+      "grad_norm": 2.3155068078689074,
+      "learning_rate": 9.000600873826558e-06,
+      "loss": 0.3376,
+      "step": 1060
+    },
+    {
+      "epoch": 0.5713519313304721,
+      "grad_norm": 2.3330209228120538,
+      "learning_rate": 8.98651496020287e-06,
+      "loss": 0.3328,
+      "step": 1065
+    },
+    {
+      "epoch": 0.5740343347639485,
+      "grad_norm": 2.3406393186838663,
+      "learning_rate": 8.972341659482666e-06,
+      "loss": 0.3316,
+      "step": 1070
+    },
+    {
+      "epoch": 0.5767167381974249,
+      "grad_norm": 2.234164678010478,
+      "learning_rate": 8.958081282354253e-06,
+      "loss": 0.3344,
+      "step": 1075
+    },
+    {
+      "epoch": 0.5793991416309013,
+      "grad_norm": 2.350491195121751,
+      "learning_rate": 8.943734141414719e-06,
+      "loss": 0.3326,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5820815450643777,
+      "grad_norm": 2.194661519862346,
+      "learning_rate": 8.929300551163068e-06,
+      "loss": 0.3287,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5847639484978541,
+      "grad_norm": 2.306271735341111,
+      "learning_rate": 8.914780827993332e-06,
+      "loss": 0.3246,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5874463519313304,
+      "grad_norm": 2.3938196044575055,
+      "learning_rate": 8.900175290187636e-06,
+      "loss": 0.3354,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5901287553648069,
+      "grad_norm": 2.393591199824571,
+      "learning_rate": 8.885484257909218e-06,
+      "loss": 0.3339,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5928111587982833,
+      "grad_norm": 2.29507846222384,
+      "learning_rate": 8.870708053195414e-06,
+      "loss": 0.3276,
+      "step": 1105
+    },
+    {
+      "epoch": 0.5954935622317596,
+      "grad_norm": 2.386042824631869,
+      "learning_rate": 8.855846999950595e-06,
+      "loss": 0.3331,
+      "step": 1110
+    },
+    {
+      "epoch": 0.598175965665236,
+      "grad_norm": 2.3737048550845308,
+      "learning_rate": 8.840901423939075e-06,
+      "loss": 0.3267,
+      "step": 1115
+    },
+    {
+      "epoch": 0.6008583690987125,
+      "grad_norm": 2.3457094796065947,
+      "learning_rate": 8.825871652777955e-06,
+      "loss": 0.332,
+      "step": 1120
+    },
+    {
+      "epoch": 0.6035407725321889,
+      "grad_norm": 2.469177839155163,
+      "learning_rate": 8.81075801592996e-06,
+      "loss": 0.3306,
+      "step": 1125
+    },
+    {
+      "epoch": 0.6062231759656652,
+      "grad_norm": 2.185053219295239,
+      "learning_rate": 8.795560844696198e-06,
+      "loss": 0.3293,
+      "step": 1130
+    },
+    {
+      "epoch": 0.6089055793991416,
+      "grad_norm": 2.387598614909069,
+      "learning_rate": 8.780280472208915e-06,
+      "loss": 0.3216,
+      "step": 1135
+    },
+    {
+      "epoch": 0.6115879828326181,
+      "grad_norm": 2.173546188624302,
+      "learning_rate": 8.764917233424179e-06,
+      "loss": 0.3238,
+      "step": 1140
+    },
+    {
+      "epoch": 0.6142703862660944,
+      "grad_norm": 2.3583055905659687,
+      "learning_rate": 8.749471465114548e-06,
+      "loss": 0.3333,
+      "step": 1145
+    },
+    {
+      "epoch": 0.6169527896995708,
+      "grad_norm": 2.360546414469146,
+      "learning_rate": 8.73394350586168e-06,
+      "loss": 0.3286,
+      "step": 1150
+    },
+    {
+      "epoch": 0.6196351931330472,
+      "grad_norm": 2.318106769491821,
+      "learning_rate": 8.71833369604891e-06,
+      "loss": 0.3311,
+      "step": 1155
+    },
+    {
+      "epoch": 0.6223175965665236,
+      "grad_norm": 2.2690977329855584,
+      "learning_rate": 8.702642377853803e-06,
+      "loss": 0.3278,
+      "step": 1160
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 2.3528719925610533,
+      "learning_rate": 8.686869895240631e-06,
+      "loss": 0.3252,
+      "step": 1165
+    },
+    {
+      "epoch": 0.6276824034334764,
+      "grad_norm": 2.3175542680444017,
+      "learning_rate": 8.671016593952853e-06,
+      "loss": 0.3307,
+      "step": 1170
+    },
+    {
+      "epoch": 0.6303648068669528,
+      "grad_norm": 2.263552176980709,
+      "learning_rate": 8.655082821505524e-06,
+      "loss": 0.3258,
+      "step": 1175
+    },
+    {
+      "epoch": 0.6330472103004292,
+      "grad_norm": 2.343567579732669,
+      "learning_rate": 8.639068927177684e-06,
+      "loss": 0.3211,
+      "step": 1180
+    },
+    {
+      "epoch": 0.6357296137339056,
+      "grad_norm": 2.2780220598628365,
+      "learning_rate": 8.622975262004694e-06,
+      "loss": 0.3298,
+      "step": 1185
+    },
+    {
+      "epoch": 0.6384120171673819,
+      "grad_norm": 2.2947818816488272,
+      "learning_rate": 8.606802178770551e-06,
+      "loss": 0.315,
+      "step": 1190
+    },
+    {
+      "epoch": 0.6410944206008584,
+      "grad_norm": 2.365849669313569,
+      "learning_rate": 8.590550032000146e-06,
+      "loss": 0.3232,
+      "step": 1195
+    },
+    {
+      "epoch": 0.6437768240343348,
+      "grad_norm": 2.144107893647448,
+      "learning_rate": 8.574219177951495e-06,
+      "loss": 0.3181,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6464592274678111,
+      "grad_norm": 2.416375076775248,
+      "learning_rate": 8.557809974607936e-06,
+      "loss": 0.3229,
+      "step": 1205
+    },
+    {
+      "epoch": 0.6491416309012875,
+      "grad_norm": 2.302421552968707,
+      "learning_rate": 8.541322781670272e-06,
+      "loss": 0.3245,
+      "step": 1210
+    },
+    {
+      "epoch": 0.651824034334764,
+      "grad_norm": 2.230585990950688,
+      "learning_rate": 8.524757960548888e-06,
+      "loss": 0.328,
+      "step": 1215
+    },
+    {
+      "epoch": 0.6545064377682404,
+      "grad_norm": 2.2888151168997597,
+      "learning_rate": 8.50811587435584e-06,
+      "loss": 0.33,
+      "step": 1220
+    },
+    {
+      "epoch": 0.6571888412017167,
+      "grad_norm": 2.2856839548113164,
+      "learning_rate": 8.491396887896878e-06,
+      "loss": 0.3224,
+      "step": 1225
+    },
+    {
+      "epoch": 0.6598712446351931,
+      "grad_norm": 2.120698181701614,
+      "learning_rate": 8.474601367663463e-06,
+      "loss": 0.3292,
+      "step": 1230
+    },
+    {
+      "epoch": 0.6625536480686696,
+      "grad_norm": 2.372814557865873,
+      "learning_rate": 8.457729681824722e-06,
+      "loss": 0.3338,
+      "step": 1235
+    },
+    {
+      "epoch": 0.6652360515021459,
+      "grad_norm": 2.2493974335307847,
+      "learning_rate": 8.440782200219391e-06,
+      "loss": 0.3316,
+      "step": 1240
+    },
+    {
+      "epoch": 0.6679184549356223,
+      "grad_norm": 2.277320210657522,
+      "learning_rate": 8.423759294347693e-06,
+      "loss": 0.3341,
+      "step": 1245
+    },
+    {
+      "epoch": 0.6706008583690987,
+      "grad_norm": 2.3582542622866898,
+      "learning_rate": 8.40666133736321e-06,
+      "loss": 0.3311,
+      "step": 1250
+    },
+    {
+      "epoch": 0.6732832618025751,
+      "grad_norm": 2.0865613012796804,
+      "learning_rate": 8.389488704064686e-06,
+      "loss": 0.3227,
+      "step": 1255
+    },
+    {
+      "epoch": 0.6759656652360515,
+      "grad_norm": 2.4350698606722823,
+      "learning_rate": 8.372241770887826e-06,
+      "loss": 0.3358,
+      "step": 1260
+    },
+    {
+      "epoch": 0.6786480686695279,
+      "grad_norm": 2.3729375593599755,
+      "learning_rate": 8.354920915897038e-06,
+      "loss": 0.3238,
+      "step": 1265
+    },
+    {
+      "epoch": 0.6813304721030042,
+      "grad_norm": 1.9989806654658056,
+      "learning_rate": 8.337526518777143e-06,
+      "loss": 0.3216,
+      "step": 1270
+    },
+    {
+      "epoch": 0.6840128755364807,
+      "grad_norm": 2.136326711892197,
+      "learning_rate": 8.32005896082506e-06,
+      "loss": 0.3226,
+      "step": 1275
+    },
+    {
+      "epoch": 0.6866952789699571,
+      "grad_norm": 2.191149877600861,
+      "learning_rate": 8.302518624941435e-06,
+      "loss": 0.3289,
+      "step": 1280
+    },
+    {
+      "epoch": 0.6893776824034334,
+      "grad_norm": 2.411570960647483,
+      "learning_rate": 8.284905895622265e-06,
+      "loss": 0.3253,
+      "step": 1285
+    },
+    {
+      "epoch": 0.6920600858369099,
+      "grad_norm": 2.0894935734976445,
+      "learning_rate": 8.26722115895045e-06,
+      "loss": 0.3188,
+      "step": 1290
+    },
+    {
+      "epoch": 0.6947424892703863,
+      "grad_norm": 2.335048531249719,
+      "learning_rate": 8.249464802587353e-06,
+      "loss": 0.3344,
+      "step": 1295
+    },
+    {
+      "epoch": 0.6974248927038627,
+      "grad_norm": 2.1431080115701286,
+      "learning_rate": 8.231637215764273e-06,
+      "loss": 0.3335,
+      "step": 1300
+    },
+    {
+      "epoch": 0.700107296137339,
+      "grad_norm": 2.1729056458527616,
+      "learning_rate": 8.21373878927394e-06,
+      "loss": 0.3142,
+      "step": 1305
+    },
+    {
+      "epoch": 0.7027896995708155,
+      "grad_norm": 2.0992512222503885,
+      "learning_rate": 8.195769915461931e-06,
+      "loss": 0.3199,
+      "step": 1310
+    },
+    {
+      "epoch": 0.7054721030042919,
+      "grad_norm": 2.4125083176551505,
+      "learning_rate": 8.177730988218083e-06,
+      "loss": 0.3222,
+      "step": 1315
+    },
+    {
+      "epoch": 0.7081545064377682,
+      "grad_norm": 2.1167993655688337,
+      "learning_rate": 8.159622402967841e-06,
+      "loss": 0.3207,
+      "step": 1320
+    },
+    {
+      "epoch": 0.7108369098712446,
+      "grad_norm": 2.132358313527919,
+      "learning_rate": 8.141444556663612e-06,
+      "loss": 0.3302,
+      "step": 1325
+    },
+    {
+      "epoch": 0.7135193133047211,
+      "grad_norm": 2.2305452399275474,
+      "learning_rate": 8.123197847776043e-06,
+      "loss": 0.3186,
+      "step": 1330
+    },
+    {
+      "epoch": 0.7162017167381974,
+      "grad_norm": 2.1240401193752465,
+      "learning_rate": 8.104882676285301e-06,
+      "loss": 0.3224,
+      "step": 1335
+    },
+    {
+      "epoch": 0.7188841201716738,
+      "grad_norm": 2.427318922218822,
+      "learning_rate": 8.086499443672297e-06,
+      "loss": 0.3273,
+      "step": 1340
+    },
+    {
+      "epoch": 0.7215665236051502,
+      "grad_norm": 2.20325872744776,
+      "learning_rate": 8.068048552909887e-06,
+      "loss": 0.3156,
+      "step": 1345
+    },
+    {
+      "epoch": 0.7242489270386266,
+      "grad_norm": 2.2501615842663214,
+      "learning_rate": 8.049530408454041e-06,
+      "loss": 0.3196,
+      "step": 1350
+    },
+    {
+      "epoch": 0.726931330472103,
+      "grad_norm": 2.01520964420955,
+      "learning_rate": 8.030945416234971e-06,
+      "loss": 0.3134,
+      "step": 1355
+    },
+    {
+      "epoch": 0.7296137339055794,
+      "grad_norm": 2.092562343745031,
+      "learning_rate": 8.012293983648247e-06,
+      "loss": 0.3109,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7322961373390557,
+      "grad_norm": 2.13112956895763,
+      "learning_rate": 7.993576519545844e-06,
+      "loss": 0.3113,
+      "step": 1365
+    },
+    {
+      "epoch": 0.7349785407725322,
+      "grad_norm": 2.342314257305073,
+      "learning_rate": 7.974793434227203e-06,
+      "loss": 0.324,
+      "step": 1370
+    },
+    {
+      "epoch": 0.7376609442060086,
+      "grad_norm": 2.199143638436571,
+      "learning_rate": 7.955945139430221e-06,
+      "loss": 0.315,
+      "step": 1375
+    },
+    {
+      "epoch": 0.740343347639485,
+      "grad_norm": 2.0119638311974386,
+      "learning_rate": 7.937032048322231e-06,
+      "loss": 0.3087,
+      "step": 1380
+    },
+    {
+      "epoch": 0.7430257510729614,
+      "grad_norm": 2.1782754033524845,
+      "learning_rate": 7.918054575490943e-06,
+      "loss": 0.3242,
+      "step": 1385
+    },
+    {
+      "epoch": 0.7457081545064378,
+      "grad_norm": 2.206496414169751,
+      "learning_rate": 7.899013136935365e-06,
+      "loss": 0.3063,
+      "step": 1390
+    },
+    {
+      "epoch": 0.7483905579399142,
+      "grad_norm": 2.0423709558066405,
+      "learning_rate": 7.879908150056668e-06,
+      "loss": 0.3117,
+      "step": 1395
+    },
+    {
+      "epoch": 0.7510729613733905,
+      "grad_norm": 2.095655750811691,
+      "learning_rate": 7.860740033649053e-06,
+      "loss": 0.3229,
+      "step": 1400
+    },
+    {
+      "epoch": 0.753755364806867,
+      "grad_norm": 2.0335948169947526,
+      "learning_rate": 7.841509207890555e-06,
+      "loss": 0.3047,
+      "step": 1405
+    },
+    {
+      "epoch": 0.7564377682403434,
+      "grad_norm": 2.21206434496724,
+      "learning_rate": 7.822216094333847e-06,
+      "loss": 0.3098,
+      "step": 1410
+    },
+    {
+      "epoch": 0.7591201716738197,
+      "grad_norm": 2.1541492290043376,
+      "learning_rate": 7.802861115896988e-06,
+      "loss": 0.3081,
+      "step": 1415
+    },
+    {
+      "epoch": 0.7618025751072961,
+      "grad_norm": 2.0550566165604756,
+      "learning_rate": 7.783444696854161e-06,
+      "loss": 0.3095,
+      "step": 1420
+    },
+    {
+      "epoch": 0.7644849785407726,
+      "grad_norm": 2.0920475443521744,
+      "learning_rate": 7.763967262826363e-06,
+      "loss": 0.3055,
+      "step": 1425
+    },
+    {
+      "epoch": 0.7671673819742489,
+      "grad_norm": 2.1615944487046708,
+      "learning_rate": 7.74442924077209e-06,
+      "loss": 0.3093,
+      "step": 1430
+    },
+    {
+      "epoch": 0.7698497854077253,
+      "grad_norm": 2.1144096218335613,
+      "learning_rate": 7.724831058977955e-06,
+      "loss": 0.3176,
+      "step": 1435
+    },
+    {
+      "epoch": 0.7725321888412017,
+      "grad_norm": 2.1346312946899455,
+      "learning_rate": 7.705173147049326e-06,
+      "loss": 0.3156,
+      "step": 1440
+    },
+    {
+      "epoch": 0.7752145922746781,
+      "grad_norm": 2.047468450694068,
+      "learning_rate": 7.685455935900886e-06,
+      "loss": 0.3052,
+      "step": 1445
+    },
+    {
+      "epoch": 0.7778969957081545,
+      "grad_norm": 2.0538961869366643,
+      "learning_rate": 7.665679857747204e-06,
+      "loss": 0.3273,
+      "step": 1450
+    },
+    {
+      "epoch": 0.7805793991416309,
+      "grad_norm": 2.1286681747583254,
+      "learning_rate": 7.645845346093246e-06,
+      "loss": 0.3179,
+      "step": 1455
+    },
+    {
+      "epoch": 0.7832618025751072,
+      "grad_norm": 1.994233749675141,
+      "learning_rate": 7.625952835724892e-06,
+      "loss": 0.3165,
+      "step": 1460
+    },
+    {
+      "epoch": 0.7859442060085837,
+      "grad_norm": 2.2976339300960373,
+      "learning_rate": 7.606002762699378e-06,
+      "loss": 0.3184,
+      "step": 1465
+    },
+    {
+      "epoch": 0.7886266094420601,
+      "grad_norm": 2.039358308364566,
+      "learning_rate": 7.585995564335764e-06,
+      "loss": 0.3109,
+      "step": 1470
+    },
+    {
+      "epoch": 0.7913090128755365,
+      "grad_norm": 2.2075630804735056,
+      "learning_rate": 7.565931679205329e-06,
+      "loss": 0.319,
+      "step": 1475
+    },
+    {
+      "epoch": 0.7939914163090128,
+      "grad_norm": 2.0106829444211733,
+      "learning_rate": 7.545811547121969e-06,
+      "loss": 0.3119,
+      "step": 1480
+    },
+    {
+      "epoch": 0.7966738197424893,
+      "grad_norm": 2.0666007146751553,
+      "learning_rate": 7.525635609132543e-06,
+      "loss": 0.3251,
+      "step": 1485
+    },
+    {
+      "epoch": 0.7993562231759657,
+      "grad_norm": 2.088216008028052,
+      "learning_rate": 7.505404307507227e-06,
+      "loss": 0.3073,
+      "step": 1490
+    },
+    {
+      "epoch": 0.802038626609442,
+      "grad_norm": 2.06365869935419,
+      "learning_rate": 7.48511808572979e-06,
+      "loss": 0.3159,
+      "step": 1495
+    },
+    {
+      "epoch": 0.8047210300429185,
+      "grad_norm": 2.1479596414620246,
+      "learning_rate": 7.464777388487899e-06,
+      "loss": 0.3135,
+      "step": 1500
+    },
+    {
+      "epoch": 0.8074034334763949,
+      "grad_norm": 2.140064633721639,
+      "learning_rate": 7.4443826616633555e-06,
+      "loss": 0.319,
+      "step": 1505
+    },
+    {
+      "epoch": 0.8100858369098712,
+      "grad_norm": 2.2172941608185814,
+      "learning_rate": 7.423934352322324e-06,
+      "loss": 0.3202,
+      "step": 1510
+    },
+    {
+      "epoch": 0.8127682403433476,
+      "grad_norm": 2.0860696548068343,
+      "learning_rate": 7.403432908705537e-06,
+      "loss": 0.315,
+      "step": 1515
+    },
+    {
+      "epoch": 0.8154506437768241,
+      "grad_norm": 2.0693450416677712,
+      "learning_rate": 7.382878780218466e-06,
+      "loss": 0.3144,
+      "step": 1520
+    },
+    {
+      "epoch": 0.8181330472103004,
+      "grad_norm": 2.2841738710721353,
+      "learning_rate": 7.362272417421467e-06,
+      "loss": 0.3063,
+      "step": 1525
+    },
+    {
+      "epoch": 0.8208154506437768,
+      "grad_norm": 2.147885334701556,
+      "learning_rate": 7.341614272019912e-06,
+      "loss": 0.3074,
+      "step": 1530
+    },
+    {
+      "epoch": 0.8234978540772532,
+      "grad_norm": 2.0237250051045392,
+      "learning_rate": 7.3209047968542815e-06,
+      "loss": 0.3148,
+      "step": 1535
+    },
+    {
+      "epoch": 0.8261802575107297,
+      "grad_norm": 2.0887045166680323,
+      "learning_rate": 7.300144445890236e-06,
+      "loss": 0.3109,
+      "step": 1540
+    },
+    {
+      "epoch": 0.828862660944206,
+      "grad_norm": 2.0283593925356262,
+      "learning_rate": 7.279333674208671e-06,
+      "loss": 0.3149,
+      "step": 1545
+    },
+    {
+      "epoch": 0.8315450643776824,
+      "grad_norm": 2.0419728278236513,
+      "learning_rate": 7.258472937995736e-06,
+      "loss": 0.303,
+      "step": 1550
+    },
+    {
+      "epoch": 0.8342274678111588,
+      "grad_norm": 2.1405345264719915,
+      "learning_rate": 7.23756269453284e-06,
+      "loss": 0.3206,
+      "step": 1555
+    },
+    {
+      "epoch": 0.8369098712446352,
+      "grad_norm": 2.1040140267852863,
+      "learning_rate": 7.216603402186618e-06,
+      "loss": 0.3168,
+      "step": 1560
+    },
+    {
+      "epoch": 0.8395922746781116,
+      "grad_norm": 2.1273226915661914,
+      "learning_rate": 7.195595520398898e-06,
+      "loss": 0.3187,
+      "step": 1565
+    },
+    {
+      "epoch": 0.842274678111588,
+      "grad_norm": 2.0095214683675424,
+      "learning_rate": 7.174539509676612e-06,
+      "loss": 0.3097,
+      "step": 1570
+    },
+    {
+      "epoch": 0.8449570815450643,
+      "grad_norm": 2.0016814081151892,
+      "learning_rate": 7.153435831581722e-06,
+      "loss": 0.3198,
+      "step": 1575
+    },
+    {
+      "epoch": 0.8476394849785408,
+      "grad_norm": 1.965454808053445,
+      "learning_rate": 7.132284948721079e-06,
+      "loss": 0.31,
+      "step": 1580
+    },
+    {
+      "epoch": 0.8503218884120172,
+      "grad_norm": 2.136352585229065,
+      "learning_rate": 7.1110873247363035e-06,
+      "loss": 0.3056,
+      "step": 1585
+    },
+    {
+      "epoch": 0.8530042918454935,
+      "grad_norm": 2.0773545777295976,
+      "learning_rate": 7.089843424293606e-06,
+      "loss": 0.3116,
+      "step": 1590
+    },
+    {
+      "epoch": 0.85568669527897,
+      "grad_norm": 1.998502422749954,
+      "learning_rate": 7.0685537130736145e-06,
+      "loss": 0.3206,
+      "step": 1595
+    },
+    {
+      "epoch": 0.8583690987124464,
+      "grad_norm": 2.0273327421570153,
+      "learning_rate": 7.047218657761156e-06,
+      "loss": 0.3061,
+      "step": 1600
+    },
+    {
+      "epoch": 0.8610515021459227,
+      "grad_norm": 2.0600139958469823,
+      "learning_rate": 7.025838726035032e-06,
+      "loss": 0.3122,
+      "step": 1605
+    },
+    {
+      "epoch": 0.8637339055793991,
+      "grad_norm": 1.9151781235741088,
+      "learning_rate": 7.004414386557765e-06,
+      "loss": 0.308,
+      "step": 1610
+    },
+    {
+      "epoch": 0.8664163090128756,
+      "grad_norm": 2.165460530379817,
+      "learning_rate": 6.982946108965326e-06,
+      "loss": 0.3129,
+      "step": 1615
+    },
+    {
+      "epoch": 0.869098712446352,
+      "grad_norm": 1.9524659521047858,
+      "learning_rate": 6.961434363856836e-06,
+      "loss": 0.3032,
+      "step": 1620
+    },
+    {
+      "epoch": 0.8717811158798283,
+      "grad_norm": 2.062634602507452,
+      "learning_rate": 6.939879622784259e-06,
+      "loss": 0.3138,
+      "step": 1625
+    },
+    {
+      "epoch": 0.8744635193133047,
+      "grad_norm": 1.9574471265790148,
+      "learning_rate": 6.918282358242053e-06,
+      "loss": 0.3074,
+      "step": 1630
+    },
+    {
+      "epoch": 0.8771459227467812,
+      "grad_norm": 1.9771976107010916,
+      "learning_rate": 6.896643043656826e-06,
+      "loss": 0.3013,
+      "step": 1635
+    },
+    {
+      "epoch": 0.8798283261802575,
+      "grad_norm": 1.937187823090658,
+      "learning_rate": 6.874962153376945e-06,
+      "loss": 0.3016,
+      "step": 1640
+    },
+    {
+      "epoch": 0.8825107296137339,
+      "grad_norm": 1.990154459364653,
+      "learning_rate": 6.853240162662149e-06,
+      "loss": 0.3017,
+      "step": 1645
+    },
+    {
+      "epoch": 0.8851931330472103,
+      "grad_norm": 1.951646105453254,
+      "learning_rate": 6.831477547673122e-06,
+      "loss": 0.3053,
+      "step": 1650
+    },
+    {
+      "epoch": 0.8878755364806867,
+      "grad_norm": 1.9520892795805238,
+      "learning_rate": 6.8096747854610634e-06,
+      "loss": 0.3089,
+      "step": 1655
+    },
+    {
+      "epoch": 0.8905579399141631,
+      "grad_norm": 2.0126176266313807,
+      "learning_rate": 6.787832353957225e-06,
+      "loss": 0.3103,
+      "step": 1660
+    },
+    {
+      "epoch": 0.8932403433476395,
+      "grad_norm": 1.9125432962564908,
+      "learning_rate": 6.7659507319624355e-06,
+      "loss": 0.3069,
+      "step": 1665
+    },
+    {
+      "epoch": 0.8959227467811158,
+      "grad_norm": 2.052079654464445,
+      "learning_rate": 6.744030399136606e-06,
+      "loss": 0.3033,
+      "step": 1670
+    },
+    {
+      "epoch": 0.8986051502145923,
+      "grad_norm": 2.050877524519969,
+      "learning_rate": 6.722071835988217e-06,
+      "loss": 0.3106,
+      "step": 1675
+    },
+    {
+      "epoch": 0.9012875536480687,
+      "grad_norm": 1.9919732658432157,
+      "learning_rate": 6.700075523863783e-06,
+      "loss": 0.3025,
+      "step": 1680
+    },
+    {
+      "epoch": 0.903969957081545,
+      "grad_norm": 1.9896839822490924,
+      "learning_rate": 6.678041944937297e-06,
+      "loss": 0.305,
+      "step": 1685
+    },
+    {
+      "epoch": 0.9066523605150214,
+      "grad_norm": 1.9158011231101228,
+      "learning_rate": 6.655971582199672e-06,
+      "loss": 0.3049,
+      "step": 1690
+    },
+    {
+      "epoch": 0.9093347639484979,
+      "grad_norm": 2.038467794641684,
+      "learning_rate": 6.633864919448143e-06,
+      "loss": 0.3139,
+      "step": 1695
+    },
+    {
+      "epoch": 0.9120171673819742,
+      "grad_norm": 1.9786294167283611,
+      "learning_rate": 6.611722441275666e-06,
+      "loss": 0.2998,
+      "step": 1700
+    },
+    {
+      "epoch": 0.9146995708154506,
+      "grad_norm": 1.9967546997105834,
+      "learning_rate": 6.589544633060298e-06,
+      "loss": 0.3061,
+      "step": 1705
+    },
+    {
+      "epoch": 0.9173819742489271,
+      "grad_norm": 1.9765034643155208,
+      "learning_rate": 6.5673319809545496e-06,
+      "loss": 0.3103,
+      "step": 1710
+    },
+    {
+      "epoch": 0.9200643776824035,
+      "grad_norm": 1.9778652823225957,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.3181,
+      "step": 1715
+    },
+    {
+      "epoch": 0.9227467811158798,
+      "grad_norm": 1.973804092462756,
+      "learning_rate": 6.522804093490305e-06,
+      "loss": 0.3051,
+      "step": 1720
+    },
+    {
+      "epoch": 0.9254291845493562,
+      "grad_norm": 1.9300897085251152,
+      "learning_rate": 6.50048983421313e-06,
+      "loss": 0.3132,
+      "step": 1725
+    },
+    {
+      "epoch": 0.9281115879828327,
+      "grad_norm": 2.0163968945583575,
+      "learning_rate": 6.478142683186827e-06,
+      "loss": 0.3029,
+      "step": 1730
+    },
+    {
+      "epoch": 0.930793991416309,
+      "grad_norm": 2.094134693771536,
+      "learning_rate": 6.455763130276019e-06,
+      "loss": 0.3089,
+      "step": 1735
+    },
+    {
+      "epoch": 0.9334763948497854,
+      "grad_norm": 1.9641907295696857,
+      "learning_rate": 6.433351666055598e-06,
+      "loss": 0.2969,
+      "step": 1740
+    },
+    {
+      "epoch": 0.9361587982832618,
+      "grad_norm": 1.9534671249367734,
+      "learning_rate": 6.410908781799974e-06,
+      "loss": 0.2992,
+      "step": 1745
+    },
+    {
+      "epoch": 0.9388412017167382,
+      "grad_norm": 1.9928634051631366,
+      "learning_rate": 6.388434969472307e-06,
+      "loss": 0.3076,
+      "step": 1750
+    },
+    {
+      "epoch": 0.9415236051502146,
+      "grad_norm": 1.8618767245463352,
+      "learning_rate": 6.365930721713718e-06,
+      "loss": 0.3073,
+      "step": 1755
+    },
+    {
+      "epoch": 0.944206008583691,
+      "grad_norm": 1.898851709477168,
+      "learning_rate": 6.343396531832497e-06,
+      "loss": 0.3117,
+      "step": 1760
+    },
+    {
+      "epoch": 0.9468884120171673,
+      "grad_norm": 1.9713962762389659,
+      "learning_rate": 6.320832893793285e-06,
+      "loss": 0.313,
+      "step": 1765
+    },
+    {
+      "epoch": 0.9495708154506438,
+      "grad_norm": 1.9747041589737182,
+      "learning_rate": 6.298240302206242e-06,
+      "loss": 0.3023,
+      "step": 1770
+    },
+    {
+      "epoch": 0.9522532188841202,
+      "grad_norm": 1.8841058711871417,
+      "learning_rate": 6.275619252316213e-06,
+      "loss": 0.3067,
+      "step": 1775
+    },
+    {
+      "epoch": 0.9549356223175965,
+      "grad_norm": 1.897284633014723,
+      "learning_rate": 6.25297023999187e-06,
+      "loss": 0.3126,
+      "step": 1780
+    },
+    {
+      "epoch": 0.9576180257510729,
+      "grad_norm": 1.9041195266763296,
+      "learning_rate": 6.2302937617148365e-06,
+      "loss": 0.3109,
+      "step": 1785
+    },
+    {
+      "epoch": 0.9603004291845494,
+      "grad_norm": 1.83771743974229,
+      "learning_rate": 6.20759031456881e-06,
+      "loss": 0.3096,
+      "step": 1790
+    },
+    {
+      "epoch": 0.9629828326180258,
+      "grad_norm": 1.80625064211799,
+      "learning_rate": 6.184860396228664e-06,
+      "loss": 0.31,
+      "step": 1795
+    },
+    {
+      "epoch": 0.9656652360515021,
+      "grad_norm": 1.8621679216479052,
+      "learning_rate": 6.1621045049495376e-06,
+      "loss": 0.2862,
+      "step": 1800
+    },
+    {
+      "epoch": 0.9683476394849786,
+      "grad_norm": 1.8652988069119067,
+      "learning_rate": 6.139323139555914e-06,
+      "loss": 0.3074,
+      "step": 1805
+    },
+    {
+      "epoch": 0.971030042918455,
+      "grad_norm": 1.8792185612098447,
+      "learning_rate": 6.116516799430689e-06,
+      "loss": 0.3022,
+      "step": 1810
+    },
+    {
+      "epoch": 0.9737124463519313,
+      "grad_norm": 1.9111804324769437,
+      "learning_rate": 6.0936859845042164e-06,
+      "loss": 0.2989,
+      "step": 1815
+    },
+    {
+      "epoch": 0.9763948497854077,
+      "grad_norm": 1.8921650943031454,
+      "learning_rate": 6.07083119524336e-06,
+      "loss": 0.3011,
+      "step": 1820
+    },
+    {
+      "epoch": 0.9790772532188842,
+      "grad_norm": 1.9316953376536945,
+      "learning_rate": 6.047952932640513e-06,
+      "loss": 0.2955,
+      "step": 1825
+    },
+    {
+      "epoch": 0.9817596566523605,
+      "grad_norm": 1.8846297197157076,
+      "learning_rate": 6.0250516982026205e-06,
+      "loss": 0.3123,
+      "step": 1830
+    },
+    {
+      "epoch": 0.9844420600858369,
+      "grad_norm": 1.8897008544103802,
+      "learning_rate": 6.002127993940187e-06,
+      "loss": 0.2921,
+      "step": 1835
+    },
+    {
+      "epoch": 0.9871244635193133,
+      "grad_norm": 1.9963273617506934,
+      "learning_rate": 5.979182322356269e-06,
+      "loss": 0.3186,
+      "step": 1840
+    },
+    {
+      "epoch": 0.9898068669527897,
+      "grad_norm": 1.9405309808504616,
+      "learning_rate": 5.956215186435464e-06,
+      "loss": 0.3043,
+      "step": 1845
+    },
+    {
+      "epoch": 0.9924892703862661,
+      "grad_norm": 1.9668200808889422,
+      "learning_rate": 5.9332270896328815e-06,
+      "loss": 0.2928,
+      "step": 1850
+    },
+    {
+      "epoch": 0.9951716738197425,
+      "grad_norm": 1.9086363268892765,
+      "learning_rate": 5.910218535863106e-06,
+      "loss": 0.2956,
+      "step": 1855
+    },
+    {
+      "epoch": 0.9978540772532188,
+      "grad_norm": 1.937226981087176,
+      "learning_rate": 5.8871900294891525e-06,
+      "loss": 0.3003,
+      "step": 1860
+    },
+    {
+      "epoch": 1.0,
+      "eval_runtime": 263.7729,
+      "eval_samples_per_second": 3.791,
+      "eval_steps_per_second": 0.948,
+      "step": 1864
+    },
+    {
+      "epoch": 1.0005364806866952,
+      "grad_norm": 2.1676471565285427,
+      "learning_rate": 5.864142075311414e-06,
+      "loss": 0.282,
+      "step": 1865
+    },
+    {
+      "epoch": 1.0032188841201717,
+      "grad_norm": 2.5924971299208486,
+      "learning_rate": 5.84107517855659e-06,
+      "loss": 0.2023,
+      "step": 1870
+    },
+    {
+      "epoch": 1.0059012875536482,
+      "grad_norm": 1.9918660104270434,
+      "learning_rate": 5.817989844866613e-06,
+      "loss": 0.1922,
+      "step": 1875
+    },
+    {
+      "epoch": 1.0085836909871244,
+      "grad_norm": 2.0582911095011767,
+      "learning_rate": 5.794886580287565e-06,
+      "loss": 0.2015,
+      "step": 1880
+    },
+    {
+      "epoch": 1.011266094420601,
+      "grad_norm": 1.9383711709784341,
+      "learning_rate": 5.77176589125859e-06,
+      "loss": 0.1984,
+      "step": 1885
+    },
+    {
+      "epoch": 1.0139484978540771,
+      "grad_norm": 2.201190788077746,
+      "learning_rate": 5.7486282846007835e-06,
+      "loss": 0.1889,
+      "step": 1890
+    },
+    {
+      "epoch": 1.0166309012875536,
+      "grad_norm": 1.9001195225997398,
+      "learning_rate": 5.725474267506088e-06,
+      "loss": 0.2041,
+      "step": 1895
+    },
+    {
+      "epoch": 1.01931330472103,
+      "grad_norm": 2.3294489874909132,
+      "learning_rate": 5.702304347526172e-06,
+      "loss": 0.1902,
+      "step": 1900
+    },
+    {
+      "epoch": 1.0219957081545064,
+      "grad_norm": 2.121399773248172,
+      "learning_rate": 5.679119032561311e-06,
+      "loss": 0.1842,
+      "step": 1905
+    },
+    {
+      "epoch": 1.0246781115879828,
+      "grad_norm": 1.9845020506511497,
+      "learning_rate": 5.655918830849243e-06,
+      "loss": 0.1882,
+      "step": 1910
+    },
+    {
+      "epoch": 1.0273605150214593,
+      "grad_norm": 1.8436034837755606,
+      "learning_rate": 5.632704250954039e-06,
+      "loss": 0.1988,
+      "step": 1915
+    },
+    {
+      "epoch": 1.0300429184549356,
+      "grad_norm": 1.9694560666463294,
+      "learning_rate": 5.6094758017549436e-06,
+      "loss": 0.1916,
+      "step": 1920
+    },
+    {
+      "epoch": 1.032725321888412,
+      "grad_norm": 1.9394642490903848,
+      "learning_rate": 5.5862339924352306e-06,
+      "loss": 0.1952,
+      "step": 1925
+    },
+    {
+      "epoch": 1.0354077253218885,
+      "grad_norm": 1.8828700417191846,
+      "learning_rate": 5.562979332471035e-06,
+      "loss": 0.1833,
+      "step": 1930
+    },
+    {
+      "epoch": 1.0380901287553648,
+      "grad_norm": 1.8908778787975598,
+      "learning_rate": 5.539712331620186e-06,
+      "loss": 0.1951,
+      "step": 1935
+    },
+    {
+      "epoch": 1.0407725321888412,
+      "grad_norm": 1.9116753445216206,
+      "learning_rate": 5.516433499911035e-06,
+      "loss": 0.1907,
+      "step": 1940
+    },
+    {
+      "epoch": 1.0434549356223175,
+      "grad_norm": 1.9940047534149794,
+      "learning_rate": 5.493143347631272e-06,
+      "loss": 0.1886,
+      "step": 1945
+    },
+    {
+      "epoch": 1.046137339055794,
+      "grad_norm": 1.9320996949249498,
+      "learning_rate": 5.4698423853167425e-06,
+      "loss": 0.1919,
+      "step": 1950
+    },
+    {
+      "epoch": 1.0488197424892705,
+      "grad_norm": 1.9586772638426304,
+      "learning_rate": 5.446531123740257e-06,
+      "loss": 0.1935,
+      "step": 1955
+    },
+    {
+      "epoch": 1.0515021459227467,
+      "grad_norm": 2.0011255021437075,
+      "learning_rate": 5.4232100739003855e-06,
+      "loss": 0.1945,
+      "step": 1960
+    },
+    {
+      "epoch": 1.0541845493562232,
+      "grad_norm": 1.9483623165568755,
+      "learning_rate": 5.399879747010275e-06,
+      "loss": 0.2003,
+      "step": 1965
+    },
+    {
+      "epoch": 1.0568669527896997,
+      "grad_norm": 1.9115878997120979,
+      "learning_rate": 5.376540654486422e-06,
+      "loss": 0.1883,
+      "step": 1970
+    },
+    {
+      "epoch": 1.059549356223176,
+      "grad_norm": 2.0116661404263145,
+      "learning_rate": 5.353193307937477e-06,
+      "loss": 0.2008,
+      "step": 1975
+    },
+    {
+      "epoch": 1.0622317596566524,
+      "grad_norm": 2.0832605757484397,
+      "learning_rate": 5.32983821915302e-06,
+      "loss": 0.1891,
+      "step": 1980
+    },
+    {
+      "epoch": 1.0649141630901287,
+      "grad_norm": 1.8531543345713375,
+      "learning_rate": 5.306475900092348e-06,
+      "loss": 0.1882,
+      "step": 1985
+    },
+    {
+      "epoch": 1.0675965665236051,
+      "grad_norm": 1.8450655041844881,
+      "learning_rate": 5.283106862873253e-06,
+      "loss": 0.1833,
+      "step": 1990
+    },
+    {
+      "epoch": 1.0702789699570816,
+      "grad_norm": 1.9044883788251383,
+      "learning_rate": 5.259731619760792e-06,
+      "loss": 0.1865,
+      "step": 1995
+    },
+    {
+      "epoch": 1.0729613733905579,
+      "grad_norm": 2.0028672116669997,
+      "learning_rate": 5.236350683156055e-06,
+      "loss": 0.1902,
+      "step": 2000
+    },
+    {
+      "epoch": 1.0756437768240343,
+      "grad_norm": 1.814402898247639,
+      "learning_rate": 5.212964565584944e-06,
+      "loss": 0.1905,
+      "step": 2005
+    },
+    {
+      "epoch": 1.0783261802575108,
+      "grad_norm": 1.8872679325335933,
+      "learning_rate": 5.189573779686929e-06,
+      "loss": 0.1949,
+      "step": 2010
+    },
+    {
+      "epoch": 1.081008583690987,
+      "grad_norm": 1.8891608815102554,
+      "learning_rate": 5.166178838203808e-06,
+      "loss": 0.1924,
+      "step": 2015
+    },
+    {
+      "epoch": 1.0836909871244635,
+      "grad_norm": 1.8837503323500486,
+      "learning_rate": 5.142780253968481e-06,
+      "loss": 0.1861,
+      "step": 2020
+    },
+    {
+      "epoch": 1.0863733905579398,
+      "grad_norm": 1.9577533659989275,
+      "learning_rate": 5.119378539893693e-06,
+      "loss": 0.1926,
+      "step": 2025
+    },
+    {
+      "epoch": 1.0890557939914163,
+      "grad_norm": 1.8789212764123366,
+      "learning_rate": 5.095974208960799e-06,
+      "loss": 0.194,
+      "step": 2030
+    },
+    {
+      "epoch": 1.0917381974248928,
+      "grad_norm": 2.0837412968641074,
+      "learning_rate": 5.072567774208518e-06,
+      "loss": 0.1938,
+      "step": 2035
+    },
+    {
+      "epoch": 1.094420600858369,
+      "grad_norm": 1.806432095417855,
+      "learning_rate": 5.049159748721685e-06,
+      "loss": 0.1875,
+      "step": 2040
+    },
+    {
+      "epoch": 1.0971030042918455,
+      "grad_norm": 2.0304775866784204,
+      "learning_rate": 5.025750645620004e-06,
+      "loss": 0.1865,
+      "step": 2045
+    },
+    {
+      "epoch": 1.099785407725322,
+      "grad_norm": 1.9275239991524047,
+      "learning_rate": 5.002340978046807e-06,
+      "loss": 0.1852,
+      "step": 2050
+    },
+    {
+      "epoch": 1.1024678111587982,
+      "grad_norm": 1.90287230101775,
+      "learning_rate": 4.978931259157791e-06,
+      "loss": 0.1835,
+      "step": 2055
+    },
+    {
+      "epoch": 1.1051502145922747,
+      "grad_norm": 1.890259840627779,
+      "learning_rate": 4.955522002109782e-06,
+      "loss": 0.189,
+      "step": 2060
+    },
+    {
+      "epoch": 1.1078326180257512,
+      "grad_norm": 1.90298316348042,
+      "learning_rate": 4.932113720049485e-06,
+      "loss": 0.1952,
+      "step": 2065
+    },
+    {
+      "epoch": 1.1105150214592274,
+      "grad_norm": 1.8449671760022834,
+      "learning_rate": 4.908706926102229e-06,
+      "loss": 0.1827,
+      "step": 2070
+    },
+    {
+      "epoch": 1.113197424892704,
+      "grad_norm": 1.9616445138070648,
+      "learning_rate": 4.885302133360722e-06,
+      "loss": 0.1952,
+      "step": 2075
+    },
+    {
+      "epoch": 1.1158798283261802,
+      "grad_norm": 1.926872317223682,
+      "learning_rate": 4.8618998548738065e-06,
+      "loss": 0.1912,
+      "step": 2080
+    },
+    {
+      "epoch": 1.1185622317596566,
+      "grad_norm": 1.8440433865572448,
+      "learning_rate": 4.8385006036352104e-06,
+      "loss": 0.1795,
+      "step": 2085
+    },
+    {
+      "epoch": 1.121244635193133,
+      "grad_norm": 1.8722843865210008,
+      "learning_rate": 4.8151048925723014e-06,
+      "loss": 0.194,
+      "step": 2090
+    },
+    {
+      "epoch": 1.1239270386266094,
+      "grad_norm": 1.9272856053757133,
+      "learning_rate": 4.791713234534844e-06,
+      "loss": 0.1846,
+      "step": 2095
+    },
+    {
+      "epoch": 1.1266094420600858,
+      "grad_norm": 1.874257158338157,
+      "learning_rate": 4.768326142283757e-06,
+      "loss": 0.1833,
+      "step": 2100
+    },
+    {
+      "epoch": 1.1292918454935623,
+      "grad_norm": 1.8978133368632915,
+      "learning_rate": 4.744944128479879e-06,
+      "loss": 0.1873,
+      "step": 2105
+    },
+    {
+      "epoch": 1.1319742489270386,
+      "grad_norm": 1.8595773129657196,
+      "learning_rate": 4.7215677056727185e-06,
+      "loss": 0.1904,
+      "step": 2110
+    },
+    {
+      "epoch": 1.134656652360515,
+      "grad_norm": 1.9790845158535326,
+      "learning_rate": 4.698197386289232e-06,
+      "loss": 0.1908,
+      "step": 2115
+    },
+    {
+      "epoch": 1.1373390557939915,
+      "grad_norm": 2.0046956900246267,
+      "learning_rate": 4.674833682622577e-06,
+      "loss": 0.1935,
+      "step": 2120
+    },
+    {
+      "epoch": 1.1400214592274678,
+      "grad_norm": 2.098142362831301,
+      "learning_rate": 4.6514771068209e-06,
+      "loss": 0.1856,
+      "step": 2125
+    },
+    {
+      "epoch": 1.1427038626609443,
+      "grad_norm": 1.8329956977422865,
+      "learning_rate": 4.628128170876093e-06,
+      "loss": 0.1896,
+      "step": 2130
+    },
+    {
+      "epoch": 1.1453862660944205,
+      "grad_norm": 1.949948729000319,
+      "learning_rate": 4.604787386612579e-06,
+      "loss": 0.1893,
+      "step": 2135
+    },
+    {
+      "epoch": 1.148068669527897,
+      "grad_norm": 1.8445942344820512,
+      "learning_rate": 4.581455265676089e-06,
+      "loss": 0.1941,
+      "step": 2140
+    },
+    {
+      "epoch": 1.1507510729613735,
+      "grad_norm": 1.805772422930948,
+      "learning_rate": 4.558132319522451e-06,
+      "loss": 0.1736,
+      "step": 2145
+    },
+    {
+      "epoch": 1.1534334763948497,
+      "grad_norm": 1.7406255756178988,
+      "learning_rate": 4.534819059406374e-06,
+      "loss": 0.1866,
+      "step": 2150
+    },
+    {
+      "epoch": 1.1561158798283262,
+      "grad_norm": 2.1014658354178644,
+      "learning_rate": 4.511515996370244e-06,
+      "loss": 0.1953,
+      "step": 2155
+    },
+    {
+      "epoch": 1.1587982832618025,
+      "grad_norm": 1.9708726994392791,
+      "learning_rate": 4.488223641232915e-06,
+      "loss": 0.1896,
+      "step": 2160
+    },
+    {
+      "epoch": 1.161480686695279,
+      "grad_norm": 1.9787947613530399,
+      "learning_rate": 4.464942504578524e-06,
+      "loss": 0.1934,
+      "step": 2165
+    },
+    {
+      "epoch": 1.1641630901287554,
+      "grad_norm": 1.992107685089113,
+      "learning_rate": 4.441673096745287e-06,
+      "loss": 0.1925,
+      "step": 2170
+    },
+    {
+      "epoch": 1.1668454935622319,
+      "grad_norm": 2.016896097316237,
+      "learning_rate": 4.418415927814315e-06,
+      "loss": 0.1848,
+      "step": 2175
+    },
+    {
+      "epoch": 1.1695278969957081,
+      "grad_norm": 1.858625974711402,
+      "learning_rate": 4.395171507598441e-06,
+      "loss": 0.1854,
+      "step": 2180
+    },
+    {
+      "epoch": 1.1722103004291846,
+      "grad_norm": 2.02254903047215,
+      "learning_rate": 4.371940345631027e-06,
+      "loss": 0.195,
+      "step": 2185
+    },
+    {
+      "epoch": 1.1748927038626609,
+      "grad_norm": 1.8176535693204507,
+      "learning_rate": 4.348722951154816e-06,
+      "loss": 0.1832,
+      "step": 2190
+    },
+    {
+      "epoch": 1.1775751072961373,
+      "grad_norm": 1.8304723547384139,
+      "learning_rate": 4.3255198331107485e-06,
+      "loss": 0.1879,
+      "step": 2195
+    },
+    {
+      "epoch": 1.1802575107296138,
+      "grad_norm": 1.8955122948963208,
+      "learning_rate": 4.302331500126824e-06,
+      "loss": 0.1871,
+      "step": 2200
+    },
+    {
+      "epoch": 1.18293991416309,
+      "grad_norm": 1.89004399640177,
+      "learning_rate": 4.279158460506939e-06,
+      "loss": 0.1819,
+      "step": 2205
+    },
+    {
+      "epoch": 1.1856223175965666,
+      "grad_norm": 1.8317609875432832,
+      "learning_rate": 4.256001222219751e-06,
+      "loss": 0.1831,
+      "step": 2210
+    },
+    {
+      "epoch": 1.1883047210300428,
+      "grad_norm": 1.9374939589512181,
+      "learning_rate": 4.232860292887537e-06,
+      "loss": 0.1806,
+      "step": 2215
+    },
+    {
+      "epoch": 1.1909871244635193,
+      "grad_norm": 2.0303723616744778,
+      "learning_rate": 4.2097361797750815e-06,
+      "loss": 0.1822,
+      "step": 2220
+    },
+    {
+      "epoch": 1.1936695278969958,
+      "grad_norm": 1.8192962113282214,
+      "learning_rate": 4.1866293897785356e-06,
+      "loss": 0.1844,
+      "step": 2225
+    },
+    {
+      "epoch": 1.196351931330472,
+      "grad_norm": 1.8604357202266497,
+      "learning_rate": 4.16354042941432e-06,
+      "loss": 0.1827,
+      "step": 2230
+    },
+    {
+      "epoch": 1.1990343347639485,
+      "grad_norm": 1.7595957688522177,
+      "learning_rate": 4.1404698048080175e-06,
+      "loss": 0.1934,
+      "step": 2235
+    },
+    {
+      "epoch": 1.201716738197425,
+      "grad_norm": 1.973110320134582,
+      "learning_rate": 4.117418021683278e-06,
+      "loss": 0.1906,
+      "step": 2240
+    },
+    {
+      "epoch": 1.2043991416309012,
+      "grad_norm": 1.8884430376988128,
+      "learning_rate": 4.094385585350736e-06,
+      "loss": 0.189,
+      "step": 2245
+    },
+    {
+      "epoch": 1.2070815450643777,
+      "grad_norm": 1.8745063485974562,
+      "learning_rate": 4.0713730006969285e-06,
+      "loss": 0.1844,
+      "step": 2250
+    },
+    {
+      "epoch": 1.2097639484978542,
+      "grad_norm": 1.8721552889135795,
+      "learning_rate": 4.048380772173231e-06,
+      "loss": 0.1922,
+      "step": 2255
+    },
+    {
+      "epoch": 1.2124463519313304,
+      "grad_norm": 2.0429736332378536,
+      "learning_rate": 4.0254094037848005e-06,
+      "loss": 0.1857,
+      "step": 2260
+    },
+    {
+      "epoch": 1.215128755364807,
+      "grad_norm": 1.8897709376195266,
+      "learning_rate": 4.002459399079523e-06,
+      "loss": 0.1853,
+      "step": 2265
+    },
+    {
+      "epoch": 1.2178111587982832,
+      "grad_norm": 1.9011418004749525,
+      "learning_rate": 3.979531261136981e-06,
+      "loss": 0.1886,
+      "step": 2270
+    },
+    {
+      "epoch": 1.2204935622317596,
+      "grad_norm": 1.969594278349951,
+      "learning_rate": 3.956625492557417e-06,
+      "loss": 0.1857,
+      "step": 2275
+    },
+    {
+      "epoch": 1.2231759656652361,
+      "grad_norm": 1.912032996448432,
+      "learning_rate": 3.933742595450733e-06,
+      "loss": 0.1849,
+      "step": 2280
+    },
+    {
+      "epoch": 1.2258583690987124,
+      "grad_norm": 1.916906298425393,
+      "learning_rate": 3.910883071425463e-06,
+      "loss": 0.1873,
+      "step": 2285
+    },
+    {
+      "epoch": 1.2285407725321889,
+      "grad_norm": 1.8578700716982628,
+      "learning_rate": 3.8880474215777915e-06,
+      "loss": 0.1838,
+      "step": 2290
+    },
+    {
+      "epoch": 1.2312231759656653,
+      "grad_norm": 1.776030564762745,
+      "learning_rate": 3.865236146480562e-06,
+      "loss": 0.1851,
+      "step": 2295
+    },
+    {
+      "epoch": 1.2339055793991416,
+      "grad_norm": 1.9137765272421998,
+      "learning_rate": 3.842449746172311e-06,
+      "loss": 0.1799,
+      "step": 2300
+    },
+    {
+      "epoch": 1.236587982832618,
+      "grad_norm": 2.0605656430444546,
+      "learning_rate": 3.8196887201463e-06,
+      "loss": 0.191,
+      "step": 2305
+    },
+    {
+      "epoch": 1.2392703862660945,
+      "grad_norm": 1.915088708869453,
+      "learning_rate": 3.796953567339571e-06,
+      "loss": 0.1952,
+      "step": 2310
+    },
+    {
+      "epoch": 1.2419527896995708,
+      "grad_norm": 1.8829918728592658,
+      "learning_rate": 3.7742447861220027e-06,
+      "loss": 0.1858,
+      "step": 2315
+    },
+    {
+      "epoch": 1.2446351931330473,
+      "grad_norm": 1.9585388606793075,
+      "learning_rate": 3.7515628742854006e-06,
+      "loss": 0.1858,
+      "step": 2320
+    },
+    {
+      "epoch": 1.2473175965665235,
+      "grad_norm": 1.8552310125655131,
+      "learning_rate": 3.7289083290325668e-06,
+      "loss": 0.1847,
+      "step": 2325
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.7754071029501433,
+      "learning_rate": 3.706281646966409e-06,
+      "loss": 0.1812,
+      "step": 2330
+    },
+    {
+      "epoch": 1.2526824034334765,
+      "grad_norm": 1.875617151757241,
+      "learning_rate": 3.6836833240790625e-06,
+      "loss": 0.185,
+      "step": 2335
+    },
+    {
+      "epoch": 1.2553648068669527,
+      "grad_norm": 1.8611799174871593,
+      "learning_rate": 3.6611138557410047e-06,
+      "loss": 0.1856,
+      "step": 2340
+    },
+    {
+      "epoch": 1.2580472103004292,
+      "grad_norm": 1.8285828739346939,
+      "learning_rate": 3.638573736690202e-06,
+      "loss": 0.1857,
+      "step": 2345
+    },
+    {
+      "epoch": 1.2607296137339055,
+      "grad_norm": 1.892962676656495,
+      "learning_rate": 3.6160634610212642e-06,
+      "loss": 0.1811,
+      "step": 2350
+    },
+    {
+      "epoch": 1.263412017167382,
+      "grad_norm": 1.9947034876105998,
+      "learning_rate": 3.5935835221746183e-06,
+      "loss": 0.1797,
+      "step": 2355
+    },
+    {
+      "epoch": 1.2660944206008584,
+      "grad_norm": 1.8965138864268372,
+      "learning_rate": 3.5711344129256832e-06,
+      "loss": 0.1858,
+      "step": 2360
+    },
+    {
+      "epoch": 1.268776824034335,
+      "grad_norm": 1.8162594165458399,
+      "learning_rate": 3.548716625374074e-06,
+      "loss": 0.1852,
+      "step": 2365
+    },
+    {
+      "epoch": 1.2714592274678111,
+      "grad_norm": 1.8466606151367597,
+      "learning_rate": 3.5263306509328103e-06,
+      "loss": 0.1905,
+      "step": 2370
+    },
+    {
+      "epoch": 1.2741416309012876,
+      "grad_norm": 1.804299478995684,
+      "learning_rate": 3.5039769803175545e-06,
+      "loss": 0.1849,
+      "step": 2375
+    },
+    {
+      "epoch": 1.2768240343347639,
+      "grad_norm": 1.8598714876367384,
+      "learning_rate": 3.481656103535839e-06,
+      "loss": 0.1863,
+      "step": 2380
+    },
+    {
+      "epoch": 1.2795064377682404,
+      "grad_norm": 1.8324237852834893,
+      "learning_rate": 3.459368509876338e-06,
+      "loss": 0.1872,
+      "step": 2385
+    },
+    {
+      "epoch": 1.2821888412017168,
+      "grad_norm": 1.9428876845890815,
+      "learning_rate": 3.437114687898132e-06,
+      "loss": 0.1848,
+      "step": 2390
+    },
+    {
+      "epoch": 1.284871244635193,
+      "grad_norm": 1.813341484571633,
+      "learning_rate": 3.414895125420013e-06,
+      "loss": 0.1794,
+      "step": 2395
+    },
+    {
+      "epoch": 1.2875536480686696,
+      "grad_norm": 1.891484957705928,
+      "learning_rate": 3.3927103095097725e-06,
+      "loss": 0.1826,
+      "step": 2400
+    },
+    {
+      "epoch": 1.2902360515021458,
+      "grad_norm": 1.8787480346224221,
+      "learning_rate": 3.370560726473537e-06,
+      "loss": 0.1871,
+      "step": 2405
+    },
+    {
+      "epoch": 1.2929184549356223,
+      "grad_norm": 1.856366211207286,
+      "learning_rate": 3.348446861845106e-06,
+      "loss": 0.1826,
+      "step": 2410
+    },
+    {
+      "epoch": 1.2956008583690988,
+      "grad_norm": 1.7873050403851052,
+      "learning_rate": 3.3263692003753056e-06,
+      "loss": 0.1805,
+      "step": 2415
+    },
+    {
+      "epoch": 1.298283261802575,
+      "grad_norm": 1.9278434650179597,
+      "learning_rate": 3.304328226021365e-06,
+      "loss": 0.1927,
+      "step": 2420
+    },
+    {
+      "epoch": 1.3009656652360515,
+      "grad_norm": 1.705467306294372,
+      "learning_rate": 3.282324421936307e-06,
+      "loss": 0.1807,
+      "step": 2425
+    },
+    {
+      "epoch": 1.3036480686695278,
+      "grad_norm": 2.0175168076056136,
+      "learning_rate": 3.2603582704583547e-06,
+      "loss": 0.1814,
+      "step": 2430
+    },
+    {
+      "epoch": 1.3063304721030042,
+      "grad_norm": 1.9599572180968123,
+      "learning_rate": 3.2384302531003676e-06,
+      "loss": 0.1809,
+      "step": 2435
+    },
+    {
+      "epoch": 1.3090128755364807,
+      "grad_norm": 1.88144040592366,
+      "learning_rate": 3.216540850539272e-06,
+      "loss": 0.1859,
+      "step": 2440
+    },
+    {
+      "epoch": 1.3116952789699572,
+      "grad_norm": 1.9414190234849964,
+      "learning_rate": 3.1946905426055353e-06,
+      "loss": 0.1855,
+      "step": 2445
+    },
+    {
+      "epoch": 1.3143776824034334,
+      "grad_norm": 1.7891027254290737,
+      "learning_rate": 3.172879808272642e-06,
+      "loss": 0.1886,
+      "step": 2450
+    },
+    {
+      "epoch": 1.31706008583691,
+      "grad_norm": 1.779801789067273,
+      "learning_rate": 3.151109125646601e-06,
+      "loss": 0.1804,
+      "step": 2455
+    },
+    {
+      "epoch": 1.3197424892703862,
+      "grad_norm": 1.6910321536471429,
+      "learning_rate": 3.1293789719554562e-06,
+      "loss": 0.1888,
+      "step": 2460
+    },
+    {
+      "epoch": 1.3224248927038627,
+      "grad_norm": 1.8978060550474511,
+      "learning_rate": 3.107689823538833e-06,
+      "loss": 0.1845,
+      "step": 2465
+    },
+    {
+      "epoch": 1.3251072961373391,
+      "grad_norm": 1.7996160906604526,
+      "learning_rate": 3.086042155837491e-06,
+      "loss": 0.182,
+      "step": 2470
+    },
+    {
+      "epoch": 1.3277896995708154,
+      "grad_norm": 1.8604318440083263,
+      "learning_rate": 3.0644364433829076e-06,
+      "loss": 0.1868,
+      "step": 2475
+    },
+    {
+      "epoch": 1.3304721030042919,
+      "grad_norm": 1.7265652913644474,
+      "learning_rate": 3.0428731597868706e-06,
+      "loss": 0.1791,
+      "step": 2480
+    },
+    {
+      "epoch": 1.3331545064377681,
+      "grad_norm": 1.8572102327854523,
+      "learning_rate": 3.021352777731096e-06,
+      "loss": 0.1889,
+      "step": 2485
+    },
+    {
+      "epoch": 1.3358369098712446,
+      "grad_norm": 1.978318614280375,
+      "learning_rate": 2.9998757689568775e-06,
+      "loss": 0.1826,
+      "step": 2490
+    },
+    {
+      "epoch": 1.338519313304721,
+      "grad_norm": 1.814421622601161,
+      "learning_rate": 2.978442604254729e-06,
+      "loss": 0.1811,
+      "step": 2495
+    },
+    {
+      "epoch": 1.3412017167381975,
+      "grad_norm": 1.8994448714761314,
+      "learning_rate": 2.9570537534540765e-06,
+      "loss": 0.1885,
+      "step": 2500
+    },
+    {
+      "epoch": 1.3438841201716738,
+      "grad_norm": 1.7626420531321445,
+      "learning_rate": 2.935709685412954e-06,
+      "loss": 0.1805,
+      "step": 2505
+    },
+    {
+      "epoch": 1.3465665236051503,
+      "grad_norm": 1.8364054864432415,
+      "learning_rate": 2.9144108680077288e-06,
+      "loss": 0.1826,
+      "step": 2510
+    },
+    {
+      "epoch": 1.3492489270386265,
+      "grad_norm": 1.847180352588691,
+      "learning_rate": 2.8931577681228407e-06,
+      "loss": 0.1807,
+      "step": 2515
+    },
+    {
+      "epoch": 1.351931330472103,
+      "grad_norm": 1.9524895693219806,
+      "learning_rate": 2.871950851640577e-06,
+      "loss": 0.1854,
+      "step": 2520
+    },
+    {
+      "epoch": 1.3546137339055795,
+      "grad_norm": 1.7963474200991176,
+      "learning_rate": 2.8507905834308417e-06,
+      "loss": 0.1877,
+      "step": 2525
+    },
+    {
+      "epoch": 1.3572961373390557,
+      "grad_norm": 1.7833300163922787,
+      "learning_rate": 2.8296774273409944e-06,
+      "loss": 0.1823,
+      "step": 2530
+    },
+    {
+      "epoch": 1.3599785407725322,
+      "grad_norm": 1.9793055958250187,
+      "learning_rate": 2.8086118461856494e-06,
+      "loss": 0.1804,
+      "step": 2535
+    },
+    {
+      "epoch": 1.3626609442060085,
+      "grad_norm": 1.8681720017812595,
+      "learning_rate": 2.787594301736556e-06,
+      "loss": 0.1898,
+      "step": 2540
+    },
+    {
+      "epoch": 1.365343347639485,
+      "grad_norm": 2.05995430381249,
+      "learning_rate": 2.7666252547124596e-06,
+      "loss": 0.1833,
+      "step": 2545
+    },
+    {
+      "epoch": 1.3680257510729614,
+      "grad_norm": 1.7854047855241022,
+      "learning_rate": 2.745705164769015e-06,
+      "loss": 0.179,
+      "step": 2550
+    },
+    {
+      "epoch": 1.370708154506438,
+      "grad_norm": 1.8166180302356945,
+      "learning_rate": 2.724834490488705e-06,
+      "loss": 0.1813,
+      "step": 2555
+    },
+    {
+      "epoch": 1.3733905579399142,
+      "grad_norm": 1.8066041043622898,
+      "learning_rate": 2.7040136893707813e-06,
+      "loss": 0.1789,
+      "step": 2560
+    },
+    {
+      "epoch": 1.3760729613733906,
+      "grad_norm": 1.903301979210332,
+      "learning_rate": 2.683243217821248e-06,
+      "loss": 0.1769,
+      "step": 2565
+    },
+    {
+      "epoch": 1.378755364806867,
+      "grad_norm": 1.7466773845577472,
+      "learning_rate": 2.66252353114285e-06,
+      "loss": 0.1841,
+      "step": 2570
+    },
+    {
+      "epoch": 1.3814377682403434,
+      "grad_norm": 1.8171236308268557,
+      "learning_rate": 2.6418550835250946e-06,
+      "loss": 0.1816,
+      "step": 2575
+    },
+    {
+      "epoch": 1.3841201716738198,
+      "grad_norm": 1.8941618390274477,
+      "learning_rate": 2.621238328034289e-06,
+      "loss": 0.1792,
+      "step": 2580
+    },
+    {
+      "epoch": 1.386802575107296,
+      "grad_norm": 1.688474610894831,
+      "learning_rate": 2.60067371660362e-06,
+      "loss": 0.172,
+      "step": 2585
+    },
+    {
+      "epoch": 1.3894849785407726,
+      "grad_norm": 1.918062296580036,
+      "learning_rate": 2.5801617000232416e-06,
+      "loss": 0.1875,
+      "step": 2590
+    },
+    {
+      "epoch": 1.3921673819742488,
+      "grad_norm": 1.8511243429388167,
+      "learning_rate": 2.559702727930386e-06,
+      "loss": 0.1765,
+      "step": 2595
+    },
+    {
+      "epoch": 1.3948497854077253,
+      "grad_norm": 1.7999901888436554,
+      "learning_rate": 2.5392972487995247e-06,
+      "loss": 0.1785,
+      "step": 2600
+    },
+    {
+      "epoch": 1.3975321888412018,
+      "grad_norm": 1.7053303808234024,
+      "learning_rate": 2.5189457099325153e-06,
+      "loss": 0.1756,
+      "step": 2605
+    },
+    {
+      "epoch": 1.400214592274678,
+      "grad_norm": 1.8663136591492098,
+      "learning_rate": 2.498648557448824e-06,
+      "loss": 0.1753,
+      "step": 2610
+    },
+    {
+      "epoch": 1.4028969957081545,
+      "grad_norm": 1.7472818518305704,
+      "learning_rate": 2.4784062362757156e-06,
+      "loss": 0.176,
+      "step": 2615
+    },
+    {
+      "epoch": 1.4055793991416308,
+      "grad_norm": 1.769076507145111,
+      "learning_rate": 2.458219190138526e-06,
+      "loss": 0.1815,
+      "step": 2620
+    },
+    {
+      "epoch": 1.4082618025751072,
+      "grad_norm": 1.8072957625179205,
+      "learning_rate": 2.4380878615509156e-06,
+      "loss": 0.1792,
+      "step": 2625
+    },
+    {
+      "epoch": 1.4109442060085837,
+      "grad_norm": 1.7819943428340819,
+      "learning_rate": 2.418012691805191e-06,
+      "loss": 0.1784,
+      "step": 2630
+    },
+    {
+      "epoch": 1.4136266094420602,
+      "grad_norm": 1.805441370100766,
+      "learning_rate": 2.3979941209626072e-06,
+      "loss": 0.1763,
+      "step": 2635
+    },
+    {
+      "epoch": 1.4163090128755365,
+      "grad_norm": 1.9782749901291634,
+      "learning_rate": 2.3780325878437415e-06,
+      "loss": 0.1781,
+      "step": 2640
+    },
+    {
+      "epoch": 1.418991416309013,
+      "grad_norm": 1.852385566702638,
+      "learning_rate": 2.358128530018858e-06,
+      "loss": 0.1836,
+      "step": 2645
+    },
+    {
+      "epoch": 1.4216738197424892,
+      "grad_norm": 2.0347102265117227,
+      "learning_rate": 2.3382823837983314e-06,
+      "loss": 0.178,
+      "step": 2650
+    },
+    {
+      "epoch": 1.4243562231759657,
+      "grad_norm": 1.8292259452694213,
+      "learning_rate": 2.318494584223072e-06,
+      "loss": 0.1863,
+      "step": 2655
+    },
+    {
+      "epoch": 1.4270386266094421,
+      "grad_norm": 1.8186682872751867,
+      "learning_rate": 2.2987655650549862e-06,
+      "loss": 0.1779,
+      "step": 2660
+    },
+    {
+      "epoch": 1.4297210300429184,
+      "grad_norm": 1.8137249215325242,
+      "learning_rate": 2.2790957587674876e-06,
+      "loss": 0.1773,
+      "step": 2665
+    },
+    {
+      "epoch": 1.4324034334763949,
+      "grad_norm": 1.657463931666108,
+      "learning_rate": 2.2594855965359906e-06,
+      "loss": 0.181,
+      "step": 2670
+    },
+    {
+      "epoch": 1.4350858369098711,
+      "grad_norm": 1.8416553188106275,
+      "learning_rate": 2.2399355082284804e-06,
+      "loss": 0.1837,
+      "step": 2675
+    },
+    {
+      "epoch": 1.4377682403433476,
+      "grad_norm": 1.8567384673546208,
+      "learning_rate": 2.2204459223960716e-06,
+      "loss": 0.1753,
+      "step": 2680
+    },
+    {
+      "epoch": 1.440450643776824,
+      "grad_norm": 1.8228824438814255,
+      "learning_rate": 2.2010172662636377e-06,
+      "loss": 0.1876,
+      "step": 2685
+    },
+    {
+      "epoch": 1.4431330472103006,
+      "grad_norm": 1.913150616422004,
+      "learning_rate": 2.1816499657204183e-06,
+      "loss": 0.1757,
+      "step": 2690
+    },
+    {
+      "epoch": 1.4458154506437768,
+      "grad_norm": 1.7948486965689003,
+      "learning_rate": 2.1623444453107067e-06,
+      "loss": 0.1811,
+      "step": 2695
+    },
+    {
+      "epoch": 1.4484978540772533,
+      "grad_norm": 1.7836453698169266,
+      "learning_rate": 2.1431011282245274e-06,
+      "loss": 0.1839,
+      "step": 2700
+    },
+    {
+      "epoch": 1.4511802575107295,
+      "grad_norm": 1.755770567268044,
+      "learning_rate": 2.12392043628837e-06,
+      "loss": 0.1772,
+      "step": 2705
+    },
+    {
+      "epoch": 1.453862660944206,
+      "grad_norm": 1.773946245819418,
+      "learning_rate": 2.10480278995594e-06,
+      "loss": 0.1793,
+      "step": 2710
+    },
+    {
+      "epoch": 1.4565450643776825,
+      "grad_norm": 1.7879606075165497,
+      "learning_rate": 2.0857486082989344e-06,
+      "loss": 0.183,
+      "step": 2715
+    },
+    {
+      "epoch": 1.4592274678111588,
+      "grad_norm": 1.7610119078458486,
+      "learning_rate": 2.0667583089978673e-06,
+      "loss": 0.1808,
+      "step": 2720
+    },
+    {
+      "epoch": 1.4619098712446352,
+      "grad_norm": 1.92829549876405,
+      "learning_rate": 2.0478323083329072e-06,
+      "loss": 0.1743,
+      "step": 2725
+    },
+    {
+      "epoch": 1.4645922746781115,
+      "grad_norm": 1.8350507362803108,
+      "learning_rate": 2.028971021174754e-06,
+      "loss": 0.1846,
+      "step": 2730
+    },
+    {
+      "epoch": 1.467274678111588,
+      "grad_norm": 1.7803185569940212,
+      "learning_rate": 2.0101748609755407e-06,
+      "loss": 0.1838,
+      "step": 2735
+    },
+    {
+      "epoch": 1.4699570815450644,
+      "grad_norm": 2.1675460899901,
+      "learning_rate": 1.9914442397597756e-06,
+      "loss": 0.176,
+      "step": 2740
+    },
+    {
+      "epoch": 1.4726394849785407,
+      "grad_norm": 1.8724514691396603,
+      "learning_rate": 1.9727795681153083e-06,
+      "loss": 0.1687,
+      "step": 2745
+    },
+    {
+      "epoch": 1.4753218884120172,
+      "grad_norm": 1.6802576933125253,
+      "learning_rate": 1.954181255184331e-06,
+      "loss": 0.1752,
+      "step": 2750
+    },
+    {
+      "epoch": 1.4780042918454936,
+      "grad_norm": 1.6820388901704724,
+      "learning_rate": 1.935649708654403e-06,
+      "loss": 0.1792,
+      "step": 2755
+    },
+    {
+      "epoch": 1.48068669527897,
+      "grad_norm": 1.7979050607058435,
+      "learning_rate": 1.9171853347495234e-06,
+      "loss": 0.1756,
+      "step": 2760
+    },
+    {
+      "epoch": 1.4833690987124464,
+      "grad_norm": 1.7614860072950995,
+      "learning_rate": 1.8987885382212235e-06,
+      "loss": 0.1793,
+      "step": 2765
+    },
+    {
+      "epoch": 1.4860515021459229,
+      "grad_norm": 1.7504341827692425,
+      "learning_rate": 1.8804597223396865e-06,
+      "loss": 0.1819,
+      "step": 2770
+    },
+    {
+      "epoch": 1.488733905579399,
+      "grad_norm": 1.8669933379497394,
+      "learning_rate": 1.8621992888849217e-06,
+      "loss": 0.1807,
+      "step": 2775
+    },
+    {
+      "epoch": 1.4914163090128756,
+      "grad_norm": 1.725707828875719,
+      "learning_rate": 1.8440076381379395e-06,
+      "loss": 0.1774,
+      "step": 2780
+    },
+    {
+      "epoch": 1.4940987124463518,
+      "grad_norm": 1.7856774718839294,
+      "learning_rate": 1.8258851688720009e-06,
+      "loss": 0.1703,
+      "step": 2785
+    },
+    {
+      "epoch": 1.4967811158798283,
+      "grad_norm": 1.9173370042151976,
+      "learning_rate": 1.807832278343849e-06,
+      "loss": 0.1825,
+      "step": 2790
+    },
+    {
+      "epoch": 1.4994635193133048,
+      "grad_norm": 1.8338932891918263,
+      "learning_rate": 1.7898493622850227e-06,
+      "loss": 0.1771,
+      "step": 2795
+    },
+    {
+      "epoch": 1.5021459227467813,
+      "grad_norm": 1.6990771812486194,
+      "learning_rate": 1.771936814893167e-06,
+      "loss": 0.1776,
+      "step": 2800
+    },
+    {
+      "epoch": 1.5048283261802575,
+      "grad_norm": 1.9186848087579995,
+      "learning_rate": 1.7540950288234033e-06,
+      "loss": 0.1868,
+      "step": 2805
+    },
+    {
+      "epoch": 1.5075107296137338,
+      "grad_norm": 1.7935654243257526,
+      "learning_rate": 1.7363243951797155e-06,
+      "loss": 0.1757,
+      "step": 2810
+    },
+    {
+      "epoch": 1.5101931330472103,
+      "grad_norm": 1.7009512554045776,
+      "learning_rate": 1.7186253035063738e-06,
+      "loss": 0.1683,
+      "step": 2815
+    },
+    {
+      "epoch": 1.5128755364806867,
+      "grad_norm": 1.6934539545605378,
+      "learning_rate": 1.7009981417794114e-06,
+      "loss": 0.1726,
+      "step": 2820
+    },
+    {
+      "epoch": 1.5155579399141632,
+      "grad_norm": 1.7582377866252281,
+      "learning_rate": 1.6834432963980957e-06,
+      "loss": 0.1778,
+      "step": 2825
+    },
+    {
+      "epoch": 1.5182403433476395,
+      "grad_norm": 1.8632756654647402,
+      "learning_rate": 1.6659611521764807e-06,
+      "loss": 0.1768,
+      "step": 2830
+    },
+    {
+      "epoch": 1.5209227467811157,
+      "grad_norm": 1.8284613621943484,
+      "learning_rate": 1.6485520923349529e-06,
+      "loss": 0.1721,
+      "step": 2835
+    },
+    {
+      "epoch": 1.5236051502145922,
+      "grad_norm": 1.7899204272020013,
+      "learning_rate": 1.6312164984918516e-06,
+      "loss": 0.1626,
+      "step": 2840
+    },
+    {
+      "epoch": 1.5262875536480687,
+      "grad_norm": 1.770033358177422,
+      "learning_rate": 1.6139547506550808e-06,
+      "loss": 0.1815,
+      "step": 2845
+    },
+    {
+      "epoch": 1.5289699570815452,
+      "grad_norm": 1.7649583624949958,
+      "learning_rate": 1.5967672272137968e-06,
+      "loss": 0.1847,
+      "step": 2850
+    },
+    {
+      "epoch": 1.5316523605150214,
+      "grad_norm": 1.7927358149656596,
+      "learning_rate": 1.5796543049301033e-06,
+      "loss": 0.1752,
+      "step": 2855
+    },
+    {
+      "epoch": 1.5343347639484979,
+      "grad_norm": 2.0042367491150577,
+      "learning_rate": 1.5626163589307991e-06,
+      "loss": 0.1822,
+      "step": 2860
+    },
+    {
+      "epoch": 1.5370171673819741,
+      "grad_norm": 1.7107730523248998,
+      "learning_rate": 1.5456537626991525e-06,
+      "loss": 0.1694,
+      "step": 2865
+    },
+    {
+      "epoch": 1.5396995708154506,
+      "grad_norm": 1.7441203072639992,
+      "learning_rate": 1.5287668880667107e-06,
+      "loss": 0.1708,
+      "step": 2870
+    },
+    {
+      "epoch": 1.542381974248927,
+      "grad_norm": 1.8476477662752824,
+      "learning_rate": 1.5119561052051546e-06,
+      "loss": 0.1735,
+      "step": 2875
+    },
+    {
+      "epoch": 1.5450643776824036,
+      "grad_norm": 1.8492021302759543,
+      "learning_rate": 1.495221782618183e-06,
+      "loss": 0.1792,
+      "step": 2880
+    },
+    {
+      "epoch": 1.5477467811158798,
+      "grad_norm": 1.8722713008127667,
+      "learning_rate": 1.4785642871334349e-06,
+      "loss": 0.1699,
+      "step": 2885
+    },
+    {
+      "epoch": 1.550429184549356,
+      "grad_norm": 1.8179023947737942,
+      "learning_rate": 1.4619839838944416e-06,
+      "loss": 0.1763,
+      "step": 2890
+    },
+    {
+      "epoch": 1.5531115879828326,
+      "grad_norm": 1.7771808696153055,
+      "learning_rate": 1.4454812363526339e-06,
+      "loss": 0.1732,
+      "step": 2895
+    },
+    {
+      "epoch": 1.555793991416309,
+      "grad_norm": 1.7067802122286628,
+      "learning_rate": 1.429056406259368e-06,
+      "loss": 0.1628,
+      "step": 2900
+    },
+    {
+      "epoch": 1.5584763948497855,
+      "grad_norm": 1.8463964562057142,
+      "learning_rate": 1.4127098536579982e-06,
+      "loss": 0.1819,
+      "step": 2905
+    },
+    {
+      "epoch": 1.5611587982832618,
+      "grad_norm": 1.7971311069451972,
+      "learning_rate": 1.3964419368759786e-06,
+      "loss": 0.1706,
+      "step": 2910
+    },
+    {
+      "epoch": 1.5638412017167382,
+      "grad_norm": 1.789561911147988,
+      "learning_rate": 1.380253012517019e-06,
+      "loss": 0.1684,
+      "step": 2915
+    },
+    {
+      "epoch": 1.5665236051502145,
+      "grad_norm": 1.7986915556625842,
+      "learning_rate": 1.3641434354532595e-06,
+      "loss": 0.1749,
+      "step": 2920
+    },
+    {
+      "epoch": 1.569206008583691,
+      "grad_norm": 1.728513882390203,
+      "learning_rate": 1.3481135588174926e-06,
+      "loss": 0.1717,
+      "step": 2925
+    },
+    {
+      "epoch": 1.5718884120171674,
+      "grad_norm": 1.8474557981862971,
+      "learning_rate": 1.332163733995427e-06,
+      "loss": 0.1797,
+      "step": 2930
+    },
+    {
+      "epoch": 1.574570815450644,
+      "grad_norm": 1.8124533163869139,
+      "learning_rate": 1.3162943106179748e-06,
+      "loss": 0.175,
+      "step": 2935
+    },
+    {
+      "epoch": 1.5772532188841202,
+      "grad_norm": 1.8363148844561816,
+      "learning_rate": 1.3005056365536067e-06,
+      "loss": 0.1763,
+      "step": 2940
+    },
+    {
+      "epoch": 1.5799356223175964,
+      "grad_norm": 1.7667627502771912,
+      "learning_rate": 1.2847980579007003e-06,
+      "loss": 0.1709,
+      "step": 2945
+    },
+    {
+      "epoch": 1.582618025751073,
+      "grad_norm": 1.8384934616186792,
+      "learning_rate": 1.2691719189799774e-06,
+      "loss": 0.1809,
+      "step": 2950
+    },
+    {
+      "epoch": 1.5853004291845494,
+      "grad_norm": 1.7432995409757972,
+      "learning_rate": 1.253627562326936e-06,
+      "loss": 0.1712,
+      "step": 2955
+    },
+    {
+      "epoch": 1.5879828326180259,
+      "grad_norm": 1.6497919957428313,
+      "learning_rate": 1.2381653286843648e-06,
+      "loss": 0.1695,
+      "step": 2960
+    },
+    {
+      "epoch": 1.5906652360515021,
+      "grad_norm": 1.8578458306330736,
+      "learning_rate": 1.2227855569948477e-06,
+      "loss": 0.1788,
+      "step": 2965
+    },
+    {
+      "epoch": 1.5933476394849786,
+      "grad_norm": 1.7101495845126193,
+      "learning_rate": 1.2074885843933542e-06,
+      "loss": 0.1668,
+      "step": 2970
+    },
+    {
+      "epoch": 1.5960300429184548,
+      "grad_norm": 1.8706776246586578,
+      "learning_rate": 1.1922747461998425e-06,
+      "loss": 0.1792,
+      "step": 2975
+    },
+    {
+      "epoch": 1.5987124463519313,
+      "grad_norm": 1.7146003209383944,
+      "learning_rate": 1.1771443759119028e-06,
+      "loss": 0.1762,
+      "step": 2980
+    },
+    {
+      "epoch": 1.6013948497854078,
+      "grad_norm": 1.7587464469782965,
+      "learning_rate": 1.162097805197459e-06,
+      "loss": 0.1721,
+      "step": 2985
+    },
+    {
+      "epoch": 1.6040772532188843,
+      "grad_norm": 1.671773670726898,
+      "learning_rate": 1.147135363887485e-06,
+      "loss": 0.1736,
+      "step": 2990
+    },
+    {
+      "epoch": 1.6067596566523605,
+      "grad_norm": 1.7831449677470894,
+      "learning_rate": 1.1322573799687904e-06,
+      "loss": 0.1821,
+      "step": 2995
+    },
+    {
+      "epoch": 1.6094420600858368,
+      "grad_norm": 1.749081297778821,
+      "learning_rate": 1.1174641795768132e-06,
+      "loss": 0.1749,
+      "step": 3000
+    },
+    {
+      "epoch": 1.6121244635193133,
+      "grad_norm": 1.6885288263413876,
+      "learning_rate": 1.1027560869884845e-06,
+      "loss": 0.1803,
+      "step": 3005
+    },
+    {
+      "epoch": 1.6148068669527897,
+      "grad_norm": 1.7789601448566963,
+      "learning_rate": 1.0881334246151114e-06,
+      "loss": 0.1792,
+      "step": 3010
+    },
+    {
+      "epoch": 1.6174892703862662,
+      "grad_norm": 1.797127788770941,
+      "learning_rate": 1.073596512995317e-06,
+      "loss": 0.1766,
+      "step": 3015
+    },
+    {
+      "epoch": 1.6201716738197425,
+      "grad_norm": 1.6913152928957935,
+      "learning_rate": 1.0591456707880077e-06,
+      "loss": 0.1677,
+      "step": 3020
+    },
+    {
+      "epoch": 1.6228540772532187,
+      "grad_norm": 1.9388861906655925,
+      "learning_rate": 1.0447812147653885e-06,
+      "loss": 0.1693,
+      "step": 3025
+    },
+    {
+      "epoch": 1.6255364806866952,
+      "grad_norm": 1.8287339322524556,
+      "learning_rate": 1.0305034598060254e-06,
+      "loss": 0.185,
+      "step": 3030
+    },
+    {
+      "epoch": 1.6282188841201717,
+      "grad_norm": 1.7359121362575998,
+      "learning_rate": 1.0163127188879352e-06,
+      "loss": 0.1696,
+      "step": 3035
+    },
+    {
+      "epoch": 1.6309012875536482,
+      "grad_norm": 1.7976891275154785,
+      "learning_rate": 1.0022093030817316e-06,
+      "loss": 0.1734,
+      "step": 3040
+    },
+    {
+      "epoch": 1.6335836909871244,
+      "grad_norm": 1.7789540175564873,
+      "learning_rate": 9.88193521543797e-07,
+      "loss": 0.1717,
+      "step": 3045
+    },
+    {
+      "epoch": 1.636266094420601,
+      "grad_norm": 1.7088064901560376,
+      "learning_rate": 9.742656815095175e-07,
+      "loss": 0.1693,
+      "step": 3050
+    },
+    {
+      "epoch": 1.6389484978540771,
+      "grad_norm": 1.6486482877435467,
+      "learning_rate": 9.604260882865395e-07,
+      "loss": 0.1743,
+      "step": 3055
+    },
+    {
+      "epoch": 1.6416309012875536,
+      "grad_norm": 1.7097523703340793,
+      "learning_rate": 9.466750452480816e-07,
+      "loss": 0.1675,
+      "step": 3060
+    },
+    {
+      "epoch": 1.64431330472103,
+      "grad_norm": 1.8142751972429312,
+      "learning_rate": 9.330128538262784e-07,
+      "loss": 0.1673,
+      "step": 3065
+    },
+    {
+      "epoch": 1.6469957081545066,
+      "grad_norm": 1.7335787796330107,
+      "learning_rate": 9.194398135055815e-07,
+      "loss": 0.1744,
+      "step": 3070
+    },
+    {
+      "epoch": 1.6496781115879828,
+      "grad_norm": 1.7933634237592693,
+      "learning_rate": 9.059562218161894e-07,
+      "loss": 0.1731,
+      "step": 3075
+    },
+    {
+      "epoch": 1.652360515021459,
+      "grad_norm": 1.7144165521860568,
+      "learning_rate": 8.925623743275235e-07,
+      "loss": 0.1794,
+      "step": 3080
+    },
+    {
+      "epoch": 1.6550429184549356,
+      "grad_norm": 1.7854348497119321,
+      "learning_rate": 8.792585646417568e-07,
+      "loss": 0.1775,
+      "step": 3085
+    },
+    {
+      "epoch": 1.657725321888412,
+      "grad_norm": 1.7455552036989075,
+      "learning_rate": 8.660450843873647e-07,
+      "loss": 0.1787,
+      "step": 3090
+    },
+    {
+      "epoch": 1.6604077253218885,
+      "grad_norm": 1.7895684168501305,
+      "learning_rate": 8.529222232127526e-07,
+      "loss": 0.1748,
+      "step": 3095
+    },
+    {
+      "epoch": 1.6630901287553648,
+      "grad_norm": 1.8436594740961931,
+      "learning_rate": 8.398902687798832e-07,
+      "loss": 0.1695,
+      "step": 3100
+    },
+    {
+      "epoch": 1.6657725321888412,
+      "grad_norm": 1.7479630576592862,
+      "learning_rate": 8.269495067579891e-07,
+      "loss": 0.184,
+      "step": 3105
+    },
+    {
+      "epoch": 1.6684549356223175,
+      "grad_norm": 1.7572581212303144,
+      "learning_rate": 8.141002208172977e-07,
+      "loss": 0.1761,
+      "step": 3110
+    },
+    {
+      "epoch": 1.671137339055794,
+      "grad_norm": 1.754681973827706,
+      "learning_rate": 8.013426926228274e-07,
+      "loss": 0.1735,
+      "step": 3115
+    },
+    {
+      "epoch": 1.6738197424892705,
+      "grad_norm": 1.9053550149600742,
+      "learning_rate": 7.886772018281969e-07,
+      "loss": 0.1796,
+      "step": 3120
+    },
+    {
+      "epoch": 1.676502145922747,
+      "grad_norm": 1.7066025328597623,
+      "learning_rate": 7.761040260695074e-07,
+      "loss": 0.168,
+      "step": 3125
+    },
+    {
+      "epoch": 1.6791845493562232,
+      "grad_norm": 1.8140596030248617,
+      "learning_rate": 7.636234409592524e-07,
+      "loss": 0.1685,
+      "step": 3130
+    },
+    {
+      "epoch": 1.6818669527896994,
+      "grad_norm": 1.823899761561944,
+      "learning_rate": 7.512357200802722e-07,
+      "loss": 0.174,
+      "step": 3135
+    },
+    {
+      "epoch": 1.684549356223176,
+      "grad_norm": 1.8364772641113434,
+      "learning_rate": 7.389411349797654e-07,
+      "loss": 0.1705,
+      "step": 3140
+    },
+    {
+      "epoch": 1.6872317596566524,
+      "grad_norm": 1.6806223651736183,
+      "learning_rate": 7.267399551633253e-07,
+      "loss": 0.1706,
+      "step": 3145
+    },
+    {
+      "epoch": 1.6899141630901289,
+      "grad_norm": 1.891605548895355,
+      "learning_rate": 7.146324480890476e-07,
+      "loss": 0.1779,
+      "step": 3150
+    },
+    {
+      "epoch": 1.6925965665236051,
+      "grad_norm": 1.788446168769966,
+      "learning_rate": 7.026188791616484e-07,
+      "loss": 0.1789,
+      "step": 3155
+    },
+    {
+      "epoch": 1.6952789699570814,
+      "grad_norm": 1.6493280124322705,
+      "learning_rate": 6.906995117266641e-07,
+      "loss": 0.1719,
+      "step": 3160
+    },
+    {
+      "epoch": 1.6979613733905579,
+      "grad_norm": 1.7790805165351633,
+      "learning_rate": 6.788746070646646e-07,
+      "loss": 0.1785,
+      "step": 3165
+    },
+    {
+      "epoch": 1.7006437768240343,
+      "grad_norm": 2.173524080131244,
+      "learning_rate": 6.671444243855368e-07,
+      "loss": 0.1694,
+      "step": 3170
+    },
+    {
+      "epoch": 1.7033261802575108,
+      "grad_norm": 1.7419492752952908,
+      "learning_rate": 6.555092208227953e-07,
+      "loss": 0.1729,
+      "step": 3175
+    },
+    {
+      "epoch": 1.7060085836909873,
+      "grad_norm": 1.7470702873021877,
+      "learning_rate": 6.439692514279516e-07,
+      "loss": 0.1664,
+      "step": 3180
+    },
+    {
+      "epoch": 1.7086909871244635,
+      "grad_norm": 1.7707011703025006,
+      "learning_rate": 6.325247691649139e-07,
+      "loss": 0.1673,
+      "step": 3185
+    },
+    {
+      "epoch": 1.7113733905579398,
+      "grad_norm": 1.8390785050997485,
+      "learning_rate": 6.211760249044535e-07,
+      "loss": 0.1677,
+      "step": 3190
+    },
+    {
+      "epoch": 1.7140557939914163,
+      "grad_norm": 1.6988621285323615,
+      "learning_rate": 6.099232674187e-07,
+      "loss": 0.1697,
+      "step": 3195
+    },
+    {
+      "epoch": 1.7167381974248928,
+      "grad_norm": 1.666246086200389,
+      "learning_rate": 5.987667433756844e-07,
+      "loss": 0.1706,
+      "step": 3200
+    },
+    {
+      "epoch": 1.7194206008583692,
+      "grad_norm": 1.8080719591700882,
+      "learning_rate": 5.877066973339379e-07,
+      "loss": 0.1745,
+      "step": 3205
+    },
+    {
+      "epoch": 1.7221030042918455,
+      "grad_norm": 1.721957786717317,
+      "learning_rate": 5.767433717371301e-07,
+      "loss": 0.1713,
+      "step": 3210
+    },
+    {
+      "epoch": 1.7247854077253217,
+      "grad_norm": 1.717789424603017,
+      "learning_rate": 5.658770069087521e-07,
+      "loss": 0.1678,
+      "step": 3215
+    },
+    {
+      "epoch": 1.7274678111587982,
+      "grad_norm": 1.7053886441436985,
+      "learning_rate": 5.551078410468486e-07,
+      "loss": 0.1672,
+      "step": 3220
+    },
+    {
+      "epoch": 1.7301502145922747,
+      "grad_norm": 1.819056657288576,
+      "learning_rate": 5.444361102187979e-07,
+      "loss": 0.1714,
+      "step": 3225
+    },
+    {
+      "epoch": 1.7328326180257512,
+      "grad_norm": 1.7995104574020357,
+      "learning_rate": 5.338620483561386e-07,
+      "loss": 0.1752,
+      "step": 3230
+    },
+    {
+      "epoch": 1.7355150214592274,
+      "grad_norm": 1.6994520700013371,
+      "learning_rate": 5.233858872494357e-07,
+      "loss": 0.1753,
+      "step": 3235
+    },
+    {
+      "epoch": 1.738197424892704,
+      "grad_norm": 1.8671476647333491,
+      "learning_rate": 5.130078565432089e-07,
+      "loss": 0.1703,
+      "step": 3240
+    },
+    {
+      "epoch": 1.7408798283261802,
+      "grad_norm": 1.6772491004527577,
+      "learning_rate": 5.027281837308873e-07,
+      "loss": 0.1628,
+      "step": 3245
+    },
+    {
+      "epoch": 1.7435622317596566,
+      "grad_norm": 1.754411882099184,
+      "learning_rate": 4.925470941498345e-07,
+      "loss": 0.1689,
+      "step": 3250
+    },
+    {
+      "epoch": 1.746244635193133,
+      "grad_norm": 1.6353039993025373,
+      "learning_rate": 4.824648109763991e-07,
+      "loss": 0.1692,
+      "step": 3255
+    },
+    {
+      "epoch": 1.7489270386266096,
+      "grad_norm": 1.7434544510607133,
+      "learning_rate": 4.724815552210288e-07,
+      "loss": 0.1704,
+      "step": 3260
+    },
+    {
+      "epoch": 1.7516094420600858,
+      "grad_norm": 1.6704471472999787,
+      "learning_rate": 4.6259754572342e-07,
+      "loss": 0.1716,
+      "step": 3265
+    },
+    {
+      "epoch": 1.754291845493562,
+      "grad_norm": 1.7516950888101692,
+      "learning_rate": 4.5281299914773146e-07,
+      "loss": 0.1741,
+      "step": 3270
+    },
+    {
+      "epoch": 1.7569742489270386,
+      "grad_norm": 1.7040325954172368,
+      "learning_rate": 4.43128129977819e-07,
+      "loss": 0.1748,
+      "step": 3275
+    },
+    {
+      "epoch": 1.759656652360515,
+      "grad_norm": 1.7021022207307568,
+      "learning_rate": 4.3354315051254927e-07,
+      "loss": 0.1765,
+      "step": 3280
+    },
+    {
+      "epoch": 1.7623390557939915,
+      "grad_norm": 1.7196523854425305,
+      "learning_rate": 4.2405827086113406e-07,
+      "loss": 0.1698,
+      "step": 3285
+    },
+    {
+      "epoch": 1.7650214592274678,
+      "grad_norm": 1.702915568531188,
+      "learning_rate": 4.146736989385336e-07,
+      "loss": 0.1598,
+      "step": 3290
+    },
+    {
+      "epoch": 1.7677038626609443,
+      "grad_norm": 1.6995494482853168,
+      "learning_rate": 4.0538964046089426e-07,
+      "loss": 0.1756,
+      "step": 3295
+    },
+    {
+      "epoch": 1.7703862660944205,
+      "grad_norm": 1.7641412694504541,
+      "learning_rate": 3.962062989410359e-07,
+      "loss": 0.1726,
+      "step": 3300
+    },
+    {
+      "epoch": 1.773068669527897,
+      "grad_norm": 1.7780397843294091,
+      "learning_rate": 3.871238756840029e-07,
+      "loss": 0.1743,
+      "step": 3305
+    },
+    {
+      "epoch": 1.7757510729613735,
+      "grad_norm": 1.812082659957161,
+      "learning_rate": 3.7814256978263465e-07,
+      "loss": 0.1701,
+      "step": 3310
+    },
+    {
+      "epoch": 1.77843347639485,
+      "grad_norm": 1.7020500555044733,
+      "learning_rate": 3.6926257811321585e-07,
+      "loss": 0.1656,
+      "step": 3315
+    },
+    {
+      "epoch": 1.7811158798283262,
+      "grad_norm": 1.8126139098321488,
+      "learning_rate": 3.604840953311506e-07,
+      "loss": 0.1676,
+      "step": 3320
+    },
+    {
+      "epoch": 1.7837982832618025,
+      "grad_norm": 1.7560277944500373,
+      "learning_rate": 3.518073138667044e-07,
+      "loss": 0.1711,
+      "step": 3325
+    },
+    {
+      "epoch": 1.786480686695279,
+      "grad_norm": 1.763432081520576,
+      "learning_rate": 3.4323242392077737e-07,
+      "loss": 0.1711,
+      "step": 3330
+    },
+    {
+      "epoch": 1.7891630901287554,
+      "grad_norm": 1.6297190806246264,
+      "learning_rate": 3.347596134607406e-07,
+      "loss": 0.1682,
+      "step": 3335
+    },
+    {
+      "epoch": 1.7918454935622319,
+      "grad_norm": 1.8062944470139457,
+      "learning_rate": 3.263890682163129e-07,
+      "loss": 0.1749,
+      "step": 3340
+    },
+    {
+      "epoch": 1.7945278969957081,
+      "grad_norm": 1.7881739648077752,
+      "learning_rate": 3.1812097167549127e-07,
+      "loss": 0.1735,
+      "step": 3345
+    },
+    {
+      "epoch": 1.7972103004291844,
+      "grad_norm": 1.6572256761747672,
+      "learning_rate": 3.0995550508052976e-07,
+      "loss": 0.1727,
+      "step": 3350
+    },
+    {
+      "epoch": 1.7998927038626609,
+      "grad_norm": 1.6259819425995783,
+      "learning_rate": 3.018928474239613e-07,
+      "loss": 0.1652,
+      "step": 3355
+    },
+    {
+      "epoch": 1.8025751072961373,
+      "grad_norm": 1.7420901379335734,
+      "learning_rate": 2.9393317544468003e-07,
+      "loss": 0.1777,
+      "step": 3360
+    },
+    {
+      "epoch": 1.8052575107296138,
+      "grad_norm": 1.7760823431379773,
+      "learning_rate": 2.860766636240636e-07,
+      "loss": 0.1739,
+      "step": 3365
+    },
+    {
+      "epoch": 1.80793991416309,
+      "grad_norm": 1.643271804266458,
+      "learning_rate": 2.7832348418215084e-07,
+      "loss": 0.1705,
+      "step": 3370
+    },
+    {
+      "epoch": 1.8106223175965666,
+      "grad_norm": 1.7104159344251477,
+      "learning_rate": 2.7067380707386235e-07,
+      "loss": 0.1735,
+      "step": 3375
+    },
+    {
+      "epoch": 1.8133047210300428,
+      "grad_norm": 1.6791868888450323,
+      "learning_rate": 2.631277999852799e-07,
+      "loss": 0.1724,
+      "step": 3380
+    },
+    {
+      "epoch": 1.8159871244635193,
+      "grad_norm": 1.7140730397456243,
+      "learning_rate": 2.556856283299691e-07,
+      "loss": 0.1676,
+      "step": 3385
+    },
+    {
+      "epoch": 1.8186695278969958,
+      "grad_norm": 1.6077185197341657,
+      "learning_rate": 2.483474552453513e-07,
+      "loss": 0.1746,
+      "step": 3390
+    },
+    {
+      "epoch": 1.8213519313304722,
+      "grad_norm": 1.7870974144659697,
+      "learning_rate": 2.4111344158912863e-07,
+      "loss": 0.1721,
+      "step": 3395
+    },
+    {
+      "epoch": 1.8240343347639485,
+      "grad_norm": 1.6827957273473264,
+      "learning_rate": 2.3398374593576022e-07,
+      "loss": 0.1614,
+      "step": 3400
+    },
+    {
+      "epoch": 1.8267167381974247,
+      "grad_norm": 1.6876082896664166,
+      "learning_rate": 2.2695852457298328e-07,
+      "loss": 0.1725,
+      "step": 3405
+    },
+    {
+      "epoch": 1.8293991416309012,
+      "grad_norm": 1.7726305091723993,
+      "learning_rate": 2.2003793149838692e-07,
+      "loss": 0.1703,
+      "step": 3410
+    },
+    {
+      "epoch": 1.8320815450643777,
+      "grad_norm": 1.8349898381032062,
+      "learning_rate": 2.1322211841604046e-07,
+      "loss": 0.1684,
+      "step": 3415
+    },
+    {
+      "epoch": 1.8347639484978542,
+      "grad_norm": 1.8816226837004817,
+      "learning_rate": 2.0651123473316103e-07,
+      "loss": 0.18,
+      "step": 3420
+    },
+    {
+      "epoch": 1.8374463519313304,
+      "grad_norm": 1.7671307777064689,
+      "learning_rate": 1.9990542755684738e-07,
+      "loss": 0.1673,
+      "step": 3425
+    },
+    {
+      "epoch": 1.840128755364807,
+      "grad_norm": 1.6997077829480403,
+      "learning_rate": 1.9340484169084627e-07,
+      "loss": 0.1695,
+      "step": 3430
+    },
+    {
+      "epoch": 1.8428111587982832,
+      "grad_norm": 1.7386958389708476,
+      "learning_rate": 1.870096196323856e-07,
+      "loss": 0.175,
+      "step": 3435
+    },
+    {
+      "epoch": 1.8454935622317596,
+      "grad_norm": 1.679414270357438,
+      "learning_rate": 1.8071990156904362e-07,
+      "loss": 0.1676,
+      "step": 3440
+    },
+    {
+      "epoch": 1.8481759656652361,
+      "grad_norm": 1.7180649273928403,
+      "learning_rate": 1.7453582537568404e-07,
+      "loss": 0.171,
+      "step": 3445
+    },
+    {
+      "epoch": 1.8508583690987126,
+      "grad_norm": 1.761958633555483,
+      "learning_rate": 1.6845752661142744e-07,
+      "loss": 0.1727,
+      "step": 3450
+    },
+    {
+      "epoch": 1.8535407725321889,
+      "grad_norm": 1.723808407240347,
+      "learning_rate": 1.624851385166809e-07,
+      "loss": 0.1688,
+      "step": 3455
+    },
+    {
+      "epoch": 1.856223175965665,
+      "grad_norm": 1.7305314750444647,
+      "learning_rate": 1.5661879201022135e-07,
+      "loss": 0.1642,
+      "step": 3460
+    },
+    {
+      "epoch": 1.8589055793991416,
+      "grad_norm": 1.8485234245901698,
+      "learning_rate": 1.5085861568631845e-07,
+      "loss": 0.1748,
+      "step": 3465
+    },
+    {
+      "epoch": 1.861587982832618,
+      "grad_norm": 1.6803556355862883,
+      "learning_rate": 1.4520473581192407e-07,
+      "loss": 0.1644,
+      "step": 3470
+    },
+    {
+      "epoch": 1.8642703862660945,
+      "grad_norm": 1.870312744732992,
+      "learning_rate": 1.396572763238957e-07,
+      "loss": 0.1699,
+      "step": 3475
+    },
+    {
+      "epoch": 1.8669527896995708,
+      "grad_norm": 1.678351269677729,
+      "learning_rate": 1.3421635882628958e-07,
+      "loss": 0.1701,
+      "step": 3480
+    },
+    {
+      "epoch": 1.869635193133047,
+      "grad_norm": 1.900918288526919,
+      "learning_rate": 1.2888210258768464e-07,
+      "loss": 0.1737,
+      "step": 3485
+    },
+    {
+      "epoch": 1.8723175965665235,
+      "grad_norm": 1.7501849452560176,
+      "learning_rate": 1.2365462453857612e-07,
+      "loss": 0.171,
+      "step": 3490
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 1.7643463297208026,
+      "learning_rate": 1.1853403926880725e-07,
+      "loss": 0.1803,
+      "step": 3495
+    },
+    {
+      "epoch": 1.8776824034334765,
+      "grad_norm": 1.7456320007378667,
+      "learning_rate": 1.1352045902506158e-07,
+      "loss": 0.1772,
+      "step": 3500
+    },
+    {
+      "epoch": 1.880364806866953,
+      "grad_norm": 1.6733217604292803,
+      "learning_rate": 1.0861399370839964e-07,
+      "loss": 0.1708,
+      "step": 3505
+    },
+    {
+      "epoch": 1.8830472103004292,
+      "grad_norm": 1.7601489763981855,
+      "learning_rate": 1.03814750871849e-07,
+      "loss": 0.1729,
+      "step": 3510
+    },
+    {
+      "epoch": 1.8857296137339055,
+      "grad_norm": 1.7554273513816643,
+      "learning_rate": 9.912283571805015e-08,
+      "loss": 0.1725,
+      "step": 3515
+    },
+    {
+      "epoch": 1.888412017167382,
+      "grad_norm": 1.8155918592259217,
+      "learning_rate": 9.45383510969472e-08,
+      "loss": 0.179,
+      "step": 3520
+    },
+    {
+      "epoch": 1.8910944206008584,
+      "grad_norm": 1.724970161445885,
+      "learning_rate": 9.006139750353526e-08,
+      "loss": 0.1719,
+      "step": 3525
+    },
+    {
+      "epoch": 1.893776824034335,
+      "grad_norm": 1.7143787726536936,
+      "learning_rate": 8.569207307565664e-08,
+      "loss": 0.178,
+      "step": 3530
+    },
+    {
+      "epoch": 1.8964592274678111,
+      "grad_norm": 1.7675589277617567,
+      "learning_rate": 8.143047359184863e-08,
+      "loss": 0.1767,
+      "step": 3535
+    },
+    {
+      "epoch": 1.8991416309012874,
+      "grad_norm": 1.7988231066894382,
+      "learning_rate": 7.727669246924697e-08,
+      "loss": 0.1721,
+      "step": 3540
+    },
+    {
+      "epoch": 1.9018240343347639,
+      "grad_norm": 1.7917443907599273,
+      "learning_rate": 7.32308207615351e-08,
+      "loss": 0.1729,
+      "step": 3545
+    },
+    {
+      "epoch": 1.9045064377682404,
+      "grad_norm": 1.7247547707337405,
+      "learning_rate": 6.929294715694923e-08,
+      "loss": 0.1675,
+      "step": 3550
+    },
+    {
+      "epoch": 1.9071888412017168,
+      "grad_norm": 1.7092676553238282,
+      "learning_rate": 6.54631579763343e-08,
+      "loss": 0.1732,
+      "step": 3555
+    },
+    {
+      "epoch": 1.909871244635193,
+      "grad_norm": 1.6963652820446242,
+      "learning_rate": 6.174153717125264e-08,
+      "loss": 0.1648,
+      "step": 3560
+    },
+    {
+      "epoch": 1.9125536480686696,
+      "grad_norm": 1.6728309532039858,
+      "learning_rate": 5.812816632214169e-08,
+      "loss": 0.1744,
+      "step": 3565
+    },
+    {
+      "epoch": 1.9152360515021458,
+      "grad_norm": 1.8234146842434622,
+      "learning_rate": 5.4623124636528635e-08,
+      "loss": 0.1743,
+      "step": 3570
+    },
+    {
+      "epoch": 1.9179184549356223,
+      "grad_norm": 1.7337294621225616,
+      "learning_rate": 5.122648894728854e-08,
+      "loss": 0.1583,
+      "step": 3575
+    },
+    {
+      "epoch": 1.9206008583690988,
+      "grad_norm": 1.8631680318842838,
+      "learning_rate": 4.7938333710969564e-08,
+      "loss": 0.1678,
+      "step": 3580
+    },
+    {
+      "epoch": 1.9232832618025753,
+      "grad_norm": 1.6286398974109075,
+      "learning_rate": 4.4758731006149804e-08,
+      "loss": 0.1655,
+      "step": 3585
+    },
+    {
+      "epoch": 1.9259656652360515,
+      "grad_norm": 1.6651386998279623,
+      "learning_rate": 4.16877505318658e-08,
+      "loss": 0.1649,
+      "step": 3590
+    },
+    {
+      "epoch": 1.9286480686695278,
+      "grad_norm": 1.7215780598096833,
+      "learning_rate": 3.872545960608099e-08,
+      "loss": 0.1716,
+      "step": 3595
+    },
+    {
+      "epoch": 1.9313304721030042,
+      "grad_norm": 1.8496948473611403,
+      "learning_rate": 3.587192316420962e-08,
+      "loss": 0.1836,
+      "step": 3600
+    },
+    {
+      "epoch": 1.9340128755364807,
+      "grad_norm": 1.789269898205407,
+      "learning_rate": 3.312720375769518e-08,
+      "loss": 0.1711,
+      "step": 3605
+    },
+    {
+      "epoch": 1.9366952789699572,
+      "grad_norm": 1.7661067152019674,
+      "learning_rate": 3.04913615526381e-08,
+      "loss": 0.1711,
+      "step": 3610
+    },
+    {
+      "epoch": 1.9393776824034334,
+      "grad_norm": 1.7277593438439514,
+      "learning_rate": 2.796445432847794e-08,
+      "loss": 0.1683,
+      "step": 3615
+    },
+    {
+      "epoch": 1.94206008583691,
+      "grad_norm": 1.666312242597901,
+      "learning_rate": 2.554653747672442e-08,
+      "loss": 0.168,
+      "step": 3620
+    },
+    {
+      "epoch": 1.9447424892703862,
+      "grad_norm": 1.7365407287758425,
+      "learning_rate": 2.323766399974614e-08,
+      "loss": 0.1703,
+      "step": 3625
+    },
+    {
+      "epoch": 1.9474248927038627,
+      "grad_norm": 1.7377335353091357,
+      "learning_rate": 2.1037884509605976e-08,
+      "loss": 0.1748,
+      "step": 3630
+    },
+    {
+      "epoch": 1.9501072961373391,
+      "grad_norm": 1.8309351566068142,
+      "learning_rate": 1.8947247226954736e-08,
+      "loss": 0.1713,
+      "step": 3635
+    },
+    {
+      "epoch": 1.9527896995708156,
+      "grad_norm": 1.776864049048742,
+      "learning_rate": 1.6965797979971442e-08,
+      "loss": 0.1773,
+      "step": 3640
+    },
+    {
+      "epoch": 1.9554721030042919,
+      "grad_norm": 1.6767580087158616,
+      "learning_rate": 1.509358020336027e-08,
+      "loss": 0.1666,
+      "step": 3645
+    },
+    {
+      "epoch": 1.9581545064377681,
+      "grad_norm": 1.817374124019115,
+      "learning_rate": 1.3330634937396835e-08,
+      "loss": 0.1616,
+      "step": 3650
+    },
+    {
+      "epoch": 1.9608369098712446,
+      "grad_norm": 1.7773594339045193,
+      "learning_rate": 1.1677000827030604e-08,
+      "loss": 0.1727,
+      "step": 3655
+    },
+    {
+      "epoch": 1.963519313304721,
+      "grad_norm": 1.7837035371769219,
+      "learning_rate": 1.0132714121037223e-08,
+      "loss": 0.175,
+      "step": 3660
+    },
+    {
+      "epoch": 1.9662017167381975,
+      "grad_norm": 1.6240013537396132,
+      "learning_rate": 8.697808671221385e-09,
+      "loss": 0.1707,
+      "step": 3665
+    },
+    {
+      "epoch": 1.9688841201716738,
+      "grad_norm": 1.6065913572501274,
+      "learning_rate": 7.3723159316796414e-09,
+      "loss": 0.1727,
+      "step": 3670
+    },
+    {
+      "epoch": 1.97156652360515,
+      "grad_norm": 1.7024659714900665,
+      "learning_rate": 6.1562649581059505e-09,
+      "loss": 0.168,
+      "step": 3675
+    },
+    {
+      "epoch": 1.9742489270386265,
+      "grad_norm": 1.6959109954027434,
+      "learning_rate": 5.049682407157752e-09,
+      "loss": 0.1685,
+      "step": 3680
+    },
+    {
+      "epoch": 1.976931330472103,
+      "grad_norm": 1.6969423589462422,
+      "learning_rate": 4.052592535871425e-09,
+      "loss": 0.1694,
+      "step": 3685
+    },
+    {
+      "epoch": 1.9796137339055795,
+      "grad_norm": 1.8395970305653224,
+      "learning_rate": 3.1650172011293834e-09,
+      "loss": 0.1794,
+      "step": 3690
+    },
+    {
+      "epoch": 1.9822961373390557,
+      "grad_norm": 1.6774442707442048,
+      "learning_rate": 2.3869758591810177e-09,
+      "loss": 0.1662,
+      "step": 3695
+    },
+    {
+      "epoch": 1.9849785407725322,
+      "grad_norm": 1.6932165167708886,
+      "learning_rate": 1.718485565218031e-09,
+      "loss": 0.1676,
+      "step": 3700
+    },
+    {
+      "epoch": 1.9876609442060085,
+      "grad_norm": 1.7427139009863861,
+      "learning_rate": 1.15956097299752e-09,
+      "loss": 0.1648,
+      "step": 3705
+    },
+    {
+      "epoch": 1.990343347639485,
+      "grad_norm": 1.701084604202847,
+      "learning_rate": 7.102143345238955e-10,
+      "loss": 0.1702,
+      "step": 3710
+    },
+    {
+      "epoch": 1.9930257510729614,
+      "grad_norm": 1.703457401005494,
+      "learning_rate": 3.7045549977909877e-10,
+      "loss": 0.1695,
+      "step": 3715
+    },
+    {
+      "epoch": 1.995708154506438,
+      "grad_norm": 1.6886515476568755,
+      "learning_rate": 1.4029191650555274e-10,
+      "loss": 0.1751,
+      "step": 3720
+    },
+    {
+      "epoch": 1.9983905579399142,
+      "grad_norm": 1.7655789088614147,
+      "learning_rate": 1.9728630044069107e-11,
+      "loss": 0.1713,
+      "step": 3725
+    },
+    {
+      "epoch": 2.0,
+      "eval_runtime": 263.6121,
+      "eval_samples_per_second": 3.793,
+      "eval_steps_per_second": 0.948,
+      "step": 3728
+    },
+    {
+      "epoch": 2.0,
+      "step": 3728,
+      "total_flos": 390283678187520.0,
+      "train_loss": 0.24746462182329704,
+      "train_runtime": 33763.0865,
+      "train_samples_per_second": 1.766,
+      "train_steps_per_second": 0.11
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 3728,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 390283678187520.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}