diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15148 @@
+{
+  "best_metric": 1.4233994483947754,
+  "best_model_checkpoint": "./results/models/checkpoint-922740",
+  "epoch": 7.0,
+  "eval_steps": 500,
+  "global_step": 1076530,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003251186683139346,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.0019998699525326743,
+      "loss": 2.8127,
+      "step": 500
+    },
+    {
+      "epoch": 0.006502373366278692,
+      "grad_norm": 67.0,
+      "learning_rate": 0.001999739905065349,
+      "loss": 2.4852,
+      "step": 1000
+    },
+    {
+      "epoch": 0.009753560049418037,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0019996098575980233,
+      "loss": 2.3924,
+      "step": 1500
+    },
+    {
+      "epoch": 0.013004746732557384,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0019994798101306975,
+      "loss": 2.3235,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01625593341569673,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.0019993497626633722,
+      "loss": 2.2647,
+      "step": 2500
+    },
+    {
+      "epoch": 0.019507120098836074,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019992197151960465,
+      "loss": 2.2077,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02275830678197542,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0019990896677287207,
+      "loss": 2.1543,
+      "step": 3500
+    },
+    {
+      "epoch": 0.026009493465114768,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0019989596202613954,
+      "loss": 2.1291,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02926068014825411,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00199882957279407,
+      "loss": 2.1703,
+      "step": 4500
+    },
+    {
+      "epoch": 0.03251186683139346,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0019986995253267444,
+      "loss": 2.1796,
+      "step": 5000
+    },
+    {
+      "epoch": 0.0357630535145328,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0019985694778594187,
+      "loss": 2.1274,
+      "step": 5500
+    },
+    {
+      "epoch": 0.03901424019767215,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.0019984394303920934,
+      "loss": 2.0801,
+      "step": 6000
+    },
+    {
+      "epoch": 0.042265426880811495,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0019983093829247676,
+      "loss": 2.0708,
+      "step": 6500
+    },
+    {
+      "epoch": 0.04551661356395084,
+      "grad_norm": 1.0,
+      "learning_rate": 0.001998179335457442,
+      "loss": 2.0459,
+      "step": 7000
+    },
+    {
+      "epoch": 0.04876780024709019,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019980492879901166,
+      "loss": 2.0342,
+      "step": 7500
+    },
+    {
+      "epoch": 0.052018986930229535,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.001997919240522791,
+      "loss": 2.0241,
+      "step": 8000
+    },
+    {
+      "epoch": 0.05527017361336888,
+      "grad_norm": 1.0,
+      "learning_rate": 0.001997789193055465,
+      "loss": 2.0065,
+      "step": 8500
+    },
+    {
+      "epoch": 0.05852136029650822,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00199765914558814,
+      "loss": 1.992,
+      "step": 9000
+    },
+    {
+      "epoch": 0.06177254697964757,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.001997529098120814,
+      "loss": 1.9743,
+      "step": 9500
+    },
+    {
+      "epoch": 0.06502373366278692,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0019973990506534883,
+      "loss": 1.9478,
+      "step": 10000
+    },
+    {
+      "epoch": 0.06827492034592626,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001997269003186163,
+      "loss": 1.943,
+      "step": 10500
+    },
+    {
+      "epoch": 0.0715261070290656,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0019971389557188377,
+      "loss": 1.92,
+      "step": 11000
+    },
+    {
+      "epoch": 0.07477729371220496,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.001997008908251512,
+      "loss": 1.9189,
+      "step": 11500
+    },
+    {
+      "epoch": 0.0780284803953443,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0019968788607841862,
+      "loss": 1.9228,
+      "step": 12000
+    },
+    {
+      "epoch": 0.08127966707848365,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.001996748813316861,
+      "loss": 1.9151,
+      "step": 12500
+    },
+    {
+      "epoch": 0.08453085376162299,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.001996618765849535,
+      "loss": 1.9197,
+      "step": 13000
+    },
+    {
+      "epoch": 0.08778204044476234,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.0019964887183822094,
+      "loss": 1.9055,
+      "step": 13500
+    },
+    {
+      "epoch": 0.09103322712790168,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001996358670914884,
+      "loss": 1.8883,
+      "step": 14000
+    },
+    {
+      "epoch": 0.09428441381104102,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.0019962286234475584,
+      "loss": 1.89,
+      "step": 14500
+    },
+    {
+      "epoch": 0.09753560049418038,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0019960985759802327,
+      "loss": 1.8763,
+      "step": 15000
+    },
+    {
+      "epoch": 0.10078678717731972,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0019959685285129074,
+      "loss": 1.8751,
+      "step": 15500
+    },
+    {
+      "epoch": 0.10403797386045907,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.0019958384810455816,
+      "loss": 1.8804,
+      "step": 16000
+    },
+    {
+      "epoch": 0.10728916054359841,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001995708433578256,
+      "loss": 1.8864,
+      "step": 16500
+    },
+    {
+      "epoch": 0.11054034722673776,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0019955783861109306,
+      "loss": 1.8726,
+      "step": 17000
+    },
+    {
+      "epoch": 0.1137915339098771,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.001995448338643605,
+      "loss": 1.8611,
+      "step": 17500
+    },
+    {
+      "epoch": 0.11704272059301644,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.001995318291176279,
+      "loss": 1.8544,
+      "step": 18000
+    },
+    {
+      "epoch": 0.1202939072761558,
+      "grad_norm": 2.34375,
+      "learning_rate": 0.001995188243708954,
+      "loss": 1.8469,
+      "step": 18500
+    },
+    {
+      "epoch": 0.12354509395929514,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0019950581962416285,
+      "loss": 1.8493,
+      "step": 19000
+    },
+    {
+      "epoch": 0.12679628064243448,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.0019949281487743028,
+      "loss": 1.8449,
+      "step": 19500
+    },
+    {
+      "epoch": 0.13004746732557385,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.001994798101306977,
+      "loss": 1.8341,
+      "step": 20000
+    },
+    {
+      "epoch": 0.13329865400871319,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0019946680538396517,
+      "loss": 1.8258,
+      "step": 20500
+    },
+    {
+      "epoch": 0.13654984069185253,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001994538006372326,
+      "loss": 1.8294,
+      "step": 21000
+    },
+    {
+      "epoch": 0.13980102737499187,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.0019944079589050002,
+      "loss": 1.8346,
+      "step": 21500
+    },
+    {
+      "epoch": 0.1430522140581312,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.001994277911437675,
+      "loss": 1.8136,
+      "step": 22000
+    },
+    {
+      "epoch": 0.14630340074127057,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.001994147863970349,
+      "loss": 1.8161,
+      "step": 22500
+    },
+    {
+      "epoch": 0.1495545874244099,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0019940178165030235,
+      "loss": 1.8165,
+      "step": 23000
+    },
+    {
+      "epoch": 0.15280577410754925,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001993887769035698,
+      "loss": 1.8134,
+      "step": 23500
+    },
+    {
+      "epoch": 0.1560569607906886,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0019937577215683724,
+      "loss": 1.8177,
+      "step": 24000
+    },
+    {
+      "epoch": 0.15930814747382796,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0019936276741010467,
+      "loss": 1.8106,
+      "step": 24500
+    },
+    {
+      "epoch": 0.1625593341569673,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.0019934976266337214,
+      "loss": 1.808,
+      "step": 25000
+    },
+    {
+      "epoch": 0.16581052084010664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.001993367579166396,
+      "loss": 1.804,
+      "step": 25500
+    },
+    {
+      "epoch": 0.16906170752324598,
+      "grad_norm": 8.6875,
+      "learning_rate": 0.0019932375316990703,
+      "loss": 1.7985,
+      "step": 26000
+    },
+    {
+      "epoch": 0.17231289420638532,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0019931074842317446,
+      "loss": 1.7876,
+      "step": 26500
+    },
+    {
+      "epoch": 0.1755640808895247,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0019929774367644193,
+      "loss": 1.787,
+      "step": 27000
+    },
+    {
+      "epoch": 0.17881526757266403,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.0019928473892970935,
+      "loss": 1.7812,
+      "step": 27500
+    },
+    {
+      "epoch": 0.18206645425580337,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.001992717341829768,
+      "loss": 1.785,
+      "step": 28000
+    },
+    {
+      "epoch": 0.1853176409389427,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0019925872943624425,
+      "loss": 1.7763,
+      "step": 28500
+    },
+    {
+      "epoch": 0.18856882762208205,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0019924572468951168,
+      "loss": 1.8051,
+      "step": 29000
+    },
+    {
+      "epoch": 0.19182001430522141,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.001992327199427791,
+      "loss": 1.7849,
+      "step": 29500
+    },
+    {
+      "epoch": 0.19507120098836075,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0019921971519604657,
+      "loss": 1.7797,
+      "step": 30000
+    },
+    {
+      "epoch": 0.1983223876715001,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00199206710449314,
+      "loss": 1.7877,
+      "step": 30500
+    },
+    {
+      "epoch": 0.20157357435463943,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0019919370570258142,
+      "loss": 1.78,
+      "step": 31000
+    },
+    {
+      "epoch": 0.2048247610377788,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.001991807009558489,
+      "loss": 1.7848,
+      "step": 31500
+    },
+    {
+      "epoch": 0.20807594772091814,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.001991676962091163,
+      "loss": 1.7725,
+      "step": 32000
+    },
+    {
+      "epoch": 0.21132713440405748,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0019915469146238375,
+      "loss": 1.7705,
+      "step": 32500
+    },
+    {
+      "epoch": 0.21457832108719682,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.001991416867156512,
+      "loss": 1.7645,
+      "step": 33000
+    },
+    {
+      "epoch": 0.21782950777033616,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.001991286819689187,
+      "loss": 1.7673,
+      "step": 33500
+    },
+    {
+      "epoch": 0.22108069445347553,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.001991156772221861,
+      "loss": 1.7758,
+      "step": 34000
+    },
+    {
+      "epoch": 0.22433188113661487,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.0019910267247545354,
+      "loss": 1.7593,
+      "step": 34500
+    },
+    {
+      "epoch": 0.2275830678197542,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.00199089667728721,
+      "loss": 1.756,
+      "step": 35000
+    },
+    {
+      "epoch": 0.23083425450289355,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0019907666298198843,
+      "loss": 1.7467,
+      "step": 35500
+    },
+    {
+      "epoch": 0.2340854411860329,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.0019906365823525586,
+      "loss": 1.7501,
+      "step": 36000
+    },
+    {
+      "epoch": 0.23733662786917226,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0019905065348852333,
+      "loss": 1.7459,
+      "step": 36500
+    },
+    {
+      "epoch": 0.2405878145523116,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0019903764874179075,
+      "loss": 1.7398,
+      "step": 37000
+    },
+    {
+      "epoch": 0.24383900123545094,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.001990246439950582,
+      "loss": 1.7416,
+      "step": 37500
+    },
+    {
+      "epoch": 0.24709018791859028,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0019901163924832565,
+      "loss": 1.7476,
+      "step": 38000
+    },
+    {
+      "epoch": 0.2503413746017296,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0019899863450159308,
+      "loss": 1.7383,
+      "step": 38500
+    },
+    {
+      "epoch": 0.25359256128486896,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001989856297548605,
+      "loss": 1.7379,
+      "step": 39000
+    },
+    {
+      "epoch": 0.2568437479680083,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0019897262500812797,
+      "loss": 1.7357,
+      "step": 39500
+    },
+    {
+      "epoch": 0.2600949346511477,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0019895962026139544,
+      "loss": 1.7301,
+      "step": 40000
+    },
+    {
+      "epoch": 0.26334612133428703,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0019894661551466287,
+      "loss": 1.7312,
+      "step": 40500
+    },
+    {
+      "epoch": 0.26659730801742637,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.001989336107679303,
+      "loss": 1.7451,
+      "step": 41000
+    },
+    {
+      "epoch": 0.2698484947005657,
+      "grad_norm": 4.5,
+      "learning_rate": 0.0019892060602119776,
+      "loss": 1.7367,
+      "step": 41500
+    },
+    {
+      "epoch": 0.27309968138370505,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.001989076012744652,
+      "loss": 1.7231,
+      "step": 42000
+    },
+    {
+      "epoch": 0.2763508680668444,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.001988945965277326,
+      "loss": 1.7225,
+      "step": 42500
+    },
+    {
+      "epoch": 0.27960205474998373,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.001988815917810001,
+      "loss": 1.7198,
+      "step": 43000
+    },
+    {
+      "epoch": 0.28285324143312307,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.001988685870342675,
+      "loss": 1.7255,
+      "step": 43500
+    },
+    {
+      "epoch": 0.2861044281162624,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0019885558228753494,
+      "loss": 1.7231,
+      "step": 44000
+    },
+    {
+      "epoch": 0.2893556147994018,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.001988425775408024,
+      "loss": 1.7365,
+      "step": 44500
+    },
+    {
+      "epoch": 0.29260680148254115,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019882957279406983,
+      "loss": 1.7237,
+      "step": 45000
+    },
+    {
+      "epoch": 0.2958579881656805,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0019881656804733726,
+      "loss": 1.7336,
+      "step": 45500
+    },
+    {
+      "epoch": 0.2991091748488198,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0019880356330060473,
+      "loss": 1.7488,
+      "step": 46000
+    },
+    {
+      "epoch": 0.30236036153195917,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0019879055855387216,
+      "loss": 1.7362,
+      "step": 46500
+    },
+    {
+      "epoch": 0.3056115482150985,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001987775538071396,
+      "loss": 1.7358,
+      "step": 47000
+    },
+    {
+      "epoch": 0.30886273489823785,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0019876454906040705,
+      "loss": 1.7274,
+      "step": 47500
+    },
+    {
+      "epoch": 0.3121139215813772,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.001987515443136745,
+      "loss": 1.7265,
+      "step": 48000
+    },
+    {
+      "epoch": 0.3153651082645165,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0019873853956694195,
+      "loss": 1.7121,
+      "step": 48500
+    },
+    {
+      "epoch": 0.3186162949476559,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0019872553482020937,
+      "loss": 1.7082,
+      "step": 49000
+    },
+    {
+      "epoch": 0.32186748163079526,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0019871253007347684,
+      "loss": 1.7064,
+      "step": 49500
+    },
+    {
+      "epoch": 0.3251186683139346,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0019869952532674427,
+      "loss": 1.7093,
+      "step": 50000
+    },
+    {
+      "epoch": 0.32836985499707394,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.001986865205800117,
+      "loss": 1.7092,
+      "step": 50500
+    },
+    {
+      "epoch": 0.3316210416802133,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0019867351583327916,
+      "loss": 1.7031,
+      "step": 51000
+    },
+    {
+      "epoch": 0.3348722283633526,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001986605110865466,
+      "loss": 1.7065,
+      "step": 51500
+    },
+    {
+      "epoch": 0.33812341504649196,
+      "grad_norm": 5.84375,
+      "learning_rate": 0.00198647506339814,
+      "loss": 1.7007,
+      "step": 52000
+    },
+    {
+      "epoch": 0.3413746017296313,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001986345015930815,
+      "loss": 1.7033,
+      "step": 52500
+    },
+    {
+      "epoch": 0.34462578841277064,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001986214968463489,
+      "loss": 1.6962,
+      "step": 53000
+    },
+    {
+      "epoch": 0.34787697509591003,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0019860849209961634,
+      "loss": 1.6886,
+      "step": 53500
+    },
+    {
+      "epoch": 0.3511281617790494,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.001985954873528838,
+      "loss": 1.6864,
+      "step": 54000
+    },
+    {
+      "epoch": 0.3543793484621887,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0019858248260615128,
+      "loss": 1.6905,
+      "step": 54500
+    },
+    {
+      "epoch": 0.35763053514532805,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.001985694778594187,
+      "loss": 1.696,
+      "step": 55000
+    },
+    {
+      "epoch": 0.3608817218284674,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0019855647311268613,
+      "loss": 1.6931,
+      "step": 55500
+    },
+    {
+      "epoch": 0.36413290851160673,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.001985434683659536,
+      "loss": 1.6919,
+      "step": 56000
+    },
+    {
+      "epoch": 0.3673840951947461,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.0019853046361922103,
+      "loss": 1.7009,
+      "step": 56500
+    },
+    {
+      "epoch": 0.3706352818778854,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0019851745887248845,
+      "loss": 1.7066,
+      "step": 57000
+    },
+    {
+      "epoch": 0.37388646856102475,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.001985044541257559,
+      "loss": 1.6974,
+      "step": 57500
+    },
+    {
+      "epoch": 0.3771376552441641,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0019849144937902335,
+      "loss": 1.695,
+      "step": 58000
+    },
+    {
+      "epoch": 0.3803888419273035,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.0019847844463229077,
+      "loss": 1.7002,
+      "step": 58500
+    },
+    {
+      "epoch": 0.38364002861044283,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0019846543988555824,
+      "loss": 1.6998,
+      "step": 59000
+    },
+    {
+      "epoch": 0.38689121529358217,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0019845243513882567,
+      "loss": 1.7108,
+      "step": 59500
+    },
+    {
+      "epoch": 0.3901424019767215,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001984394303920931,
+      "loss": 1.6777,
+      "step": 60000
+    },
+    {
+      "epoch": 0.39339358865986085,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0019842642564536057,
+      "loss": 1.6866,
+      "step": 60500
+    },
+    {
+      "epoch": 0.3966447753430002,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.00198413420898628,
+      "loss": 1.6838,
+      "step": 61000
+    },
+    {
+      "epoch": 0.39989596202613953,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.001984004161518954,
+      "loss": 1.6834,
+      "step": 61500
+    },
+    {
+      "epoch": 0.40314714870927887,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.001983874114051629,
+      "loss": 1.6837,
+      "step": 62000
+    },
+    {
+      "epoch": 0.4063983353924182,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0019837440665843036,
+      "loss": 1.6869,
+      "step": 62500
+    },
+    {
+      "epoch": 0.4096495220755576,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.001983614019116978,
+      "loss": 1.7305,
+      "step": 63000
+    },
+    {
+      "epoch": 0.41290070875869694,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.001983483971649652,
+      "loss": 1.746,
+      "step": 63500
+    },
+    {
+      "epoch": 0.4161518954418363,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.001983353924182327,
+      "loss": 1.7387,
+      "step": 64000
+    },
+    {
+      "epoch": 0.4194030821249756,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.001983223876715001,
+      "loss": 1.7394,
+      "step": 64500
+    },
+    {
+      "epoch": 0.42265426880811496,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0019830938292476753,
+      "loss": 1.7914,
+      "step": 65000
+    },
+    {
+      "epoch": 0.4259054554912543,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00198296378178035,
+      "loss": 1.7304,
+      "step": 65500
+    },
+    {
+      "epoch": 0.42915664217439364,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019828337343130243,
+      "loss": 1.7173,
+      "step": 66000
+    },
+    {
+      "epoch": 0.432407828857533,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0019827036868456985,
+      "loss": 1.7409,
+      "step": 66500
+    },
+    {
+      "epoch": 0.4356590155406723,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0019825736393783732,
+      "loss": 1.7136,
+      "step": 67000
+    },
+    {
+      "epoch": 0.4389102022238117,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0019824435919110475,
+      "loss": 1.7297,
+      "step": 67500
+    },
+    {
+      "epoch": 0.44216138890695106,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0019823135444437217,
+      "loss": 1.7409,
+      "step": 68000
+    },
+    {
+      "epoch": 0.4454125755900904,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.0019821834969763964,
+      "loss": 1.7818,
+      "step": 68500
+    },
+    {
+      "epoch": 0.44866376227322974,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001982053449509071,
+      "loss": 1.7189,
+      "step": 69000
+    },
+    {
+      "epoch": 0.4519149489563691,
+      "grad_norm": 6.21875,
+      "learning_rate": 0.0019819234020417454,
+      "loss": 1.7128,
+      "step": 69500
+    },
+    {
+      "epoch": 0.4551661356395084,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0019817933545744197,
+      "loss": 1.6918,
+      "step": 70000
+    },
+    {
+      "epoch": 0.45841732232264776,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0019816633071070944,
+      "loss": 1.6941,
+      "step": 70500
+    },
+    {
+      "epoch": 0.4616685090057871,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0019815332596397686,
+      "loss": 1.6888,
+      "step": 71000
+    },
+    {
+      "epoch": 0.46491969568892644,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.001981403212172443,
+      "loss": 1.689,
+      "step": 71500
+    },
+    {
+      "epoch": 0.4681708823720658,
+      "grad_norm": 4.75,
+      "learning_rate": 0.0019812731647051176,
+      "loss": 1.69,
+      "step": 72000
+    },
+    {
+      "epoch": 0.4714220690552052,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.001981143117237792,
+      "loss": 1.6951,
+      "step": 72500
+    },
+    {
+      "epoch": 0.4746732557383445,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.001981013069770466,
+      "loss": 1.6849,
+      "step": 73000
+    },
+    {
+      "epoch": 0.47792444242148385,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001980883022303141,
+      "loss": 1.7031,
+      "step": 73500
+    },
+    {
+      "epoch": 0.4811756291046232,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001980752974835815,
+      "loss": 1.6962,
+      "step": 74000
+    },
+    {
+      "epoch": 0.48442681578776253,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0019806229273684893,
+      "loss": 1.7377,
+      "step": 74500
+    },
+    {
+      "epoch": 0.4876780024709019,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.001980492879901164,
+      "loss": 1.7273,
+      "step": 75000
+    },
+    {
+      "epoch": 0.4909291891540412,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0019803628324338383,
+      "loss": 1.703,
+      "step": 75500
+    },
+    {
+      "epoch": 0.49418037583718055,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0019802327849665125,
+      "loss": 1.7018,
+      "step": 76000
+    },
+    {
+      "epoch": 0.4974315625203199,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.0019801027374991872,
+      "loss": 1.7036,
+      "step": 76500
+    },
+    {
+      "epoch": 0.5006827492034592,
+      "grad_norm": 1.25,
+      "learning_rate": 0.001979972690031862,
+      "loss": 1.7149,
+      "step": 77000
+    },
+    {
+      "epoch": 0.5039339358865986,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.001979842642564536,
+      "loss": 1.7483,
+      "step": 77500
+    },
+    {
+      "epoch": 0.5071851225697379,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.0019797125950972105,
+      "loss": 1.731,
+      "step": 78000
+    },
+    {
+      "epoch": 0.5104363092528773,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.001979582547629885,
+      "loss": 1.7228,
+      "step": 78500
+    },
+    {
+      "epoch": 0.5136874959360166,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0019794525001625594,
+      "loss": 1.7164,
+      "step": 79000
+    },
+    {
+      "epoch": 0.516938682619156,
+      "grad_norm": 6.75,
+      "learning_rate": 0.0019793224526952337,
+      "loss": 1.7132,
+      "step": 79500
+    },
+    {
+      "epoch": 0.5201898693022954,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0019791924052279084,
+      "loss": 1.7004,
+      "step": 80000
+    },
+    {
+      "epoch": 0.5234410559854347,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0019790623577605826,
+      "loss": 1.7019,
+      "step": 80500
+    },
+    {
+      "epoch": 0.5266922426685741,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001978932310293257,
+      "loss": 1.7002,
+      "step": 81000
+    },
+    {
+      "epoch": 0.5299434293517133,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0019788022628259316,
+      "loss": 1.6903,
+      "step": 81500
+    },
+    {
+      "epoch": 0.5331946160348527,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.001978672215358606,
+      "loss": 1.6837,
+      "step": 82000
+    },
+    {
+      "epoch": 0.536445802717992,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00197854216789128,
+      "loss": 1.6843,
+      "step": 82500
+    },
+    {
+      "epoch": 0.5396969894011314,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.001978412120423955,
+      "loss": 1.6899,
+      "step": 83000
+    },
+    {
+      "epoch": 0.5429481760842707,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0019782820729566295,
+      "loss": 1.685,
+      "step": 83500
+    },
+    {
+      "epoch": 0.5461993627674101,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0019781520254893038,
+      "loss": 1.6823,
+      "step": 84000
+    },
+    {
+      "epoch": 0.5494505494505495,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.001978021978021978,
+      "loss": 1.6779,
+      "step": 84500
+    },
+    {
+      "epoch": 0.5527017361336888,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0019778919305546527,
+      "loss": 1.6857,
+      "step": 85000
+    },
+    {
+      "epoch": 0.5559529228168282,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001977761883087327,
+      "loss": 1.6778,
+      "step": 85500
+    },
+    {
+      "epoch": 0.5592041094999675,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.0019776318356200012,
+      "loss": 1.6724,
+      "step": 86000
+    },
+    {
+      "epoch": 0.5624552961831069,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.001977501788152676,
+      "loss": 1.6823,
+      "step": 86500
+    },
+    {
+      "epoch": 0.5657064828662461,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.00197737174068535,
+      "loss": 1.6812,
+      "step": 87000
+    },
+    {
+      "epoch": 0.5689576695493855,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0019772416932180245,
+      "loss": 1.6774,
+      "step": 87500
+    },
+    {
+      "epoch": 0.5722088562325248,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.001977111645750699,
+      "loss": 1.6614,
+      "step": 88000
+    },
+    {
+      "epoch": 0.5754600429156642,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0019769815982833734,
+      "loss": 1.657,
+      "step": 88500
+    },
+    {
+      "epoch": 0.5787112295988036,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0019768515508160477,
+      "loss": 1.6616,
+      "step": 89000
+    },
+    {
+      "epoch": 0.5819624162819429,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.0019767215033487224,
+      "loss": 1.6561,
+      "step": 89500
+    },
+    {
+      "epoch": 0.5852136029650823,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0019765914558813966,
+      "loss": 1.6612,
+      "step": 90000
+    },
+    {
+      "epoch": 0.5884647896482216,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.001976461408414071,
+      "loss": 1.6658,
+      "step": 90500
+    },
+    {
+      "epoch": 0.591715976331361,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0019763313609467456,
+      "loss": 1.6678,
+      "step": 91000
+    },
+    {
+      "epoch": 0.5949671630145003,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.0019762013134794203,
+      "loss": 1.6657,
+      "step": 91500
+    },
+    {
+      "epoch": 0.5982183496976397,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019760712660120945,
+      "loss": 1.6659,
+      "step": 92000
+    },
+    {
+      "epoch": 0.6014695363807789,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.001975941218544769,
+      "loss": 1.6694,
+      "step": 92500
+    },
+    {
+      "epoch": 0.6047207230639183,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.0019758111710774435,
+      "loss": 1.6644,
+      "step": 93000
+    },
+    {
+      "epoch": 0.6079719097470577,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0019756811236101178,
+      "loss": 1.6526,
+      "step": 93500
+    },
+    {
+      "epoch": 0.611223096430197,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.001975551076142792,
+      "loss": 1.6593,
+      "step": 94000
+    },
+    {
+      "epoch": 0.6144742831133364,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0019754210286754667,
+      "loss": 1.6524,
+      "step": 94500
+    },
+    {
+      "epoch": 0.6177254697964757,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.001975290981208141,
+      "loss": 1.6447,
+      "step": 95000
+    },
+    {
+      "epoch": 0.6209766564796151,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0019751609337408152,
+      "loss": 1.6478,
+      "step": 95500
+    },
+    {
+      "epoch": 0.6242278431627544,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00197503088627349,
+      "loss": 1.6488,
+      "step": 96000
+    },
+    {
+      "epoch": 0.6274790298458938,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001974900838806164,
+      "loss": 1.6573,
+      "step": 96500
+    },
+    {
+      "epoch": 0.630730216529033,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.0019747707913388385,
+      "loss": 1.6593,
+      "step": 97000
+    },
+    {
+      "epoch": 0.6339814032121724,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.001974640743871513,
+      "loss": 1.6694,
+      "step": 97500
+    },
+    {
+      "epoch": 0.6372325898953118,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.001974510696404188,
+      "loss": 1.6712,
+      "step": 98000
+    },
+    {
+      "epoch": 0.6404837765784511,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.001974380648936862,
+      "loss": 1.6678,
+      "step": 98500
+    },
+    {
+      "epoch": 0.6437349632615905,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0019742506014695364,
+      "loss": 1.662,
+      "step": 99000
+    },
+    {
+      "epoch": 0.6469861499447298,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.001974120554002211,
+      "loss": 1.6648,
+      "step": 99500
+    },
+    {
+      "epoch": 0.6502373366278692,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0019739905065348853,
+      "loss": 1.6518,
+      "step": 100000
+    },
+    {
+      "epoch": 0.6534885233110085,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.0019738604590675596,
+      "loss": 1.6528,
+      "step": 100500
+    },
+    {
+      "epoch": 0.6567397099941479,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0019737304116002343,
+      "loss": 1.6754,
+      "step": 101000
+    },
+    {
+      "epoch": 0.6599908966772872,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.0019736003641329086,
+      "loss": 1.68,
+      "step": 101500
+    },
+    {
+      "epoch": 0.6632420833604266,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.001973470316665583,
+      "loss": 1.6636,
+      "step": 102000
+    },
+    {
+      "epoch": 0.666493270043566,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0019733402691982575,
+      "loss": 1.6543,
+      "step": 102500
+    },
+    {
+      "epoch": 0.6697444567267052,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.0019732102217309318,
+      "loss": 1.6503,
+      "step": 103000
+    },
+    {
+      "epoch": 0.6729956434098446,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.001973080174263606,
+      "loss": 1.6522,
+      "step": 103500
+    },
+    {
+      "epoch": 0.6762468300929839,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0019729501267962807,
+      "loss": 1.6545,
+      "step": 104000
+    },
+    {
+      "epoch": 0.6794980167761233,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.001972820079328955,
+      "loss": 1.651,
+      "step": 104500
+    },
+    {
+      "epoch": 0.6827492034592626,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0019726900318616293,
+      "loss": 1.6601,
+      "step": 105000
+    },
+    {
+      "epoch": 0.686000390142402,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001972559984394304,
+      "loss": 1.6394,
+      "step": 105500
+    },
+    {
+      "epoch": 0.6892515768255413,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0019724299369269786,
+      "loss": 1.639,
+      "step": 106000
+    },
+    {
+      "epoch": 0.6925027635086807,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001972299889459653,
+      "loss": 1.629,
+      "step": 106500
+    },
+    {
+      "epoch": 0.6957539501918201,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.001972169841992327,
+      "loss": 1.645,
+      "step": 107000
+    },
+    {
+      "epoch": 0.6990051368749594,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.001972039794525002,
+      "loss": 1.6692,
+      "step": 107500
+    },
+    {
+      "epoch": 0.7022563235580987,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.001971909747057676,
+      "loss": 1.6579,
+      "step": 108000
+    },
+    {
+      "epoch": 0.705507510241238,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0019717796995903504,
+      "loss": 1.6385,
+      "step": 108500
+    },
+    {
+      "epoch": 0.7087586969243774,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001971649652123025,
+      "loss": 1.648,
+      "step": 109000
+    },
+    {
+      "epoch": 0.7120098836075167,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.0019715196046556993,
+      "loss": 1.6436,
+      "step": 109500
+    },
+    {
+      "epoch": 0.7152610702906561,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0019713895571883736,
+      "loss": 1.6362,
+      "step": 110000
+    },
+    {
+      "epoch": 0.7185122569737954,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0019712595097210483,
+      "loss": 1.645,
+      "step": 110500
+    },
+    {
+      "epoch": 0.7217634436569348,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0019711294622537226,
+      "loss": 1.642,
+      "step": 111000
+    },
+    {
+      "epoch": 0.7250146303400741,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001970999414786397,
+      "loss": 1.6355,
+      "step": 111500
+    },
+    {
+      "epoch": 0.7282658170232135,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0019708693673190715,
+      "loss": 1.6339,
+      "step": 112000
+    },
+    {
+      "epoch": 0.7315170037063529,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.001970739319851746,
+      "loss": 1.6294,
+      "step": 112500
+    },
+    {
+      "epoch": 0.7347681903894921,
+      "grad_norm": 3.84375,
+      "learning_rate": 0.0019706092723844205,
+      "loss": 1.6307,
+      "step": 113000
+    },
+    {
+      "epoch": 0.7380193770726315,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0019704792249170947,
+      "loss": 1.6373,
+      "step": 113500
+    },
+    {
+      "epoch": 0.7412705637557708,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0019703491774497694,
+      "loss": 1.6368,
+      "step": 114000
+    },
+    {
+      "epoch": 0.7445217504389102,
+      "grad_norm": 42.75,
+      "learning_rate": 0.0019702191299824437,
+      "loss": 1.6278,
+      "step": 114500
+    },
+    {
+      "epoch": 0.7477729371220495,
+      "grad_norm": 3.53125,
+      "learning_rate": 0.001970089082515118,
+      "loss": 1.6316,
+      "step": 115000
+    },
+    {
+      "epoch": 0.7510241238051889,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.0019699590350477927,
+      "loss": 1.6202,
+      "step": 115500
+    },
+    {
+      "epoch": 0.7542753104883282,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.001969828987580467,
+      "loss": 1.6194,
+      "step": 116000
+    },
+    {
+      "epoch": 0.7575264971714676,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.001969698940113141,
+      "loss": 1.6219,
+      "step": 116500
+    },
+    {
+      "epoch": 0.760777683854607,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.001969568892645816,
+      "loss": 1.6229,
+      "step": 117000
+    },
+    {
+      "epoch": 0.7640288705377463,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00196943884517849,
+      "loss": 1.6253,
+      "step": 117500
+    },
+    {
+      "epoch": 0.7672800572208857,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0019693087977111644,
+      "loss": 1.6203,
+      "step": 118000
+    },
+    {
+      "epoch": 0.7705312439040249,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.001969178750243839,
+      "loss": 1.6222,
+      "step": 118500
+    },
+    {
+      "epoch": 0.7737824305871643,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0019690487027765134,
+      "loss": 1.6241,
+      "step": 119000
+    },
+    {
+      "epoch": 0.7770336172703036,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0019689186553091876,
+      "loss": 1.6173,
+      "step": 119500
+    },
+    {
+      "epoch": 0.780284803953443,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0019687886078418623,
+      "loss": 1.6136,
+      "step": 120000
+    },
+    {
+      "epoch": 0.7835359906365823,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001968658560374537,
+      "loss": 1.6079,
+      "step": 120500
+    },
+    {
+      "epoch": 0.7867871773197217,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0019685285129072113,
+      "loss": 1.6035,
+      "step": 121000
+    },
+    {
+      "epoch": 0.7900383640028611,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0019683984654398855,
+      "loss": 1.6098,
+      "step": 121500
+    },
+    {
+      "epoch": 0.7932895506860004,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0019682684179725602,
+      "loss": 1.6112,
+      "step": 122000
+    },
+    {
+      "epoch": 0.7965407373691398,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0019681383705052345,
+      "loss": 1.6075,
+      "step": 122500
+    },
+    {
+      "epoch": 0.7997919240522791,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0019680083230379087,
+      "loss": 1.6124,
+      "step": 123000
+    },
+    {
+      "epoch": 0.8030431107354185,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0019678782755705834,
+      "loss": 1.6121,
+      "step": 123500
+    },
+    {
+      "epoch": 0.8062942974185577,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0019677482281032577,
+      "loss": 1.5979,
+      "step": 124000
+    },
+    {
+      "epoch": 0.8095454841016971,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001967618180635932,
+      "loss": 1.6022,
+      "step": 124500
+    },
+    {
+      "epoch": 0.8127966707848364,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.0019674881331686067,
+      "loss": 1.6035,
+      "step": 125000
+    },
+    {
+      "epoch": 0.8160478574679758,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.001967358085701281,
+      "loss": 1.5999,
+      "step": 125500
+    },
+    {
+      "epoch": 0.8192990441511152,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001967228038233955,
+      "loss": 1.6004,
+      "step": 126000
+    },
+    {
+      "epoch": 0.8225502308342545,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00196709799076663,
+      "loss": 1.5918,
+      "step": 126500
+    },
+    {
+      "epoch": 0.8258014175173939,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0019669679432993046,
+      "loss": 1.6031,
+      "step": 127000
+    },
+    {
+      "epoch": 0.8290526042005332,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.001966837895831979,
+      "loss": 1.6028,
+      "step": 127500
+    },
+    {
+      "epoch": 0.8323037908836726,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.001966707848364653,
+      "loss": 1.5984,
+      "step": 128000
+    },
+    {
+      "epoch": 0.8355549775668119,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.001966577800897328,
+      "loss": 1.5988,
+      "step": 128500
+    },
+    {
+      "epoch": 0.8388061642499512,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.001966447753430002,
+      "loss": 1.6013,
+      "step": 129000
+    },
+    {
+      "epoch": 0.8420573509330905,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0019663177059626763,
+      "loss": 1.6111,
+      "step": 129500
+    },
+    {
+      "epoch": 0.8453085376162299,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.001966187658495351,
+      "loss": 1.6014,
+      "step": 130000
+    },
+    {
+      "epoch": 0.8485597242993693,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.0019660576110280253,
+      "loss": 1.5989,
+      "step": 130500
+    },
+    {
+      "epoch": 0.8518109109825086,
+      "grad_norm": 5.5,
+      "learning_rate": 0.0019659275635606995,
+      "loss": 1.5957,
+      "step": 131000
+    },
+    {
+      "epoch": 0.855062097665648,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.0019657975160933742,
+      "loss": 1.5993,
+      "step": 131500
+    },
+    {
+      "epoch": 0.8583132843487873,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0019656674686260485,
+      "loss": 1.6048,
+      "step": 132000
+    },
+    {
+      "epoch": 0.8615644710319267,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0019655374211587228,
+      "loss": 1.6039,
+      "step": 132500
+    },
+    {
+      "epoch": 0.864815657715066,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0019654073736913974,
+      "loss": 1.6012,
+      "step": 133000
+    },
+    {
+      "epoch": 0.8680668443982054,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0019652773262240717,
+      "loss": 1.5923,
+      "step": 133500
+    },
+    {
+      "epoch": 0.8713180310813446,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.001965147278756746,
+      "loss": 1.585,
+      "step": 134000
+    },
+    {
+      "epoch": 0.874569217764484,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0019650172312894207,
+      "loss": 1.5838,
+      "step": 134500
+    },
+    {
+      "epoch": 0.8778204044476234,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0019648871838220954,
+      "loss": 1.5818,
+      "step": 135000
+    },
+    {
+      "epoch": 0.8810715911307627,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0019647571363547696,
+      "loss": 1.5889,
+      "step": 135500
+    },
+    {
+      "epoch": 0.8843227778139021,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.001964627088887444,
+      "loss": 1.587,
+      "step": 136000
+    },
+    {
+      "epoch": 0.8875739644970414,
+      "grad_norm": 5.25,
+      "learning_rate": 0.0019644970414201186,
+      "loss": 1.5835,
+      "step": 136500
+    },
+    {
+      "epoch": 0.8908251511801808,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.001964366993952793,
+      "loss": 1.5802,
+      "step": 137000
+    },
+    {
+      "epoch": 0.8940763378633201,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.001964236946485467,
+      "loss": 1.5836,
+      "step": 137500
+    },
+    {
+      "epoch": 0.8973275245464595,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001964106899018142,
+      "loss": 1.5861,
+      "step": 138000
+    },
+    {
+      "epoch": 0.9005787112295988,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.001963976851550816,
+      "loss": 1.5777,
+      "step": 138500
+    },
+    {
+      "epoch": 0.9038298979127382,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019638468040834903,
+      "loss": 1.5759,
+      "step": 139000
+    },
+    {
+      "epoch": 0.9070810845958774,
+      "grad_norm": 2.625,
+      "learning_rate": 0.001963716756616165,
+      "loss": 1.5863,
+      "step": 139500
+    },
+    {
+      "epoch": 0.9103322712790168,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0019635867091488393,
+      "loss": 1.5756,
+      "step": 140000
+    },
+    {
+      "epoch": 0.9135834579621562,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0019634566616815135,
+      "loss": 1.5774,
+      "step": 140500
+    },
+    {
+      "epoch": 0.9168346446452955,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.0019633266142141882,
+      "loss": 1.5707,
+      "step": 141000
+    },
+    {
+      "epoch": 0.9200858313284349,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001963196566746863,
+      "loss": 1.5849,
+      "step": 141500
+    },
+    {
+      "epoch": 0.9233370180115742,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.001963066519279537,
+      "loss": 1.5774,
+      "step": 142000
+    },
+    {
+      "epoch": 0.9265882046947136,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0019629364718122115,
+      "loss": 1.5774,
+      "step": 142500
+    },
+    {
+      "epoch": 0.9298393913778529,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.001962806424344886,
+      "loss": 1.5733,
+      "step": 143000
+    },
+    {
+      "epoch": 0.9330905780609923,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0019626763768775604,
+      "loss": 1.5819,
+      "step": 143500
+    },
+    {
+      "epoch": 0.9363417647441316,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0019625463294102347,
+      "loss": 1.5717,
+      "step": 144000
+    },
+    {
+      "epoch": 0.939592951427271,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0019624162819429094,
+      "loss": 1.5737,
+      "step": 144500
+    },
+    {
+      "epoch": 0.9428441381104103,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.0019622862344755836,
+      "loss": 1.577,
+      "step": 145000
+    },
+    {
+      "epoch": 0.9460953247935496,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.001962156187008258,
+      "loss": 1.5795,
+      "step": 145500
+    },
+    {
+      "epoch": 0.949346511476689,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0019620261395409326,
+      "loss": 1.5723,
+      "step": 146000
+    },
+    {
+      "epoch": 0.9525976981598283,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.001961896092073607,
+      "loss": 1.5775,
+      "step": 146500
+    },
+    {
+      "epoch": 0.9558488848429677,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.001961766044606281,
+      "loss": 1.5767,
+      "step": 147000
+    },
+    {
+      "epoch": 0.959100071526107,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.001961635997138956,
+      "loss": 1.5762,
+      "step": 147500
+    },
+    {
+      "epoch": 0.9623512582092464,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00196150594967163,
+      "loss": 1.5697,
+      "step": 148000
+    },
+    {
+      "epoch": 0.9656024448923857,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0019613759022043043,
+      "loss": 1.5608,
+      "step": 148500
+    },
+    {
+      "epoch": 0.9688536315755251,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.001961245854736979,
+      "loss": 1.5679,
+      "step": 149000
+    },
+    {
+      "epoch": 0.9721048182586645,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0019611158072696537,
+      "loss": 1.5627,
+      "step": 149500
+    },
+    {
+      "epoch": 0.9753560049418037,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.001960985759802328,
+      "loss": 1.5678,
+      "step": 150000
+    },
+    {
+      "epoch": 0.9786071916249431,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0019608557123350022,
+      "loss": 1.5679,
+      "step": 150500
+    },
+    {
+      "epoch": 0.9818583783080824,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.001960725664867677,
+      "loss": 1.5662,
+      "step": 151000
+    },
+    {
+      "epoch": 0.9851095649912218,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.001960595617400351,
+      "loss": 1.5687,
+      "step": 151500
+    },
+    {
+      "epoch": 0.9883607516743611,
+      "grad_norm": 3.328125,
+      "learning_rate": 0.0019604655699330255,
+      "loss": 1.5623,
+      "step": 152000
+    },
+    {
+      "epoch": 0.9916119383575005,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0019603355224657,
+      "loss": 1.5668,
+      "step": 152500
+    },
+    {
+      "epoch": 0.9948631250406398,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0019602054749983744,
+      "loss": 1.5668,
+      "step": 153000
+    },
+    {
+      "epoch": 0.9981143117237792,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0019600754275310487,
+      "loss": 1.5688,
+      "step": 153500
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.5614691972732544,
+      "eval_runtime": 0.5382,
+      "eval_samples_per_second": 1857.967,
+      "eval_steps_per_second": 29.727,
+      "step": 153790
+    },
+    {
+      "epoch": 1.0013654984069185,
+      "grad_norm": 3.140625,
+      "learning_rate": 0.0019599453800637234,
+      "loss": 1.5728,
+      "step": 154000
+    },
+    {
+      "epoch": 1.004616685090058,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0019598153325963976,
+      "loss": 1.5747,
+      "step": 154500
+    },
+    {
+      "epoch": 1.0078678717731973,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.001959685285129072,
+      "loss": 1.572,
+      "step": 155000
+    },
+    {
+      "epoch": 1.0111190584563365,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0019595552376617466,
+      "loss": 1.5646,
+      "step": 155500
+    },
+    {
+      "epoch": 1.0143702451394758,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.0019594251901944213,
+      "loss": 1.5695,
+      "step": 156000
+    },
+    {
+      "epoch": 1.0176214318226153,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0019592951427270956,
+      "loss": 1.5595,
+      "step": 156500
+    },
+    {
+      "epoch": 1.0208726185057546,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00195916509525977,
+      "loss": 1.5574,
+      "step": 157000
+    },
+    {
+      "epoch": 1.024123805188894,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0019590350477924445,
+      "loss": 1.5624,
+      "step": 157500
+    },
+    {
+      "epoch": 1.0273749918720332,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0019589050003251188,
+      "loss": 1.5611,
+      "step": 158000
+    },
+    {
+      "epoch": 1.0306261785551727,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.001958774952857793,
+      "loss": 1.5626,
+      "step": 158500
+    },
+    {
+      "epoch": 1.033877365238312,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0019586449053904677,
+      "loss": 1.5657,
+      "step": 159000
+    },
+    {
+      "epoch": 1.0371285519214513,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.001958514857923142,
+      "loss": 1.5622,
+      "step": 159500
+    },
+    {
+      "epoch": 1.0403797386045908,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0019583848104558163,
+      "loss": 1.5667,
+      "step": 160000
+    },
+    {
+      "epoch": 1.04363092528773,
+      "grad_norm": 7.78125,
+      "learning_rate": 0.001958254762988491,
+      "loss": 1.5694,
+      "step": 160500
+    },
+    {
+      "epoch": 1.0468821119708693,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.001958124715521165,
+      "loss": 1.5726,
+      "step": 161000
+    },
+    {
+      "epoch": 1.0501332986540086,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0019579946680538395,
+      "loss": 1.5669,
+      "step": 161500
+    },
+    {
+      "epoch": 1.0533844853371481,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.001957864620586514,
+      "loss": 1.567,
+      "step": 162000
+    },
+    {
+      "epoch": 1.0566356720202874,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0019577345731191884,
+      "loss": 1.5696,
+      "step": 162500
+    },
+    {
+      "epoch": 1.0598868587034267,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0019576045256518627,
+      "loss": 1.5617,
+      "step": 163000
+    },
+    {
+      "epoch": 1.0631380453865662,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0019574744781845374,
+      "loss": 1.5626,
+      "step": 163500
+    },
+    {
+      "epoch": 1.0663892320697055,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001957344430717212,
+      "loss": 1.5643,
+      "step": 164000
+    },
+    {
+      "epoch": 1.0696404187528448,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0019572143832498863,
+      "loss": 1.5572,
+      "step": 164500
+    },
+    {
+      "epoch": 1.072891605435984,
+      "grad_norm": 3.140625,
+      "learning_rate": 0.0019570843357825606,
+      "loss": 1.5707,
+      "step": 165000
+    },
+    {
+      "epoch": 1.0761427921191236,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0019569542883152353,
+      "loss": 1.5728,
+      "step": 165500
+    },
+    {
+      "epoch": 1.0793939788022628,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0019568242408479096,
+      "loss": 1.5795,
+      "step": 166000
+    },
+    {
+      "epoch": 1.0826451654854021,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001956694193380584,
+      "loss": 1.5778,
+      "step": 166500
+    },
+    {
+      "epoch": 1.0858963521685414,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0019565641459132585,
+      "loss": 1.5702,
+      "step": 167000
+    },
+    {
+      "epoch": 1.089147538851681,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0019564340984459328,
+      "loss": 1.5654,
+      "step": 167500
+    },
+    {
+      "epoch": 1.0923987255348202,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001956304050978607,
+      "loss": 1.5677,
+      "step": 168000
+    },
+    {
+      "epoch": 1.0956499122179595,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0019561740035112817,
+      "loss": 1.5646,
+      "step": 168500
+    },
+    {
+      "epoch": 1.098901098901099,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.001956043956043956,
+      "loss": 1.583,
+      "step": 169000
+    },
+    {
+      "epoch": 1.1021522855842383,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0019559139085766303,
+      "loss": 1.5675,
+      "step": 169500
+    },
+    {
+      "epoch": 1.1054034722673776,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.001955783861109305,
+      "loss": 1.5986,
+      "step": 170000
+    },
+    {
+      "epoch": 1.1086546589505168,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0019556538136419797,
+      "loss": 1.6581,
+      "step": 170500
+    },
+    {
+      "epoch": 1.1119058456336564,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001955523766174654,
+      "loss": 1.6557,
+      "step": 171000
+    },
+    {
+      "epoch": 1.1151570323167956,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.001955393718707328,
+      "loss": 1.6413,
+      "step": 171500
+    },
+    {
+      "epoch": 1.118408218999935,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.001955263671240003,
+      "loss": 1.6369,
+      "step": 172000
+    },
+    {
+      "epoch": 1.1216594056830744,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001955133623772677,
+      "loss": 1.637,
+      "step": 172500
+    },
+    {
+      "epoch": 1.1249105923662137,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0019550035763053514,
+      "loss": 1.6328,
+      "step": 173000
+    },
+    {
+      "epoch": 1.128161779049353,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.001954873528838026,
+      "loss": 1.6243,
+      "step": 173500
+    },
+    {
+      "epoch": 1.1314129657324923,
+      "grad_norm": 3.828125,
+      "learning_rate": 0.0019547434813707004,
+      "loss": 1.6165,
+      "step": 174000
+    },
+    {
+      "epoch": 1.1346641524156318,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0019546134339033746,
+      "loss": 1.6187,
+      "step": 174500
+    },
+    {
+      "epoch": 1.137915339098771,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0019544833864360493,
+      "loss": 1.6138,
+      "step": 175000
+    },
+    {
+      "epoch": 1.1411665257819104,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0019543533389687236,
+      "loss": 1.6141,
+      "step": 175500
+    },
+    {
+      "epoch": 1.1444177124650499,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.001954223291501398,
+      "loss": 1.6027,
+      "step": 176000
+    },
+    {
+      "epoch": 1.1476688991481891,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0019540932440340725,
+      "loss": 1.6105,
+      "step": 176500
+    },
+    {
+      "epoch": 1.1509200858313284,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001953963196566747,
+      "loss": 1.6119,
+      "step": 177000
+    },
+    {
+      "epoch": 1.1541712725144677,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001953833149099421,
+      "loss": 1.6046,
+      "step": 177500
+    },
+    {
+      "epoch": 1.1574224591976072,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0019537031016320957,
+      "loss": 1.6093,
+      "step": 178000
+    },
+    {
+      "epoch": 1.1606736458807465,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0019535730541647704,
+      "loss": 1.6047,
+      "step": 178500
+    },
+    {
+      "epoch": 1.1639248325638858,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0019534430066974447,
+      "loss": 1.5999,
+      "step": 179000
+    },
+    {
+      "epoch": 1.167176019247025,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.001953312959230119,
+      "loss": 1.5978,
+      "step": 179500
+    },
+    {
+      "epoch": 1.1704272059301646,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0019531829117627937,
+      "loss": 1.5949,
+      "step": 180000
+    },
+    {
+      "epoch": 1.1736783926133039,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.001953052864295468,
+      "loss": 1.5913,
+      "step": 180500
+    },
+    {
+      "epoch": 1.1769295792964432,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0019529228168281424,
+      "loss": 1.5873,
+      "step": 181000
+    },
+    {
+      "epoch": 1.1801807659795824,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0019527927693608167,
+      "loss": 1.5831,
+      "step": 181500
+    },
+    {
+      "epoch": 1.183431952662722,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0019526627218934911,
+      "loss": 1.575,
+      "step": 182000
+    },
+    {
+      "epoch": 1.1866831393458612,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.0019525326744261656,
+      "loss": 1.583,
+      "step": 182500
+    },
+    {
+      "epoch": 1.1899343260290005,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0019524026269588399,
+      "loss": 1.5749,
+      "step": 183000
+    },
+    {
+      "epoch": 1.19318551271214,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0019522725794915144,
+      "loss": 1.5779,
+      "step": 183500
+    },
+    {
+      "epoch": 1.1964366993952793,
+      "grad_norm": 3.5625,
+      "learning_rate": 0.0019521425320241888,
+      "loss": 1.5732,
+      "step": 184000
+    },
+    {
+      "epoch": 1.1996878860784186,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001952012484556863,
+      "loss": 1.5771,
+      "step": 184500
+    },
+    {
+      "epoch": 1.2029390727615579,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0019518824370895378,
+      "loss": 1.58,
+      "step": 185000
+    },
+    {
+      "epoch": 1.2061902594446974,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0019517523896222123,
+      "loss": 1.5675,
+      "step": 185500
+    },
+    {
+      "epoch": 1.2094414461278367,
+      "grad_norm": 4.25,
+      "learning_rate": 0.0019516223421548868,
+      "loss": 1.563,
+      "step": 186000
+    },
+    {
+      "epoch": 1.212692632810976,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.001951492294687561,
+      "loss": 1.5731,
+      "step": 186500
+    },
+    {
+      "epoch": 1.2159438194941155,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0019513622472202355,
+      "loss": 1.569,
+      "step": 187000
+    },
+    {
+      "epoch": 1.2191950061772547,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00195123219975291,
+      "loss": 1.5707,
+      "step": 187500
+    },
+    {
+      "epoch": 1.222446192860394,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.0019511021522855842,
+      "loss": 1.5635,
+      "step": 188000
+    },
+    {
+      "epoch": 1.2256973795435333,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0019509721048182587,
+      "loss": 1.5603,
+      "step": 188500
+    },
+    {
+      "epoch": 1.2289485662266728,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0019508420573509332,
+      "loss": 1.5645,
+      "step": 189000
+    },
+    {
+      "epoch": 1.232199752909812,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0019507120098836074,
+      "loss": 1.5594,
+      "step": 189500
+    },
+    {
+      "epoch": 1.2354509395929514,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.001950581962416282,
+      "loss": 1.5594,
+      "step": 190000
+    },
+    {
+      "epoch": 1.2387021262760909,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.0019504519149489564,
+      "loss": 1.5585,
+      "step": 190500
+    },
+    {
+      "epoch": 1.2419533129592302,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0019503218674816307,
+      "loss": 1.5542,
+      "step": 191000
+    },
+    {
+      "epoch": 1.2452044996423695,
+      "grad_norm": 2.671875,
+      "learning_rate": 0.0019501918200143051,
+      "loss": 1.554,
+      "step": 191500
+    },
+    {
+      "epoch": 1.2484556863255087,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.0019500617725469796,
+      "loss": 1.5572,
+      "step": 192000
+    },
+    {
+      "epoch": 1.251706873008648,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0019499317250796543,
+      "loss": 1.5587,
+      "step": 192500
+    },
+    {
+      "epoch": 1.2549580596917875,
+      "grad_norm": 8.4375,
+      "learning_rate": 0.0019498016776123286,
+      "loss": 1.5515,
+      "step": 193000
+    },
+    {
+      "epoch": 1.2582092463749268,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001949671630145003,
+      "loss": 1.5584,
+      "step": 193500
+    },
+    {
+      "epoch": 1.2614604330580663,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0019495415826776775,
+      "loss": 1.5514,
+      "step": 194000
+    },
+    {
+      "epoch": 1.2647116197412056,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0019494115352103518,
+      "loss": 1.5492,
+      "step": 194500
+    },
+    {
+      "epoch": 1.267962806424345,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0019492814877430263,
+      "loss": 1.5522,
+      "step": 195000
+    },
+    {
+      "epoch": 1.2712139931074842,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0019491514402757008,
+      "loss": 1.5548,
+      "step": 195500
+    },
+    {
+      "epoch": 1.2744651797906235,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.001949021392808375,
+      "loss": 1.55,
+      "step": 196000
+    },
+    {
+      "epoch": 1.277716366473763,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0019488913453410495,
+      "loss": 1.5467,
+      "step": 196500
+    },
+    {
+      "epoch": 1.2809675531569022,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001948761297873724,
+      "loss": 1.5492,
+      "step": 197000
+    },
+    {
+      "epoch": 1.2842187398400415,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0019486312504063982,
+      "loss": 1.5494,
+      "step": 197500
+    },
+    {
+      "epoch": 1.287469926523181,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0019485012029390727,
+      "loss": 1.5533,
+      "step": 198000
+    },
+    {
+      "epoch": 1.2907211132063203,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0019483711554717472,
+      "loss": 1.5472,
+      "step": 198500
+    },
+    {
+      "epoch": 1.2939722998894596,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0019482411080044215,
+      "loss": 1.5491,
+      "step": 199000
+    },
+    {
+      "epoch": 1.297223486572599,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0019481110605370962,
+      "loss": 1.5507,
+      "step": 199500
+    },
+    {
+      "epoch": 1.3004746732557384,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.0019479810130697706,
+      "loss": 1.5485,
+      "step": 200000
+    },
+    {
+      "epoch": 1.3037258599388777,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0019478509656024451,
+      "loss": 1.5496,
+      "step": 200500
+    },
+    {
+      "epoch": 1.306977046622017,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0019477209181351194,
+      "loss": 1.5401,
+      "step": 201000
+    },
+    {
+      "epoch": 1.3102282333051565,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0019475908706677938,
+      "loss": 1.5444,
+      "step": 201500
+    },
+    {
+      "epoch": 1.3134794199882958,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0019474608232004683,
+      "loss": 1.5402,
+      "step": 202000
+    },
+    {
+      "epoch": 1.316730606671435,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.0019473307757331426,
+      "loss": 1.5417,
+      "step": 202500
+    },
+    {
+      "epoch": 1.3199817933545743,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.001947200728265817,
+      "loss": 1.5397,
+      "step": 203000
+    },
+    {
+      "epoch": 1.3232329800377138,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0019470706807984915,
+      "loss": 1.5408,
+      "step": 203500
+    },
+    {
+      "epoch": 1.3264841667208531,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0019469406333311658,
+      "loss": 1.54,
+      "step": 204000
+    },
+    {
+      "epoch": 1.3297353534039924,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.0019468105858638403,
+      "loss": 1.5395,
+      "step": 204500
+    },
+    {
+      "epoch": 1.332986540087132,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.0019466805383965148,
+      "loss": 1.5428,
+      "step": 205000
+    },
+    {
+      "epoch": 1.3362377267702712,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.001946550490929189,
+      "loss": 1.533,
+      "step": 205500
+    },
+    {
+      "epoch": 1.3394889134534105,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0019464204434618635,
+      "loss": 1.5305,
+      "step": 206000
+    },
+    {
+      "epoch": 1.3427401001365498,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.001946290395994538,
+      "loss": 1.5351,
+      "step": 206500
+    },
+    {
+      "epoch": 1.345991286819689,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0019461603485272127,
+      "loss": 1.5327,
+      "step": 207000
+    },
+    {
+      "epoch": 1.3492424735028286,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.001946030301059887,
+      "loss": 1.5305,
+      "step": 207500
+    },
+    {
+      "epoch": 1.3524936601859678,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0019459002535925614,
+      "loss": 1.5374,
+      "step": 208000
+    },
+    {
+      "epoch": 1.3557448468691073,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001945770206125236,
+      "loss": 1.5349,
+      "step": 208500
+    },
+    {
+      "epoch": 1.3589960335522466,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0019456401586579102,
+      "loss": 1.5313,
+      "step": 209000
+    },
+    {
+      "epoch": 1.362247220235386,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019455101111905846,
+      "loss": 1.5399,
+      "step": 209500
+    },
+    {
+      "epoch": 1.3654984069185252,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0019453800637232591,
+      "loss": 1.5343,
+      "step": 210000
+    },
+    {
+      "epoch": 1.3687495936016645,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0019452500162559334,
+      "loss": 1.5308,
+      "step": 210500
+    },
+    {
+      "epoch": 1.372000780284804,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0019451199687886079,
+      "loss": 1.5316,
+      "step": 211000
+    },
+    {
+      "epoch": 1.3752519669679433,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0019449899213212823,
+      "loss": 1.5338,
+      "step": 211500
+    },
+    {
+      "epoch": 1.3785031536510828,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0019448598738539566,
+      "loss": 1.5312,
+      "step": 212000
+    },
+    {
+      "epoch": 1.381754340334222,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001944729826386631,
+      "loss": 1.5325,
+      "step": 212500
+    },
+    {
+      "epoch": 1.3850055270173613,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0019445997789193056,
+      "loss": 1.5329,
+      "step": 213000
+    },
+    {
+      "epoch": 1.3882567137005006,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0019444697314519798,
+      "loss": 1.5323,
+      "step": 213500
+    },
+    {
+      "epoch": 1.39150790038364,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.0019443396839846543,
+      "loss": 1.5399,
+      "step": 214000
+    },
+    {
+      "epoch": 1.3947590870667794,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.001944209636517329,
+      "loss": 1.5351,
+      "step": 214500
+    },
+    {
+      "epoch": 1.3980102737499187,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0019440795890500035,
+      "loss": 1.5353,
+      "step": 215000
+    },
+    {
+      "epoch": 1.401261460433058,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0019439495415826777,
+      "loss": 1.5367,
+      "step": 215500
+    },
+    {
+      "epoch": 1.4045126471161975,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0019438194941153522,
+      "loss": 1.5354,
+      "step": 216000
+    },
+    {
+      "epoch": 1.4077638337993368,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0019436894466480267,
+      "loss": 1.5261,
+      "step": 216500
+    },
+    {
+      "epoch": 1.411015020482476,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.001943559399180701,
+      "loss": 1.5305,
+      "step": 217000
+    },
+    {
+      "epoch": 1.4142662071656154,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0019434293517133754,
+      "loss": 1.5362,
+      "step": 217500
+    },
+    {
+      "epoch": 1.4175173938487549,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00194329930424605,
+      "loss": 1.5272,
+      "step": 218000
+    },
+    {
+      "epoch": 1.4207685805318941,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0019431692567787242,
+      "loss": 1.5304,
+      "step": 218500
+    },
+    {
+      "epoch": 1.4240197672150334,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0019430392093113986,
+      "loss": 1.5318,
+      "step": 219000
+    },
+    {
+      "epoch": 1.427270953898173,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0019429091618440731,
+      "loss": 1.5312,
+      "step": 219500
+    },
+    {
+      "epoch": 1.4305221405813122,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0019427791143767474,
+      "loss": 1.5338,
+      "step": 220000
+    },
+    {
+      "epoch": 1.4337733272644515,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0019426490669094219,
+      "loss": 1.5308,
+      "step": 220500
+    },
+    {
+      "epoch": 1.4370245139475908,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0019425190194420963,
+      "loss": 1.5243,
+      "step": 221000
+    },
+    {
+      "epoch": 1.4402757006307303,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.001942388971974771,
+      "loss": 1.528,
+      "step": 221500
+    },
+    {
+      "epoch": 1.4435268873138696,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0019422589245074453,
+      "loss": 1.5385,
+      "step": 222000
+    },
+    {
+      "epoch": 1.4467780739970089,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0019421288770401198,
+      "loss": 1.5363,
+      "step": 222500
+    },
+    {
+      "epoch": 1.4500292606801484,
+      "grad_norm": 6.3125,
+      "learning_rate": 0.0019419988295727943,
+      "loss": 1.528,
+      "step": 223000
+    },
+    {
+      "epoch": 1.4532804473632877,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0019418687821054685,
+      "loss": 1.5238,
+      "step": 223500
+    },
+    {
+      "epoch": 1.456531634046427,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.001941738734638143,
+      "loss": 1.5261,
+      "step": 224000
+    },
+    {
+      "epoch": 1.4597828207295662,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0019416086871708175,
+      "loss": 1.5223,
+      "step": 224500
+    },
+    {
+      "epoch": 1.4630340074127055,
+      "grad_norm": 1.796875,
+      "learning_rate": 0.0019414786397034917,
+      "loss": 1.5345,
+      "step": 225000
+    },
+    {
+      "epoch": 1.466285194095845,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0019413485922361662,
+      "loss": 1.5254,
+      "step": 225500
+    },
+    {
+      "epoch": 1.4695363807789843,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0019412185447688407,
+      "loss": 1.5274,
+      "step": 226000
+    },
+    {
+      "epoch": 1.4727875674621238,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001941088497301515,
+      "loss": 1.5266,
+      "step": 226500
+    },
+    {
+      "epoch": 1.476038754145263,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0019409584498341894,
+      "loss": 1.5234,
+      "step": 227000
+    },
+    {
+      "epoch": 1.4792899408284024,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.001940828402366864,
+      "loss": 1.5274,
+      "step": 227500
+    },
+    {
+      "epoch": 1.4825411275115417,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0019406983548995382,
+      "loss": 1.5308,
+      "step": 228000
+    },
+    {
+      "epoch": 1.485792314194681,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0019405683074322127,
+      "loss": 1.5266,
+      "step": 228500
+    },
+    {
+      "epoch": 1.4890435008778204,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0019404382599648873,
+      "loss": 1.5297,
+      "step": 229000
+    },
+    {
+      "epoch": 1.4922946875609597,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0019403082124975618,
+      "loss": 1.5334,
+      "step": 229500
+    },
+    {
+      "epoch": 1.495545874244099,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001940178165030236,
+      "loss": 1.5235,
+      "step": 230000
+    },
+    {
+      "epoch": 1.4987970609272385,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0019400481175629106,
+      "loss": 1.52,
+      "step": 230500
+    },
+    {
+      "epoch": 1.5020482476103778,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.001939918070095585,
+      "loss": 1.5206,
+      "step": 231000
+    },
+    {
+      "epoch": 1.505299434293517,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.0019397880226282593,
+      "loss": 1.5168,
+      "step": 231500
+    },
+    {
+      "epoch": 1.5085506209766564,
+      "grad_norm": 2.796875,
+      "learning_rate": 0.0019396579751609338,
+      "loss": 1.5241,
+      "step": 232000
+    },
+    {
+      "epoch": 1.5118018076597957,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0019395279276936083,
+      "loss": 1.5297,
+      "step": 232500
+    },
+    {
+      "epoch": 1.5150529943429352,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0019393978802262825,
+      "loss": 1.526,
+      "step": 233000
+    },
+    {
+      "epoch": 1.5183041810260747,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.001939267832758957,
+      "loss": 1.5229,
+      "step": 233500
+    },
+    {
+      "epoch": 1.521555367709214,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.0019391377852916315,
+      "loss": 1.5213,
+      "step": 234000
+    },
+    {
+      "epoch": 1.5248065543923532,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0019390077378243057,
+      "loss": 1.5181,
+      "step": 234500
+    },
+    {
+      "epoch": 1.5280577410754925,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0019388776903569802,
+      "loss": 1.5245,
+      "step": 235000
+    },
+    {
+      "epoch": 1.5313089277586318,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0019387476428896547,
+      "loss": 1.5236,
+      "step": 235500
+    },
+    {
+      "epoch": 1.534560114441771,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0019386175954223294,
+      "loss": 1.5121,
+      "step": 236000
+    },
+    {
+      "epoch": 1.5378113011249106,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.0019384875479550037,
+      "loss": 1.5178,
+      "step": 236500
+    },
+    {
+      "epoch": 1.5410624878080499,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0019383575004876781,
+      "loss": 1.5149,
+      "step": 237000
+    },
+    {
+      "epoch": 1.5443136744911894,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.0019382274530203526,
+      "loss": 1.5149,
+      "step": 237500
+    },
+    {
+      "epoch": 1.5475648611743287,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0019380974055530269,
+      "loss": 1.5182,
+      "step": 238000
+    },
+    {
+      "epoch": 1.550816047857468,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.0019379673580857014,
+      "loss": 1.5145,
+      "step": 238500
+    },
+    {
+      "epoch": 1.5540672345406072,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0019378373106183758,
+      "loss": 1.5163,
+      "step": 239000
+    },
+    {
+      "epoch": 1.5573184212237465,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00193770726315105,
+      "loss": 1.5128,
+      "step": 239500
+    },
+    {
+      "epoch": 1.560569607906886,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0019375772156837246,
+      "loss": 1.5102,
+      "step": 240000
+    },
+    {
+      "epoch": 1.5638207945900253,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.001937447168216399,
+      "loss": 1.5091,
+      "step": 240500
+    },
+    {
+      "epoch": 1.5670719812731648,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0019373171207490733,
+      "loss": 1.5095,
+      "step": 241000
+    },
+    {
+      "epoch": 1.5703231679563041,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0019371870732817478,
+      "loss": 1.5146,
+      "step": 241500
+    },
+    {
+      "epoch": 1.5735743546394434,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0019370570258144223,
+      "loss": 1.5162,
+      "step": 242000
+    },
+    {
+      "epoch": 1.5768255413225827,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0019369269783470965,
+      "loss": 1.5214,
+      "step": 242500
+    },
+    {
+      "epoch": 1.580076728005722,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.001936796930879771,
+      "loss": 1.5162,
+      "step": 243000
+    },
+    {
+      "epoch": 1.5833279146888615,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0019366668834124457,
+      "loss": 1.5258,
+      "step": 243500
+    },
+    {
+      "epoch": 1.5865791013720008,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0019365368359451202,
+      "loss": 1.5242,
+      "step": 244000
+    },
+    {
+      "epoch": 1.5898302880551403,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0019364067884777944,
+      "loss": 1.5162,
+      "step": 244500
+    },
+    {
+      "epoch": 1.5930814747382795,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.001936276741010469,
+      "loss": 1.5204,
+      "step": 245000
+    },
+    {
+      "epoch": 1.5963326614214188,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0019361466935431434,
+      "loss": 1.5206,
+      "step": 245500
+    },
+    {
+      "epoch": 1.5995838481045581,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0019360166460758177,
+      "loss": 1.5266,
+      "step": 246000
+    },
+    {
+      "epoch": 1.6028350347876974,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.0019358865986084921,
+      "loss": 1.5218,
+      "step": 246500
+    },
+    {
+      "epoch": 1.606086221470837,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0019357565511411666,
+      "loss": 1.5187,
+      "step": 247000
+    },
+    {
+      "epoch": 1.6093374081539762,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.0019356265036738409,
+      "loss": 1.527,
+      "step": 247500
+    },
+    {
+      "epoch": 1.6125885948371157,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0019354964562065154,
+      "loss": 1.519,
+      "step": 248000
+    },
+    {
+      "epoch": 1.615839781520255,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0019353664087391898,
+      "loss": 1.5178,
+      "step": 248500
+    },
+    {
+      "epoch": 1.6190909682033943,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.001935236361271864,
+      "loss": 1.5131,
+      "step": 249000
+    },
+    {
+      "epoch": 1.6223421548865335,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0019351063138045386,
+      "loss": 1.5118,
+      "step": 249500
+    },
+    {
+      "epoch": 1.6255933415696728,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001934976266337213,
+      "loss": 1.5144,
+      "step": 250000
+    },
+    {
+      "epoch": 1.6288445282528121,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0019348462188698878,
+      "loss": 1.5197,
+      "step": 250500
+    },
+    {
+      "epoch": 1.6320957149359516,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.001934716171402562,
+      "loss": 1.5172,
+      "step": 251000
+    },
+    {
+      "epoch": 1.6353469016190911,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.0019345861239352365,
+      "loss": 1.5207,
+      "step": 251500
+    },
+    {
+      "epoch": 1.6385980883022304,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.001934456076467911,
+      "loss": 1.5156,
+      "step": 252000
+    },
+    {
+      "epoch": 1.6418492749853697,
+      "grad_norm": 2.71875,
+      "learning_rate": 0.0019343260290005852,
+      "loss": 1.5161,
+      "step": 252500
+    },
+    {
+      "epoch": 1.645100461668509,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0019341959815332597,
+      "loss": 1.5193,
+      "step": 253000
+    },
+    {
+      "epoch": 1.6483516483516483,
+      "grad_norm": 33.5,
+      "learning_rate": 0.0019340659340659342,
+      "loss": 1.5211,
+      "step": 253500
+    },
+    {
+      "epoch": 1.6516028350347876,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0019339358865986085,
+      "loss": 1.5132,
+      "step": 254000
+    },
+    {
+      "epoch": 1.654854021717927,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.001933805839131283,
+      "loss": 1.5106,
+      "step": 254500
+    },
+    {
+      "epoch": 1.6581052084010663,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0019336757916639574,
+      "loss": 1.5101,
+      "step": 255000
+    },
+    {
+      "epoch": 1.6613563950842059,
+      "grad_norm": 27.125,
+      "learning_rate": 0.0019335457441966317,
+      "loss": 1.5107,
+      "step": 255500
+    },
+    {
+      "epoch": 1.6646075817673451,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.0019334156967293062,
+      "loss": 1.5099,
+      "step": 256000
+    },
+    {
+      "epoch": 1.6678587684504844,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0019332856492619806,
+      "loss": 1.5145,
+      "step": 256500
+    },
+    {
+      "epoch": 1.6711099551336237,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001933155601794655,
+      "loss": 1.5125,
+      "step": 257000
+    },
+    {
+      "epoch": 1.674361141816763,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0019330255543273294,
+      "loss": 1.5107,
+      "step": 257500
+    },
+    {
+      "epoch": 1.6776123284999025,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.001932895506860004,
+      "loss": 1.5076,
+      "step": 258000
+    },
+    {
+      "epoch": 1.6808635151830418,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0019327654593926785,
+      "loss": 1.5095,
+      "step": 258500
+    },
+    {
+      "epoch": 1.6841147018661813,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0019326354119253528,
+      "loss": 1.5088,
+      "step": 259000
+    },
+    {
+      "epoch": 1.6873658885493206,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0019325053644580273,
+      "loss": 1.5099,
+      "step": 259500
+    },
+    {
+      "epoch": 1.6906170752324599,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0019323753169907018,
+      "loss": 1.5095,
+      "step": 260000
+    },
+    {
+      "epoch": 1.6938682619155991,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.001932245269523376,
+      "loss": 1.514,
+      "step": 260500
+    },
+    {
+      "epoch": 1.6971194485987384,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.0019321152220560505,
+      "loss": 1.5228,
+      "step": 261000
+    },
+    {
+      "epoch": 1.700370635281878,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001931985174588725,
+      "loss": 1.5193,
+      "step": 261500
+    },
+    {
+      "epoch": 1.7036218219650172,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0019318551271213992,
+      "loss": 1.5193,
+      "step": 262000
+    },
+    {
+      "epoch": 1.7068730086481567,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019317250796540737,
+      "loss": 1.5236,
+      "step": 262500
+    },
+    {
+      "epoch": 1.710124195331296,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.0019315950321867482,
+      "loss": 1.51,
+      "step": 263000
+    },
+    {
+      "epoch": 1.7133753820144353,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0019314649847194225,
+      "loss": 1.5129,
+      "step": 263500
+    },
+    {
+      "epoch": 1.7166265686975746,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.001931334937252097,
+      "loss": 1.5101,
+      "step": 264000
+    },
+    {
+      "epoch": 1.7198777553807139,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0019312048897847714,
+      "loss": 1.5112,
+      "step": 264500
+    },
+    {
+      "epoch": 1.7231289420638531,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0019310748423174461,
+      "loss": 1.5131,
+      "step": 265000
+    },
+    {
+      "epoch": 1.7263801287469926,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0019309447948501204,
+      "loss": 1.5135,
+      "step": 265500
+    },
+    {
+      "epoch": 1.7296313154301322,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0019308147473827949,
+      "loss": 1.5105,
+      "step": 266000
+    },
+    {
+      "epoch": 1.7328825021132714,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019306846999154693,
+      "loss": 1.5133,
+      "step": 266500
+    },
+    {
+      "epoch": 1.7361336887964107,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0019305546524481436,
+      "loss": 1.5098,
+      "step": 267000
+    },
+    {
+      "epoch": 1.73938487547955,
+      "grad_norm": 5.84375,
+      "learning_rate": 0.001930424604980818,
+      "loss": 1.5059,
+      "step": 267500
+    },
+    {
+      "epoch": 1.7426360621626893,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0019302945575134926,
+      "loss": 1.514,
+      "step": 268000
+    },
+    {
+      "epoch": 1.7458872488458286,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0019301645100461668,
+      "loss": 1.4967,
+      "step": 268500
+    },
+    {
+      "epoch": 1.749138435528968,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0019300344625788413,
+      "loss": 1.505,
+      "step": 269000
+    },
+    {
+      "epoch": 1.7523896222121074,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.0019299044151115158,
+      "loss": 1.5132,
+      "step": 269500
+    },
+    {
+      "epoch": 1.7556408088952469,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00192977436764419,
+      "loss": 1.5124,
+      "step": 270000
+    },
+    {
+      "epoch": 1.7588919955783862,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0019296443201768645,
+      "loss": 1.509,
+      "step": 270500
+    },
+    {
+      "epoch": 1.7621431822615254,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.001929514272709539,
+      "loss": 1.5049,
+      "step": 271000
+    },
+    {
+      "epoch": 1.7653943689446647,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.0019293842252422133,
+      "loss": 1.5067,
+      "step": 271500
+    },
+    {
+      "epoch": 1.768645555627804,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0019292541777748877,
+      "loss": 1.5076,
+      "step": 272000
+    },
+    {
+      "epoch": 1.7718967423109435,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.0019291241303075624,
+      "loss": 1.5,
+      "step": 272500
+    },
+    {
+      "epoch": 1.7751479289940828,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.001928994082840237,
+      "loss": 1.5139,
+      "step": 273000
+    },
+    {
+      "epoch": 1.7783991156772223,
+      "grad_norm": 0.875,
+      "learning_rate": 0.0019288640353729112,
+      "loss": 1.5092,
+      "step": 273500
+    },
+    {
+      "epoch": 1.7816503023603616,
+      "grad_norm": 4.34375,
+      "learning_rate": 0.0019287339879055856,
+      "loss": 1.5087,
+      "step": 274000
+    },
+    {
+      "epoch": 1.7849014890435009,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0019286039404382601,
+      "loss": 1.5178,
+      "step": 274500
+    },
+    {
+      "epoch": 1.7881526757266402,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0019284738929709344,
+      "loss": 1.5253,
+      "step": 275000
+    },
+    {
+      "epoch": 1.7914038624097794,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.0019283438455036089,
+      "loss": 1.5293,
+      "step": 275500
+    },
+    {
+      "epoch": 1.794655049092919,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.0019282137980362833,
+      "loss": 1.528,
+      "step": 276000
+    },
+    {
+      "epoch": 1.7979062357760582,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019280837505689576,
+      "loss": 1.5178,
+      "step": 276500
+    },
+    {
+      "epoch": 1.8011574224591977,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001927953703101632,
+      "loss": 1.5362,
+      "step": 277000
+    },
+    {
+      "epoch": 1.804408609142337,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0019278236556343066,
+      "loss": 1.5321,
+      "step": 277500
+    },
+    {
+      "epoch": 1.8076597958254763,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.0019276936081669808,
+      "loss": 1.5199,
+      "step": 278000
+    },
+    {
+      "epoch": 1.8109109825086156,
+      "grad_norm": 3.921875,
+      "learning_rate": 0.0019275635606996553,
+      "loss": 1.52,
+      "step": 278500
+    },
+    {
+      "epoch": 1.8141621691917549,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0019274335132323298,
+      "loss": 1.5242,
+      "step": 279000
+    },
+    {
+      "epoch": 1.8174133558748944,
+      "grad_norm": 2.125,
+      "learning_rate": 0.0019273034657650045,
+      "loss": 1.5241,
+      "step": 279500
+    },
+    {
+      "epoch": 1.8206645425580337,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0019271734182976787,
+      "loss": 1.5294,
+      "step": 280000
+    },
+    {
+      "epoch": 1.8239157292411732,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0019270433708303532,
+      "loss": 1.5173,
+      "step": 280500
+    },
+    {
+      "epoch": 1.8271669159243125,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.0019269133233630277,
+      "loss": 1.5224,
+      "step": 281000
+    },
+    {
+      "epoch": 1.8304181026074517,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.001926783275895702,
+      "loss": 1.5304,
+      "step": 281500
+    },
+    {
+      "epoch": 1.833669289290591,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0019266532284283764,
+      "loss": 1.5289,
+      "step": 282000
+    },
+    {
+      "epoch": 1.8369204759737303,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.001926523180961051,
+      "loss": 1.5198,
+      "step": 282500
+    },
+    {
+      "epoch": 1.8401716626568696,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0019263931334937252,
+      "loss": 1.5292,
+      "step": 283000
+    },
+    {
+      "epoch": 1.843422849340009,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0019262630860263997,
+      "loss": 1.5336,
+      "step": 283500
+    },
+    {
+      "epoch": 1.8466740360231486,
+      "grad_norm": 3.90625,
+      "learning_rate": 0.0019261330385590741,
+      "loss": 1.5415,
+      "step": 284000
+    },
+    {
+      "epoch": 1.849925222706288,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0019260029910917484,
+      "loss": 1.5339,
+      "step": 284500
+    },
+    {
+      "epoch": 1.8531764093894272,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0019258729436244229,
+      "loss": 1.5423,
+      "step": 285000
+    },
+    {
+      "epoch": 1.8564275960725665,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0019257428961570973,
+      "loss": 1.5517,
+      "step": 285500
+    },
+    {
+      "epoch": 1.8596787827557058,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0019256128486897716,
+      "loss": 1.5582,
+      "step": 286000
+    },
+    {
+      "epoch": 1.862929969438845,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.001925482801222446,
+      "loss": 1.5598,
+      "step": 286500
+    },
+    {
+      "epoch": 1.8661811561219845,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0019253527537551208,
+      "loss": 1.5541,
+      "step": 287000
+    },
+    {
+      "epoch": 1.8694323428051238,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0019252227062877953,
+      "loss": 1.551,
+      "step": 287500
+    },
+    {
+      "epoch": 1.8726835294882633,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0019250926588204695,
+      "loss": 1.5573,
+      "step": 288000
+    },
+    {
+      "epoch": 1.8759347161714026,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.001924962611353144,
+      "loss": 1.5485,
+      "step": 288500
+    },
+    {
+      "epoch": 1.879185902854542,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0019248325638858185,
+      "loss": 1.5529,
+      "step": 289000
+    },
+    {
+      "epoch": 1.8824370895376812,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0019247025164184927,
+      "loss": 1.5538,
+      "step": 289500
+    },
+    {
+      "epoch": 1.8856882762208205,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0019245724689511672,
+      "loss": 1.5524,
+      "step": 290000
+    },
+    {
+      "epoch": 1.88893946290396,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0019244424214838417,
+      "loss": 1.5582,
+      "step": 290500
+    },
+    {
+      "epoch": 1.8921906495870993,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.001924312374016516,
+      "loss": 1.5682,
+      "step": 291000
+    },
+    {
+      "epoch": 1.8954418362702388,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0019241823265491904,
+      "loss": 1.5639,
+      "step": 291500
+    },
+    {
+      "epoch": 1.898693022953378,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001924052279081865,
+      "loss": 1.5732,
+      "step": 292000
+    },
+    {
+      "epoch": 1.9019442096365173,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0019239222316145392,
+      "loss": 1.5634,
+      "step": 292500
+    },
+    {
+      "epoch": 1.9051953963196566,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0019237921841472137,
+      "loss": 1.5599,
+      "step": 293000
+    },
+    {
+      "epoch": 1.908446583002796,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0019236621366798881,
+      "loss": 1.5507,
+      "step": 293500
+    },
+    {
+      "epoch": 1.9116977696859354,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0019235320892125628,
+      "loss": 1.5467,
+      "step": 294000
+    },
+    {
+      "epoch": 1.9149489563690747,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001923402041745237,
+      "loss": 1.5465,
+      "step": 294500
+    },
+    {
+      "epoch": 1.9182001430522142,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.0019232719942779116,
+      "loss": 1.5408,
+      "step": 295000
+    },
+    {
+      "epoch": 1.9214513297353535,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.001923141946810586,
+      "loss": 1.5481,
+      "step": 295500
+    },
+    {
+      "epoch": 1.9247025164184928,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0019230118993432603,
+      "loss": 1.5488,
+      "step": 296000
+    },
+    {
+      "epoch": 1.927953703101632,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.0019228818518759348,
+      "loss": 1.5443,
+      "step": 296500
+    },
+    {
+      "epoch": 1.9312048897847713,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.0019227518044086093,
+      "loss": 1.5579,
+      "step": 297000
+    },
+    {
+      "epoch": 1.9344560764679106,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.0019226217569412835,
+      "loss": 1.5481,
+      "step": 297500
+    },
+    {
+      "epoch": 1.9377072631510501,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001922491709473958,
+      "loss": 1.5384,
+      "step": 298000
+    },
+    {
+      "epoch": 1.9409584498341896,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0019223616620066325,
+      "loss": 1.5387,
+      "step": 298500
+    },
+    {
+      "epoch": 1.944209636517329,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.0019222316145393068,
+      "loss": 1.5439,
+      "step": 299000
+    },
+    {
+      "epoch": 1.9474608232004682,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0019221015670719812,
+      "loss": 1.5443,
+      "step": 299500
+    },
+    {
+      "epoch": 1.9507120098836075,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0019219715196046557,
+      "loss": 1.5366,
+      "step": 300000
+    },
+    {
+      "epoch": 1.9539631965667468,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00192184147213733,
+      "loss": 1.5391,
+      "step": 300500
+    },
+    {
+      "epoch": 1.957214383249886,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0019217114246700044,
+      "loss": 1.5394,
+      "step": 301000
+    },
+    {
+      "epoch": 1.9604655699330256,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0019215813772026791,
+      "loss": 1.5341,
+      "step": 301500
+    },
+    {
+      "epoch": 1.9637167566161648,
+      "grad_norm": 1.375,
+      "learning_rate": 0.0019214513297353536,
+      "loss": 1.5414,
+      "step": 302000
+    },
+    {
+      "epoch": 1.9669679432993044,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0019213212822680279,
+      "loss": 1.5436,
+      "step": 302500
+    },
+    {
+      "epoch": 1.9702191299824436,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0019211912348007024,
+      "loss": 1.5284,
+      "step": 303000
+    },
+    {
+      "epoch": 1.973470316665583,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0019210611873333768,
+      "loss": 1.5208,
+      "step": 303500
+    },
+    {
+      "epoch": 1.9767215033487222,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.001920931139866051,
+      "loss": 1.5234,
+      "step": 304000
+    },
+    {
+      "epoch": 1.9799726900318615,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0019208010923987256,
+      "loss": 1.5299,
+      "step": 304500
+    },
+    {
+      "epoch": 1.983223876715001,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0019206710449314,
+      "loss": 1.5367,
+      "step": 305000
+    },
+    {
+      "epoch": 1.9864750633981403,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0019205409974640743,
+      "loss": 1.5328,
+      "step": 305500
+    },
+    {
+      "epoch": 1.9897262500812798,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0019204109499967488,
+      "loss": 1.5326,
+      "step": 306000
+    },
+    {
+      "epoch": 1.992977436764419,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0019202809025294233,
+      "loss": 1.5357,
+      "step": 306500
+    },
+    {
+      "epoch": 1.9962286234475584,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0019201508550620975,
+      "loss": 1.5266,
+      "step": 307000
+    },
+    {
+      "epoch": 1.9994798101306976,
+      "grad_norm": 5.8125,
+      "learning_rate": 0.001920020807594772,
+      "loss": 1.5319,
+      "step": 307500
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 1.5122345685958862,
+      "eval_runtime": 0.5402,
+      "eval_samples_per_second": 1851.06,
+      "eval_steps_per_second": 29.617,
+      "step": 307580
+    },
+    {
+      "epoch": 2.002730996813837,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0019198907601274465,
+      "loss": 1.53,
+      "step": 308000
+    },
+    {
+      "epoch": 2.005982183496976,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0019197607126601212,
+      "loss": 1.5226,
+      "step": 308500
+    },
+    {
+      "epoch": 2.009233370180116,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0019196306651927955,
+      "loss": 1.5161,
+      "step": 309000
+    },
+    {
+      "epoch": 2.0124845568632552,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00191950061772547,
+      "loss": 1.5224,
+      "step": 309500
+    },
+    {
+      "epoch": 2.0157357435463945,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0019193705702581444,
+      "loss": 1.5189,
+      "step": 310000
+    },
+    {
+      "epoch": 2.018986930229534,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0019192405227908187,
+      "loss": 1.5231,
+      "step": 310500
+    },
+    {
+      "epoch": 2.022238116912673,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.0019191104753234932,
+      "loss": 1.5372,
+      "step": 311000
+    },
+    {
+      "epoch": 2.0254893035958124,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.0019189804278561676,
+      "loss": 1.5421,
+      "step": 311500
+    },
+    {
+      "epoch": 2.0287404902789516,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.001918850380388842,
+      "loss": 1.5291,
+      "step": 312000
+    },
+    {
+      "epoch": 2.0319916769620914,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0019187203329215164,
+      "loss": 1.534,
+      "step": 312500
+    },
+    {
+      "epoch": 2.0352428636452307,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.0019185902854541908,
+      "loss": 1.5307,
+      "step": 313000
+    },
+    {
+      "epoch": 2.03849405032837,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0019184602379868651,
+      "loss": 1.5325,
+      "step": 313500
+    },
+    {
+      "epoch": 2.0417452370115092,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.0019183301905195396,
+      "loss": 1.5389,
+      "step": 314000
+    },
+    {
+      "epoch": 2.0449964236946485,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.001918200143052214,
+      "loss": 1.5328,
+      "step": 314500
+    },
+    {
+      "epoch": 2.048247610377788,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.0019180700955848883,
+      "loss": 1.5415,
+      "step": 315000
+    },
+    {
+      "epoch": 2.051498797060927,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0019179400481175628,
+      "loss": 1.5549,
+      "step": 315500
+    },
+    {
+      "epoch": 2.0547499837440664,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0019178100006502375,
+      "loss": 1.5597,
+      "step": 316000
+    },
+    {
+      "epoch": 2.058001170427206,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001917679953182912,
+      "loss": 1.5947,
+      "step": 316500
+    },
+    {
+      "epoch": 2.0612523571103454,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0019175499057155862,
+      "loss": 1.5896,
+      "step": 317000
+    },
+    {
+      "epoch": 2.0645035437934847,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0019174198582482607,
+      "loss": 1.5825,
+      "step": 317500
+    },
+    {
+      "epoch": 2.067754730476624,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0019172898107809352,
+      "loss": 1.5415,
+      "step": 318000
+    },
+    {
+      "epoch": 2.0710059171597632,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0019171597633136095,
+      "loss": 1.5405,
+      "step": 318500
+    },
+    {
+      "epoch": 2.0742571038429025,
+      "grad_norm": 1.5,
+      "learning_rate": 0.001917029715846284,
+      "loss": 1.5333,
+      "step": 319000
+    },
+    {
+      "epoch": 2.077508290526042,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0019168996683789584,
+      "loss": 1.5374,
+      "step": 319500
+    },
+    {
+      "epoch": 2.0807594772091815,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0019167696209116327,
+      "loss": 1.5327,
+      "step": 320000
+    },
+    {
+      "epoch": 2.084010663892321,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0019166395734443072,
+      "loss": 1.5281,
+      "step": 320500
+    },
+    {
+      "epoch": 2.08726185057546,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0019165095259769816,
+      "loss": 1.529,
+      "step": 321000
+    },
+    {
+      "epoch": 2.0905130372585994,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.001916379478509656,
+      "loss": 1.5322,
+      "step": 321500
+    },
+    {
+      "epoch": 2.0937642239417387,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0019162494310423304,
+      "loss": 1.5284,
+      "step": 322000
+    },
+    {
+      "epoch": 2.097015410624878,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0019161193835750049,
+      "loss": 1.5338,
+      "step": 322500
+    },
+    {
+      "epoch": 2.1002665973080172,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.0019159893361076796,
+      "loss": 1.524,
+      "step": 323000
+    },
+    {
+      "epoch": 2.103517783991157,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0019158592886403538,
+      "loss": 1.5225,
+      "step": 323500
+    },
+    {
+      "epoch": 2.1067689706742962,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0019157292411730283,
+      "loss": 1.5322,
+      "step": 324000
+    },
+    {
+      "epoch": 2.1100201573574355,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0019155991937057028,
+      "loss": 1.5216,
+      "step": 324500
+    },
+    {
+      "epoch": 2.113271344040575,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.001915469146238377,
+      "loss": 1.5237,
+      "step": 325000
+    },
+    {
+      "epoch": 2.116522530723714,
+      "grad_norm": 5.28125,
+      "learning_rate": 0.0019153390987710515,
+      "loss": 1.5298,
+      "step": 325500
+    },
+    {
+      "epoch": 2.1197737174068534,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001915209051303726,
+      "loss": 1.5249,
+      "step": 326000
+    },
+    {
+      "epoch": 2.1230249040899927,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.0019150790038364003,
+      "loss": 1.5262,
+      "step": 326500
+    },
+    {
+      "epoch": 2.1262760907731324,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0019149489563690747,
+      "loss": 1.5222,
+      "step": 327000
+    },
+    {
+      "epoch": 2.1295272774562717,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0019148189089017492,
+      "loss": 1.5148,
+      "step": 327500
+    },
+    {
+      "epoch": 2.132778464139411,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0019146888614344235,
+      "loss": 1.5151,
+      "step": 328000
+    },
+    {
+      "epoch": 2.1360296508225503,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.001914558813967098,
+      "loss": 1.5202,
+      "step": 328500
+    },
+    {
+      "epoch": 2.1392808375056895,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0019144287664997724,
+      "loss": 1.5299,
+      "step": 329000
+    },
+    {
+      "epoch": 2.142532024188829,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.0019142987190324467,
+      "loss": 1.5327,
+      "step": 329500
+    },
+    {
+      "epoch": 2.145783210871968,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0019141686715651212,
+      "loss": 1.5325,
+      "step": 330000
+    },
+    {
+      "epoch": 2.1490343975551074,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0019140386240977959,
+      "loss": 1.5311,
+      "step": 330500
+    },
+    {
+      "epoch": 2.152285584238247,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0019139085766304703,
+      "loss": 1.5228,
+      "step": 331000
+    },
+    {
+      "epoch": 2.1555367709213864,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0019137785291631446,
+      "loss": 1.5265,
+      "step": 331500
+    },
+    {
+      "epoch": 2.1587879576045257,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001913648481695819,
+      "loss": 1.5246,
+      "step": 332000
+    },
+    {
+      "epoch": 2.162039144287665,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0019135184342284936,
+      "loss": 1.5268,
+      "step": 332500
+    },
+    {
+      "epoch": 2.1652903309708043,
+      "grad_norm": 1.8046875,
+      "learning_rate": 0.0019133883867611678,
+      "loss": 1.5286,
+      "step": 333000
+    },
+    {
+      "epoch": 2.1685415176539435,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0019132583392938423,
+      "loss": 1.5289,
+      "step": 333500
+    },
+    {
+      "epoch": 2.171792704337083,
+      "grad_norm": 3.296875,
+      "learning_rate": 0.0019131282918265168,
+      "loss": 1.5281,
+      "step": 334000
+    },
+    {
+      "epoch": 2.1750438910202226,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.001912998244359191,
+      "loss": 1.5205,
+      "step": 334500
+    },
+    {
+      "epoch": 2.178295077703362,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0019128681968918655,
+      "loss": 1.518,
+      "step": 335000
+    },
+    {
+      "epoch": 2.181546264386501,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00191273814942454,
+      "loss": 1.5153,
+      "step": 335500
+    },
+    {
+      "epoch": 2.1847974510696404,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0019126081019572143,
+      "loss": 1.5212,
+      "step": 336000
+    },
+    {
+      "epoch": 2.1880486377527797,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.0019124780544898887,
+      "loss": 1.5129,
+      "step": 336500
+    },
+    {
+      "epoch": 2.191299824435919,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.0019123480070225632,
+      "loss": 1.523,
+      "step": 337000
+    },
+    {
+      "epoch": 2.1945510111190583,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.001912217959555238,
+      "loss": 1.5178,
+      "step": 337500
+    },
+    {
+      "epoch": 2.197802197802198,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0019120879120879122,
+      "loss": 1.5259,
+      "step": 338000
+    },
+    {
+      "epoch": 2.2010533844853373,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0019119578646205867,
+      "loss": 1.5123,
+      "step": 338500
+    },
+    {
+      "epoch": 2.2043045711684766,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0019118278171532611,
+      "loss": 1.5192,
+      "step": 339000
+    },
+    {
+      "epoch": 2.207555757851616,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0019116977696859354,
+      "loss": 1.5165,
+      "step": 339500
+    },
+    {
+      "epoch": 2.210806944534755,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.0019115677222186099,
+      "loss": 1.517,
+      "step": 340000
+    },
+    {
+      "epoch": 2.2140581312178944,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0019114376747512843,
+      "loss": 1.5276,
+      "step": 340500
+    },
+    {
+      "epoch": 2.2173093179010337,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0019113076272839586,
+      "loss": 1.5122,
+      "step": 341000
+    },
+    {
+      "epoch": 2.2205605045841734,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.001911177579816633,
+      "loss": 1.5157,
+      "step": 341500
+    },
+    {
+      "epoch": 2.2238116912673127,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0019110475323493076,
+      "loss": 1.5224,
+      "step": 342000
+    },
+    {
+      "epoch": 2.227062877950452,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0019109174848819818,
+      "loss": 1.5313,
+      "step": 342500
+    },
+    {
+      "epoch": 2.2303140646335913,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.0019107874374146563,
+      "loss": 1.5349,
+      "step": 343000
+    },
+    {
+      "epoch": 2.2335652513167306,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0019106573899473308,
+      "loss": 1.5298,
+      "step": 343500
+    },
+    {
+      "epoch": 2.23681643799987,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.001910527342480005,
+      "loss": 1.5312,
+      "step": 344000
+    },
+    {
+      "epoch": 2.240067624683009,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0019103972950126795,
+      "loss": 1.5363,
+      "step": 344500
+    },
+    {
+      "epoch": 2.243318811366149,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0019102672475453542,
+      "loss": 1.535,
+      "step": 345000
+    },
+    {
+      "epoch": 2.246569998049288,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0019101372000780287,
+      "loss": 1.5359,
+      "step": 345500
+    },
+    {
+      "epoch": 2.2498211847324274,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.001910007152610703,
+      "loss": 1.5286,
+      "step": 346000
+    },
+    {
+      "epoch": 2.2530723714155667,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0019098771051433774,
+      "loss": 1.5244,
+      "step": 346500
+    },
+    {
+      "epoch": 2.256323558098706,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001909747057676052,
+      "loss": 1.5219,
+      "step": 347000
+    },
+    {
+      "epoch": 2.2595747447818453,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.0019096170102087262,
+      "loss": 1.5239,
+      "step": 347500
+    },
+    {
+      "epoch": 2.2628259314649846,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.0019094869627414007,
+      "loss": 1.5331,
+      "step": 348000
+    },
+    {
+      "epoch": 2.2660771181481243,
+      "grad_norm": 3.984375,
+      "learning_rate": 0.0019093569152740751,
+      "loss": 1.5267,
+      "step": 348500
+    },
+    {
+      "epoch": 2.2693283048312636,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0019092268678067494,
+      "loss": 1.5241,
+      "step": 349000
+    },
+    {
+      "epoch": 2.272579491514403,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0019090968203394239,
+      "loss": 1.5222,
+      "step": 349500
+    },
+    {
+      "epoch": 2.275830678197542,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0019089667728720984,
+      "loss": 1.5208,
+      "step": 350000
+    },
+    {
+      "epoch": 2.2790818648806814,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0019088367254047726,
+      "loss": 1.5276,
+      "step": 350500
+    },
+    {
+      "epoch": 2.2823330515638207,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.001908706677937447,
+      "loss": 1.5304,
+      "step": 351000
+    },
+    {
+      "epoch": 2.28558423824696,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0019085766304701216,
+      "loss": 1.5276,
+      "step": 351500
+    },
+    {
+      "epoch": 2.2888354249300997,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0019084465830027963,
+      "loss": 1.5292,
+      "step": 352000
+    },
+    {
+      "epoch": 2.292086611613239,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0019083165355354705,
+      "loss": 1.5298,
+      "step": 352500
+    },
+    {
+      "epoch": 2.2953377982963783,
+      "grad_norm": 3.234375,
+      "learning_rate": 0.001908186488068145,
+      "loss": 1.5257,
+      "step": 353000
+    },
+    {
+      "epoch": 2.2985889849795176,
+      "grad_norm": 3.953125,
+      "learning_rate": 0.0019080564406008195,
+      "loss": 1.5232,
+      "step": 353500
+    },
+    {
+      "epoch": 2.301840171662657,
+      "grad_norm": 14.6875,
+      "learning_rate": 0.0019079263931334938,
+      "loss": 1.519,
+      "step": 354000
+    },
+    {
+      "epoch": 2.305091358345796,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0019077963456661682,
+      "loss": 1.5199,
+      "step": 354500
+    },
+    {
+      "epoch": 2.3083425450289354,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0019076662981988427,
+      "loss": 1.5216,
+      "step": 355000
+    },
+    {
+      "epoch": 2.3115937317120747,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.001907536250731517,
+      "loss": 1.522,
+      "step": 355500
+    },
+    {
+      "epoch": 2.3148449183952144,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.0019074062032641914,
+      "loss": 1.5132,
+      "step": 356000
+    },
+    {
+      "epoch": 2.3180961050783537,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.001907276155796866,
+      "loss": 1.5111,
+      "step": 356500
+    },
+    {
+      "epoch": 2.321347291761493,
+      "grad_norm": 2.78125,
+      "learning_rate": 0.0019071461083295402,
+      "loss": 1.5108,
+      "step": 357000
+    },
+    {
+      "epoch": 2.3245984784446323,
+      "grad_norm": 3.3125,
+      "learning_rate": 0.0019070160608622147,
+      "loss": 1.507,
+      "step": 357500
+    },
+    {
+      "epoch": 2.3278496651277716,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0019068860133948891,
+      "loss": 1.5052,
+      "step": 358000
+    },
+    {
+      "epoch": 2.331100851810911,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0019067559659275634,
+      "loss": 1.5098,
+      "step": 358500
+    },
+    {
+      "epoch": 2.33435203849405,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0019066259184602379,
+      "loss": 1.5149,
+      "step": 359000
+    },
+    {
+      "epoch": 2.3376032251771894,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0019064958709929126,
+      "loss": 1.5081,
+      "step": 359500
+    },
+    {
+      "epoch": 2.340854411860329,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.001906365823525587,
+      "loss": 1.5098,
+      "step": 360000
+    },
+    {
+      "epoch": 2.3441055985434684,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0019062357760582613,
+      "loss": 1.5113,
+      "step": 360500
+    },
+    {
+      "epoch": 2.3473567852266077,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.0019061057285909358,
+      "loss": 1.5087,
+      "step": 361000
+    },
+    {
+      "epoch": 2.350607971909747,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0019059756811236103,
+      "loss": 1.5004,
+      "step": 361500
+    },
+    {
+      "epoch": 2.3538591585928863,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0019058456336562845,
+      "loss": 1.5032,
+      "step": 362000
+    },
+    {
+      "epoch": 2.3571103452760256,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001905715586188959,
+      "loss": 1.5072,
+      "step": 362500
+    },
+    {
+      "epoch": 2.360361531959165,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0019055855387216335,
+      "loss": 1.5085,
+      "step": 363000
+    },
+    {
+      "epoch": 2.3636127186423046,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0019054554912543078,
+      "loss": 1.5041,
+      "step": 363500
+    },
+    {
+      "epoch": 2.366863905325444,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0019053254437869822,
+      "loss": 1.5088,
+      "step": 364000
+    },
+    {
+      "epoch": 2.370115092008583,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0019051953963196567,
+      "loss": 1.5057,
+      "step": 364500
+    },
+    {
+      "epoch": 2.3733662786917225,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001905065348852331,
+      "loss": 1.5074,
+      "step": 365000
+    },
+    {
+      "epoch": 2.3766174653748617,
+      "grad_norm": 7.75,
+      "learning_rate": 0.0019049353013850055,
+      "loss": 1.5067,
+      "step": 365500
+    },
+    {
+      "epoch": 2.379868652058001,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00190480525391768,
+      "loss": 1.5077,
+      "step": 366000
+    },
+    {
+      "epoch": 2.3831198387411403,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0019046752064503546,
+      "loss": 1.5026,
+      "step": 366500
+    },
+    {
+      "epoch": 2.38637102542428,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.0019045451589830289,
+      "loss": 1.5061,
+      "step": 367000
+    },
+    {
+      "epoch": 2.3896222121074193,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0019044151115157034,
+      "loss": 1.5035,
+      "step": 367500
+    },
+    {
+      "epoch": 2.3928733987905586,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0019042850640483778,
+      "loss": 1.5019,
+      "step": 368000
+    },
+    {
+      "epoch": 2.396124585473698,
+      "grad_norm": 1.5,
+      "learning_rate": 0.0019041550165810521,
+      "loss": 1.4978,
+      "step": 368500
+    },
+    {
+      "epoch": 2.399375772156837,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0019040249691137266,
+      "loss": 1.4969,
+      "step": 369000
+    },
+    {
+      "epoch": 2.4026269588399765,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.001903894921646401,
+      "loss": 1.4945,
+      "step": 369500
+    },
+    {
+      "epoch": 2.4058781455231157,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0019037648741790753,
+      "loss": 1.4858,
+      "step": 370000
+    },
+    {
+      "epoch": 2.4091293322062555,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.0019036348267117498,
+      "loss": 1.4978,
+      "step": 370500
+    },
+    {
+      "epoch": 2.4123805188893948,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.0019035047792444243,
+      "loss": 1.491,
+      "step": 371000
+    },
+    {
+      "epoch": 2.415631705572534,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.0019033747317770985,
+      "loss": 1.498,
+      "step": 371500
+    },
+    {
+      "epoch": 2.4188828922556733,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.001903244684309773,
+      "loss": 1.4916,
+      "step": 372000
+    },
+    {
+      "epoch": 2.4221340789388126,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0019031146368424475,
+      "loss": 1.4962,
+      "step": 372500
+    },
+    {
+      "epoch": 2.425385265621952,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0019029845893751218,
+      "loss": 1.4885,
+      "step": 373000
+    },
+    {
+      "epoch": 2.428636452305091,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.0019028545419077962,
+      "loss": 1.4952,
+      "step": 373500
+    },
+    {
+      "epoch": 2.431887638988231,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.001902724494440471,
+      "loss": 1.4921,
+      "step": 374000
+    },
+    {
+      "epoch": 2.43513882567137,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0019025944469731454,
+      "loss": 1.4982,
+      "step": 374500
+    },
+    {
+      "epoch": 2.4383900123545095,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0019024643995058197,
+      "loss": 1.5005,
+      "step": 375000
+    },
+    {
+      "epoch": 2.4416411990376488,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0019023343520384942,
+      "loss": 1.4998,
+      "step": 375500
+    },
+    {
+      "epoch": 2.444892385720788,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0019022043045711686,
+      "loss": 1.4939,
+      "step": 376000
+    },
+    {
+      "epoch": 2.4481435724039273,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.001902074257103843,
+      "loss": 1.4971,
+      "step": 376500
+    },
+    {
+      "epoch": 2.4513947590870666,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0019019442096365174,
+      "loss": 1.4931,
+      "step": 377000
+    },
+    {
+      "epoch": 2.4546459457702063,
+      "grad_norm": 2.890625,
+      "learning_rate": 0.0019018141621691919,
+      "loss": 1.4884,
+      "step": 377500
+    },
+    {
+      "epoch": 2.4578971324533456,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.0019016841147018661,
+      "loss": 1.4982,
+      "step": 378000
+    },
+    {
+      "epoch": 2.461148319136485,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0019015540672345406,
+      "loss": 1.496,
+      "step": 378500
+    },
+    {
+      "epoch": 2.464399505819624,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001901424019767215,
+      "loss": 1.4937,
+      "step": 379000
+    },
+    {
+      "epoch": 2.4676506925027635,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0019012939722998893,
+      "loss": 1.4894,
+      "step": 379500
+    },
+    {
+      "epoch": 2.4709018791859028,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0019011639248325638,
+      "loss": 1.4896,
+      "step": 380000
+    },
+    {
+      "epoch": 2.474153065869042,
+      "grad_norm": 3.859375,
+      "learning_rate": 0.0019010338773652383,
+      "loss": 1.492,
+      "step": 380500
+    },
+    {
+      "epoch": 2.4774042525521818,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.001900903829897913,
+      "loss": 1.499,
+      "step": 381000
+    },
+    {
+      "epoch": 2.480655439235321,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0019007737824305873,
+      "loss": 1.4932,
+      "step": 381500
+    },
+    {
+      "epoch": 2.4839066259184603,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0019006437349632617,
+      "loss": 1.4962,
+      "step": 382000
+    },
+    {
+      "epoch": 2.4871578126015996,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0019005136874959362,
+      "loss": 1.4942,
+      "step": 382500
+    },
+    {
+      "epoch": 2.490408999284739,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0019003836400286105,
+      "loss": 1.487,
+      "step": 383000
+    },
+    {
+      "epoch": 2.493660185967878,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001900253592561285,
+      "loss": 1.4892,
+      "step": 383500
+    },
+    {
+      "epoch": 2.4969113726510175,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.0019001235450939594,
+      "loss": 1.4959,
+      "step": 384000
+    },
+    {
+      "epoch": 2.500162559334157,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0018999934976266337,
+      "loss": 1.4904,
+      "step": 384500
+    },
+    {
+      "epoch": 2.503413746017296,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0018998634501593082,
+      "loss": 1.4926,
+      "step": 385000
+    },
+    {
+      "epoch": 2.5066649327004358,
+      "grad_norm": 2.390625,
+      "learning_rate": 0.0018997334026919826,
+      "loss": 1.4885,
+      "step": 385500
+    },
+    {
+      "epoch": 2.509916119383575,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001899603355224657,
+      "loss": 1.489,
+      "step": 386000
+    },
+    {
+      "epoch": 2.5131673060667143,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0018994733077573314,
+      "loss": 1.4856,
+      "step": 386500
+    },
+    {
+      "epoch": 2.5164184927498536,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.0018993432602900059,
+      "loss": 1.4854,
+      "step": 387000
+    },
+    {
+      "epoch": 2.519669679432993,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0018992132128226801,
+      "loss": 1.487,
+      "step": 387500
+    },
+    {
+      "epoch": 2.5229208661161326,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0018990831653553546,
+      "loss": 1.4987,
+      "step": 388000
+    },
+    {
+      "epoch": 2.5261720527992715,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0018989531178880293,
+      "loss": 1.4923,
+      "step": 388500
+    },
+    {
+      "epoch": 2.529423239482411,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0018988230704207038,
+      "loss": 1.4905,
+      "step": 389000
+    },
+    {
+      "epoch": 2.5326744261655505,
+      "grad_norm": 4.875,
+      "learning_rate": 0.001898693022953378,
+      "loss": 1.4929,
+      "step": 389500
+    },
+    {
+      "epoch": 2.53592561284869,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0018985629754860525,
+      "loss": 1.4889,
+      "step": 390000
+    },
+    {
+      "epoch": 2.539176799531829,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.001898432928018727,
+      "loss": 1.4915,
+      "step": 390500
+    },
+    {
+      "epoch": 2.5424279862149683,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0018983028805514013,
+      "loss": 1.484,
+      "step": 391000
+    },
+    {
+      "epoch": 2.545679172898108,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0018981728330840757,
+      "loss": 1.4877,
+      "step": 391500
+    },
+    {
+      "epoch": 2.548930359581247,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018980427856167502,
+      "loss": 1.4979,
+      "step": 392000
+    },
+    {
+      "epoch": 2.5521815462643866,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018979127381494245,
+      "loss": 1.4893,
+      "step": 392500
+    },
+    {
+      "epoch": 2.555432732947526,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.001897782690682099,
+      "loss": 1.4904,
+      "step": 393000
+    },
+    {
+      "epoch": 2.558683919630665,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0018976526432147734,
+      "loss": 1.4982,
+      "step": 393500
+    },
+    {
+      "epoch": 2.5619351063138045,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.0018975225957474477,
+      "loss": 1.4952,
+      "step": 394000
+    },
+    {
+      "epoch": 2.565186292996944,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0018973925482801222,
+      "loss": 1.4959,
+      "step": 394500
+    },
+    {
+      "epoch": 2.568437479680083,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0018972625008127967,
+      "loss": 1.488,
+      "step": 395000
+    },
+    {
+      "epoch": 2.5716886663632224,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.0018971324533454713,
+      "loss": 1.491,
+      "step": 395500
+    },
+    {
+      "epoch": 2.574939853046362,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.0018970024058781456,
+      "loss": 1.4935,
+      "step": 396000
+    },
+    {
+      "epoch": 2.5781910397295014,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00189687235841082,
+      "loss": 1.4867,
+      "step": 396500
+    },
+    {
+      "epoch": 2.5814422264126407,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0018967423109434946,
+      "loss": 1.4899,
+      "step": 397000
+    },
+    {
+      "epoch": 2.58469341309578,
+      "grad_norm": 2.890625,
+      "learning_rate": 0.0018966122634761688,
+      "loss": 1.4949,
+      "step": 397500
+    },
+    {
+      "epoch": 2.587944599778919,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0018964822160088433,
+      "loss": 1.4928,
+      "step": 398000
+    },
+    {
+      "epoch": 2.5911957864620585,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0018963521685415178,
+      "loss": 1.4994,
+      "step": 398500
+    },
+    {
+      "epoch": 2.594446973145198,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.001896222121074192,
+      "loss": 1.4884,
+      "step": 399000
+    },
+    {
+      "epoch": 2.5976981598283375,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0018960920736068665,
+      "loss": 1.479,
+      "step": 399500
+    },
+    {
+      "epoch": 2.600949346511477,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.001895962026139541,
+      "loss": 1.4862,
+      "step": 400000
+    },
+    {
+      "epoch": 2.604200533194616,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0018958319786722153,
+      "loss": 1.4912,
+      "step": 400500
+    },
+    {
+      "epoch": 2.6074517198777554,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0018957019312048897,
+      "loss": 1.4863,
+      "step": 401000
+    },
+    {
+      "epoch": 2.6107029065608947,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0018955718837375642,
+      "loss": 1.4892,
+      "step": 401500
+    },
+    {
+      "epoch": 2.613954093244034,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0018954418362702385,
+      "loss": 1.4843,
+      "step": 402000
+    },
+    {
+      "epoch": 2.6172052799271732,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.001895311788802913,
+      "loss": 1.4803,
+      "step": 402500
+    },
+    {
+      "epoch": 2.620456466610313,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.0018951817413355877,
+      "loss": 1.4781,
+      "step": 403000
+    },
+    {
+      "epoch": 2.6237076532934522,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0018950516938682621,
+      "loss": 1.484,
+      "step": 403500
+    },
+    {
+      "epoch": 2.6269588399765915,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0018949216464009364,
+      "loss": 1.4841,
+      "step": 404000
+    },
+    {
+      "epoch": 2.630210026659731,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0018947915989336109,
+      "loss": 1.4807,
+      "step": 404500
+    },
+    {
+      "epoch": 2.63346121334287,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0018946615514662854,
+      "loss": 1.4828,
+      "step": 405000
+    },
+    {
+      "epoch": 2.6367124000260094,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0018945315039989596,
+      "loss": 1.4817,
+      "step": 405500
+    },
+    {
+      "epoch": 2.6399635867091487,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.001894401456531634,
+      "loss": 1.4844,
+      "step": 406000
+    },
+    {
+      "epoch": 2.6432147733922884,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0018942714090643086,
+      "loss": 1.4832,
+      "step": 406500
+    },
+    {
+      "epoch": 2.6464659600754277,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018941413615969828,
+      "loss": 1.4784,
+      "step": 407000
+    },
+    {
+      "epoch": 2.649717146758567,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0018940113141296573,
+      "loss": 1.4795,
+      "step": 407500
+    },
+    {
+      "epoch": 2.6529683334417062,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0018938812666623318,
+      "loss": 1.4795,
+      "step": 408000
+    },
+    {
+      "epoch": 2.6562195201248455,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.001893751219195006,
+      "loss": 1.4835,
+      "step": 408500
+    },
+    {
+      "epoch": 2.659470706807985,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0018936211717276805,
+      "loss": 1.4775,
+      "step": 409000
+    },
+    {
+      "epoch": 2.662721893491124,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.001893491124260355,
+      "loss": 1.4808,
+      "step": 409500
+    },
+    {
+      "epoch": 2.665973080174264,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0018933610767930297,
+      "loss": 1.4789,
+      "step": 410000
+    },
+    {
+      "epoch": 2.669224266857403,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.001893231029325704,
+      "loss": 1.4741,
+      "step": 410500
+    },
+    {
+      "epoch": 2.6724754535405424,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0018931009818583784,
+      "loss": 1.4803,
+      "step": 411000
+    },
+    {
+      "epoch": 2.6757266402236817,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001892970934391053,
+      "loss": 1.4856,
+      "step": 411500
+    },
+    {
+      "epoch": 2.678977826906821,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.0018928408869237272,
+      "loss": 1.4796,
+      "step": 412000
+    },
+    {
+      "epoch": 2.6822290135899602,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0018927108394564017,
+      "loss": 1.4844,
+      "step": 412500
+    },
+    {
+      "epoch": 2.6854802002730995,
+      "grad_norm": 3.4375,
+      "learning_rate": 0.0018925807919890761,
+      "loss": 1.4866,
+      "step": 413000
+    },
+    {
+      "epoch": 2.6887313869562393,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0018924507445217504,
+      "loss": 1.496,
+      "step": 413500
+    },
+    {
+      "epoch": 2.691982573639378,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0018923206970544249,
+      "loss": 1.4882,
+      "step": 414000
+    },
+    {
+      "epoch": 2.695233760322518,
+      "grad_norm": 2.734375,
+      "learning_rate": 0.0018921906495870994,
+      "loss": 1.4902,
+      "step": 414500
+    },
+    {
+      "epoch": 2.698484947005657,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0018920606021197736,
+      "loss": 1.4878,
+      "step": 415000
+    },
+    {
+      "epoch": 2.7017361336887964,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001891930554652448,
+      "loss": 1.4858,
+      "step": 415500
+    },
+    {
+      "epoch": 2.7049873203719357,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0018918005071851226,
+      "loss": 1.4857,
+      "step": 416000
+    },
+    {
+      "epoch": 2.708238507055075,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0018916704597177968,
+      "loss": 1.4811,
+      "step": 416500
+    },
+    {
+      "epoch": 2.7114896937382147,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0018915404122504713,
+      "loss": 1.481,
+      "step": 417000
+    },
+    {
+      "epoch": 2.7147408804213535,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001891410364783146,
+      "loss": 1.4776,
+      "step": 417500
+    },
+    {
+      "epoch": 2.7179920671044933,
+      "grad_norm": 15.0625,
+      "learning_rate": 0.0018912803173158205,
+      "loss": 1.4783,
+      "step": 418000
+    },
+    {
+      "epoch": 2.7212432537876325,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0018911502698484948,
+      "loss": 1.4751,
+      "step": 418500
+    },
+    {
+      "epoch": 2.724494440470772,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.0018910202223811692,
+      "loss": 1.4733,
+      "step": 419000
+    },
+    {
+      "epoch": 2.727745627153911,
+      "grad_norm": 1.8359375,
+      "learning_rate": 0.0018908901749138437,
+      "loss": 1.4783,
+      "step": 419500
+    },
+    {
+      "epoch": 2.7309968138370504,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.001890760127446518,
+      "loss": 1.4742,
+      "step": 420000
+    },
+    {
+      "epoch": 2.73424800052019,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018906300799791925,
+      "loss": 1.4728,
+      "step": 420500
+    },
+    {
+      "epoch": 2.737499187203329,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001890500032511867,
+      "loss": 1.4777,
+      "step": 421000
+    },
+    {
+      "epoch": 2.7407503738864687,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0018903699850445412,
+      "loss": 1.4702,
+      "step": 421500
+    },
+    {
+      "epoch": 2.744001560569608,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0018902399375772157,
+      "loss": 1.474,
+      "step": 422000
+    },
+    {
+      "epoch": 2.7472527472527473,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0018901098901098902,
+      "loss": 1.4753,
+      "step": 422500
+    },
+    {
+      "epoch": 2.7505039339358865,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0018899798426425644,
+      "loss": 1.4768,
+      "step": 423000
+    },
+    {
+      "epoch": 2.753755120619026,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.001889849795175239,
+      "loss": 1.4777,
+      "step": 423500
+    },
+    {
+      "epoch": 2.7570063073021656,
+      "grad_norm": 1.921875,
+      "learning_rate": 0.0018897197477079134,
+      "loss": 1.4726,
+      "step": 424000
+    },
+    {
+      "epoch": 2.7602574939853044,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.001889589700240588,
+      "loss": 1.477,
+      "step": 424500
+    },
+    {
+      "epoch": 2.763508680668444,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0018894596527732623,
+      "loss": 1.4721,
+      "step": 425000
+    },
+    {
+      "epoch": 2.7667598673515834,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0018893296053059368,
+      "loss": 1.4787,
+      "step": 425500
+    },
+    {
+      "epoch": 2.7700110540347227,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0018891995578386113,
+      "loss": 1.4727,
+      "step": 426000
+    },
+    {
+      "epoch": 2.773262240717862,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0018890695103712855,
+      "loss": 1.4694,
+      "step": 426500
+    },
+    {
+      "epoch": 2.7765134274010013,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00188893946290396,
+      "loss": 1.476,
+      "step": 427000
+    },
+    {
+      "epoch": 2.7797646140841406,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0018888094154366345,
+      "loss": 1.4783,
+      "step": 427500
+    },
+    {
+      "epoch": 2.78301580076728,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0018886793679693088,
+      "loss": 1.4707,
+      "step": 428000
+    },
+    {
+      "epoch": 2.7862669874504196,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0018885493205019832,
+      "loss": 1.4717,
+      "step": 428500
+    },
+    {
+      "epoch": 2.789518174133559,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0018884192730346577,
+      "loss": 1.4716,
+      "step": 429000
+    },
+    {
+      "epoch": 2.792769360816698,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001888289225567332,
+      "loss": 1.4667,
+      "step": 429500
+    },
+    {
+      "epoch": 2.7960205474998374,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0018881591781000065,
+      "loss": 1.4683,
+      "step": 430000
+    },
+    {
+      "epoch": 2.7992717341829767,
+      "grad_norm": 12.0625,
+      "learning_rate": 0.001888029130632681,
+      "loss": 1.4719,
+      "step": 430500
+    },
+    {
+      "epoch": 2.802522920866116,
+      "grad_norm": 1.9296875,
+      "learning_rate": 0.0018878990831653552,
+      "loss": 1.4711,
+      "step": 431000
+    },
+    {
+      "epoch": 2.8057741075492553,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.0018877690356980297,
+      "loss": 1.4727,
+      "step": 431500
+    },
+    {
+      "epoch": 2.809025294232395,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0018876389882307044,
+      "loss": 1.472,
+      "step": 432000
+    },
+    {
+      "epoch": 2.8122764809155343,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0018875089407633789,
+      "loss": 1.4796,
+      "step": 432500
+    },
+    {
+      "epoch": 2.8155276675986736,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.0018873788932960531,
+      "loss": 1.4719,
+      "step": 433000
+    },
+    {
+      "epoch": 2.818778854281813,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0018872488458287276,
+      "loss": 1.4764,
+      "step": 433500
+    },
+    {
+      "epoch": 2.822030040964952,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.001887118798361402,
+      "loss": 1.4712,
+      "step": 434000
+    },
+    {
+      "epoch": 2.8252812276480914,
+      "grad_norm": 12.875,
+      "learning_rate": 0.0018869887508940763,
+      "loss": 1.4664,
+      "step": 434500
+    },
+    {
+      "epoch": 2.8285324143312307,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0018868587034267508,
+      "loss": 1.4708,
+      "step": 435000
+    },
+    {
+      "epoch": 2.8317836010143704,
+      "grad_norm": 3.203125,
+      "learning_rate": 0.0018867286559594253,
+      "loss": 1.4718,
+      "step": 435500
+    },
+    {
+      "epoch": 2.8350347876975097,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0018865986084920996,
+      "loss": 1.4719,
+      "step": 436000
+    },
+    {
+      "epoch": 2.838285974380649,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.001886468561024774,
+      "loss": 1.4692,
+      "step": 436500
+    },
+    {
+      "epoch": 2.8415371610637883,
+      "grad_norm": 2.953125,
+      "learning_rate": 0.0018863385135574485,
+      "loss": 1.4735,
+      "step": 437000
+    },
+    {
+      "epoch": 2.8447883477469276,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0018862084660901228,
+      "loss": 1.4724,
+      "step": 437500
+    },
+    {
+      "epoch": 2.848039534430067,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0018860784186227973,
+      "loss": 1.4687,
+      "step": 438000
+    },
+    {
+      "epoch": 2.851290721113206,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0018859483711554717,
+      "loss": 1.4734,
+      "step": 438500
+    },
+    {
+      "epoch": 2.854541907796346,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0018858183236881464,
+      "loss": 1.4705,
+      "step": 439000
+    },
+    {
+      "epoch": 2.857793094479485,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0018856882762208207,
+      "loss": 1.4702,
+      "step": 439500
+    },
+    {
+      "epoch": 2.8610442811626244,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0018855582287534952,
+      "loss": 1.4725,
+      "step": 440000
+    },
+    {
+      "epoch": 2.8642954678457637,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0018854281812861696,
+      "loss": 1.4784,
+      "step": 440500
+    },
+    {
+      "epoch": 2.867546654528903,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.001885298133818844,
+      "loss": 1.4737,
+      "step": 441000
+    },
+    {
+      "epoch": 2.8707978412120423,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0018851680863515184,
+      "loss": 1.4762,
+      "step": 441500
+    },
+    {
+      "epoch": 2.8740490278951816,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0018850380388841929,
+      "loss": 1.4729,
+      "step": 442000
+    },
+    {
+      "epoch": 2.8773002145783213,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0018849079914168671,
+      "loss": 1.4756,
+      "step": 442500
+    },
+    {
+      "epoch": 2.8805514012614606,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0018847779439495416,
+      "loss": 1.4757,
+      "step": 443000
+    },
+    {
+      "epoch": 2.8838025879446,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001884647896482216,
+      "loss": 1.4769,
+      "step": 443500
+    },
+    {
+      "epoch": 2.887053774627739,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0018845178490148903,
+      "loss": 1.4805,
+      "step": 444000
+    },
+    {
+      "epoch": 2.8903049613108784,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0018843878015475648,
+      "loss": 1.4756,
+      "step": 444500
+    },
+    {
+      "epoch": 2.8935561479940177,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0018842577540802393,
+      "loss": 1.4753,
+      "step": 445000
+    },
+    {
+      "epoch": 2.896807334677157,
+      "grad_norm": 2.875,
+      "learning_rate": 0.0018841277066129136,
+      "loss": 1.474,
+      "step": 445500
+    },
+    {
+      "epoch": 2.9000585213602967,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.001883997659145588,
+      "loss": 1.4762,
+      "step": 446000
+    },
+    {
+      "epoch": 2.9033097080434356,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0018838676116782627,
+      "loss": 1.4703,
+      "step": 446500
+    },
+    {
+      "epoch": 2.9065608947265753,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0018837375642109372,
+      "loss": 1.4681,
+      "step": 447000
+    },
+    {
+      "epoch": 2.9098120814097146,
+      "grad_norm": 2.6875,
+      "learning_rate": 0.0018836075167436115,
+      "loss": 1.4768,
+      "step": 447500
+    },
+    {
+      "epoch": 2.913063268092854,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.001883477469276286,
+      "loss": 1.4705,
+      "step": 448000
+    },
+    {
+      "epoch": 2.916314454775993,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.0018833474218089604,
+      "loss": 1.4775,
+      "step": 448500
+    },
+    {
+      "epoch": 2.9195656414591324,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0018832173743416347,
+      "loss": 1.4738,
+      "step": 449000
+    },
+    {
+      "epoch": 2.922816828142272,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0018830873268743092,
+      "loss": 1.4776,
+      "step": 449500
+    },
+    {
+      "epoch": 2.926068014825411,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0018829572794069837,
+      "loss": 1.4748,
+      "step": 450000
+    },
+    {
+      "epoch": 2.9293192015085507,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.001882827231939658,
+      "loss": 1.4722,
+      "step": 450500
+    },
+    {
+      "epoch": 2.93257038819169,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0018826971844723324,
+      "loss": 1.476,
+      "step": 451000
+    },
+    {
+      "epoch": 2.9358215748748293,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0018825671370050069,
+      "loss": 1.4676,
+      "step": 451500
+    },
+    {
+      "epoch": 2.9390727615579686,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0018824370895376811,
+      "loss": 1.4714,
+      "step": 452000
+    },
+    {
+      "epoch": 2.942323948241108,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0018823070420703556,
+      "loss": 1.468,
+      "step": 452500
+    },
+    {
+      "epoch": 2.9455751349242476,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.00188217699460303,
+      "loss": 1.472,
+      "step": 453000
+    },
+    {
+      "epoch": 2.9488263216073864,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0018820469471357048,
+      "loss": 1.4727,
+      "step": 453500
+    },
+    {
+      "epoch": 2.952077508290526,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.001881916899668379,
+      "loss": 1.4722,
+      "step": 454000
+    },
+    {
+      "epoch": 2.9553286949736655,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0018817868522010535,
+      "loss": 1.4691,
+      "step": 454500
+    },
+    {
+      "epoch": 2.9585798816568047,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001881656804733728,
+      "loss": 1.4649,
+      "step": 455000
+    },
+    {
+      "epoch": 2.961831068339944,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0018815267572664023,
+      "loss": 1.4633,
+      "step": 455500
+    },
+    {
+      "epoch": 2.9650822550230833,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0018813967097990767,
+      "loss": 1.461,
+      "step": 456000
+    },
+    {
+      "epoch": 2.968333441706223,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0018812666623317512,
+      "loss": 1.4665,
+      "step": 456500
+    },
+    {
+      "epoch": 2.971584628389362,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0018811366148644255,
+      "loss": 1.4658,
+      "step": 457000
+    },
+    {
+      "epoch": 2.9748358150725016,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0018810065673971,
+      "loss": 1.4662,
+      "step": 457500
+    },
+    {
+      "epoch": 2.978087001755641,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0018808765199297744,
+      "loss": 1.4632,
+      "step": 458000
+    },
+    {
+      "epoch": 2.98133818843878,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0018807464724624487,
+      "loss": 1.4701,
+      "step": 458500
+    },
+    {
+      "epoch": 2.9845893751219195,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0018806164249951232,
+      "loss": 1.4664,
+      "step": 459000
+    },
+    {
+      "epoch": 2.9878405618050587,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0018804863775277977,
+      "loss": 1.4679,
+      "step": 459500
+    },
+    {
+      "epoch": 2.991091748488198,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.001880356330060472,
+      "loss": 1.4674,
+      "step": 460000
+    },
+    {
+      "epoch": 2.9943429351713373,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.0018802262825931464,
+      "loss": 1.4695,
+      "step": 460500
+    },
+    {
+      "epoch": 2.997594121854477,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.001880096235125821,
+      "loss": 1.464,
+      "step": 461000
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 1.4555257558822632,
+      "eval_runtime": 0.5334,
+      "eval_samples_per_second": 1874.801,
+      "eval_steps_per_second": 29.997,
+      "step": 461370
+    },
+    {
+      "epoch": 3.0008453085376163,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0018799661876584956,
+      "loss": 1.4686,
+      "step": 461500
+    },
+    {
+      "epoch": 3.0040964952207556,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0018798361401911698,
+      "loss": 1.4627,
+      "step": 462000
+    },
+    {
+      "epoch": 3.007347681903895,
+      "grad_norm": 2.734375,
+      "learning_rate": 0.0018797060927238443,
+      "loss": 1.4682,
+      "step": 462500
+    },
+    {
+      "epoch": 3.010598868587034,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0018795760452565188,
+      "loss": 1.4645,
+      "step": 463000
+    },
+    {
+      "epoch": 3.0138500552701735,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001879445997789193,
+      "loss": 1.4582,
+      "step": 463500
+    },
+    {
+      "epoch": 3.0171012419533128,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0018793159503218675,
+      "loss": 1.462,
+      "step": 464000
+    },
+    {
+      "epoch": 3.0203524286364525,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.001879185902854542,
+      "loss": 1.4655,
+      "step": 464500
+    },
+    {
+      "epoch": 3.0236036153195918,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0018790558553872163,
+      "loss": 1.4624,
+      "step": 465000
+    },
+    {
+      "epoch": 3.026854802002731,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0018789258079198908,
+      "loss": 1.4621,
+      "step": 465500
+    },
+    {
+      "epoch": 3.0301059886858703,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0018787957604525652,
+      "loss": 1.4682,
+      "step": 466000
+    },
+    {
+      "epoch": 3.0333571753690096,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0018786657129852395,
+      "loss": 1.4636,
+      "step": 466500
+    },
+    {
+      "epoch": 3.036608362052149,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001878535665517914,
+      "loss": 1.4661,
+      "step": 467000
+    },
+    {
+      "epoch": 3.039859548735288,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0018784056180505884,
+      "loss": 1.4604,
+      "step": 467500
+    },
+    {
+      "epoch": 3.043110735418428,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018782755705832631,
+      "loss": 1.4597,
+      "step": 468000
+    },
+    {
+      "epoch": 3.046361922101567,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018781455231159374,
+      "loss": 1.4634,
+      "step": 468500
+    },
+    {
+      "epoch": 3.0496131087847065,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0018780154756486119,
+      "loss": 1.4617,
+      "step": 469000
+    },
+    {
+      "epoch": 3.0528642954678458,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018778854281812864,
+      "loss": 1.4653,
+      "step": 469500
+    },
+    {
+      "epoch": 3.056115482150985,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0018777553807139606,
+      "loss": 1.4659,
+      "step": 470000
+    },
+    {
+      "epoch": 3.0593666688341243,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.001877625333246635,
+      "loss": 1.4602,
+      "step": 470500
+    },
+    {
+      "epoch": 3.0626178555172636,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0018774952857793096,
+      "loss": 1.4627,
+      "step": 471000
+    },
+    {
+      "epoch": 3.0658690422004033,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0018773652383119838,
+      "loss": 1.4682,
+      "step": 471500
+    },
+    {
+      "epoch": 3.0691202288835426,
+      "grad_norm": 2.609375,
+      "learning_rate": 0.0018772351908446583,
+      "loss": 1.4656,
+      "step": 472000
+    },
+    {
+      "epoch": 3.072371415566682,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0018771051433773328,
+      "loss": 1.4598,
+      "step": 472500
+    },
+    {
+      "epoch": 3.075622602249821,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001876975095910007,
+      "loss": 1.4602,
+      "step": 473000
+    },
+    {
+      "epoch": 3.0788737889329605,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0018768450484426815,
+      "loss": 1.4621,
+      "step": 473500
+    },
+    {
+      "epoch": 3.0821249756160998,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001876715000975356,
+      "loss": 1.4628,
+      "step": 474000
+    },
+    {
+      "epoch": 3.085376162299239,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0018765849535080303,
+      "loss": 1.4626,
+      "step": 474500
+    },
+    {
+      "epoch": 3.088627348982379,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0018764549060407048,
+      "loss": 1.466,
+      "step": 475000
+    },
+    {
+      "epoch": 3.091878535665518,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0018763248585733795,
+      "loss": 1.462,
+      "step": 475500
+    },
+    {
+      "epoch": 3.0951297223486574,
+      "grad_norm": 7.71875,
+      "learning_rate": 0.001876194811106054,
+      "loss": 1.464,
+      "step": 476000
+    },
+    {
+      "epoch": 3.0983809090317966,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0018760647636387282,
+      "loss": 1.4624,
+      "step": 476500
+    },
+    {
+      "epoch": 3.101632095714936,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0018759347161714027,
+      "loss": 1.4656,
+      "step": 477000
+    },
+    {
+      "epoch": 3.104883282398075,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0018758046687040772,
+      "loss": 1.464,
+      "step": 477500
+    },
+    {
+      "epoch": 3.1081344690812145,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.0018756746212367514,
+      "loss": 1.461,
+      "step": 478000
+    },
+    {
+      "epoch": 3.1113856557643538,
+      "grad_norm": 2.328125,
+      "learning_rate": 0.0018755445737694259,
+      "loss": 1.4651,
+      "step": 478500
+    },
+    {
+      "epoch": 3.1146368424474935,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0018754145263021004,
+      "loss": 1.4645,
+      "step": 479000
+    },
+    {
+      "epoch": 3.117888029130633,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0018752844788347746,
+      "loss": 1.4721,
+      "step": 479500
+    },
+    {
+      "epoch": 3.121139215813772,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0018751544313674491,
+      "loss": 1.4656,
+      "step": 480000
+    },
+    {
+      "epoch": 3.1243904024969114,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0018750243839001236,
+      "loss": 1.464,
+      "step": 480500
+    },
+    {
+      "epoch": 3.1276415891800506,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0018748943364327978,
+      "loss": 1.4671,
+      "step": 481000
+    },
+    {
+      "epoch": 3.13089277586319,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.0018747642889654723,
+      "loss": 1.4645,
+      "step": 481500
+    },
+    {
+      "epoch": 3.134143962546329,
+      "grad_norm": 1.8359375,
+      "learning_rate": 0.0018746342414981468,
+      "loss": 1.4666,
+      "step": 482000
+    },
+    {
+      "epoch": 3.137395149229469,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0018745041940308215,
+      "loss": 1.4619,
+      "step": 482500
+    },
+    {
+      "epoch": 3.1406463359126082,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0018743741465634958,
+      "loss": 1.4618,
+      "step": 483000
+    },
+    {
+      "epoch": 3.1438975225957475,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0018742440990961702,
+      "loss": 1.4534,
+      "step": 483500
+    },
+    {
+      "epoch": 3.147148709278887,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0018741140516288447,
+      "loss": 1.4633,
+      "step": 484000
+    },
+    {
+      "epoch": 3.150399895962026,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001873984004161519,
+      "loss": 1.4596,
+      "step": 484500
+    },
+    {
+      "epoch": 3.1536510826451654,
+      "grad_norm": 1.9453125,
+      "learning_rate": 0.0018738539566941935,
+      "loss": 1.4652,
+      "step": 485000
+    },
+    {
+      "epoch": 3.1569022693283046,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.001873723909226868,
+      "loss": 1.4645,
+      "step": 485500
+    },
+    {
+      "epoch": 3.1601534560114444,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0018735938617595422,
+      "loss": 1.4604,
+      "step": 486000
+    },
+    {
+      "epoch": 3.1634046426945837,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0018734638142922167,
+      "loss": 1.4634,
+      "step": 486500
+    },
+    {
+      "epoch": 3.166655829377723,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0018733337668248912,
+      "loss": 1.4572,
+      "step": 487000
+    },
+    {
+      "epoch": 3.1699070160608622,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0018732037193575654,
+      "loss": 1.4651,
+      "step": 487500
+    },
+    {
+      "epoch": 3.1731582027440015,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.00187307367189024,
+      "loss": 1.465,
+      "step": 488000
+    },
+    {
+      "epoch": 3.176409389427141,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0018729436244229144,
+      "loss": 1.4643,
+      "step": 488500
+    },
+    {
+      "epoch": 3.17966057611028,
+      "grad_norm": 2.234375,
+      "learning_rate": 0.0018728135769555886,
+      "loss": 1.4622,
+      "step": 489000
+    },
+    {
+      "epoch": 3.1829117627934194,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0018726835294882631,
+      "loss": 1.4618,
+      "step": 489500
+    },
+    {
+      "epoch": 3.186162949476559,
+      "grad_norm": 4.8125,
+      "learning_rate": 0.0018725534820209378,
+      "loss": 1.4592,
+      "step": 490000
+    },
+    {
+      "epoch": 3.1894141361596984,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018724234345536123,
+      "loss": 1.4562,
+      "step": 490500
+    },
+    {
+      "epoch": 3.1926653228428377,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0018722933870862866,
+      "loss": 1.4643,
+      "step": 491000
+    },
+    {
+      "epoch": 3.195916509525977,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001872163339618961,
+      "loss": 1.4631,
+      "step": 491500
+    },
+    {
+      "epoch": 3.1991676962091162,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0018720332921516355,
+      "loss": 1.4638,
+      "step": 492000
+    },
+    {
+      "epoch": 3.2024188828922555,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0018719032446843098,
+      "loss": 1.4637,
+      "step": 492500
+    },
+    {
+      "epoch": 3.205670069575395,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0018717731972169842,
+      "loss": 1.4646,
+      "step": 493000
+    },
+    {
+      "epoch": 3.2089212562585345,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0018716431497496587,
+      "loss": 1.4647,
+      "step": 493500
+    },
+    {
+      "epoch": 3.212172442941674,
+      "grad_norm": 4.71875,
+      "learning_rate": 0.001871513102282333,
+      "loss": 1.4617,
+      "step": 494000
+    },
+    {
+      "epoch": 3.215423629624813,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0018713830548150075,
+      "loss": 1.4578,
+      "step": 494500
+    },
+    {
+      "epoch": 3.2186748163079524,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.001871253007347682,
+      "loss": 1.4573,
+      "step": 495000
+    },
+    {
+      "epoch": 3.2219260029910917,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0018711229598803562,
+      "loss": 1.4608,
+      "step": 495500
+    },
+    {
+      "epoch": 3.225177189674231,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0018709929124130307,
+      "loss": 1.4641,
+      "step": 496000
+    },
+    {
+      "epoch": 3.2284283763573702,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0018708628649457052,
+      "loss": 1.4638,
+      "step": 496500
+    },
+    {
+      "epoch": 3.23167956304051,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0018707328174783799,
+      "loss": 1.4615,
+      "step": 497000
+    },
+    {
+      "epoch": 3.2349307497236492,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0018706027700110541,
+      "loss": 1.4576,
+      "step": 497500
+    },
+    {
+      "epoch": 3.2381819364067885,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0018704727225437286,
+      "loss": 1.4655,
+      "step": 498000
+    },
+    {
+      "epoch": 3.241433123089928,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001870342675076403,
+      "loss": 1.4625,
+      "step": 498500
+    },
+    {
+      "epoch": 3.244684309773067,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0018702126276090773,
+      "loss": 1.4666,
+      "step": 499000
+    },
+    {
+      "epoch": 3.2479354964562064,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0018700825801417518,
+      "loss": 1.4697,
+      "step": 499500
+    },
+    {
+      "epoch": 3.2511866831393457,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0018699525326744263,
+      "loss": 1.4659,
+      "step": 500000
+    },
+    {
+      "epoch": 3.2544378698224854,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0018698224852071006,
+      "loss": 1.4683,
+      "step": 500500
+    },
+    {
+      "epoch": 3.2576890565056247,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001869692437739775,
+      "loss": 1.4658,
+      "step": 501000
+    },
+    {
+      "epoch": 3.260940243188764,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0018695623902724495,
+      "loss": 1.4642,
+      "step": 501500
+    },
+    {
+      "epoch": 3.2641914298719032,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0018694323428051238,
+      "loss": 1.4566,
+      "step": 502000
+    },
+    {
+      "epoch": 3.2674426165550425,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0018693022953377983,
+      "loss": 1.4618,
+      "step": 502500
+    },
+    {
+      "epoch": 3.270693803238182,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0018691722478704727,
+      "loss": 1.4692,
+      "step": 503000
+    },
+    {
+      "epoch": 3.273944989921321,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.001869042200403147,
+      "loss": 1.4591,
+      "step": 503500
+    },
+    {
+      "epoch": 3.277196176604461,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0018689121529358215,
+      "loss": 1.4591,
+      "step": 504000
+    },
+    {
+      "epoch": 3.2804473632876,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0018687821054684962,
+      "loss": 1.4616,
+      "step": 504500
+    },
+    {
+      "epoch": 3.2836985499707394,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0018686520580011706,
+      "loss": 1.4581,
+      "step": 505000
+    },
+    {
+      "epoch": 3.2869497366538787,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.001868522010533845,
+      "loss": 1.4645,
+      "step": 505500
+    },
+    {
+      "epoch": 3.290200923337018,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0018683919630665194,
+      "loss": 1.468,
+      "step": 506000
+    },
+    {
+      "epoch": 3.2934521100201573,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0018682619155991939,
+      "loss": 1.4623,
+      "step": 506500
+    },
+    {
+      "epoch": 3.2967032967032965,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.0018681318681318681,
+      "loss": 1.4576,
+      "step": 507000
+    },
+    {
+      "epoch": 3.2999544833864363,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0018680018206645426,
+      "loss": 1.4624,
+      "step": 507500
+    },
+    {
+      "epoch": 3.3032056700695756,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.001867871773197217,
+      "loss": 1.4646,
+      "step": 508000
+    },
+    {
+      "epoch": 3.306456856752715,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0018677417257298913,
+      "loss": 1.4571,
+      "step": 508500
+    },
+    {
+      "epoch": 3.309708043435854,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0018676116782625658,
+      "loss": 1.4602,
+      "step": 509000
+    },
+    {
+      "epoch": 3.3129592301189934,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0018674816307952403,
+      "loss": 1.4644,
+      "step": 509500
+    },
+    {
+      "epoch": 3.3162104168021327,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0018673515833279146,
+      "loss": 1.4602,
+      "step": 510000
+    },
+    {
+      "epoch": 3.319461603485272,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.001867221535860589,
+      "loss": 1.4605,
+      "step": 510500
+    },
+    {
+      "epoch": 3.3227127901684117,
+      "grad_norm": 8.0625,
+      "learning_rate": 0.0018670914883932635,
+      "loss": 1.4608,
+      "step": 511000
+    },
+    {
+      "epoch": 3.325963976851551,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0018669614409259382,
+      "loss": 1.4653,
+      "step": 511500
+    },
+    {
+      "epoch": 3.3292151635346903,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0018668313934586125,
+      "loss": 1.4617,
+      "step": 512000
+    },
+    {
+      "epoch": 3.3324663502178296,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.001866701345991287,
+      "loss": 1.4624,
+      "step": 512500
+    },
+    {
+      "epoch": 3.335717536900969,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0018665712985239614,
+      "loss": 1.4737,
+      "step": 513000
+    },
+    {
+      "epoch": 3.338968723584108,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0018664412510566357,
+      "loss": 1.4689,
+      "step": 513500
+    },
+    {
+      "epoch": 3.3422199102672474,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0018663112035893102,
+      "loss": 1.4679,
+      "step": 514000
+    },
+    {
+      "epoch": 3.345471096950387,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0018661811561219847,
+      "loss": 1.4666,
+      "step": 514500
+    },
+    {
+      "epoch": 3.3487222836335264,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001866051108654659,
+      "loss": 1.4707,
+      "step": 515000
+    },
+    {
+      "epoch": 3.3519734703166657,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0018659210611873334,
+      "loss": 1.4658,
+      "step": 515500
+    },
+    {
+      "epoch": 3.355224656999805,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0018657910137200079,
+      "loss": 1.4621,
+      "step": 516000
+    },
+    {
+      "epoch": 3.3584758436829443,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.0018656609662526821,
+      "loss": 1.4627,
+      "step": 516500
+    },
+    {
+      "epoch": 3.3617270303660836,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.0018655309187853566,
+      "loss": 1.466,
+      "step": 517000
+    },
+    {
+      "epoch": 3.364978217049223,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.001865400871318031,
+      "loss": 1.4665,
+      "step": 517500
+    },
+    {
+      "epoch": 3.368229403732362,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0018652708238507054,
+      "loss": 1.4685,
+      "step": 518000
+    },
+    {
+      "epoch": 3.3714805904155014,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.0018651407763833798,
+      "loss": 1.4629,
+      "step": 518500
+    },
+    {
+      "epoch": 3.374731777098641,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0018650107289160545,
+      "loss": 1.4591,
+      "step": 519000
+    },
+    {
+      "epoch": 3.3779829637817804,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.001864880681448729,
+      "loss": 1.4592,
+      "step": 519500
+    },
+    {
+      "epoch": 3.3812341504649197,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0018647506339814033,
+      "loss": 1.4595,
+      "step": 520000
+    },
+    {
+      "epoch": 3.384485337148059,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0018646205865140777,
+      "loss": 1.454,
+      "step": 520500
+    },
+    {
+      "epoch": 3.3877365238311983,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0018644905390467522,
+      "loss": 1.4567,
+      "step": 521000
+    },
+    {
+      "epoch": 3.3909877105143376,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0018643604915794265,
+      "loss": 1.4625,
+      "step": 521500
+    },
+    {
+      "epoch": 3.394238897197477,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.001864230444112101,
+      "loss": 1.4636,
+      "step": 522000
+    },
+    {
+      "epoch": 3.3974900838806166,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0018641003966447754,
+      "loss": 1.4584,
+      "step": 522500
+    },
+    {
+      "epoch": 3.400741270563756,
+      "grad_norm": 12.1875,
+      "learning_rate": 0.0018639703491774497,
+      "loss": 1.4628,
+      "step": 523000
+    },
+    {
+      "epoch": 3.403992457246895,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0018638403017101242,
+      "loss": 1.4558,
+      "step": 523500
+    },
+    {
+      "epoch": 3.4072436439300344,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0018637102542427987,
+      "loss": 1.4548,
+      "step": 524000
+    },
+    {
+      "epoch": 3.4104948306131737,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.001863580206775473,
+      "loss": 1.4562,
+      "step": 524500
+    },
+    {
+      "epoch": 3.413746017296313,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.0018634501593081474,
+      "loss": 1.4587,
+      "step": 525000
+    },
+    {
+      "epoch": 3.4169972039794523,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018633201118408219,
+      "loss": 1.4576,
+      "step": 525500
+    },
+    {
+      "epoch": 3.420248390662592,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0018631900643734966,
+      "loss": 1.4567,
+      "step": 526000
+    },
+    {
+      "epoch": 3.4234995773457313,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0018630600169061708,
+      "loss": 1.4547,
+      "step": 526500
+    },
+    {
+      "epoch": 3.4267507640288706,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0018629299694388453,
+      "loss": 1.4625,
+      "step": 527000
+    },
+    {
+      "epoch": 3.43000195071201,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0018627999219715198,
+      "loss": 1.4562,
+      "step": 527500
+    },
+    {
+      "epoch": 3.433253137395149,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.001862669874504194,
+      "loss": 1.4553,
+      "step": 528000
+    },
+    {
+      "epoch": 3.4365043240782884,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0018625398270368685,
+      "loss": 1.4575,
+      "step": 528500
+    },
+    {
+      "epoch": 3.4397555107614277,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001862409779569543,
+      "loss": 1.4557,
+      "step": 529000
+    },
+    {
+      "epoch": 3.4430066974445674,
+      "grad_norm": 7.375,
+      "learning_rate": 0.0018622797321022173,
+      "loss": 1.4513,
+      "step": 529500
+    },
+    {
+      "epoch": 3.4462578841277067,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0018621496846348918,
+      "loss": 1.4547,
+      "step": 530000
+    },
+    {
+      "epoch": 3.449509070810846,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0018620196371675662,
+      "loss": 1.4489,
+      "step": 530500
+    },
+    {
+      "epoch": 3.4527602574939853,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0018618895897002405,
+      "loss": 1.4588,
+      "step": 531000
+    },
+    {
+      "epoch": 3.4560114441771246,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.001861759542232915,
+      "loss": 1.4543,
+      "step": 531500
+    },
+    {
+      "epoch": 3.459262630860264,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0018616294947655895,
+      "loss": 1.4538,
+      "step": 532000
+    },
+    {
+      "epoch": 3.462513817543403,
+      "grad_norm": 1.8515625,
+      "learning_rate": 0.0018614994472982637,
+      "loss": 1.4574,
+      "step": 532500
+    },
+    {
+      "epoch": 3.465765004226543,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0018613693998309382,
+      "loss": 1.4542,
+      "step": 533000
+    },
+    {
+      "epoch": 3.469016190909682,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0018612393523636129,
+      "loss": 1.4478,
+      "step": 533500
+    },
+    {
+      "epoch": 3.4722673775928214,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0018611093048962874,
+      "loss": 1.4491,
+      "step": 534000
+    },
+    {
+      "epoch": 3.4755185642759607,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0018609792574289616,
+      "loss": 1.4509,
+      "step": 534500
+    },
+    {
+      "epoch": 3.4787697509591,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.001860849209961636,
+      "loss": 1.4515,
+      "step": 535000
+    },
+    {
+      "epoch": 3.4820209376422393,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.0018607191624943106,
+      "loss": 1.4496,
+      "step": 535500
+    },
+    {
+      "epoch": 3.4852721243253786,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0018605891150269848,
+      "loss": 1.4531,
+      "step": 536000
+    },
+    {
+      "epoch": 3.4885233110085183,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0018604590675596593,
+      "loss": 1.4456,
+      "step": 536500
+    },
+    {
+      "epoch": 3.4917744976916576,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0018603290200923338,
+      "loss": 1.445,
+      "step": 537000
+    },
+    {
+      "epoch": 3.495025684374797,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001860198972625008,
+      "loss": 1.4524,
+      "step": 537500
+    },
+    {
+      "epoch": 3.498276871057936,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0018600689251576825,
+      "loss": 1.4515,
+      "step": 538000
+    },
+    {
+      "epoch": 3.5015280577410755,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.001859938877690357,
+      "loss": 1.452,
+      "step": 538500
+    },
+    {
+      "epoch": 3.5047792444242147,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0018598088302230313,
+      "loss": 1.4509,
+      "step": 539000
+    },
+    {
+      "epoch": 3.508030431107354,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0018596787827557058,
+      "loss": 1.4488,
+      "step": 539500
+    },
+    {
+      "epoch": 3.5112816177904937,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0018595487352883802,
+      "loss": 1.4503,
+      "step": 540000
+    },
+    {
+      "epoch": 3.514532804473633,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.001859418687821055,
+      "loss": 1.4458,
+      "step": 540500
+    },
+    {
+      "epoch": 3.5177839911567723,
+      "grad_norm": 3.40625,
+      "learning_rate": 0.0018592886403537292,
+      "loss": 1.448,
+      "step": 541000
+    },
+    {
+      "epoch": 3.5210351778399116,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.0018591585928864037,
+      "loss": 1.4465,
+      "step": 541500
+    },
+    {
+      "epoch": 3.524286364523051,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0018590285454190782,
+      "loss": 1.4497,
+      "step": 542000
+    },
+    {
+      "epoch": 3.52753755120619,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0018588984979517524,
+      "loss": 1.4566,
+      "step": 542500
+    },
+    {
+      "epoch": 3.5307887378893295,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.001858768450484427,
+      "loss": 1.446,
+      "step": 543000
+    },
+    {
+      "epoch": 3.534039924572469,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0018586384030171014,
+      "loss": 1.4494,
+      "step": 543500
+    },
+    {
+      "epoch": 3.537291111255608,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.0018585083555497756,
+      "loss": 1.4507,
+      "step": 544000
+    },
+    {
+      "epoch": 3.5405422979387478,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0018583783080824501,
+      "loss": 1.4519,
+      "step": 544500
+    },
+    {
+      "epoch": 3.543793484621887,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0018582482606151246,
+      "loss": 1.4519,
+      "step": 545000
+    },
+    {
+      "epoch": 3.5470446713050263,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0018581182131477989,
+      "loss": 1.4518,
+      "step": 545500
+    },
+    {
+      "epoch": 3.5502958579881656,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0018579881656804733,
+      "loss": 1.4468,
+      "step": 546000
+    },
+    {
+      "epoch": 3.553547044671305,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.0018578581182131478,
+      "loss": 1.4491,
+      "step": 546500
+    },
+    {
+      "epoch": 3.5567982313544446,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001857728070745822,
+      "loss": 1.4475,
+      "step": 547000
+    },
+    {
+      "epoch": 3.5600494180375835,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.0018575980232784966,
+      "loss": 1.4443,
+      "step": 547500
+    },
+    {
+      "epoch": 3.563300604720723,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0018574679758111712,
+      "loss": 1.4528,
+      "step": 548000
+    },
+    {
+      "epoch": 3.5665517914038625,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0018573379283438457,
+      "loss": 1.4504,
+      "step": 548500
+    },
+    {
+      "epoch": 3.5698029780870018,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00185720788087652,
+      "loss": 1.4515,
+      "step": 549000
+    },
+    {
+      "epoch": 3.573054164770141,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0018570778334091945,
+      "loss": 1.4514,
+      "step": 549500
+    },
+    {
+      "epoch": 3.5763053514532803,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.001856947785941869,
+      "loss": 1.4486,
+      "step": 550000
+    },
+    {
+      "epoch": 3.57955653813642,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0018568177384745432,
+      "loss": 1.4558,
+      "step": 550500
+    },
+    {
+      "epoch": 3.582807724819559,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.0018566876910072177,
+      "loss": 1.4568,
+      "step": 551000
+    },
+    {
+      "epoch": 3.5860589115026986,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0018565576435398922,
+      "loss": 1.4611,
+      "step": 551500
+    },
+    {
+      "epoch": 3.589310098185838,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0018564275960725664,
+      "loss": 1.4546,
+      "step": 552000
+    },
+    {
+      "epoch": 3.592561284868977,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.001856297548605241,
+      "loss": 1.4468,
+      "step": 552500
+    },
+    {
+      "epoch": 3.5958124715521165,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0018561675011379154,
+      "loss": 1.4537,
+      "step": 553000
+    },
+    {
+      "epoch": 3.5990636582352558,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.0018560374536705896,
+      "loss": 1.4525,
+      "step": 553500
+    },
+    {
+      "epoch": 3.6023148449183955,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0018559074062032641,
+      "loss": 1.4554,
+      "step": 554000
+    },
+    {
+      "epoch": 3.6055660316015343,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0018557773587359386,
+      "loss": 1.4624,
+      "step": 554500
+    },
+    {
+      "epoch": 3.608817218284674,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0018556473112686133,
+      "loss": 1.45,
+      "step": 555000
+    },
+    {
+      "epoch": 3.6120684049678133,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0018555172638012876,
+      "loss": 1.4379,
+      "step": 555500
+    },
+    {
+      "epoch": 3.6153195916509526,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001855387216333962,
+      "loss": 1.4456,
+      "step": 556000
+    },
+    {
+      "epoch": 3.618570778334092,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.0018552571688666365,
+      "loss": 1.4477,
+      "step": 556500
+    },
+    {
+      "epoch": 3.621821965017231,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0018551271213993108,
+      "loss": 1.4479,
+      "step": 557000
+    },
+    {
+      "epoch": 3.6250731517003705,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0018549970739319853,
+      "loss": 1.4461,
+      "step": 557500
+    },
+    {
+      "epoch": 3.6283243383835098,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0018548670264646597,
+      "loss": 1.4451,
+      "step": 558000
+    },
+    {
+      "epoch": 3.6315755250666495,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.001854736978997334,
+      "loss": 1.4457,
+      "step": 558500
+    },
+    {
+      "epoch": 3.6348267117497888,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0018546069315300085,
+      "loss": 1.4474,
+      "step": 559000
+    },
+    {
+      "epoch": 3.638077898432928,
+      "grad_norm": 0.875,
+      "learning_rate": 0.001854476884062683,
+      "loss": 1.4448,
+      "step": 559500
+    },
+    {
+      "epoch": 3.6413290851160673,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0018543468365953572,
+      "loss": 1.4462,
+      "step": 560000
+    },
+    {
+      "epoch": 3.6445802717992066,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0018542167891280317,
+      "loss": 1.446,
+      "step": 560500
+    },
+    {
+      "epoch": 3.647831458482346,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.0018540867416607062,
+      "loss": 1.4452,
+      "step": 561000
+    },
+    {
+      "epoch": 3.651082645165485,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0018539566941933804,
+      "loss": 1.4469,
+      "step": 561500
+    },
+    {
+      "epoch": 3.654333831848625,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.001853826646726055,
+      "loss": 1.4522,
+      "step": 562000
+    },
+    {
+      "epoch": 3.657585018531764,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0018536965992587296,
+      "loss": 1.4482,
+      "step": 562500
+    },
+    {
+      "epoch": 3.6608362052149035,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.001853566551791404,
+      "loss": 1.4472,
+      "step": 563000
+    },
+    {
+      "epoch": 3.664087391898043,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0018534365043240783,
+      "loss": 1.443,
+      "step": 563500
+    },
+    {
+      "epoch": 3.667338578581182,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0018533064568567528,
+      "loss": 1.4432,
+      "step": 564000
+    },
+    {
+      "epoch": 3.6705897652643213,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0018531764093894273,
+      "loss": 1.4502,
+      "step": 564500
+    },
+    {
+      "epoch": 3.6738409519474606,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.0018530463619221016,
+      "loss": 1.4471,
+      "step": 565000
+    },
+    {
+      "epoch": 3.6770921386306004,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.001852916314454776,
+      "loss": 1.4464,
+      "step": 565500
+    },
+    {
+      "epoch": 3.6803433253137396,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.0018527862669874505,
+      "loss": 1.4476,
+      "step": 566000
+    },
+    {
+      "epoch": 3.683594511996879,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0018526562195201248,
+      "loss": 1.4494,
+      "step": 566500
+    },
+    {
+      "epoch": 3.686845698680018,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0018525261720527993,
+      "loss": 1.4483,
+      "step": 567000
+    },
+    {
+      "epoch": 3.6900968853631575,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018523961245854737,
+      "loss": 1.4481,
+      "step": 567500
+    },
+    {
+      "epoch": 3.693348072046297,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.001852266077118148,
+      "loss": 1.4484,
+      "step": 568000
+    },
+    {
+      "epoch": 3.696599258729436,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0018521360296508225,
+      "loss": 1.4434,
+      "step": 568500
+    },
+    {
+      "epoch": 3.699850445412576,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001852005982183497,
+      "loss": 1.4446,
+      "step": 569000
+    },
+    {
+      "epoch": 3.703101632095715,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0018518759347161717,
+      "loss": 1.443,
+      "step": 569500
+    },
+    {
+      "epoch": 3.7063528187788544,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001851745887248846,
+      "loss": 1.4453,
+      "step": 570000
+    },
+    {
+      "epoch": 3.7096040054619936,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0018516158397815204,
+      "loss": 1.4477,
+      "step": 570500
+    },
+    {
+      "epoch": 3.712855192145133,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0018514857923141949,
+      "loss": 1.441,
+      "step": 571000
+    },
+    {
+      "epoch": 3.716106378828272,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0018513557448468691,
+      "loss": 1.4444,
+      "step": 571500
+    },
+    {
+      "epoch": 3.7193575655114115,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0018512256973795436,
+      "loss": 1.4424,
+      "step": 572000
+    },
+    {
+      "epoch": 3.7226087521945512,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.001851095649912218,
+      "loss": 1.4451,
+      "step": 572500
+    },
+    {
+      "epoch": 3.7258599388776905,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0018509656024448924,
+      "loss": 1.4453,
+      "step": 573000
+    },
+    {
+      "epoch": 3.72911112556083,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0018508355549775668,
+      "loss": 1.446,
+      "step": 573500
+    },
+    {
+      "epoch": 3.732362312243969,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0018507055075102413,
+      "loss": 1.4498,
+      "step": 574000
+    },
+    {
+      "epoch": 3.7356134989271084,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0018505754600429156,
+      "loss": 1.4429,
+      "step": 574500
+    },
+    {
+      "epoch": 3.7388646856102477,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00185044541257559,
+      "loss": 1.4459,
+      "step": 575000
+    },
+    {
+      "epoch": 3.742115872293387,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0018503153651082645,
+      "loss": 1.4468,
+      "step": 575500
+    },
+    {
+      "epoch": 3.7453670589765267,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0018501853176409388,
+      "loss": 1.4479,
+      "step": 576000
+    },
+    {
+      "epoch": 3.7486182456596655,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0018500552701736133,
+      "loss": 1.4447,
+      "step": 576500
+    },
+    {
+      "epoch": 3.7518694323428052,
+      "grad_norm": 7.25,
+      "learning_rate": 0.001849925222706288,
+      "loss": 1.447,
+      "step": 577000
+    },
+    {
+      "epoch": 3.7551206190259445,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0018497951752389624,
+      "loss": 1.443,
+      "step": 577500
+    },
+    {
+      "epoch": 3.758371805709084,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.0018496651277716367,
+      "loss": 1.4386,
+      "step": 578000
+    },
+    {
+      "epoch": 3.761622992392223,
+      "grad_norm": 3.828125,
+      "learning_rate": 0.0018495350803043112,
+      "loss": 1.4434,
+      "step": 578500
+    },
+    {
+      "epoch": 3.7648741790753624,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0018494050328369857,
+      "loss": 1.4513,
+      "step": 579000
+    },
+    {
+      "epoch": 3.768125365758502,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00184927498536966,
+      "loss": 1.444,
+      "step": 579500
+    },
+    {
+      "epoch": 3.771376552441641,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0018491449379023344,
+      "loss": 1.444,
+      "step": 580000
+    },
+    {
+      "epoch": 3.7746277391247807,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0018490148904350089,
+      "loss": 1.449,
+      "step": 580500
+    },
+    {
+      "epoch": 3.77787892580792,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0018488848429676831,
+      "loss": 1.447,
+      "step": 581000
+    },
+    {
+      "epoch": 3.7811301124910592,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0018487547955003576,
+      "loss": 1.4415,
+      "step": 581500
+    },
+    {
+      "epoch": 3.7843812991741985,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.001848624748033032,
+      "loss": 1.443,
+      "step": 582000
+    },
+    {
+      "epoch": 3.787632485857338,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0018484947005657064,
+      "loss": 1.4442,
+      "step": 582500
+    },
+    {
+      "epoch": 3.7908836725404775,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0018483646530983808,
+      "loss": 1.4434,
+      "step": 583000
+    },
+    {
+      "epoch": 3.7941348592236164,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.0018482346056310553,
+      "loss": 1.4487,
+      "step": 583500
+    },
+    {
+      "epoch": 3.797386045906756,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00184810455816373,
+      "loss": 1.4402,
+      "step": 584000
+    },
+    {
+      "epoch": 3.8006372325898954,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.0018479745106964043,
+      "loss": 1.4463,
+      "step": 584500
+    },
+    {
+      "epoch": 3.8038884192730347,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0018478444632290788,
+      "loss": 1.4456,
+      "step": 585000
+    },
+    {
+      "epoch": 3.807139605956174,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0018477144157617532,
+      "loss": 1.4485,
+      "step": 585500
+    },
+    {
+      "epoch": 3.8103907926393132,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.0018475843682944275,
+      "loss": 1.4514,
+      "step": 586000
+    },
+    {
+      "epoch": 3.813641979322453,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.001847454320827102,
+      "loss": 1.4494,
+      "step": 586500
+    },
+    {
+      "epoch": 3.816893166005592,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0018473242733597765,
+      "loss": 1.4423,
+      "step": 587000
+    },
+    {
+      "epoch": 3.8201443526887315,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.0018471942258924507,
+      "loss": 1.454,
+      "step": 587500
+    },
+    {
+      "epoch": 3.823395539371871,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0018470641784251252,
+      "loss": 1.4479,
+      "step": 588000
+    },
+    {
+      "epoch": 3.82664672605501,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0018469341309577997,
+      "loss": 1.449,
+      "step": 588500
+    },
+    {
+      "epoch": 3.8298979127381494,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.001846804083490474,
+      "loss": 1.4458,
+      "step": 589000
+    },
+    {
+      "epoch": 3.8331490994212887,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0018466740360231484,
+      "loss": 1.4449,
+      "step": 589500
+    },
+    {
+      "epoch": 3.836400286104428,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018465439885558229,
+      "loss": 1.4478,
+      "step": 590000
+    },
+    {
+      "epoch": 3.8396514727875672,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0018464139410884972,
+      "loss": 1.4433,
+      "step": 590500
+    },
+    {
+      "epoch": 3.842902659470707,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0018462838936211716,
+      "loss": 1.442,
+      "step": 591000
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 7.4375,
+      "learning_rate": 0.0018461538461538463,
+      "loss": 1.4469,
+      "step": 591500
+    },
+    {
+      "epoch": 3.8494050328369855,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0018460237986865208,
+      "loss": 1.4432,
+      "step": 592000
+    },
+    {
+      "epoch": 3.852656219520125,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001845893751219195,
+      "loss": 1.4445,
+      "step": 592500
+    },
+    {
+      "epoch": 3.855907406203264,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0018457637037518695,
+      "loss": 1.4427,
+      "step": 593000
+    },
+    {
+      "epoch": 3.8591585928864034,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.001845633656284544,
+      "loss": 1.4461,
+      "step": 593500
+    },
+    {
+      "epoch": 3.8624097795695427,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0018455036088172183,
+      "loss": 1.4441,
+      "step": 594000
+    },
+    {
+      "epoch": 3.8656609662526824,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0018453735613498928,
+      "loss": 1.4397,
+      "step": 594500
+    },
+    {
+      "epoch": 3.8689121529358217,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0018452435138825672,
+      "loss": 1.4462,
+      "step": 595000
+    },
+    {
+      "epoch": 3.872163339618961,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0018451134664152415,
+      "loss": 1.4442,
+      "step": 595500
+    },
+    {
+      "epoch": 3.8754145263021003,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.001844983418947916,
+      "loss": 1.4517,
+      "step": 596000
+    },
+    {
+      "epoch": 3.8786657129852395,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0018448533714805905,
+      "loss": 1.4499,
+      "step": 596500
+    },
+    {
+      "epoch": 3.881916899668379,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0018447233240132647,
+      "loss": 1.4482,
+      "step": 597000
+    },
+    {
+      "epoch": 3.885168086351518,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0018445932765459392,
+      "loss": 1.4522,
+      "step": 597500
+    },
+    {
+      "epoch": 3.888419273034658,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0018444632290786137,
+      "loss": 1.4533,
+      "step": 598000
+    },
+    {
+      "epoch": 3.891670459717797,
+      "grad_norm": 1.75,
+      "learning_rate": 0.0018443331816112884,
+      "loss": 1.4641,
+      "step": 598500
+    },
+    {
+      "epoch": 3.8949216464009364,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0018442031341439626,
+      "loss": 1.4639,
+      "step": 599000
+    },
+    {
+      "epoch": 3.8981728330840757,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018440730866766371,
+      "loss": 1.4723,
+      "step": 599500
+    },
+    {
+      "epoch": 3.901424019767215,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0018439430392093116,
+      "loss": 1.47,
+      "step": 600000
+    },
+    {
+      "epoch": 3.9046752064503543,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0018438129917419859,
+      "loss": 1.4575,
+      "step": 600500
+    },
+    {
+      "epoch": 3.9079263931334935,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0018436829442746603,
+      "loss": 1.4491,
+      "step": 601000
+    },
+    {
+      "epoch": 3.9111775798166333,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0018435528968073348,
+      "loss": 1.4543,
+      "step": 601500
+    },
+    {
+      "epoch": 3.9144287664997726,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.001843422849340009,
+      "loss": 1.4517,
+      "step": 602000
+    },
+    {
+      "epoch": 3.917679953182912,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018432928018726836,
+      "loss": 1.4506,
+      "step": 602500
+    },
+    {
+      "epoch": 3.920931139866051,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001843162754405358,
+      "loss": 1.4549,
+      "step": 603000
+    },
+    {
+      "epoch": 3.9241823265491904,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.0018430327069380323,
+      "loss": 1.4574,
+      "step": 603500
+    },
+    {
+      "epoch": 3.9274335132323297,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.0018429026594707068,
+      "loss": 1.4556,
+      "step": 604000
+    },
+    {
+      "epoch": 3.930684699915469,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0018427726120033812,
+      "loss": 1.4504,
+      "step": 604500
+    },
+    {
+      "epoch": 3.9339358865986087,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018426425645360555,
+      "loss": 1.4489,
+      "step": 605000
+    },
+    {
+      "epoch": 3.937187073281748,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00184251251706873,
+      "loss": 1.4473,
+      "step": 605500
+    },
+    {
+      "epoch": 3.9404382599648873,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0018423824696014047,
+      "loss": 1.4407,
+      "step": 606000
+    },
+    {
+      "epoch": 3.9436894466480266,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0018422524221340792,
+      "loss": 1.4481,
+      "step": 606500
+    },
+    {
+      "epoch": 3.946940633331166,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0018421223746667534,
+      "loss": 1.4457,
+      "step": 607000
+    },
+    {
+      "epoch": 3.950191820014305,
+      "grad_norm": 1.0,
+      "learning_rate": 0.001841992327199428,
+      "loss": 1.4491,
+      "step": 607500
+    },
+    {
+      "epoch": 3.9534430066974444,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0018418622797321024,
+      "loss": 1.443,
+      "step": 608000
+    },
+    {
+      "epoch": 3.956694193380584,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0018417322322647766,
+      "loss": 1.4424,
+      "step": 608500
+    },
+    {
+      "epoch": 3.959945380063723,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0018416021847974511,
+      "loss": 1.4371,
+      "step": 609000
+    },
+    {
+      "epoch": 3.9631965667468627,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0018414721373301256,
+      "loss": 1.439,
+      "step": 609500
+    },
+    {
+      "epoch": 3.966447753430002,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0018413420898627999,
+      "loss": 1.4452,
+      "step": 610000
+    },
+    {
+      "epoch": 3.9696989401131413,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0018412120423954743,
+      "loss": 1.4444,
+      "step": 610500
+    },
+    {
+      "epoch": 3.9729501267962806,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0018410819949281488,
+      "loss": 1.4489,
+      "step": 611000
+    },
+    {
+      "epoch": 3.97620131347942,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.001840951947460823,
+      "loss": 1.4499,
+      "step": 611500
+    },
+    {
+      "epoch": 3.9794525001625596,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0018408218999934976,
+      "loss": 1.4398,
+      "step": 612000
+    },
+    {
+      "epoch": 3.9827036868456984,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.001840691852526172,
+      "loss": 1.4407,
+      "step": 612500
+    },
+    {
+      "epoch": 3.985954873528838,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0018405618050588467,
+      "loss": 1.4432,
+      "step": 613000
+    },
+    {
+      "epoch": 3.9892060602119774,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.001840431757591521,
+      "loss": 1.4482,
+      "step": 613500
+    },
+    {
+      "epoch": 3.9924572468951167,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.0018403017101241955,
+      "loss": 1.4452,
+      "step": 614000
+    },
+    {
+      "epoch": 3.995708433578256,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00184017166265687,
+      "loss": 1.4436,
+      "step": 614500
+    },
+    {
+      "epoch": 3.9989596202613953,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0018400416151895442,
+      "loss": 1.4495,
+      "step": 615000
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 1.4305795431137085,
+      "eval_runtime": 0.5339,
+      "eval_samples_per_second": 1873.085,
+      "eval_steps_per_second": 29.969,
+      "step": 615160
+    },
+    {
+      "epoch": 4.002210806944535,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0018399115677222187,
+      "loss": 1.4472,
+      "step": 615500
+    },
+    {
+      "epoch": 4.005461993627674,
+      "grad_norm": 2.375,
+      "learning_rate": 0.0018397815202548932,
+      "loss": 1.4458,
+      "step": 616000
+    },
+    {
+      "epoch": 4.008713180310814,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0018396514727875674,
+      "loss": 1.4438,
+      "step": 616500
+    },
+    {
+      "epoch": 4.011964366993952,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.001839521425320242,
+      "loss": 1.4449,
+      "step": 617000
+    },
+    {
+      "epoch": 4.015215553677092,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0018393913778529164,
+      "loss": 1.4383,
+      "step": 617500
+    },
+    {
+      "epoch": 4.018466740360232,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0018392613303855907,
+      "loss": 1.4475,
+      "step": 618000
+    },
+    {
+      "epoch": 4.021717927043371,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0018391312829182651,
+      "loss": 1.4417,
+      "step": 618500
+    },
+    {
+      "epoch": 4.0249691137265105,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0018390012354509396,
+      "loss": 1.4449,
+      "step": 619000
+    },
+    {
+      "epoch": 4.028220300409649,
+      "grad_norm": 2.34375,
+      "learning_rate": 0.0018388711879836139,
+      "loss": 1.4429,
+      "step": 619500
+    },
+    {
+      "epoch": 4.031471487092789,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0018387411405162883,
+      "loss": 1.4423,
+      "step": 620000
+    },
+    {
+      "epoch": 4.034722673775928,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001838611093048963,
+      "loss": 1.4394,
+      "step": 620500
+    },
+    {
+      "epoch": 4.037973860459068,
+      "grad_norm": 3.515625,
+      "learning_rate": 0.0018384810455816375,
+      "loss": 1.4429,
+      "step": 621000
+    },
+    {
+      "epoch": 4.041225047142207,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0018383509981143118,
+      "loss": 1.4435,
+      "step": 621500
+    },
+    {
+      "epoch": 4.044476233825346,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0018382209506469863,
+      "loss": 1.444,
+      "step": 622000
+    },
+    {
+      "epoch": 4.047727420508486,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0018380909031796607,
+      "loss": 1.4436,
+      "step": 622500
+    },
+    {
+      "epoch": 4.050978607191625,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001837960855712335,
+      "loss": 1.4432,
+      "step": 623000
+    },
+    {
+      "epoch": 4.0542297938747645,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.0018378308082450095,
+      "loss": 1.4503,
+      "step": 623500
+    },
+    {
+      "epoch": 4.057480980557903,
+      "grad_norm": 4.625,
+      "learning_rate": 0.001837700760777684,
+      "loss": 1.4511,
+      "step": 624000
+    },
+    {
+      "epoch": 4.060732167241043,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0018375707133103582,
+      "loss": 1.444,
+      "step": 624500
+    },
+    {
+      "epoch": 4.063983353924183,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0018374406658430327,
+      "loss": 1.4435,
+      "step": 625000
+    },
+    {
+      "epoch": 4.067234540607322,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0018373106183757072,
+      "loss": 1.4421,
+      "step": 625500
+    },
+    {
+      "epoch": 4.070485727290461,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0018371805709083814,
+      "loss": 1.4455,
+      "step": 626000
+    },
+    {
+      "epoch": 4.0737369139736,
+      "grad_norm": 1.25,
+      "learning_rate": 0.001837050523441056,
+      "loss": 1.4422,
+      "step": 626500
+    },
+    {
+      "epoch": 4.07698810065674,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0018369204759737304,
+      "loss": 1.4459,
+      "step": 627000
+    },
+    {
+      "epoch": 4.080239287339879,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.001836790428506405,
+      "loss": 1.4457,
+      "step": 627500
+    },
+    {
+      "epoch": 4.0834904740230185,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0018366603810390794,
+      "loss": 1.4458,
+      "step": 628000
+    },
+    {
+      "epoch": 4.086741660706158,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0018365303335717538,
+      "loss": 1.4424,
+      "step": 628500
+    },
+    {
+      "epoch": 4.089992847389297,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0018364002861044283,
+      "loss": 1.4486,
+      "step": 629000
+    },
+    {
+      "epoch": 4.093244034072437,
+      "grad_norm": 1.828125,
+      "learning_rate": 0.0018362702386371026,
+      "loss": 1.4482,
+      "step": 629500
+    },
+    {
+      "epoch": 4.096495220755576,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.001836140191169777,
+      "loss": 1.4395,
+      "step": 630000
+    },
+    {
+      "epoch": 4.099746407438715,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0018360101437024515,
+      "loss": 1.4456,
+      "step": 630500
+    },
+    {
+      "epoch": 4.102997594121854,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0018358800962351258,
+      "loss": 1.4444,
+      "step": 631000
+    },
+    {
+      "epoch": 4.106248780804994,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0018357500487678003,
+      "loss": 1.443,
+      "step": 631500
+    },
+    {
+      "epoch": 4.109499967488133,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0018356200013004747,
+      "loss": 1.4445,
+      "step": 632000
+    },
+    {
+      "epoch": 4.1127511541712725,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.001835489953833149,
+      "loss": 1.4599,
+      "step": 632500
+    },
+    {
+      "epoch": 4.116002340854412,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.0018353599063658235,
+      "loss": 1.448,
+      "step": 633000
+    },
+    {
+      "epoch": 4.119253527537551,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001835229858898498,
+      "loss": 1.4514,
+      "step": 633500
+    },
+    {
+      "epoch": 4.122504714220691,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0018350998114311722,
+      "loss": 1.4508,
+      "step": 634000
+    },
+    {
+      "epoch": 4.12575590090383,
+      "grad_norm": 9.625,
+      "learning_rate": 0.0018349697639638467,
+      "loss": 1.4499,
+      "step": 634500
+    },
+    {
+      "epoch": 4.129007087586969,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0018348397164965214,
+      "loss": 1.4394,
+      "step": 635000
+    },
+    {
+      "epoch": 4.132258274270108,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0018347096690291959,
+      "loss": 1.4405,
+      "step": 635500
+    },
+    {
+      "epoch": 4.135509460953248,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018345796215618701,
+      "loss": 1.4426,
+      "step": 636000
+    },
+    {
+      "epoch": 4.138760647636388,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.0018344495740945446,
+      "loss": 1.4407,
+      "step": 636500
+    },
+    {
+      "epoch": 4.1420118343195265,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.001834319526627219,
+      "loss": 1.4451,
+      "step": 637000
+    },
+    {
+      "epoch": 4.145263021002666,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.0018341894791598934,
+      "loss": 1.4492,
+      "step": 637500
+    },
+    {
+      "epoch": 4.148514207685805,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0018340594316925678,
+      "loss": 1.4399,
+      "step": 638000
+    },
+    {
+      "epoch": 4.151765394368945,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.0018339293842252423,
+      "loss": 1.4504,
+      "step": 638500
+    },
+    {
+      "epoch": 4.155016581052084,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.0018337993367579166,
+      "loss": 1.4557,
+      "step": 639000
+    },
+    {
+      "epoch": 4.158267767735223,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.001833669289290591,
+      "loss": 1.4523,
+      "step": 639500
+    },
+    {
+      "epoch": 4.161518954418363,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018335392418232655,
+      "loss": 1.4516,
+      "step": 640000
+    },
+    {
+      "epoch": 4.164770141101502,
+      "grad_norm": 2.984375,
+      "learning_rate": 0.0018334091943559398,
+      "loss": 1.4507,
+      "step": 640500
+    },
+    {
+      "epoch": 4.168021327784642,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0018332791468886143,
+      "loss": 1.4546,
+      "step": 641000
+    },
+    {
+      "epoch": 4.1712725144677805,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0018331490994212888,
+      "loss": 1.4501,
+      "step": 641500
+    },
+    {
+      "epoch": 4.17452370115092,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001833019051953963,
+      "loss": 1.4429,
+      "step": 642000
+    },
+    {
+      "epoch": 4.177774887834059,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0018328890044866377,
+      "loss": 1.4459,
+      "step": 642500
+    },
+    {
+      "epoch": 4.181026074517199,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0018327589570193122,
+      "loss": 1.4485,
+      "step": 643000
+    },
+    {
+      "epoch": 4.1842772612003385,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0018326289095519867,
+      "loss": 1.4511,
+      "step": 643500
+    },
+    {
+      "epoch": 4.187528447883477,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.001832498862084661,
+      "loss": 1.4479,
+      "step": 644000
+    },
+    {
+      "epoch": 4.190779634566617,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0018323688146173354,
+      "loss": 1.451,
+      "step": 644500
+    },
+    {
+      "epoch": 4.194030821249756,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0018322387671500099,
+      "loss": 1.4497,
+      "step": 645000
+    },
+    {
+      "epoch": 4.197282007932896,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0018321087196826842,
+      "loss": 1.4446,
+      "step": 645500
+    },
+    {
+      "epoch": 4.2005331946160345,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0018319786722153586,
+      "loss": 1.4413,
+      "step": 646000
+    },
+    {
+      "epoch": 4.203784381299174,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001831848624748033,
+      "loss": 1.445,
+      "step": 646500
+    },
+    {
+      "epoch": 4.207035567982314,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0018317185772807074,
+      "loss": 1.4409,
+      "step": 647000
+    },
+    {
+      "epoch": 4.210286754665453,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0018315885298133818,
+      "loss": 1.4424,
+      "step": 647500
+    },
+    {
+      "epoch": 4.2135379413485925,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0018314584823460563,
+      "loss": 1.4467,
+      "step": 648000
+    },
+    {
+      "epoch": 4.216789128031731,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0018313284348787306,
+      "loss": 1.4442,
+      "step": 648500
+    },
+    {
+      "epoch": 4.220040314714871,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.001831198387411405,
+      "loss": 1.4404,
+      "step": 649000
+    },
+    {
+      "epoch": 4.22329150139801,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.0018310683399440798,
+      "loss": 1.441,
+      "step": 649500
+    },
+    {
+      "epoch": 4.22654268808115,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0018309382924767542,
+      "loss": 1.442,
+      "step": 650000
+    },
+    {
+      "epoch": 4.229793874764289,
+      "grad_norm": 3.046875,
+      "learning_rate": 0.0018308082450094285,
+      "loss": 1.4432,
+      "step": 650500
+    },
+    {
+      "epoch": 4.233045061447428,
+      "grad_norm": 2.25,
+      "learning_rate": 0.001830678197542103,
+      "loss": 1.4433,
+      "step": 651000
+    },
+    {
+      "epoch": 4.236296248130568,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0018305481500747775,
+      "loss": 1.4452,
+      "step": 651500
+    },
+    {
+      "epoch": 4.239547434813707,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0018304181026074517,
+      "loss": 1.4406,
+      "step": 652000
+    },
+    {
+      "epoch": 4.2427986214968465,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0018302880551401262,
+      "loss": 1.4419,
+      "step": 652500
+    },
+    {
+      "epoch": 4.246049808179985,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.0018301580076728007,
+      "loss": 1.4422,
+      "step": 653000
+    },
+    {
+      "epoch": 4.249300994863125,
+      "grad_norm": 3.53125,
+      "learning_rate": 0.001830027960205475,
+      "loss": 1.4436,
+      "step": 653500
+    },
+    {
+      "epoch": 4.252552181546265,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0018298979127381494,
+      "loss": 1.4412,
+      "step": 654000
+    },
+    {
+      "epoch": 4.255803368229404,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.001829767865270824,
+      "loss": 1.4406,
+      "step": 654500
+    },
+    {
+      "epoch": 4.259054554912543,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0018296378178034982,
+      "loss": 1.4354,
+      "step": 655000
+    },
+    {
+      "epoch": 4.262305741595682,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0018295077703361726,
+      "loss": 1.4452,
+      "step": 655500
+    },
+    {
+      "epoch": 4.265556928278822,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0018293777228688471,
+      "loss": 1.4366,
+      "step": 656000
+    },
+    {
+      "epoch": 4.268808114961961,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0018292476754015214,
+      "loss": 1.4403,
+      "step": 656500
+    },
+    {
+      "epoch": 4.2720593016451005,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.001829117627934196,
+      "loss": 1.4437,
+      "step": 657000
+    },
+    {
+      "epoch": 4.275310488328239,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0018289875804668706,
+      "loss": 1.4443,
+      "step": 657500
+    },
+    {
+      "epoch": 4.278561675011379,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.001828857532999545,
+      "loss": 1.4399,
+      "step": 658000
+    },
+    {
+      "epoch": 4.281812861694519,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0018287274855322193,
+      "loss": 1.4402,
+      "step": 658500
+    },
+    {
+      "epoch": 4.285064048377658,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0018285974380648938,
+      "loss": 1.4419,
+      "step": 659000
+    },
+    {
+      "epoch": 4.288315235060797,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0018284673905975682,
+      "loss": 1.441,
+      "step": 659500
+    },
+    {
+      "epoch": 4.291566421743936,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018283373431302425,
+      "loss": 1.4357,
+      "step": 660000
+    },
+    {
+      "epoch": 4.294817608427076,
+      "grad_norm": 2.453125,
+      "learning_rate": 0.001828207295662917,
+      "loss": 1.4341,
+      "step": 660500
+    },
+    {
+      "epoch": 4.298068795110215,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.0018280772481955915,
+      "loss": 1.4512,
+      "step": 661000
+    },
+    {
+      "epoch": 4.3013199817933545,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0018279472007282657,
+      "loss": 1.4393,
+      "step": 661500
+    },
+    {
+      "epoch": 4.304571168476494,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0018278171532609402,
+      "loss": 1.4444,
+      "step": 662000
+    },
+    {
+      "epoch": 4.307822355159633,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018276871057936147,
+      "loss": 1.4476,
+      "step": 662500
+    },
+    {
+      "epoch": 4.311073541842773,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.001827557058326289,
+      "loss": 1.4516,
+      "step": 663000
+    },
+    {
+      "epoch": 4.314324728525912,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0018274270108589634,
+      "loss": 1.4479,
+      "step": 663500
+    },
+    {
+      "epoch": 4.317575915209051,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0018272969633916381,
+      "loss": 1.4526,
+      "step": 664000
+    },
+    {
+      "epoch": 4.32082710189219,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0018271669159243126,
+      "loss": 1.4503,
+      "step": 664500
+    },
+    {
+      "epoch": 4.32407828857533,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0018270368684569869,
+      "loss": 1.4487,
+      "step": 665000
+    },
+    {
+      "epoch": 4.32732947525847,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0018269068209896613,
+      "loss": 1.4459,
+      "step": 665500
+    },
+    {
+      "epoch": 4.3305806619416085,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0018267767735223358,
+      "loss": 1.4458,
+      "step": 666000
+    },
+    {
+      "epoch": 4.333831848624748,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.00182664672605501,
+      "loss": 1.4371,
+      "step": 666500
+    },
+    {
+      "epoch": 4.337083035307887,
+      "grad_norm": 1.625,
+      "learning_rate": 0.0018265166785876846,
+      "loss": 1.447,
+      "step": 667000
+    },
+    {
+      "epoch": 4.340334221991027,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.001826386631120359,
+      "loss": 1.4428,
+      "step": 667500
+    },
+    {
+      "epoch": 4.343585408674166,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0018262565836530333,
+      "loss": 1.4406,
+      "step": 668000
+    },
+    {
+      "epoch": 4.346836595357305,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0018261265361857078,
+      "loss": 1.4477,
+      "step": 668500
+    },
+    {
+      "epoch": 4.350087782040445,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.0018259964887183823,
+      "loss": 1.445,
+      "step": 669000
+    },
+    {
+      "epoch": 4.353338968723584,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0018258664412510565,
+      "loss": 1.442,
+      "step": 669500
+    },
+    {
+      "epoch": 4.356590155406724,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001825736393783731,
+      "loss": 1.449,
+      "step": 670000
+    },
+    {
+      "epoch": 4.3598413420898625,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0018256063463164055,
+      "loss": 1.4543,
+      "step": 670500
+    },
+    {
+      "epoch": 4.363092528773002,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0018254762988490797,
+      "loss": 1.4485,
+      "step": 671000
+    },
+    {
+      "epoch": 4.366343715456141,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.0018253462513817544,
+      "loss": 1.4508,
+      "step": 671500
+    },
+    {
+      "epoch": 4.369594902139281,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.001825216203914429,
+      "loss": 1.4501,
+      "step": 672000
+    },
+    {
+      "epoch": 4.3728460888224205,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.0018250861564471034,
+      "loss": 1.4442,
+      "step": 672500
+    },
+    {
+      "epoch": 4.376097275505559,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.0018249561089797777,
+      "loss": 1.4462,
+      "step": 673000
+    },
+    {
+      "epoch": 4.379348462188699,
+      "grad_norm": 3.125,
+      "learning_rate": 0.0018248260615124521,
+      "loss": 1.4463,
+      "step": 673500
+    },
+    {
+      "epoch": 4.382599648871838,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0018246960140451266,
+      "loss": 1.4506,
+      "step": 674000
+    },
+    {
+      "epoch": 4.385850835554978,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0018245659665778009,
+      "loss": 1.4447,
+      "step": 674500
+    },
+    {
+      "epoch": 4.3891020222381165,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0018244359191104753,
+      "loss": 1.4543,
+      "step": 675000
+    },
+    {
+      "epoch": 4.392353208921256,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0018243058716431498,
+      "loss": 1.4488,
+      "step": 675500
+    },
+    {
+      "epoch": 4.395604395604396,
+      "grad_norm": 3.1875,
+      "learning_rate": 0.001824175824175824,
+      "loss": 1.4386,
+      "step": 676000
+    },
+    {
+      "epoch": 4.398855582287535,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0018240457767084986,
+      "loss": 1.4409,
+      "step": 676500
+    },
+    {
+      "epoch": 4.4021067689706745,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001823915729241173,
+      "loss": 1.4445,
+      "step": 677000
+    },
+    {
+      "epoch": 4.405357955653813,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0018237856817738473,
+      "loss": 1.4439,
+      "step": 677500
+    },
+    {
+      "epoch": 4.408609142336953,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0018236556343065218,
+      "loss": 1.4467,
+      "step": 678000
+    },
+    {
+      "epoch": 4.411860329020092,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0018235255868391965,
+      "loss": 1.4435,
+      "step": 678500
+    },
+    {
+      "epoch": 4.415111515703232,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.001823395539371871,
+      "loss": 1.4387,
+      "step": 679000
+    },
+    {
+      "epoch": 4.418362702386371,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0018232654919045452,
+      "loss": 1.4442,
+      "step": 679500
+    },
+    {
+      "epoch": 4.42161388906951,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0018231354444372197,
+      "loss": 1.4385,
+      "step": 680000
+    },
+    {
+      "epoch": 4.42486507575265,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.0018230053969698942,
+      "loss": 1.4436,
+      "step": 680500
+    },
+    {
+      "epoch": 4.428116262435789,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0018228753495025684,
+      "loss": 1.4407,
+      "step": 681000
+    },
+    {
+      "epoch": 4.4313674491189285,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001822745302035243,
+      "loss": 1.4402,
+      "step": 681500
+    },
+    {
+      "epoch": 4.434618635802067,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018226152545679174,
+      "loss": 1.4414,
+      "step": 682000
+    },
+    {
+      "epoch": 4.437869822485207,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.0018224852071005917,
+      "loss": 1.4379,
+      "step": 682500
+    },
+    {
+      "epoch": 4.441121009168347,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0018223551596332661,
+      "loss": 1.4403,
+      "step": 683000
+    },
+    {
+      "epoch": 4.444372195851486,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0018222251121659406,
+      "loss": 1.4432,
+      "step": 683500
+    },
+    {
+      "epoch": 4.447623382534625,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0018220950646986149,
+      "loss": 1.4378,
+      "step": 684000
+    },
+    {
+      "epoch": 4.450874569217764,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0018219650172312894,
+      "loss": 1.439,
+      "step": 684500
+    },
+    {
+      "epoch": 4.454125755900904,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0018218349697639638,
+      "loss": 1.435,
+      "step": 685000
+    },
+    {
+      "epoch": 4.457376942584043,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.001821704922296638,
+      "loss": 1.4391,
+      "step": 685500
+    },
+    {
+      "epoch": 4.4606281292671826,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0018215748748293128,
+      "loss": 1.4392,
+      "step": 686000
+    },
+    {
+      "epoch": 4.463879315950322,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0018214448273619873,
+      "loss": 1.4382,
+      "step": 686500
+    },
+    {
+      "epoch": 4.467130502633461,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0018213147798946617,
+      "loss": 1.4411,
+      "step": 687000
+    },
+    {
+      "epoch": 4.470381689316601,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001821184732427336,
+      "loss": 1.4439,
+      "step": 687500
+    },
+    {
+      "epoch": 4.47363287599974,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0018210546849600105,
+      "loss": 1.4478,
+      "step": 688000
+    },
+    {
+      "epoch": 4.476884062682879,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.001820924637492685,
+      "loss": 1.4437,
+      "step": 688500
+    },
+    {
+      "epoch": 4.480135249366018,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0018207945900253592,
+      "loss": 1.4461,
+      "step": 689000
+    },
+    {
+      "epoch": 4.483386436049158,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0018206645425580337,
+      "loss": 1.4388,
+      "step": 689500
+    },
+    {
+      "epoch": 4.486637622732298,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.0018205344950907082,
+      "loss": 1.4433,
+      "step": 690000
+    },
+    {
+      "epoch": 4.4898888094154366,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0018204044476233824,
+      "loss": 1.4442,
+      "step": 690500
+    },
+    {
+      "epoch": 4.493139996098576,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001820274400156057,
+      "loss": 1.4312,
+      "step": 691000
+    },
+    {
+      "epoch": 4.496391182781715,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0018201443526887314,
+      "loss": 1.4469,
+      "step": 691500
+    },
+    {
+      "epoch": 4.499642369464855,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0018200143052214057,
+      "loss": 1.4389,
+      "step": 692000
+    },
+    {
+      "epoch": 4.502893556147994,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0018198842577540801,
+      "loss": 1.4402,
+      "step": 692500
+    },
+    {
+      "epoch": 4.506144742831133,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0018197542102867548,
+      "loss": 1.4366,
+      "step": 693000
+    },
+    {
+      "epoch": 4.509395929514273,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0018196241628194293,
+      "loss": 1.4418,
+      "step": 693500
+    },
+    {
+      "epoch": 4.512647116197412,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018194941153521036,
+      "loss": 1.439,
+      "step": 694000
+    },
+    {
+      "epoch": 4.515898302880552,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001819364067884778,
+      "loss": 1.4341,
+      "step": 694500
+    },
+    {
+      "epoch": 4.519149489563691,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0018192340204174525,
+      "loss": 1.4412,
+      "step": 695000
+    },
+    {
+      "epoch": 4.52240067624683,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0018191039729501268,
+      "loss": 1.4458,
+      "step": 695500
+    },
+    {
+      "epoch": 4.525651862929969,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0018189739254828013,
+      "loss": 1.4464,
+      "step": 696000
+    },
+    {
+      "epoch": 4.528903049613109,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0018188438780154758,
+      "loss": 1.4463,
+      "step": 696500
+    },
+    {
+      "epoch": 4.532154236296249,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.00181871383054815,
+      "loss": 1.4492,
+      "step": 697000
+    },
+    {
+      "epoch": 4.535405422979387,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018185837830808245,
+      "loss": 1.4423,
+      "step": 697500
+    },
+    {
+      "epoch": 4.538656609662527,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.001818453735613499,
+      "loss": 1.4375,
+      "step": 698000
+    },
+    {
+      "epoch": 4.541907796345666,
+      "grad_norm": 4.5,
+      "learning_rate": 0.0018183236881461732,
+      "loss": 1.445,
+      "step": 698500
+    },
+    {
+      "epoch": 4.545158983028806,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.0018181936406788477,
+      "loss": 1.4405,
+      "step": 699000
+    },
+    {
+      "epoch": 4.548410169711945,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018180635932115222,
+      "loss": 1.4406,
+      "step": 699500
+    },
+    {
+      "epoch": 4.551661356395084,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0018179335457441965,
+      "loss": 1.442,
+      "step": 700000
+    },
+    {
+      "epoch": 4.554912543078224,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0018178034982768711,
+      "loss": 1.4395,
+      "step": 700500
+    },
+    {
+      "epoch": 4.558163729761363,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018176734508095456,
+      "loss": 1.4396,
+      "step": 701000
+    },
+    {
+      "epoch": 4.561414916444503,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00181754340334222,
+      "loss": 1.4441,
+      "step": 701500
+    },
+    {
+      "epoch": 4.564666103127641,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.0018174133558748944,
+      "loss": 1.4435,
+      "step": 702000
+    },
+    {
+      "epoch": 4.567917289810781,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0018172833084075688,
+      "loss": 1.4422,
+      "step": 702500
+    },
+    {
+      "epoch": 4.57116847649392,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0018171532609402433,
+      "loss": 1.4483,
+      "step": 703000
+    },
+    {
+      "epoch": 4.57441966317706,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0018170232134729176,
+      "loss": 1.4467,
+      "step": 703500
+    },
+    {
+      "epoch": 4.5776708498601995,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001816893166005592,
+      "loss": 1.4474,
+      "step": 704000
+    },
+    {
+      "epoch": 4.580922036543338,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0018167631185382665,
+      "loss": 1.4412,
+      "step": 704500
+    },
+    {
+      "epoch": 4.584173223226478,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0018166330710709408,
+      "loss": 1.4412,
+      "step": 705000
+    },
+    {
+      "epoch": 4.587424409909617,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0018165030236036153,
+      "loss": 1.4442,
+      "step": 705500
+    },
+    {
+      "epoch": 4.590675596592757,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0018163729761362898,
+      "loss": 1.4428,
+      "step": 706000
+    },
+    {
+      "epoch": 4.593926783275895,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.001816242928668964,
+      "loss": 1.4454,
+      "step": 706500
+    },
+    {
+      "epoch": 4.597177969959035,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0018161128812016385,
+      "loss": 1.4456,
+      "step": 707000
+    },
+    {
+      "epoch": 4.600429156642175,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0018159828337343132,
+      "loss": 1.4355,
+      "step": 707500
+    },
+    {
+      "epoch": 4.603680343325314,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.0018158527862669877,
+      "loss": 1.4477,
+      "step": 708000
+    },
+    {
+      "epoch": 4.6069315300084535,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.001815722738799662,
+      "loss": 1.4491,
+      "step": 708500
+    },
+    {
+      "epoch": 4.610182716691592,
+      "grad_norm": 3.078125,
+      "learning_rate": 0.0018155926913323364,
+      "loss": 1.4513,
+      "step": 709000
+    },
+    {
+      "epoch": 4.613433903374732,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001815462643865011,
+      "loss": 1.4504,
+      "step": 709500
+    },
+    {
+      "epoch": 4.616685090057871,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0018153325963976852,
+      "loss": 1.455,
+      "step": 710000
+    },
+    {
+      "epoch": 4.619936276741011,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0018152025489303596,
+      "loss": 1.4542,
+      "step": 710500
+    },
+    {
+      "epoch": 4.623187463424149,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0018150725014630341,
+      "loss": 1.4562,
+      "step": 711000
+    },
+    {
+      "epoch": 4.626438650107289,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0018149424539957084,
+      "loss": 1.4514,
+      "step": 711500
+    },
+    {
+      "epoch": 4.629689836790429,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0018148124065283829,
+      "loss": 1.444,
+      "step": 712000
+    },
+    {
+      "epoch": 4.632941023473568,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0018146823590610573,
+      "loss": 1.4446,
+      "step": 712500
+    },
+    {
+      "epoch": 4.6361922101567075,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0018145523115937316,
+      "loss": 1.4515,
+      "step": 713000
+    },
+    {
+      "epoch": 4.639443396839846,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001814422264126406,
+      "loss": 1.449,
+      "step": 713500
+    },
+    {
+      "epoch": 4.642694583522986,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0018142922166590806,
+      "loss": 1.4478,
+      "step": 714000
+    },
+    {
+      "epoch": 4.645945770206125,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0018141621691917548,
+      "loss": 1.4463,
+      "step": 714500
+    },
+    {
+      "epoch": 4.649196956889265,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0018140321217244295,
+      "loss": 1.4528,
+      "step": 715000
+    },
+    {
+      "epoch": 4.652448143572403,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.001813902074257104,
+      "loss": 1.4528,
+      "step": 715500
+    },
+    {
+      "epoch": 4.655699330255543,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018137720267897785,
+      "loss": 1.4492,
+      "step": 716000
+    },
+    {
+      "epoch": 4.658950516938683,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0018136419793224527,
+      "loss": 1.4516,
+      "step": 716500
+    },
+    {
+      "epoch": 4.662201703621822,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0018135119318551272,
+      "loss": 1.4532,
+      "step": 717000
+    },
+    {
+      "epoch": 4.6654528903049615,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0018133818843878017,
+      "loss": 1.4468,
+      "step": 717500
+    },
+    {
+      "epoch": 4.6687040769881,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001813251836920476,
+      "loss": 1.4498,
+      "step": 718000
+    },
+    {
+      "epoch": 4.67195526367124,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0018131217894531504,
+      "loss": 1.4473,
+      "step": 718500
+    },
+    {
+      "epoch": 4.675206450354379,
+      "grad_norm": 1.0,
+      "learning_rate": 0.001812991741985825,
+      "loss": 1.4454,
+      "step": 719000
+    },
+    {
+      "epoch": 4.678457637037519,
+      "grad_norm": 2.921875,
+      "learning_rate": 0.0018128616945184992,
+      "loss": 1.4448,
+      "step": 719500
+    },
+    {
+      "epoch": 4.681708823720658,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0018127316470511736,
+      "loss": 1.4431,
+      "step": 720000
+    },
+    {
+      "epoch": 4.684960010403797,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0018126015995838481,
+      "loss": 1.4483,
+      "step": 720500
+    },
+    {
+      "epoch": 4.688211197086937,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.0018124715521165224,
+      "loss": 1.4487,
+      "step": 721000
+    },
+    {
+      "epoch": 4.691462383770076,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0018123415046491969,
+      "loss": 1.4505,
+      "step": 721500
+    },
+    {
+      "epoch": 4.6947135704532155,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0018122114571818716,
+      "loss": 1.4404,
+      "step": 722000
+    },
+    {
+      "epoch": 4.697964757136354,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.001812081409714546,
+      "loss": 1.4502,
+      "step": 722500
+    },
+    {
+      "epoch": 4.701215943819494,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0018119513622472203,
+      "loss": 1.4484,
+      "step": 723000
+    },
+    {
+      "epoch": 4.704467130502634,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0018118213147798948,
+      "loss": 1.4488,
+      "step": 723500
+    },
+    {
+      "epoch": 4.707718317185773,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.0018116912673125693,
+      "loss": 1.4532,
+      "step": 724000
+    },
+    {
+      "epoch": 4.710969503868912,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0018115612198452435,
+      "loss": 1.4492,
+      "step": 724500
+    },
+    {
+      "epoch": 4.714220690552051,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.001811431172377918,
+      "loss": 1.4479,
+      "step": 725000
+    },
+    {
+      "epoch": 4.717471877235191,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0018113011249105925,
+      "loss": 1.4552,
+      "step": 725500
+    },
+    {
+      "epoch": 4.72072306391833,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0018111710774432667,
+      "loss": 1.4542,
+      "step": 726000
+    },
+    {
+      "epoch": 4.7239742506014695,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0018110410299759412,
+      "loss": 1.4606,
+      "step": 726500
+    },
+    {
+      "epoch": 4.727225437284609,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0018109109825086157,
+      "loss": 1.4569,
+      "step": 727000
+    },
+    {
+      "epoch": 4.730476623967748,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00181078093504129,
+      "loss": 1.4634,
+      "step": 727500
+    },
+    {
+      "epoch": 4.733727810650888,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018106508875739644,
+      "loss": 1.461,
+      "step": 728000
+    },
+    {
+      "epoch": 4.736978997334027,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001810520840106639,
+      "loss": 1.4565,
+      "step": 728500
+    },
+    {
+      "epoch": 4.740230184017166,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0018103907926393132,
+      "loss": 1.4479,
+      "step": 729000
+    },
+    {
+      "epoch": 4.743481370700305,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0018102607451719879,
+      "loss": 1.452,
+      "step": 729500
+    },
+    {
+      "epoch": 4.746732557383445,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0018101306977046623,
+      "loss": 1.4529,
+      "step": 730000
+    },
+    {
+      "epoch": 4.749983744066585,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018100006502373368,
+      "loss": 1.4525,
+      "step": 730500
+    },
+    {
+      "epoch": 4.7532349307497235,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.001809870602770011,
+      "loss": 1.4505,
+      "step": 731000
+    },
+    {
+      "epoch": 4.756486117432863,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0018097405553026856,
+      "loss": 1.4475,
+      "step": 731500
+    },
+    {
+      "epoch": 4.759737304116002,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00180961050783536,
+      "loss": 1.4491,
+      "step": 732000
+    },
+    {
+      "epoch": 4.762988490799142,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0018094804603680343,
+      "loss": 1.4475,
+      "step": 732500
+    },
+    {
+      "epoch": 4.766239677482281,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0018093504129007088,
+      "loss": 1.4493,
+      "step": 733000
+    },
+    {
+      "epoch": 4.76949086416542,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0018092203654333833,
+      "loss": 1.446,
+      "step": 733500
+    },
+    {
+      "epoch": 4.77274205084856,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018090903179660575,
+      "loss": 1.4519,
+      "step": 734000
+    },
+    {
+      "epoch": 4.775993237531699,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.001808960270498732,
+      "loss": 1.4519,
+      "step": 734500
+    },
+    {
+      "epoch": 4.779244424214839,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0018088302230314065,
+      "loss": 1.4519,
+      "step": 735000
+    },
+    {
+      "epoch": 4.7824956108979775,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0018087001755640807,
+      "loss": 1.4519,
+      "step": 735500
+    },
+    {
+      "epoch": 4.785746797581117,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0018085701280967552,
+      "loss": 1.4503,
+      "step": 736000
+    },
+    {
+      "epoch": 4.788997984264256,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00180844008062943,
+      "loss": 1.4491,
+      "step": 736500
+    },
+    {
+      "epoch": 4.792249170947396,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.0018083100331621044,
+      "loss": 1.4455,
+      "step": 737000
+    },
+    {
+      "epoch": 4.7955003576305355,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0018081799856947787,
+      "loss": 1.4475,
+      "step": 737500
+    },
+    {
+      "epoch": 4.798751544313674,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.0018080499382274531,
+      "loss": 1.4454,
+      "step": 738000
+    },
+    {
+      "epoch": 4.802002730996814,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0018079198907601276,
+      "loss": 1.4446,
+      "step": 738500
+    },
+    {
+      "epoch": 4.805253917679953,
+      "grad_norm": 3.140625,
+      "learning_rate": 0.0018077898432928019,
+      "loss": 1.4439,
+      "step": 739000
+    },
+    {
+      "epoch": 4.808505104363093,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0018076597958254764,
+      "loss": 1.4392,
+      "step": 739500
+    },
+    {
+      "epoch": 4.8117562910462315,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0018075297483581508,
+      "loss": 1.4405,
+      "step": 740000
+    },
+    {
+      "epoch": 4.815007477729371,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.001807399700890825,
+      "loss": 1.4409,
+      "step": 740500
+    },
+    {
+      "epoch": 4.818258664412511,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018072696534234996,
+      "loss": 1.4425,
+      "step": 741000
+    },
+    {
+      "epoch": 4.82150985109565,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.001807139605956174,
+      "loss": 1.4373,
+      "step": 741500
+    },
+    {
+      "epoch": 4.8247610377787895,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0018070095584888483,
+      "loss": 1.4415,
+      "step": 742000
+    },
+    {
+      "epoch": 4.828012224461928,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0018068795110215228,
+      "loss": 1.4401,
+      "step": 742500
+    },
+    {
+      "epoch": 4.831263411145068,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0018067494635541973,
+      "loss": 1.4448,
+      "step": 743000
+    },
+    {
+      "epoch": 4.834514597828207,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0018066194160868715,
+      "loss": 1.4407,
+      "step": 743500
+    },
+    {
+      "epoch": 4.837765784511347,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0018064893686195462,
+      "loss": 1.4454,
+      "step": 744000
+    },
+    {
+      "epoch": 4.841016971194486,
+      "grad_norm": 3.75,
+      "learning_rate": 0.0018063593211522207,
+      "loss": 1.4433,
+      "step": 744500
+    },
+    {
+      "epoch": 4.844268157877625,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018062292736848952,
+      "loss": 1.441,
+      "step": 745000
+    },
+    {
+      "epoch": 4.847519344560765,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.0018060992262175694,
+      "loss": 1.4374,
+      "step": 745500
+    },
+    {
+      "epoch": 4.850770531243904,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.001805969178750244,
+      "loss": 1.4399,
+      "step": 746000
+    },
+    {
+      "epoch": 4.8540217179270435,
+      "grad_norm": 4.46875,
+      "learning_rate": 0.0018058391312829184,
+      "loss": 1.4392,
+      "step": 746500
+    },
+    {
+      "epoch": 4.857272904610182,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.0018057090838155927,
+      "loss": 1.4403,
+      "step": 747000
+    },
+    {
+      "epoch": 4.860524091293322,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0018055790363482671,
+      "loss": 1.4378,
+      "step": 747500
+    },
+    {
+      "epoch": 4.863775277976462,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0018054489888809416,
+      "loss": 1.4415,
+      "step": 748000
+    },
+    {
+      "epoch": 4.867026464659601,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0018053189414136159,
+      "loss": 1.4444,
+      "step": 748500
+    },
+    {
+      "epoch": 4.87027765134274,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0018051888939462904,
+      "loss": 1.441,
+      "step": 749000
+    },
+    {
+      "epoch": 4.873528838025879,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0018050588464789648,
+      "loss": 1.4387,
+      "step": 749500
+    },
+    {
+      "epoch": 4.876780024709019,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001804928799011639,
+      "loss": 1.4346,
+      "step": 750000
+    },
+    {
+      "epoch": 4.880031211392158,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0018047987515443136,
+      "loss": 1.4347,
+      "step": 750500
+    },
+    {
+      "epoch": 4.8832823980752975,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.0018046687040769883,
+      "loss": 1.4413,
+      "step": 751000
+    },
+    {
+      "epoch": 4.886533584758437,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0018045386566096628,
+      "loss": 1.4479,
+      "step": 751500
+    },
+    {
+      "epoch": 4.889784771441576,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.001804408609142337,
+      "loss": 1.446,
+      "step": 752000
+    },
+    {
+      "epoch": 4.893035958124716,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0018042785616750115,
+      "loss": 1.4515,
+      "step": 752500
+    },
+    {
+      "epoch": 4.896287144807855,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.001804148514207686,
+      "loss": 1.448,
+      "step": 753000
+    },
+    {
+      "epoch": 4.899538331490994,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0018040184667403602,
+      "loss": 1.4473,
+      "step": 753500
+    },
+    {
+      "epoch": 4.902789518174133,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018038884192730347,
+      "loss": 1.448,
+      "step": 754000
+    },
+    {
+      "epoch": 4.906040704857273,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.0018037583718057092,
+      "loss": 1.4512,
+      "step": 754500
+    },
+    {
+      "epoch": 4.909291891540413,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0018036283243383835,
+      "loss": 1.4474,
+      "step": 755000
+    },
+    {
+      "epoch": 4.9125430782235515,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.001803498276871058,
+      "loss": 1.4573,
+      "step": 755500
+    },
+    {
+      "epoch": 4.915794264906691,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0018033682294037324,
+      "loss": 1.4442,
+      "step": 756000
+    },
+    {
+      "epoch": 4.91904545158983,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.0018032381819364067,
+      "loss": 1.4525,
+      "step": 756500
+    },
+    {
+      "epoch": 4.92229663827297,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0018031081344690812,
+      "loss": 1.4477,
+      "step": 757000
+    },
+    {
+      "epoch": 4.925547824956109,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0018029780870017556,
+      "loss": 1.4477,
+      "step": 757500
+    },
+    {
+      "epoch": 4.928799011639248,
+      "grad_norm": 2.0,
+      "learning_rate": 0.0018028480395344299,
+      "loss": 1.4513,
+      "step": 758000
+    },
+    {
+      "epoch": 4.932050198322388,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0018027179920671046,
+      "loss": 1.4502,
+      "step": 758500
+    },
+    {
+      "epoch": 4.935301385005527,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.001802587944599779,
+      "loss": 1.4486,
+      "step": 759000
+    },
+    {
+      "epoch": 4.938552571688667,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0018024578971324535,
+      "loss": 1.4453,
+      "step": 759500
+    },
+    {
+      "epoch": 4.9418037583718055,
+      "grad_norm": 4.0,
+      "learning_rate": 0.0018023278496651278,
+      "loss": 1.4467,
+      "step": 760000
+    },
+    {
+      "epoch": 4.945054945054945,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0018021978021978023,
+      "loss": 1.4401,
+      "step": 760500
+    },
+    {
+      "epoch": 4.948306131738084,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0018020677547304768,
+      "loss": 1.4461,
+      "step": 761000
+    },
+    {
+      "epoch": 4.951557318421224,
+      "grad_norm": 0.875,
+      "learning_rate": 0.001801937707263151,
+      "loss": 1.4454,
+      "step": 761500
+    },
+    {
+      "epoch": 4.9548085051043635,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0018018076597958255,
+      "loss": 1.4468,
+      "step": 762000
+    },
+    {
+      "epoch": 4.958059691787502,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0018016776123285,
+      "loss": 1.4547,
+      "step": 762500
+    },
+    {
+      "epoch": 4.961310878470642,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0018015475648611742,
+      "loss": 1.4528,
+      "step": 763000
+    },
+    {
+      "epoch": 4.964562065153781,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0018014175173938487,
+      "loss": 1.4477,
+      "step": 763500
+    },
+    {
+      "epoch": 4.967813251836921,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018012874699265232,
+      "loss": 1.4547,
+      "step": 764000
+    },
+    {
+      "epoch": 4.9710644385200595,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0018011574224591975,
+      "loss": 1.4555,
+      "step": 764500
+    },
+    {
+      "epoch": 4.974315625203199,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001801027374991872,
+      "loss": 1.4538,
+      "step": 765000
+    },
+    {
+      "epoch": 4.977566811886339,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0018008973275245466,
+      "loss": 1.4523,
+      "step": 765500
+    },
+    {
+      "epoch": 4.980817998569478,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0018007672800572211,
+      "loss": 1.4484,
+      "step": 766000
+    },
+    {
+      "epoch": 4.9840691852526176,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0018006372325898954,
+      "loss": 1.4489,
+      "step": 766500
+    },
+    {
+      "epoch": 4.987320371935756,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0018005071851225699,
+      "loss": 1.4465,
+      "step": 767000
+    },
+    {
+      "epoch": 4.990571558618896,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0018003771376552443,
+      "loss": 1.4522,
+      "step": 767500
+    },
+    {
+      "epoch": 4.993822745302035,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0018002470901879186,
+      "loss": 1.453,
+      "step": 768000
+    },
+    {
+      "epoch": 4.997073931985175,
+      "grad_norm": 1.375,
+      "learning_rate": 0.001800117042720593,
+      "loss": 1.4571,
+      "step": 768500
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 1.441786766052246,
+      "eval_runtime": 0.5442,
+      "eval_samples_per_second": 1837.612,
+      "eval_steps_per_second": 29.402,
+      "step": 768950
+    },
+    {
+      "epoch": 5.0003251186683135,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0017999869952532676,
+      "loss": 1.45,
+      "step": 769000
+    },
+    {
+      "epoch": 5.003576305351453,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.0017998569477859418,
+      "loss": 1.4452,
+      "step": 769500
+    },
+    {
+      "epoch": 5.006827492034593,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0017997269003186163,
+      "loss": 1.4399,
+      "step": 770000
+    },
+    {
+      "epoch": 5.010078678717732,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.0017995968528512908,
+      "loss": 1.4414,
+      "step": 770500
+    },
+    {
+      "epoch": 5.0133298654008716,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.001799466805383965,
+      "loss": 1.4448,
+      "step": 771000
+    },
+    {
+      "epoch": 5.01658105208401,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0017993367579166395,
+      "loss": 1.4429,
+      "step": 771500
+    },
+    {
+      "epoch": 5.01983223876715,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.001799206710449314,
+      "loss": 1.4388,
+      "step": 772000
+    },
+    {
+      "epoch": 5.023083425450289,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0017990766629819882,
+      "loss": 1.442,
+      "step": 772500
+    },
+    {
+      "epoch": 5.026334612133429,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.001798946615514663,
+      "loss": 1.4433,
+      "step": 773000
+    },
+    {
+      "epoch": 5.029585798816568,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0017988165680473374,
+      "loss": 1.443,
+      "step": 773500
+    },
+    {
+      "epoch": 5.032836985499707,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001798686520580012,
+      "loss": 1.448,
+      "step": 774000
+    },
+    {
+      "epoch": 5.036088172182847,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0017985564731126862,
+      "loss": 1.4465,
+      "step": 774500
+    },
+    {
+      "epoch": 5.039339358865986,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0017984264256453606,
+      "loss": 1.4448,
+      "step": 775000
+    },
+    {
+      "epoch": 5.042590545549126,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0017982963781780351,
+      "loss": 1.4418,
+      "step": 775500
+    },
+    {
+      "epoch": 5.045841732232264,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0017981663307107094,
+      "loss": 1.4425,
+      "step": 776000
+    },
+    {
+      "epoch": 5.049092918915404,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.0017980362832433839,
+      "loss": 1.444,
+      "step": 776500
+    },
+    {
+      "epoch": 5.052344105598544,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0017979062357760583,
+      "loss": 1.4431,
+      "step": 777000
+    },
+    {
+      "epoch": 5.055595292281683,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.0017977761883087326,
+      "loss": 1.4459,
+      "step": 777500
+    },
+    {
+      "epoch": 5.058846478964822,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.001797646140841407,
+      "loss": 1.4453,
+      "step": 778000
+    },
+    {
+      "epoch": 5.062097665647961,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0017975160933740816,
+      "loss": 1.452,
+      "step": 778500
+    },
+    {
+      "epoch": 5.065348852331101,
+      "grad_norm": 1.5,
+      "learning_rate": 0.0017973860459067558,
+      "loss": 1.4492,
+      "step": 779000
+    },
+    {
+      "epoch": 5.06860003901424,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0017972559984394303,
+      "loss": 1.4533,
+      "step": 779500
+    },
+    {
+      "epoch": 5.07185122569738,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001797125950972105,
+      "loss": 1.4462,
+      "step": 780000
+    },
+    {
+      "epoch": 5.075102412380519,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0017969959035047795,
+      "loss": 1.4468,
+      "step": 780500
+    },
+    {
+      "epoch": 5.078353599063658,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0017968658560374537,
+      "loss": 1.4464,
+      "step": 781000
+    },
+    {
+      "epoch": 5.081604785746798,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.0017967358085701282,
+      "loss": 1.4535,
+      "step": 781500
+    },
+    {
+      "epoch": 5.084855972429937,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0017966057611028027,
+      "loss": 1.4473,
+      "step": 782000
+    },
+    {
+      "epoch": 5.088107159113076,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.001796475713635477,
+      "loss": 1.4522,
+      "step": 782500
+    },
+    {
+      "epoch": 5.091358345796215,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0017963456661681514,
+      "loss": 1.4525,
+      "step": 783000
+    },
+    {
+      "epoch": 5.094609532479355,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.001796215618700826,
+      "loss": 1.447,
+      "step": 783500
+    },
+    {
+      "epoch": 5.097860719162495,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0017960855712335002,
+      "loss": 1.4464,
+      "step": 784000
+    },
+    {
+      "epoch": 5.101111905845634,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0017959555237661746,
+      "loss": 1.4419,
+      "step": 784500
+    },
+    {
+      "epoch": 5.104363092528773,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0017958254762988491,
+      "loss": 1.4459,
+      "step": 785000
+    },
+    {
+      "epoch": 5.107614279211912,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0017956954288315234,
+      "loss": 1.4475,
+      "step": 785500
+    },
+    {
+      "epoch": 5.110865465895052,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0017955653813641979,
+      "loss": 1.4389,
+      "step": 786000
+    },
+    {
+      "epoch": 5.114116652578191,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0017954353338968723,
+      "loss": 1.4455,
+      "step": 786500
+    },
+    {
+      "epoch": 5.11736783926133,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0017953052864295466,
+      "loss": 1.4381,
+      "step": 787000
+    },
+    {
+      "epoch": 5.12061902594447,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0017951752389622213,
+      "loss": 1.4444,
+      "step": 787500
+    },
+    {
+      "epoch": 5.123870212627609,
+      "grad_norm": 9.1875,
+      "learning_rate": 0.0017950451914948958,
+      "loss": 1.4436,
+      "step": 788000
+    },
+    {
+      "epoch": 5.127121399310749,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0017949151440275703,
+      "loss": 1.4437,
+      "step": 788500
+    },
+    {
+      "epoch": 5.130372585993888,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0017947850965602445,
+      "loss": 1.4385,
+      "step": 789000
+    },
+    {
+      "epoch": 5.133623772677027,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.001794655049092919,
+      "loss": 1.4424,
+      "step": 789500
+    },
+    {
+      "epoch": 5.136874959360166,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0017945250016255935,
+      "loss": 1.4419,
+      "step": 790000
+    },
+    {
+      "epoch": 5.140126146043306,
+      "grad_norm": 2.765625,
+      "learning_rate": 0.0017943949541582677,
+      "loss": 1.4416,
+      "step": 790500
+    },
+    {
+      "epoch": 5.143377332726445,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.0017942649066909422,
+      "loss": 1.4391,
+      "step": 791000
+    },
+    {
+      "epoch": 5.146628519409584,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0017941348592236167,
+      "loss": 1.4361,
+      "step": 791500
+    },
+    {
+      "epoch": 5.149879706092724,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.001794004811756291,
+      "loss": 1.4397,
+      "step": 792000
+    },
+    {
+      "epoch": 5.153130892775863,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0017938747642889654,
+      "loss": 1.4351,
+      "step": 792500
+    },
+    {
+      "epoch": 5.156382079459003,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00179374471682164,
+      "loss": 1.4365,
+      "step": 793000
+    },
+    {
+      "epoch": 5.159633266142142,
+      "grad_norm": 6.6875,
+      "learning_rate": 0.0017936146693543142,
+      "loss": 1.4359,
+      "step": 793500
+    },
+    {
+      "epoch": 5.162884452825281,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0017934846218869887,
+      "loss": 1.4403,
+      "step": 794000
+    },
+    {
+      "epoch": 5.16613563950842,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0017933545744196634,
+      "loss": 1.4368,
+      "step": 794500
+    },
+    {
+      "epoch": 5.16938682619156,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0017932245269523378,
+      "loss": 1.4316,
+      "step": 795000
+    },
+    {
+      "epoch": 5.1726380128747,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.001793094479485012,
+      "loss": 1.4363,
+      "step": 795500
+    },
+    {
+      "epoch": 5.175889199557838,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0017929644320176866,
+      "loss": 1.4426,
+      "step": 796000
+    },
+    {
+      "epoch": 5.179140386240978,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.001792834384550361,
+      "loss": 1.4405,
+      "step": 796500
+    },
+    {
+      "epoch": 5.182391572924117,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0017927043370830353,
+      "loss": 1.4329,
+      "step": 797000
+    },
+    {
+      "epoch": 5.185642759607257,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0017925742896157098,
+      "loss": 1.4337,
+      "step": 797500
+    },
+    {
+      "epoch": 5.188893946290396,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0017924442421483843,
+      "loss": 1.4318,
+      "step": 798000
+    },
+    {
+      "epoch": 5.192145132973535,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0017923141946810585,
+      "loss": 1.4362,
+      "step": 798500
+    },
+    {
+      "epoch": 5.195396319656675,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.001792184147213733,
+      "loss": 1.4352,
+      "step": 799000
+    },
+    {
+      "epoch": 5.198647506339814,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0017920540997464075,
+      "loss": 1.4379,
+      "step": 799500
+    },
+    {
+      "epoch": 5.201898693022954,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.0017919240522790817,
+      "loss": 1.4319,
+      "step": 800000
+    },
+    {
+      "epoch": 5.205149879706092,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0017917940048117562,
+      "loss": 1.4339,
+      "step": 800500
+    },
+    {
+      "epoch": 5.208401066389232,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0017916639573444307,
+      "loss": 1.4374,
+      "step": 801000
+    },
+    {
+      "epoch": 5.211652253072371,
+      "grad_norm": 1.5,
+      "learning_rate": 0.001791533909877105,
+      "loss": 1.4314,
+      "step": 801500
+    },
+    {
+      "epoch": 5.214903439755511,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0017914038624097797,
+      "loss": 1.4339,
+      "step": 802000
+    },
+    {
+      "epoch": 5.2181546264386505,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0017912738149424541,
+      "loss": 1.4345,
+      "step": 802500
+    },
+    {
+      "epoch": 5.221405813121789,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0017911437674751286,
+      "loss": 1.4352,
+      "step": 803000
+    },
+    {
+      "epoch": 5.224656999804929,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0017910137200078029,
+      "loss": 1.4291,
+      "step": 803500
+    },
+    {
+      "epoch": 5.227908186488068,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0017908836725404774,
+      "loss": 1.4321,
+      "step": 804000
+    },
+    {
+      "epoch": 5.231159373171208,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0017907536250731518,
+      "loss": 1.4279,
+      "step": 804500
+    },
+    {
+      "epoch": 5.2344105598543464,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001790623577605826,
+      "loss": 1.4371,
+      "step": 805000
+    },
+    {
+      "epoch": 5.237661746537486,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0017904935301385006,
+      "loss": 1.431,
+      "step": 805500
+    },
+    {
+      "epoch": 5.240912933220626,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001790363482671175,
+      "loss": 1.4389,
+      "step": 806000
+    },
+    {
+      "epoch": 5.244164119903765,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.0017902334352038493,
+      "loss": 1.433,
+      "step": 806500
+    },
+    {
+      "epoch": 5.2474153065869045,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017901033877365238,
+      "loss": 1.4342,
+      "step": 807000
+    },
+    {
+      "epoch": 5.250666493270043,
+      "grad_norm": 4.71875,
+      "learning_rate": 0.0017899733402691983,
+      "loss": 1.4353,
+      "step": 807500
+    },
+    {
+      "epoch": 5.253917679953183,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0017898432928018725,
+      "loss": 1.4306,
+      "step": 808000
+    },
+    {
+      "epoch": 5.257168866636322,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.001789713245334547,
+      "loss": 1.4336,
+      "step": 808500
+    },
+    {
+      "epoch": 5.260420053319462,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0017895831978672217,
+      "loss": 1.4366,
+      "step": 809000
+    },
+    {
+      "epoch": 5.263671240002601,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0017894531503998962,
+      "loss": 1.4348,
+      "step": 809500
+    },
+    {
+      "epoch": 5.26692242668574,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0017893231029325705,
+      "loss": 1.4361,
+      "step": 810000
+    },
+    {
+      "epoch": 5.27017361336888,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.001789193055465245,
+      "loss": 1.4286,
+      "step": 810500
+    },
+    {
+      "epoch": 5.273424800052019,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0017890630079979194,
+      "loss": 1.4307,
+      "step": 811000
+    },
+    {
+      "epoch": 5.2766759867351585,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0017889329605305937,
+      "loss": 1.4324,
+      "step": 811500
+    },
+    {
+      "epoch": 5.279927173418297,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0017888029130632681,
+      "loss": 1.4263,
+      "step": 812000
+    },
+    {
+      "epoch": 5.283178360101437,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0017886728655959426,
+      "loss": 1.4316,
+      "step": 812500
+    },
+    {
+      "epoch": 5.286429546784577,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0017885428181286169,
+      "loss": 1.4346,
+      "step": 813000
+    },
+    {
+      "epoch": 5.289680733467716,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0017884127706612914,
+      "loss": 1.4345,
+      "step": 813500
+    },
+    {
+      "epoch": 5.292931920150855,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0017882827231939658,
+      "loss": 1.4316,
+      "step": 814000
+    },
+    {
+      "epoch": 5.296183106833994,
+      "grad_norm": 1.8671875,
+      "learning_rate": 0.00178815267572664,
+      "loss": 1.4313,
+      "step": 814500
+    },
+    {
+      "epoch": 5.299434293517134,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0017880226282593146,
+      "loss": 1.4309,
+      "step": 815000
+    },
+    {
+      "epoch": 5.302685480200273,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.001787892580791989,
+      "loss": 1.4336,
+      "step": 815500
+    },
+    {
+      "epoch": 5.3059366668834125,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017877625333246633,
+      "loss": 1.4283,
+      "step": 816000
+    },
+    {
+      "epoch": 5.309187853566552,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.001787632485857338,
+      "loss": 1.43,
+      "step": 816500
+    },
+    {
+      "epoch": 5.312439040249691,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0017875024383900125,
+      "loss": 1.4354,
+      "step": 817000
+    },
+    {
+      "epoch": 5.315690226932831,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.001787372390922687,
+      "loss": 1.4303,
+      "step": 817500
+    },
+    {
+      "epoch": 5.31894141361597,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0017872423434553612,
+      "loss": 1.4338,
+      "step": 818000
+    },
+    {
+      "epoch": 5.322192600299109,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0017871122959880357,
+      "loss": 1.4386,
+      "step": 818500
+    },
+    {
+      "epoch": 5.325443786982248,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0017869822485207102,
+      "loss": 1.4364,
+      "step": 819000
+    },
+    {
+      "epoch": 5.328694973665388,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0017868522010533845,
+      "loss": 1.4372,
+      "step": 819500
+    },
+    {
+      "epoch": 5.331946160348528,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.001786722153586059,
+      "loss": 1.4403,
+      "step": 820000
+    },
+    {
+      "epoch": 5.3351973470316665,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0017865921061187334,
+      "loss": 1.432,
+      "step": 820500
+    },
+    {
+      "epoch": 5.338448533714806,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.0017864620586514077,
+      "loss": 1.429,
+      "step": 821000
+    },
+    {
+      "epoch": 5.341699720397945,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0017863320111840822,
+      "loss": 1.4301,
+      "step": 821500
+    },
+    {
+      "epoch": 5.344950907081085,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0017862019637167566,
+      "loss": 1.4332,
+      "step": 822000
+    },
+    {
+      "epoch": 5.348202093764224,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001786071916249431,
+      "loss": 1.4334,
+      "step": 822500
+    },
+    {
+      "epoch": 5.351453280447363,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0017859418687821054,
+      "loss": 1.4321,
+      "step": 823000
+    },
+    {
+      "epoch": 5.354704467130503,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00178581182131478,
+      "loss": 1.4315,
+      "step": 823500
+    },
+    {
+      "epoch": 5.357955653813642,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0017856817738474545,
+      "loss": 1.4369,
+      "step": 824000
+    },
+    {
+      "epoch": 5.361206840496782,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.0017855517263801288,
+      "loss": 1.4303,
+      "step": 824500
+    },
+    {
+      "epoch": 5.3644580271799205,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0017854216789128033,
+      "loss": 1.4386,
+      "step": 825000
+    },
+    {
+      "epoch": 5.36770921386306,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0017852916314454778,
+      "loss": 1.4326,
+      "step": 825500
+    },
+    {
+      "epoch": 5.370960400546199,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001785161583978152,
+      "loss": 1.4357,
+      "step": 826000
+    },
+    {
+      "epoch": 5.374211587229339,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0017850315365108265,
+      "loss": 1.4397,
+      "step": 826500
+    },
+    {
+      "epoch": 5.3774627739124785,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.001784901489043501,
+      "loss": 1.4392,
+      "step": 827000
+    },
+    {
+      "epoch": 5.380713960595617,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0017847714415761752,
+      "loss": 1.4344,
+      "step": 827500
+    },
+    {
+      "epoch": 5.383965147278757,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0017846413941088497,
+      "loss": 1.4347,
+      "step": 828000
+    },
+    {
+      "epoch": 5.387216333961896,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0017845113466415242,
+      "loss": 1.4388,
+      "step": 828500
+    },
+    {
+      "epoch": 5.390467520645036,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0017843812991741985,
+      "loss": 1.4361,
+      "step": 829000
+    },
+    {
+      "epoch": 5.3937187073281745,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.001784251251706873,
+      "loss": 1.4407,
+      "step": 829500
+    },
+    {
+      "epoch": 5.396969894011314,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0017841212042395474,
+      "loss": 1.4423,
+      "step": 830000
+    },
+    {
+      "epoch": 5.400221080694454,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0017839911567722217,
+      "loss": 1.4417,
+      "step": 830500
+    },
+    {
+      "epoch": 5.403472267377593,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.0017838611093048964,
+      "loss": 1.4446,
+      "step": 831000
+    },
+    {
+      "epoch": 5.4067234540607325,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0017837310618375709,
+      "loss": 1.4417,
+      "step": 831500
+    },
+    {
+      "epoch": 5.409974640743871,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0017836010143702453,
+      "loss": 1.4407,
+      "step": 832000
+    },
+    {
+      "epoch": 5.413225827427011,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0017834709669029196,
+      "loss": 1.4398,
+      "step": 832500
+    },
+    {
+      "epoch": 5.41647701411015,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.001783340919435594,
+      "loss": 1.444,
+      "step": 833000
+    },
+    {
+      "epoch": 5.41972820079329,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0017832108719682686,
+      "loss": 1.4429,
+      "step": 833500
+    },
+    {
+      "epoch": 5.4229793874764285,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0017830808245009428,
+      "loss": 1.448,
+      "step": 834000
+    },
+    {
+      "epoch": 5.426230574159568,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017829507770336173,
+      "loss": 1.4424,
+      "step": 834500
+    },
+    {
+      "epoch": 5.429481760842708,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.0017828207295662918,
+      "loss": 1.4442,
+      "step": 835000
+    },
+    {
+      "epoch": 5.432732947525847,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.001782690682098966,
+      "loss": 1.4364,
+      "step": 835500
+    },
+    {
+      "epoch": 5.4359841342089865,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0017825606346316405,
+      "loss": 1.4351,
+      "step": 836000
+    },
+    {
+      "epoch": 5.439235320892125,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.001782430587164315,
+      "loss": 1.4341,
+      "step": 836500
+    },
+    {
+      "epoch": 5.442486507575265,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0017823005396969893,
+      "loss": 1.4395,
+      "step": 837000
+    },
+    {
+      "epoch": 5.445737694258404,
+      "grad_norm": 2.65625,
+      "learning_rate": 0.0017821704922296637,
+      "loss": 1.4416,
+      "step": 837500
+    },
+    {
+      "epoch": 5.448988880941544,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0017820404447623384,
+      "loss": 1.4371,
+      "step": 838000
+    },
+    {
+      "epoch": 5.452240067624683,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.001781910397295013,
+      "loss": 1.4336,
+      "step": 838500
+    },
+    {
+      "epoch": 5.455491254307822,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0017817803498276872,
+      "loss": 1.4373,
+      "step": 839000
+    },
+    {
+      "epoch": 5.458742440990962,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0017816503023603616,
+      "loss": 1.4375,
+      "step": 839500
+    },
+    {
+      "epoch": 5.461993627674101,
+      "grad_norm": 3.96875,
+      "learning_rate": 0.0017815202548930361,
+      "loss": 1.4356,
+      "step": 840000
+    },
+    {
+      "epoch": 5.4652448143572405,
+      "grad_norm": 3.21875,
+      "learning_rate": 0.0017813902074257104,
+      "loss": 1.4346,
+      "step": 840500
+    },
+    {
+      "epoch": 5.468496001040379,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0017812601599583849,
+      "loss": 1.4327,
+      "step": 841000
+    },
+    {
+      "epoch": 5.471747187723519,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0017811301124910593,
+      "loss": 1.4347,
+      "step": 841500
+    },
+    {
+      "epoch": 5.474998374406659,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0017810000650237336,
+      "loss": 1.439,
+      "step": 842000
+    },
+    {
+      "epoch": 5.478249561089798,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.001780870017556408,
+      "loss": 1.4383,
+      "step": 842500
+    },
+    {
+      "epoch": 5.481500747772937,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0017807399700890826,
+      "loss": 1.4395,
+      "step": 843000
+    },
+    {
+      "epoch": 5.484751934456076,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0017806099226217568,
+      "loss": 1.4374,
+      "step": 843500
+    },
+    {
+      "epoch": 5.488003121139216,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0017804798751544313,
+      "loss": 1.445,
+      "step": 844000
+    },
+    {
+      "epoch": 5.491254307822355,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0017803498276871058,
+      "loss": 1.4482,
+      "step": 844500
+    },
+    {
+      "epoch": 5.4945054945054945,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00178021978021978,
+      "loss": 1.4499,
+      "step": 845000
+    },
+    {
+      "epoch": 5.497756681188634,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0017800897327524547,
+      "loss": 1.4462,
+      "step": 845500
+    },
+    {
+      "epoch": 5.501007867871773,
+      "grad_norm": 3.71875,
+      "learning_rate": 0.0017799596852851292,
+      "loss": 1.4478,
+      "step": 846000
+    },
+    {
+      "epoch": 5.504259054554913,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0017798296378178037,
+      "loss": 1.4479,
+      "step": 846500
+    },
+    {
+      "epoch": 5.507510241238052,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001779699590350478,
+      "loss": 1.4523,
+      "step": 847000
+    },
+    {
+      "epoch": 5.510761427921191,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0017795695428831524,
+      "loss": 1.4498,
+      "step": 847500
+    },
+    {
+      "epoch": 5.51401261460433,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.001779439495415827,
+      "loss": 1.4493,
+      "step": 848000
+    },
+    {
+      "epoch": 5.51726380128747,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0017793094479485012,
+      "loss": 1.454,
+      "step": 848500
+    },
+    {
+      "epoch": 5.520514987970609,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0017791794004811757,
+      "loss": 1.4481,
+      "step": 849000
+    },
+    {
+      "epoch": 5.5237661746537485,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0017790493530138501,
+      "loss": 1.4474,
+      "step": 849500
+    },
+    {
+      "epoch": 5.527017361336888,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0017789193055465244,
+      "loss": 1.4459,
+      "step": 850000
+    },
+    {
+      "epoch": 5.530268548020027,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0017787892580791989,
+      "loss": 1.4437,
+      "step": 850500
+    },
+    {
+      "epoch": 5.533519734703167,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0017786592106118734,
+      "loss": 1.4456,
+      "step": 851000
+    },
+    {
+      "epoch": 5.536770921386306,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0017785291631445476,
+      "loss": 1.44,
+      "step": 851500
+    },
+    {
+      "epoch": 5.540022108069445,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.001778399115677222,
+      "loss": 1.4362,
+      "step": 852000
+    },
+    {
+      "epoch": 5.543273294752584,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0017782690682098968,
+      "loss": 1.4389,
+      "step": 852500
+    },
+    {
+      "epoch": 5.546524481435724,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0017781390207425713,
+      "loss": 1.4365,
+      "step": 853000
+    },
+    {
+      "epoch": 5.549775668118864,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0017780089732752455,
+      "loss": 1.4347,
+      "step": 853500
+    },
+    {
+      "epoch": 5.5530268548020025,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00177787892580792,
+      "loss": 1.433,
+      "step": 854000
+    },
+    {
+      "epoch": 5.556278041485142,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0017777488783405945,
+      "loss": 1.4389,
+      "step": 854500
+    },
+    {
+      "epoch": 5.559529228168281,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0017776188308732687,
+      "loss": 1.4277,
+      "step": 855000
+    },
+    {
+      "epoch": 5.562780414851421,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0017774887834059432,
+      "loss": 1.4311,
+      "step": 855500
+    },
+    {
+      "epoch": 5.56603160153456,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0017773587359386177,
+      "loss": 1.4269,
+      "step": 856000
+    },
+    {
+      "epoch": 5.569282788217699,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001777228688471292,
+      "loss": 1.4299,
+      "step": 856500
+    },
+    {
+      "epoch": 5.572533974900839,
+      "grad_norm": 4.78125,
+      "learning_rate": 0.0017770986410039664,
+      "loss": 1.4351,
+      "step": 857000
+    },
+    {
+      "epoch": 5.575785161583978,
+      "grad_norm": 2.921875,
+      "learning_rate": 0.001776968593536641,
+      "loss": 1.4311,
+      "step": 857500
+    },
+    {
+      "epoch": 5.579036348267118,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0017768385460693152,
+      "loss": 1.4392,
+      "step": 858000
+    },
+    {
+      "epoch": 5.5822875349502565,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.0017767084986019897,
+      "loss": 1.4358,
+      "step": 858500
+    },
+    {
+      "epoch": 5.585538721633396,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0017765784511346641,
+      "loss": 1.4362,
+      "step": 859000
+    },
+    {
+      "epoch": 5.588789908316535,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0017764484036673384,
+      "loss": 1.4434,
+      "step": 859500
+    },
+    {
+      "epoch": 5.592041094999675,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001776318356200013,
+      "loss": 1.4344,
+      "step": 860000
+    },
+    {
+      "epoch": 5.595292281682815,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0017761883087326876,
+      "loss": 1.4339,
+      "step": 860500
+    },
+    {
+      "epoch": 5.598543468365953,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.001776058261265362,
+      "loss": 1.4316,
+      "step": 861000
+    },
+    {
+      "epoch": 5.601794655049093,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0017759282137980363,
+      "loss": 1.4352,
+      "step": 861500
+    },
+    {
+      "epoch": 5.605045841732232,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0017757981663307108,
+      "loss": 1.4341,
+      "step": 862000
+    },
+    {
+      "epoch": 5.608297028415372,
+      "grad_norm": 3.8125,
+      "learning_rate": 0.0017756681188633853,
+      "loss": 1.4367,
+      "step": 862500
+    },
+    {
+      "epoch": 5.6115482150985105,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0017755380713960595,
+      "loss": 1.4371,
+      "step": 863000
+    },
+    {
+      "epoch": 5.61479940178165,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.001775408023928734,
+      "loss": 1.434,
+      "step": 863500
+    },
+    {
+      "epoch": 5.61805058846479,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0017752779764614085,
+      "loss": 1.4328,
+      "step": 864000
+    },
+    {
+      "epoch": 5.621301775147929,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.0017751479289940828,
+      "loss": 1.4349,
+      "step": 864500
+    },
+    {
+      "epoch": 5.624552961831069,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0017750178815267572,
+      "loss": 1.4316,
+      "step": 865000
+    },
+    {
+      "epoch": 5.627804148514207,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0017748878340594317,
+      "loss": 1.4332,
+      "step": 865500
+    },
+    {
+      "epoch": 5.631055335197347,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.001774757786592106,
+      "loss": 1.4338,
+      "step": 866000
+    },
+    {
+      "epoch": 5.634306521880486,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0017746277391247805,
+      "loss": 1.4292,
+      "step": 866500
+    },
+    {
+      "epoch": 5.637557708563626,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0017744976916574551,
+      "loss": 1.4338,
+      "step": 867000
+    },
+    {
+      "epoch": 5.640808895246765,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0017743676441901296,
+      "loss": 1.4284,
+      "step": 867500
+    },
+    {
+      "epoch": 5.644060081929904,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0017742375967228039,
+      "loss": 1.4261,
+      "step": 868000
+    },
+    {
+      "epoch": 5.647311268613044,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.0017741075492554784,
+      "loss": 1.4281,
+      "step": 868500
+    },
+    {
+      "epoch": 5.650562455296183,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017739775017881528,
+      "loss": 1.4258,
+      "step": 869000
+    },
+    {
+      "epoch": 5.653813641979323,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001773847454320827,
+      "loss": 1.4271,
+      "step": 869500
+    },
+    {
+      "epoch": 5.657064828662461,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0017737174068535016,
+      "loss": 1.4319,
+      "step": 870000
+    },
+    {
+      "epoch": 5.660316015345601,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.001773587359386176,
+      "loss": 1.4392,
+      "step": 870500
+    },
+    {
+      "epoch": 5.663567202028741,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0017734573119188503,
+      "loss": 1.4391,
+      "step": 871000
+    },
+    {
+      "epoch": 5.66681838871188,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0017733272644515248,
+      "loss": 1.4384,
+      "step": 871500
+    },
+    {
+      "epoch": 5.670069575395019,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0017731972169841993,
+      "loss": 1.433,
+      "step": 872000
+    },
+    {
+      "epoch": 5.673320762078158,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0017730671695168735,
+      "loss": 1.4417,
+      "step": 872500
+    },
+    {
+      "epoch": 5.676571948761298,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001772937122049548,
+      "loss": 1.4341,
+      "step": 873000
+    },
+    {
+      "epoch": 5.679823135444437,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0017728070745822225,
+      "loss": 1.4381,
+      "step": 873500
+    },
+    {
+      "epoch": 5.683074322127577,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0017726770271148968,
+      "loss": 1.4375,
+      "step": 874000
+    },
+    {
+      "epoch": 5.686325508810716,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0017725469796475715,
+      "loss": 1.4401,
+      "step": 874500
+    },
+    {
+      "epoch": 5.689576695493855,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.001772416932180246,
+      "loss": 1.4388,
+      "step": 875000
+    },
+    {
+      "epoch": 5.692827882176995,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.0017722868847129204,
+      "loss": 1.4357,
+      "step": 875500
+    },
+    {
+      "epoch": 5.696079068860134,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0017721568372455947,
+      "loss": 1.4377,
+      "step": 876000
+    },
+    {
+      "epoch": 5.699330255543273,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.0017720267897782692,
+      "loss": 1.433,
+      "step": 876500
+    },
+    {
+      "epoch": 5.702581442226412,
+      "grad_norm": 4.15625,
+      "learning_rate": 0.0017718967423109436,
+      "loss": 1.4328,
+      "step": 877000
+    },
+    {
+      "epoch": 5.705832628909552,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.001771766694843618,
+      "loss": 1.4349,
+      "step": 877500
+    },
+    {
+      "epoch": 5.709083815592692,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017716366473762924,
+      "loss": 1.4363,
+      "step": 878000
+    },
+    {
+      "epoch": 5.712335002275831,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.0017715065999089669,
+      "loss": 1.4388,
+      "step": 878500
+    },
+    {
+      "epoch": 5.71558618895897,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0017713765524416411,
+      "loss": 1.4395,
+      "step": 879000
+    },
+    {
+      "epoch": 5.718837375642109,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0017712465049743156,
+      "loss": 1.4353,
+      "step": 879500
+    },
+    {
+      "epoch": 5.722088562325249,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00177111645750699,
+      "loss": 1.4373,
+      "step": 880000
+    },
+    {
+      "epoch": 5.725339749008388,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0017709864100396643,
+      "loss": 1.4349,
+      "step": 880500
+    },
+    {
+      "epoch": 5.728590935691527,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0017708563625723388,
+      "loss": 1.4382,
+      "step": 881000
+    },
+    {
+      "epoch": 5.731842122374667,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0017707263151050135,
+      "loss": 1.4353,
+      "step": 881500
+    },
+    {
+      "epoch": 5.735093309057806,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.001770596267637688,
+      "loss": 1.4369,
+      "step": 882000
+    },
+    {
+      "epoch": 5.738344495740946,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0017704662201703622,
+      "loss": 1.4347,
+      "step": 882500
+    },
+    {
+      "epoch": 5.741595682424085,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0017703361727030367,
+      "loss": 1.4345,
+      "step": 883000
+    },
+    {
+      "epoch": 5.744846869107224,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0017702061252357112,
+      "loss": 1.4395,
+      "step": 883500
+    },
+    {
+      "epoch": 5.748098055790363,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0017700760777683855,
+      "loss": 1.4383,
+      "step": 884000
+    },
+    {
+      "epoch": 5.751349242473503,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00176994603030106,
+      "loss": 1.4386,
+      "step": 884500
+    },
+    {
+      "epoch": 5.754600429156643,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0017698159828337344,
+      "loss": 1.4323,
+      "step": 885000
+    },
+    {
+      "epoch": 5.7578516158397814,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0017696859353664087,
+      "loss": 1.4471,
+      "step": 885500
+    },
+    {
+      "epoch": 5.761102802522921,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0017695558878990832,
+      "loss": 1.437,
+      "step": 886000
+    },
+    {
+      "epoch": 5.76435398920606,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0017694258404317576,
+      "loss": 1.4366,
+      "step": 886500
+    },
+    {
+      "epoch": 5.7676051758892,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.001769295792964432,
+      "loss": 1.4318,
+      "step": 887000
+    },
+    {
+      "epoch": 5.770856362572339,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0017691657454971064,
+      "loss": 1.4392,
+      "step": 887500
+    },
+    {
+      "epoch": 5.774107549255478,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0017690356980297809,
+      "loss": 1.437,
+      "step": 888000
+    },
+    {
+      "epoch": 5.777358735938618,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0017689056505624551,
+      "loss": 1.4328,
+      "step": 888500
+    },
+    {
+      "epoch": 5.780609922621757,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0017687756030951298,
+      "loss": 1.4395,
+      "step": 889000
+    },
+    {
+      "epoch": 5.783861109304897,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0017686455556278043,
+      "loss": 1.4352,
+      "step": 889500
+    },
+    {
+      "epoch": 5.7871122959880354,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0017685155081604788,
+      "loss": 1.4312,
+      "step": 890000
+    },
+    {
+      "epoch": 5.790363482671175,
+      "grad_norm": 0.875,
+      "learning_rate": 0.001768385460693153,
+      "loss": 1.4387,
+      "step": 890500
+    },
+    {
+      "epoch": 5.793614669354314,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0017682554132258275,
+      "loss": 1.4412,
+      "step": 891000
+    },
+    {
+      "epoch": 5.796865856037454,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001768125365758502,
+      "loss": 1.4367,
+      "step": 891500
+    },
+    {
+      "epoch": 5.8001170427205935,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0017679953182911763,
+      "loss": 1.4418,
+      "step": 892000
+    },
+    {
+      "epoch": 5.803368229403732,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0017678652708238507,
+      "loss": 1.4393,
+      "step": 892500
+    },
+    {
+      "epoch": 5.806619416086872,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0017677352233565252,
+      "loss": 1.4412,
+      "step": 893000
+    },
+    {
+      "epoch": 5.809870602770011,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0017676051758891995,
+      "loss": 1.4506,
+      "step": 893500
+    },
+    {
+      "epoch": 5.813121789453151,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.001767475128421874,
+      "loss": 1.4387,
+      "step": 894000
+    },
+    {
+      "epoch": 5.8163729761362895,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0017673450809545484,
+      "loss": 1.4368,
+      "step": 894500
+    },
+    {
+      "epoch": 5.819624162819429,
+      "grad_norm": 2.6875,
+      "learning_rate": 0.0017672150334872227,
+      "loss": 1.4371,
+      "step": 895000
+    },
+    {
+      "epoch": 5.822875349502569,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0017670849860198972,
+      "loss": 1.4433,
+      "step": 895500
+    },
+    {
+      "epoch": 5.826126536185708,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0017669549385525719,
+      "loss": 1.4398,
+      "step": 896000
+    },
+    {
+      "epoch": 5.8293777228688475,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0017668248910852463,
+      "loss": 1.4315,
+      "step": 896500
+    },
+    {
+      "epoch": 5.832628909551986,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0017666948436179206,
+      "loss": 1.4369,
+      "step": 897000
+    },
+    {
+      "epoch": 5.835880096235126,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.001766564796150595,
+      "loss": 1.4403,
+      "step": 897500
+    },
+    {
+      "epoch": 5.839131282918265,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0017664347486832696,
+      "loss": 1.4389,
+      "step": 898000
+    },
+    {
+      "epoch": 5.842382469601405,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0017663047012159438,
+      "loss": 1.44,
+      "step": 898500
+    },
+    {
+      "epoch": 5.845633656284544,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0017661746537486183,
+      "loss": 1.4428,
+      "step": 899000
+    },
+    {
+      "epoch": 5.848884842967683,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0017660446062812928,
+      "loss": 1.4401,
+      "step": 899500
+    },
+    {
+      "epoch": 5.852136029650823,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.001765914558813967,
+      "loss": 1.4453,
+      "step": 900000
+    },
+    {
+      "epoch": 5.855387216333962,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0017657845113466415,
+      "loss": 1.4403,
+      "step": 900500
+    },
+    {
+      "epoch": 5.8586384030171015,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001765654463879316,
+      "loss": 1.4371,
+      "step": 901000
+    },
+    {
+      "epoch": 5.86188958970024,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0017655244164119903,
+      "loss": 1.4415,
+      "step": 901500
+    },
+    {
+      "epoch": 5.86514077638338,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0017653943689446647,
+      "loss": 1.4538,
+      "step": 902000
+    },
+    {
+      "epoch": 5.86839196306652,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0017652643214773392,
+      "loss": 1.4442,
+      "step": 902500
+    },
+    {
+      "epoch": 5.871643149749659,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0017651342740100135,
+      "loss": 1.4423,
+      "step": 903000
+    },
+    {
+      "epoch": 5.874894336432798,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0017650042265426882,
+      "loss": 1.4468,
+      "step": 903500
+    },
+    {
+      "epoch": 5.878145523115937,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0017648741790753627,
+      "loss": 1.4504,
+      "step": 904000
+    },
+    {
+      "epoch": 5.881396709799077,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0017647441316080371,
+      "loss": 1.4504,
+      "step": 904500
+    },
+    {
+      "epoch": 5.884647896482216,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0017646140841407114,
+      "loss": 1.4506,
+      "step": 905000
+    },
+    {
+      "epoch": 5.8878990831653555,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0017644840366733859,
+      "loss": 1.4464,
+      "step": 905500
+    },
+    {
+      "epoch": 5.891150269848494,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0017643539892060604,
+      "loss": 1.4466,
+      "step": 906000
+    },
+    {
+      "epoch": 5.894401456531634,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0017642239417387346,
+      "loss": 1.4492,
+      "step": 906500
+    },
+    {
+      "epoch": 5.897652643214773,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.001764093894271409,
+      "loss": 1.4484,
+      "step": 907000
+    },
+    {
+      "epoch": 5.900903829897913,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0017639638468040836,
+      "loss": 1.453,
+      "step": 907500
+    },
+    {
+      "epoch": 5.904155016581052,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0017638337993367578,
+      "loss": 1.4505,
+      "step": 908000
+    },
+    {
+      "epoch": 5.907406203264191,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.0017637037518694323,
+      "loss": 1.4462,
+      "step": 908500
+    },
+    {
+      "epoch": 5.910657389947331,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0017635737044021068,
+      "loss": 1.4504,
+      "step": 909000
+    },
+    {
+      "epoch": 5.91390857663047,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001763443656934781,
+      "loss": 1.4482,
+      "step": 909500
+    },
+    {
+      "epoch": 5.9171597633136095,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0017633136094674555,
+      "loss": 1.448,
+      "step": 910000
+    },
+    {
+      "epoch": 5.920410949996748,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0017631835620001302,
+      "loss": 1.4487,
+      "step": 910500
+    },
+    {
+      "epoch": 5.923662136679888,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0017630535145328047,
+      "loss": 1.4487,
+      "step": 911000
+    },
+    {
+      "epoch": 5.926913323363028,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.001762923467065479,
+      "loss": 1.451,
+      "step": 911500
+    },
+    {
+      "epoch": 5.930164510046167,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0017627934195981534,
+      "loss": 1.4477,
+      "step": 912000
+    },
+    {
+      "epoch": 5.933415696729306,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.001762663372130828,
+      "loss": 1.4422,
+      "step": 912500
+    },
+    {
+      "epoch": 5.936666883412445,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0017625333246635022,
+      "loss": 1.4406,
+      "step": 913000
+    },
+    {
+      "epoch": 5.939918070095585,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0017624032771961767,
+      "loss": 1.4463,
+      "step": 913500
+    },
+    {
+      "epoch": 5.943169256778724,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0017622732297288511,
+      "loss": 1.4441,
+      "step": 914000
+    },
+    {
+      "epoch": 5.9464204434618635,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.0017621431822615254,
+      "loss": 1.4461,
+      "step": 914500
+    },
+    {
+      "epoch": 5.949671630145003,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0017620131347941999,
+      "loss": 1.4471,
+      "step": 915000
+    },
+    {
+      "epoch": 5.952922816828142,
+      "grad_norm": 3.171875,
+      "learning_rate": 0.0017618830873268744,
+      "loss": 1.4448,
+      "step": 915500
+    },
+    {
+      "epoch": 5.956174003511282,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0017617530398595486,
+      "loss": 1.4441,
+      "step": 916000
+    },
+    {
+      "epoch": 5.959425190194421,
+      "grad_norm": 0.875,
+      "learning_rate": 0.001761622992392223,
+      "loss": 1.4413,
+      "step": 916500
+    },
+    {
+      "epoch": 5.96267637687756,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0017614929449248976,
+      "loss": 1.4439,
+      "step": 917000
+    },
+    {
+      "epoch": 5.965927563560699,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0017613628974575718,
+      "loss": 1.4338,
+      "step": 917500
+    },
+    {
+      "epoch": 5.969178750243839,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0017612328499902465,
+      "loss": 1.4402,
+      "step": 918000
+    },
+    {
+      "epoch": 5.972429936926979,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.001761102802522921,
+      "loss": 1.4474,
+      "step": 918500
+    },
+    {
+      "epoch": 5.9756811236101175,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0017609727550555955,
+      "loss": 1.4411,
+      "step": 919000
+    },
+    {
+      "epoch": 5.978932310293257,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.0017608427075882698,
+      "loss": 1.4379,
+      "step": 919500
+    },
+    {
+      "epoch": 5.982183496976396,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.0017607126601209442,
+      "loss": 1.4487,
+      "step": 920000
+    },
+    {
+      "epoch": 5.985434683659536,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0017605826126536187,
+      "loss": 1.4472,
+      "step": 920500
+    },
+    {
+      "epoch": 5.988685870342675,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.001760452565186293,
+      "loss": 1.4468,
+      "step": 921000
+    },
+    {
+      "epoch": 5.991937057025814,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0017603225177189675,
+      "loss": 1.443,
+      "step": 921500
+    },
+    {
+      "epoch": 5.995188243708954,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001760192470251642,
+      "loss": 1.4388,
+      "step": 922000
+    },
+    {
+      "epoch": 5.998439430392093,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0017600624227843162,
+      "loss": 1.443,
+      "step": 922500
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 1.4233994483947754,
+      "eval_runtime": 0.5293,
+      "eval_samples_per_second": 1889.415,
+      "eval_steps_per_second": 30.231,
+      "step": 922740
+    },
+    {
+      "epoch": 6.001690617075233,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0017599323753169907,
+      "loss": 1.4447,
+      "step": 923000
+    },
+    {
+      "epoch": 6.0049418037583715,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0017598023278496651,
+      "loss": 1.4429,
+      "step": 923500
+    },
+    {
+      "epoch": 6.008192990441511,
+      "grad_norm": 3.890625,
+      "learning_rate": 0.0017596722803823394,
+      "loss": 1.4379,
+      "step": 924000
+    },
+    {
+      "epoch": 6.01144417712465,
+      "grad_norm": 3.109375,
+      "learning_rate": 0.0017595422329150139,
+      "loss": 1.4337,
+      "step": 924500
+    },
+    {
+      "epoch": 6.01469536380779,
+      "grad_norm": 2.984375,
+      "learning_rate": 0.0017594121854476886,
+      "loss": 1.4372,
+      "step": 925000
+    },
+    {
+      "epoch": 6.0179465504909295,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.001759282137980363,
+      "loss": 1.4368,
+      "step": 925500
+    },
+    {
+      "epoch": 6.021197737174068,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.0017591520905130373,
+      "loss": 1.4414,
+      "step": 926000
+    },
+    {
+      "epoch": 6.024448923857208,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0017590220430457118,
+      "loss": 1.4377,
+      "step": 926500
+    },
+    {
+      "epoch": 6.027700110540347,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0017588919955783863,
+      "loss": 1.4385,
+      "step": 927000
+    },
+    {
+      "epoch": 6.030951297223487,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0017587619481110605,
+      "loss": 1.437,
+      "step": 927500
+    },
+    {
+      "epoch": 6.0342024839066255,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001758631900643735,
+      "loss": 1.4363,
+      "step": 928000
+    },
+    {
+      "epoch": 6.037453670589765,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0017585018531764095,
+      "loss": 1.4425,
+      "step": 928500
+    },
+    {
+      "epoch": 6.040704857272905,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0017583718057090838,
+      "loss": 1.4346,
+      "step": 929000
+    },
+    {
+      "epoch": 6.043956043956044,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0017582417582417582,
+      "loss": 1.4351,
+      "step": 929500
+    },
+    {
+      "epoch": 6.0472072306391835,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0017581117107744327,
+      "loss": 1.4381,
+      "step": 930000
+    },
+    {
+      "epoch": 6.050458417322322,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.001757981663307107,
+      "loss": 1.4369,
+      "step": 930500
+    },
+    {
+      "epoch": 6.053709604005462,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0017578516158397815,
+      "loss": 1.4339,
+      "step": 931000
+    },
+    {
+      "epoch": 6.056960790688601,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.001757721568372456,
+      "loss": 1.432,
+      "step": 931500
+    },
+    {
+      "epoch": 6.060211977371741,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0017575915209051302,
+      "loss": 1.429,
+      "step": 932000
+    },
+    {
+      "epoch": 6.06346316405488,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001757461473437805,
+      "loss": 1.4366,
+      "step": 932500
+    },
+    {
+      "epoch": 6.066714350738019,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0017573314259704794,
+      "loss": 1.4358,
+      "step": 933000
+    },
+    {
+      "epoch": 6.069965537421159,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0017572013785031539,
+      "loss": 1.4321,
+      "step": 933500
+    },
+    {
+      "epoch": 6.073216724104298,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0017570713310358281,
+      "loss": 1.4307,
+      "step": 934000
+    },
+    {
+      "epoch": 6.0764679107874375,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0017569412835685026,
+      "loss": 1.4316,
+      "step": 934500
+    },
+    {
+      "epoch": 6.079719097470576,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001756811236101177,
+      "loss": 1.4334,
+      "step": 935000
+    },
+    {
+      "epoch": 6.082970284153716,
+      "grad_norm": 15.6875,
+      "learning_rate": 0.0017566811886338513,
+      "loss": 1.4387,
+      "step": 935500
+    },
+    {
+      "epoch": 6.086221470836856,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0017565511411665258,
+      "loss": 1.4378,
+      "step": 936000
+    },
+    {
+      "epoch": 6.089472657519995,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.0017564210936992003,
+      "loss": 1.438,
+      "step": 936500
+    },
+    {
+      "epoch": 6.092723844203134,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.0017562910462318746,
+      "loss": 1.4442,
+      "step": 937000
+    },
+    {
+      "epoch": 6.095975030886273,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001756160998764549,
+      "loss": 1.4395,
+      "step": 937500
+    },
+    {
+      "epoch": 6.099226217569413,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0017560309512972235,
+      "loss": 1.4356,
+      "step": 938000
+    },
+    {
+      "epoch": 6.102477404252552,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0017559009038298978,
+      "loss": 1.4354,
+      "step": 938500
+    },
+    {
+      "epoch": 6.1057285909356915,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.0017557708563625722,
+      "loss": 1.4383,
+      "step": 939000
+    },
+    {
+      "epoch": 6.108979777618831,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.001755640808895247,
+      "loss": 1.4348,
+      "step": 939500
+    },
+    {
+      "epoch": 6.11223096430197,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0017555107614279214,
+      "loss": 1.4367,
+      "step": 940000
+    },
+    {
+      "epoch": 6.11548215098511,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0017553807139605957,
+      "loss": 1.4381,
+      "step": 940500
+    },
+    {
+      "epoch": 6.118733337668249,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0017552506664932702,
+      "loss": 1.4336,
+      "step": 941000
+    },
+    {
+      "epoch": 6.121984524351388,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.0017551206190259446,
+      "loss": 1.4387,
+      "step": 941500
+    },
+    {
+      "epoch": 6.125235711034527,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001754990571558619,
+      "loss": 1.4386,
+      "step": 942000
+    },
+    {
+      "epoch": 6.128486897717667,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0017548605240912934,
+      "loss": 1.4418,
+      "step": 942500
+    },
+    {
+      "epoch": 6.131738084400807,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0017547304766239679,
+      "loss": 1.4401,
+      "step": 943000
+    },
+    {
+      "epoch": 6.1349892710839455,
+      "grad_norm": 4.8125,
+      "learning_rate": 0.0017546004291566421,
+      "loss": 1.4466,
+      "step": 943500
+    },
+    {
+      "epoch": 6.138240457767085,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0017544703816893166,
+      "loss": 1.4419,
+      "step": 944000
+    },
+    {
+      "epoch": 6.141491644450224,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.001754340334221991,
+      "loss": 1.4366,
+      "step": 944500
+    },
+    {
+      "epoch": 6.144742831133364,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.0017542102867546653,
+      "loss": 1.4378,
+      "step": 945000
+    },
+    {
+      "epoch": 6.147994017816503,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.0017540802392873398,
+      "loss": 1.4372,
+      "step": 945500
+    },
+    {
+      "epoch": 6.151245204499642,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.0017539501918200143,
+      "loss": 1.4418,
+      "step": 946000
+    },
+    {
+      "epoch": 6.154496391182782,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0017538201443526886,
+      "loss": 1.4448,
+      "step": 946500
+    },
+    {
+      "epoch": 6.157747577865921,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0017536900968853633,
+      "loss": 1.4402,
+      "step": 947000
+    },
+    {
+      "epoch": 6.160998764549061,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0017535600494180377,
+      "loss": 1.44,
+      "step": 947500
+    },
+    {
+      "epoch": 6.1642499512321995,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0017534300019507122,
+      "loss": 1.4443,
+      "step": 948000
+    },
+    {
+      "epoch": 6.167501137915339,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0017532999544833865,
+      "loss": 1.4419,
+      "step": 948500
+    },
+    {
+      "epoch": 6.170752324598478,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.001753169907016061,
+      "loss": 1.4433,
+      "step": 949000
+    },
+    {
+      "epoch": 6.174003511281618,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0017530398595487354,
+      "loss": 1.4432,
+      "step": 949500
+    },
+    {
+      "epoch": 6.177254697964758,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.0017529098120814097,
+      "loss": 1.4434,
+      "step": 950000
+    },
+    {
+      "epoch": 6.180505884647896,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0017527797646140842,
+      "loss": 1.4415,
+      "step": 950500
+    },
+    {
+      "epoch": 6.183757071331036,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.0017526497171467586,
+      "loss": 1.4471,
+      "step": 951000
+    },
+    {
+      "epoch": 6.187008258014175,
+      "grad_norm": 0.875,
+      "learning_rate": 0.001752519669679433,
+      "loss": 1.4471,
+      "step": 951500
+    },
+    {
+      "epoch": 6.190259444697315,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0017523896222121074,
+      "loss": 1.4398,
+      "step": 952000
+    },
+    {
+      "epoch": 6.1935106313804535,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0017522595747447819,
+      "loss": 1.4425,
+      "step": 952500
+    },
+    {
+      "epoch": 6.196761818063593,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0017521295272774561,
+      "loss": 1.4445,
+      "step": 953000
+    },
+    {
+      "epoch": 6.200013004746733,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0017519994798101306,
+      "loss": 1.447,
+      "step": 953500
+    },
+    {
+      "epoch": 6.203264191429872,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0017518694323428053,
+      "loss": 1.4354,
+      "step": 954000
+    },
+    {
+      "epoch": 6.206515378113012,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0017517393848754798,
+      "loss": 1.436,
+      "step": 954500
+    },
+    {
+      "epoch": 6.20976656479615,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.001751609337408154,
+      "loss": 1.4396,
+      "step": 955000
+    },
+    {
+      "epoch": 6.21301775147929,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0017514792899408285,
+      "loss": 1.4427,
+      "step": 955500
+    },
+    {
+      "epoch": 6.216268938162429,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.001751349242473503,
+      "loss": 1.4362,
+      "step": 956000
+    },
+    {
+      "epoch": 6.219520124845569,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0017512191950061773,
+      "loss": 1.4419,
+      "step": 956500
+    },
+    {
+      "epoch": 6.2227713115287075,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0017510891475388517,
+      "loss": 1.4488,
+      "step": 957000
+    },
+    {
+      "epoch": 6.226022498211847,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0017509591000715262,
+      "loss": 1.4408,
+      "step": 957500
+    },
+    {
+      "epoch": 6.229273684894987,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0017508290526042005,
+      "loss": 1.4444,
+      "step": 958000
+    },
+    {
+      "epoch": 6.232524871578126,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.001750699005136875,
+      "loss": 1.448,
+      "step": 958500
+    },
+    {
+      "epoch": 6.235776058261266,
+      "grad_norm": 2.6875,
+      "learning_rate": 0.0017505689576695494,
+      "loss": 1.4426,
+      "step": 959000
+    },
+    {
+      "epoch": 6.239027244944404,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0017504389102022237,
+      "loss": 1.4409,
+      "step": 959500
+    },
+    {
+      "epoch": 6.242278431627544,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0017503088627348982,
+      "loss": 1.4349,
+      "step": 960000
+    },
+    {
+      "epoch": 6.245529618310683,
+      "grad_norm": 16.875,
+      "learning_rate": 0.0017501788152675727,
+      "loss": 1.4433,
+      "step": 960500
+    },
+    {
+      "epoch": 6.248780804993823,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.001750048767800247,
+      "loss": 1.4417,
+      "step": 961000
+    },
+    {
+      "epoch": 6.252031991676962,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0017499187203329216,
+      "loss": 1.4414,
+      "step": 961500
+    },
+    {
+      "epoch": 6.255283178360101,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.001749788672865596,
+      "loss": 1.4424,
+      "step": 962000
+    },
+    {
+      "epoch": 6.258534365043241,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0017496586253982706,
+      "loss": 1.4417,
+      "step": 962500
+    },
+    {
+      "epoch": 6.26178555172638,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0017495285779309448,
+      "loss": 1.4416,
+      "step": 963000
+    },
+    {
+      "epoch": 6.26503673840952,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.0017493985304636193,
+      "loss": 1.4448,
+      "step": 963500
+    },
+    {
+      "epoch": 6.268287925092658,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0017492684829962938,
+      "loss": 1.4438,
+      "step": 964000
+    },
+    {
+      "epoch": 6.271539111775798,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.001749138435528968,
+      "loss": 1.443,
+      "step": 964500
+    },
+    {
+      "epoch": 6.274790298458938,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.0017490083880616425,
+      "loss": 1.4432,
+      "step": 965000
+    },
+    {
+      "epoch": 6.278041485142077,
+      "grad_norm": 1.8046875,
+      "learning_rate": 0.001748878340594317,
+      "loss": 1.4418,
+      "step": 965500
+    },
+    {
+      "epoch": 6.2812926718252164,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0017487482931269913,
+      "loss": 1.4437,
+      "step": 966000
+    },
+    {
+      "epoch": 6.284543858508355,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0017486182456596657,
+      "loss": 1.4367,
+      "step": 966500
+    },
+    {
+      "epoch": 6.287795045191495,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0017484881981923402,
+      "loss": 1.4362,
+      "step": 967000
+    },
+    {
+      "epoch": 6.291046231874634,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0017483581507250145,
+      "loss": 1.443,
+      "step": 967500
+    },
+    {
+      "epoch": 6.294297418557774,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.001748228103257689,
+      "loss": 1.4402,
+      "step": 968000
+    },
+    {
+      "epoch": 6.297548605240913,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0017480980557903637,
+      "loss": 1.4419,
+      "step": 968500
+    },
+    {
+      "epoch": 6.300799791924052,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0017479680083230381,
+      "loss": 1.4365,
+      "step": 969000
+    },
+    {
+      "epoch": 6.304050978607192,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0017478379608557124,
+      "loss": 1.436,
+      "step": 969500
+    },
+    {
+      "epoch": 6.307302165290331,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017477079133883869,
+      "loss": 1.4406,
+      "step": 970000
+    },
+    {
+      "epoch": 6.3105533519734704,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0017475778659210614,
+      "loss": 1.4384,
+      "step": 970500
+    },
+    {
+      "epoch": 6.313804538656609,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0017474478184537356,
+      "loss": 1.4433,
+      "step": 971000
+    },
+    {
+      "epoch": 6.317055725339749,
+      "grad_norm": 3.265625,
+      "learning_rate": 0.00174731777098641,
+      "loss": 1.4363,
+      "step": 971500
+    },
+    {
+      "epoch": 6.320306912022889,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017471877235190846,
+      "loss": 1.4395,
+      "step": 972000
+    },
+    {
+      "epoch": 6.323558098706028,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0017470576760517588,
+      "loss": 1.4361,
+      "step": 972500
+    },
+    {
+      "epoch": 6.326809285389167,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0017469276285844333,
+      "loss": 1.4345,
+      "step": 973000
+    },
+    {
+      "epoch": 6.330060472072306,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0017467975811171078,
+      "loss": 1.4347,
+      "step": 973500
+    },
+    {
+      "epoch": 6.333311658755446,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.001746667533649782,
+      "loss": 1.4364,
+      "step": 974000
+    },
+    {
+      "epoch": 6.336562845438585,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0017465374861824565,
+      "loss": 1.4367,
+      "step": 974500
+    },
+    {
+      "epoch": 6.3398140321217245,
+      "grad_norm": 5.71875,
+      "learning_rate": 0.001746407438715131,
+      "loss": 1.4332,
+      "step": 975000
+    },
+    {
+      "epoch": 6.343065218804864,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.0017462773912478053,
+      "loss": 1.4422,
+      "step": 975500
+    },
+    {
+      "epoch": 6.346316405488003,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00174614734378048,
+      "loss": 1.4358,
+      "step": 976000
+    },
+    {
+      "epoch": 6.349567592171143,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.0017460172963131545,
+      "loss": 1.4357,
+      "step": 976500
+    },
+    {
+      "epoch": 6.352818778854282,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.001745887248845829,
+      "loss": 1.4352,
+      "step": 977000
+    },
+    {
+      "epoch": 6.356069965537421,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.0017457572013785032,
+      "loss": 1.4334,
+      "step": 977500
+    },
+    {
+      "epoch": 6.35932115222056,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0017456271539111777,
+      "loss": 1.4317,
+      "step": 978000
+    },
+    {
+      "epoch": 6.3625723389037,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0017454971064438521,
+      "loss": 1.4379,
+      "step": 978500
+    },
+    {
+      "epoch": 6.365823525586839,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0017453670589765264,
+      "loss": 1.4336,
+      "step": 979000
+    },
+    {
+      "epoch": 6.3690747122699785,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0017452370115092009,
+      "loss": 1.4379,
+      "step": 979500
+    },
+    {
+      "epoch": 6.372325898953118,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017451069640418754,
+      "loss": 1.4391,
+      "step": 980000
+    },
+    {
+      "epoch": 6.375577085636257,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0017449769165745496,
+      "loss": 1.4391,
+      "step": 980500
+    },
+    {
+      "epoch": 6.378828272319397,
+      "grad_norm": 4.0625,
+      "learning_rate": 0.001744846869107224,
+      "loss": 1.4431,
+      "step": 981000
+    },
+    {
+      "epoch": 6.382079459002536,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0017447168216398986,
+      "loss": 1.4351,
+      "step": 981500
+    },
+    {
+      "epoch": 6.385330645685675,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0017445867741725728,
+      "loss": 1.4387,
+      "step": 982000
+    },
+    {
+      "epoch": 6.388581832368814,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0017444567267052473,
+      "loss": 1.4328,
+      "step": 982500
+    },
+    {
+      "epoch": 6.391833019051954,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001744326679237922,
+      "loss": 1.4343,
+      "step": 983000
+    },
+    {
+      "epoch": 6.395084205735094,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0017441966317705965,
+      "loss": 1.4335,
+      "step": 983500
+    },
+    {
+      "epoch": 6.3983353924182325,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0017440665843032708,
+      "loss": 1.435,
+      "step": 984000
+    },
+    {
+      "epoch": 6.401586579101372,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0017439365368359452,
+      "loss": 1.4415,
+      "step": 984500
+    },
+    {
+      "epoch": 6.404837765784511,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0017438064893686197,
+      "loss": 1.443,
+      "step": 985000
+    },
+    {
+      "epoch": 6.408088952467651,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.001743676441901294,
+      "loss": 1.4426,
+      "step": 985500
+    },
+    {
+      "epoch": 6.41134013915079,
+      "grad_norm": 15.4375,
+      "learning_rate": 0.0017435463944339685,
+      "loss": 1.4405,
+      "step": 986000
+    },
+    {
+      "epoch": 6.414591325833929,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.001743416346966643,
+      "loss": 1.4408,
+      "step": 986500
+    },
+    {
+      "epoch": 6.417842512517069,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0017432862994993172,
+      "loss": 1.4509,
+      "step": 987000
+    },
+    {
+      "epoch": 6.421093699200208,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0017431562520319917,
+      "loss": 1.4435,
+      "step": 987500
+    },
+    {
+      "epoch": 6.424344885883348,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0017430262045646662,
+      "loss": 1.4417,
+      "step": 988000
+    },
+    {
+      "epoch": 6.4275960725664865,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0017428961570973404,
+      "loss": 1.4566,
+      "step": 988500
+    },
+    {
+      "epoch": 6.430847259249626,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001742766109630015,
+      "loss": 1.4679,
+      "step": 989000
+    },
+    {
+      "epoch": 6.434098445932765,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0017426360621626894,
+      "loss": 1.4754,
+      "step": 989500
+    },
+    {
+      "epoch": 6.437349632615905,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0017425060146953636,
+      "loss": 1.4804,
+      "step": 990000
+    },
+    {
+      "epoch": 6.4406008192990445,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0017423759672280383,
+      "loss": 1.4847,
+      "step": 990500
+    },
+    {
+      "epoch": 6.443852005982183,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.0017422459197607128,
+      "loss": 1.4757,
+      "step": 991000
+    },
+    {
+      "epoch": 6.447103192665323,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.0017421158722933873,
+      "loss": 1.4749,
+      "step": 991500
+    },
+    {
+      "epoch": 6.450354379348462,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0017419858248260615,
+      "loss": 1.4674,
+      "step": 992000
+    },
+    {
+      "epoch": 6.453605566031602,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.001741855777358736,
+      "loss": 1.4648,
+      "step": 992500
+    },
+    {
+      "epoch": 6.4568567527147405,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0017417257298914105,
+      "loss": 1.453,
+      "step": 993000
+    },
+    {
+      "epoch": 6.46010793939788,
+      "grad_norm": 3.25,
+      "learning_rate": 0.0017415956824240848,
+      "loss": 1.4638,
+      "step": 993500
+    },
+    {
+      "epoch": 6.46335912608102,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0017414656349567592,
+      "loss": 1.4796,
+      "step": 994000
+    },
+    {
+      "epoch": 6.466610312764159,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0017413355874894337,
+      "loss": 1.4722,
+      "step": 994500
+    },
+    {
+      "epoch": 6.4698614994472985,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001741205540022108,
+      "loss": 1.474,
+      "step": 995000
+    },
+    {
+      "epoch": 6.473112686130437,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0017410754925547825,
+      "loss": 1.4831,
+      "step": 995500
+    },
+    {
+      "epoch": 6.476363872813577,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.001740945445087457,
+      "loss": 1.4832,
+      "step": 996000
+    },
+    {
+      "epoch": 6.479615059496716,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.0017408153976201312,
+      "loss": 1.4926,
+      "step": 996500
+    },
+    {
+      "epoch": 6.482866246179856,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0017406853501528057,
+      "loss": 1.489,
+      "step": 997000
+    },
+    {
+      "epoch": 6.486117432862995,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017405553026854804,
+      "loss": 1.494,
+      "step": 997500
+    },
+    {
+      "epoch": 6.489368619546134,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0017404252552181549,
+      "loss": 1.4693,
+      "step": 998000
+    },
+    {
+      "epoch": 6.492619806229274,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0017402952077508291,
+      "loss": 1.457,
+      "step": 998500
+    },
+    {
+      "epoch": 6.495870992912413,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0017401651602835036,
+      "loss": 1.4526,
+      "step": 999000
+    },
+    {
+      "epoch": 6.4991221795955525,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.001740035112816178,
+      "loss": 1.4506,
+      "step": 999500
+    },
+    {
+      "epoch": 6.502373366278691,
+      "grad_norm": 19.5,
+      "learning_rate": 0.0017399050653488523,
+      "loss": 1.4474,
+      "step": 1000000
+    },
+    {
+      "epoch": 6.505624552961831,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0017397750178815268,
+      "loss": 1.4499,
+      "step": 1000500
+    },
+    {
+      "epoch": 6.508875739644971,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0017396449704142013,
+      "loss": 1.4509,
+      "step": 1001000
+    },
+    {
+      "epoch": 6.51212692632811,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0017395149229468756,
+      "loss": 1.4505,
+      "step": 1001500
+    },
+    {
+      "epoch": 6.515378113011249,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.00173938487547955,
+      "loss": 1.4481,
+      "step": 1002000
+    },
+    {
+      "epoch": 6.518629299694388,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0017392548280122245,
+      "loss": 1.4461,
+      "step": 1002500
+    },
+    {
+      "epoch": 6.521880486377528,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0017391247805448988,
+      "loss": 1.4411,
+      "step": 1003000
+    },
+    {
+      "epoch": 6.525131673060667,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0017389947330775733,
+      "loss": 1.4491,
+      "step": 1003500
+    },
+    {
+      "epoch": 6.5283828597438065,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0017388646856102477,
+      "loss": 1.4502,
+      "step": 1004000
+    },
+    {
+      "epoch": 6.531634046426946,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.001738734638142922,
+      "loss": 1.4517,
+      "step": 1004500
+    },
+    {
+      "epoch": 6.534885233110085,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.0017386045906755967,
+      "loss": 1.4522,
+      "step": 1005000
+    },
+    {
+      "epoch": 6.538136419793225,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0017384745432082712,
+      "loss": 1.4466,
+      "step": 1005500
+    },
+    {
+      "epoch": 6.541387606476364,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.0017383444957409456,
+      "loss": 1.4474,
+      "step": 1006000
+    },
+    {
+      "epoch": 6.544638793159503,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00173821444827362,
+      "loss": 1.4479,
+      "step": 1006500
+    },
+    {
+      "epoch": 6.547889979842642,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0017380844008062944,
+      "loss": 1.4444,
+      "step": 1007000
+    },
+    {
+      "epoch": 6.551141166525782,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.0017379543533389689,
+      "loss": 1.4431,
+      "step": 1007500
+    },
+    {
+      "epoch": 6.554392353208922,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0017378243058716431,
+      "loss": 1.4487,
+      "step": 1008000
+    },
+    {
+      "epoch": 6.5576435398920605,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0017376942584043176,
+      "loss": 1.4422,
+      "step": 1008500
+    },
+    {
+      "epoch": 6.5608947265752,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.001737564210936992,
+      "loss": 1.44,
+      "step": 1009000
+    },
+    {
+      "epoch": 6.564145913258339,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.0017374341634696663,
+      "loss": 1.4404,
+      "step": 1009500
+    },
+    {
+      "epoch": 6.567397099941479,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.0017373041160023408,
+      "loss": 1.4407,
+      "step": 1010000
+    },
+    {
+      "epoch": 6.570648286624618,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.0017371740685350153,
+      "loss": 1.446,
+      "step": 1010500
+    },
+    {
+      "epoch": 6.573899473307757,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0017370440210676896,
+      "loss": 1.449,
+      "step": 1011000
+    },
+    {
+      "epoch": 6.577150659990897,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001736913973600364,
+      "loss": 1.449,
+      "step": 1011500
+    },
+    {
+      "epoch": 6.580401846674036,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0017367839261330387,
+      "loss": 1.4324,
+      "step": 1012000
+    },
+    {
+      "epoch": 6.583653033357176,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0017366538786657132,
+      "loss": 1.4414,
+      "step": 1012500
+    },
+    {
+      "epoch": 6.5869042200403145,
+      "grad_norm": 1.8984375,
+      "learning_rate": 0.0017365238311983875,
+      "loss": 1.4451,
+      "step": 1013000
+    },
+    {
+      "epoch": 6.590155406723454,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.001736393783731062,
+      "loss": 1.4398,
+      "step": 1013500
+    },
+    {
+      "epoch": 6.593406593406593,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0017362637362637364,
+      "loss": 1.446,
+      "step": 1014000
+    },
+    {
+      "epoch": 6.596657780089733,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0017361336887964107,
+      "loss": 1.4452,
+      "step": 1014500
+    },
+    {
+      "epoch": 6.5999089667728725,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0017360036413290852,
+      "loss": 1.4497,
+      "step": 1015000
+    },
+    {
+      "epoch": 6.603160153456011,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0017358735938617597,
+      "loss": 1.4545,
+      "step": 1015500
+    },
+    {
+      "epoch": 6.606411340139151,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001735743546394434,
+      "loss": 1.4642,
+      "step": 1016000
+    },
+    {
+      "epoch": 6.60966252682229,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0017356134989271084,
+      "loss": 1.4545,
+      "step": 1016500
+    },
+    {
+      "epoch": 6.61291371350543,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0017354834514597829,
+      "loss": 1.4551,
+      "step": 1017000
+    },
+    {
+      "epoch": 6.6161649001885685,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.0017353534039924571,
+      "loss": 1.4558,
+      "step": 1017500
+    },
+    {
+      "epoch": 6.619416086871708,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0017352233565251316,
+      "loss": 1.4509,
+      "step": 1018000
+    },
+    {
+      "epoch": 6.622667273554848,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001735093309057806,
+      "loss": 1.4542,
+      "step": 1018500
+    },
+    {
+      "epoch": 6.625918460237987,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0017349632615904804,
+      "loss": 1.448,
+      "step": 1019000
+    },
+    {
+      "epoch": 6.6291696469211265,
+      "grad_norm": 2.78125,
+      "learning_rate": 0.001734833214123155,
+      "loss": 1.4441,
+      "step": 1019500
+    },
+    {
+      "epoch": 6.632420833604265,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0017347031666558295,
+      "loss": 1.4479,
+      "step": 1020000
+    },
+    {
+      "epoch": 6.635672020287405,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.001734573119188504,
+      "loss": 1.4445,
+      "step": 1020500
+    },
+    {
+      "epoch": 6.638923206970544,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0017344430717211783,
+      "loss": 1.4484,
+      "step": 1021000
+    },
+    {
+      "epoch": 6.642174393653684,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0017343130242538527,
+      "loss": 1.4464,
+      "step": 1021500
+    },
+    {
+      "epoch": 6.645425580336823,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.0017341829767865272,
+      "loss": 1.4434,
+      "step": 1022000
+    },
+    {
+      "epoch": 6.648676767019962,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0017340529293192015,
+      "loss": 1.4531,
+      "step": 1022500
+    },
+    {
+      "epoch": 6.651927953703102,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.001733922881851876,
+      "loss": 1.454,
+      "step": 1023000
+    },
+    {
+      "epoch": 6.655179140386241,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0017337928343845504,
+      "loss": 1.4707,
+      "step": 1023500
+    },
+    {
+      "epoch": 6.6584303270693805,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0017336627869172247,
+      "loss": 1.4892,
+      "step": 1024000
+    },
+    {
+      "epoch": 6.661681513752519,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0017335327394498992,
+      "loss": 1.5023,
+      "step": 1024500
+    },
+    {
+      "epoch": 6.664932700435659,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0017334026919825737,
+      "loss": 1.5322,
+      "step": 1025000
+    },
+    {
+      "epoch": 6.668183887118799,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001733272644515248,
+      "loss": 1.5527,
+      "step": 1025500
+    },
+    {
+      "epoch": 6.671435073801938,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0017331425970479224,
+      "loss": 1.523,
+      "step": 1026000
+    },
+    {
+      "epoch": 6.674686260485077,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.001733012549580597,
+      "loss": 1.5219,
+      "step": 1026500
+    },
+    {
+      "epoch": 6.677937447168216,
+      "grad_norm": 1.375,
+      "learning_rate": 0.0017328825021132716,
+      "loss": 1.5343,
+      "step": 1027000
+    },
+    {
+      "epoch": 6.681188633851356,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0017327524546459458,
+      "loss": 1.5245,
+      "step": 1027500
+    },
+    {
+      "epoch": 6.684439820534495,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0017326224071786203,
+      "loss": 1.5177,
+      "step": 1028000
+    },
+    {
+      "epoch": 6.6876910072176345,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0017324923597112948,
+      "loss": 1.513,
+      "step": 1028500
+    },
+    {
+      "epoch": 6.690942193900774,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001732362312243969,
+      "loss": 1.5196,
+      "step": 1029000
+    },
+    {
+      "epoch": 6.694193380583913,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0017322322647766435,
+      "loss": 1.5089,
+      "step": 1029500
+    },
+    {
+      "epoch": 6.697444567267053,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001732102217309318,
+      "loss": 1.5089,
+      "step": 1030000
+    },
+    {
+      "epoch": 6.700695753950192,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0017319721698419923,
+      "loss": 1.5037,
+      "step": 1030500
+    },
+    {
+      "epoch": 6.703946940633331,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.0017318421223746668,
+      "loss": 1.4995,
+      "step": 1031000
+    },
+    {
+      "epoch": 6.70719812731647,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0017317120749073412,
+      "loss": 1.4928,
+      "step": 1031500
+    },
+    {
+      "epoch": 6.71044931399961,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0017315820274400155,
+      "loss": 1.4962,
+      "step": 1032000
+    },
+    {
+      "epoch": 6.713700500682749,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00173145197997269,
+      "loss": 1.4943,
+      "step": 1032500
+    },
+    {
+      "epoch": 6.7169516873658885,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0017313219325053645,
+      "loss": 1.4925,
+      "step": 1033000
+    },
+    {
+      "epoch": 6.720202874049028,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0017311918850380387,
+      "loss": 1.4849,
+      "step": 1033500
+    },
+    {
+      "epoch": 6.723454060732167,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0017310618375707134,
+      "loss": 1.479,
+      "step": 1034000
+    },
+    {
+      "epoch": 6.726705247415307,
+      "grad_norm": 0.875,
+      "learning_rate": 0.0017309317901033879,
+      "loss": 1.4838,
+      "step": 1034500
+    },
+    {
+      "epoch": 6.729956434098446,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0017308017426360624,
+      "loss": 1.4778,
+      "step": 1035000
+    },
+    {
+      "epoch": 6.733207620781585,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0017306716951687366,
+      "loss": 1.4825,
+      "step": 1035500
+    },
+    {
+      "epoch": 6.736458807464724,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.001730541647701411,
+      "loss": 1.4769,
+      "step": 1036000
+    },
+    {
+      "epoch": 6.739709994147864,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0017304116002340856,
+      "loss": 1.4705,
+      "step": 1036500
+    },
+    {
+      "epoch": 6.742961180831003,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.0017302815527667598,
+      "loss": 1.4729,
+      "step": 1037000
+    },
+    {
+      "epoch": 6.7462123675141425,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0017301515052994343,
+      "loss": 1.4738,
+      "step": 1037500
+    },
+    {
+      "epoch": 6.749463554197282,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0017300214578321088,
+      "loss": 1.4775,
+      "step": 1038000
+    },
+    {
+      "epoch": 6.752714740880421,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.001729891410364783,
+      "loss": 1.4787,
+      "step": 1038500
+    },
+    {
+      "epoch": 6.755965927563561,
+      "grad_norm": 1.796875,
+      "learning_rate": 0.0017297613628974575,
+      "loss": 1.4751,
+      "step": 1039000
+    },
+    {
+      "epoch": 6.7592171142467,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.001729631315430132,
+      "loss": 1.4772,
+      "step": 1039500
+    },
+    {
+      "epoch": 6.762468300929839,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0017295012679628063,
+      "loss": 1.4735,
+      "step": 1040000
+    },
+    {
+      "epoch": 6.765719487612978,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0017293712204954808,
+      "loss": 1.4709,
+      "step": 1040500
+    },
+    {
+      "epoch": 6.768970674296118,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0017292411730281555,
+      "loss": 1.4708,
+      "step": 1041000
+    },
+    {
+      "epoch": 6.772221860979258,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00172911112556083,
+      "loss": 1.4705,
+      "step": 1041500
+    },
+    {
+      "epoch": 6.7754730476623966,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0017289810780935042,
+      "loss": 1.4681,
+      "step": 1042000
+    },
+    {
+      "epoch": 6.778724234345536,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0017288510306261787,
+      "loss": 1.4618,
+      "step": 1042500
+    },
+    {
+      "epoch": 6.781975421028675,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0017287209831588532,
+      "loss": 1.4675,
+      "step": 1043000
+    },
+    {
+      "epoch": 6.785226607711815,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0017285909356915274,
+      "loss": 1.472,
+      "step": 1043500
+    },
+    {
+      "epoch": 6.788477794394954,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001728460888224202,
+      "loss": 1.4722,
+      "step": 1044000
+    },
+    {
+      "epoch": 6.791728981078093,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0017283308407568764,
+      "loss": 1.4638,
+      "step": 1044500
+    },
+    {
+      "epoch": 6.794980167761233,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0017282007932895506,
+      "loss": 1.4743,
+      "step": 1045000
+    },
+    {
+      "epoch": 6.798231354444372,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0017280707458222251,
+      "loss": 1.472,
+      "step": 1045500
+    },
+    {
+      "epoch": 6.801482541127512,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.0017279406983548996,
+      "loss": 1.4667,
+      "step": 1046000
+    },
+    {
+      "epoch": 6.804733727810651,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0017278106508875739,
+      "loss": 1.4613,
+      "step": 1046500
+    },
+    {
+      "epoch": 6.80798491449379,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0017276806034202483,
+      "loss": 1.4652,
+      "step": 1047000
+    },
+    {
+      "epoch": 6.811236101176929,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.0017275505559529228,
+      "loss": 1.4641,
+      "step": 1047500
+    },
+    {
+      "epoch": 6.814487287860069,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.001727420508485597,
+      "loss": 1.4573,
+      "step": 1048000
+    },
+    {
+      "epoch": 6.817738474543209,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0017272904610182718,
+      "loss": 1.4655,
+      "step": 1048500
+    },
+    {
+      "epoch": 6.820989661226347,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0017271604135509462,
+      "loss": 1.4632,
+      "step": 1049000
+    },
+    {
+      "epoch": 6.824240847909487,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0017270303660836207,
+      "loss": 1.461,
+      "step": 1049500
+    },
+    {
+      "epoch": 6.827492034592626,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.001726900318616295,
+      "loss": 1.4637,
+      "step": 1050000
+    },
+    {
+      "epoch": 6.830743221275766,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0017267702711489695,
+      "loss": 1.4603,
+      "step": 1050500
+    },
+    {
+      "epoch": 6.833994407958905,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001726640223681644,
+      "loss": 1.4622,
+      "step": 1051000
+    },
+    {
+      "epoch": 6.837245594642044,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0017265101762143182,
+      "loss": 1.4617,
+      "step": 1051500
+    },
+    {
+      "epoch": 6.840496781325184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0017263801287469927,
+      "loss": 1.4625,
+      "step": 1052000
+    },
+    {
+      "epoch": 6.843747968008323,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0017262500812796672,
+      "loss": 1.459,
+      "step": 1052500
+    },
+    {
+      "epoch": 6.846999154691463,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0017261200338123414,
+      "loss": 1.4576,
+      "step": 1053000
+    },
+    {
+      "epoch": 6.850250341374601,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.001725989986345016,
+      "loss": 1.4597,
+      "step": 1053500
+    },
+    {
+      "epoch": 6.853501528057741,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.0017258599388776904,
+      "loss": 1.4643,
+      "step": 1054000
+    },
+    {
+      "epoch": 6.85675271474088,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017257298914103646,
+      "loss": 1.4575,
+      "step": 1054500
+    },
+    {
+      "epoch": 6.86000390142402,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0017255998439430391,
+      "loss": 1.4588,
+      "step": 1055000
+    },
+    {
+      "epoch": 6.8632550881071595,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0017254697964757138,
+      "loss": 1.4612,
+      "step": 1055500
+    },
+    {
+      "epoch": 6.866506274790298,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0017253397490083883,
+      "loss": 1.4574,
+      "step": 1056000
+    },
+    {
+      "epoch": 6.869757461473438,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0017252097015410626,
+      "loss": 1.4544,
+      "step": 1056500
+    },
+    {
+      "epoch": 6.873008648156577,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.001725079654073737,
+      "loss": 1.4566,
+      "step": 1057000
+    },
+    {
+      "epoch": 6.876259834839717,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0017249496066064115,
+      "loss": 1.4514,
+      "step": 1057500
+    },
+    {
+      "epoch": 6.879511021522855,
+      "grad_norm": 1.9453125,
+      "learning_rate": 0.0017248195591390858,
+      "loss": 1.4564,
+      "step": 1058000
+    },
+    {
+      "epoch": 6.882762208205995,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0017246895116717603,
+      "loss": 1.452,
+      "step": 1058500
+    },
+    {
+      "epoch": 6.886013394889135,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0017245594642044347,
+      "loss": 1.4514,
+      "step": 1059000
+    },
+    {
+      "epoch": 6.889264581572274,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001724429416737109,
+      "loss": 1.4594,
+      "step": 1059500
+    },
+    {
+      "epoch": 6.8925157682554135,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0017242993692697835,
+      "loss": 1.4545,
+      "step": 1060000
+    },
+    {
+      "epoch": 6.895766954938552,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001724169321802458,
+      "loss": 1.4563,
+      "step": 1060500
+    },
+    {
+      "epoch": 6.899018141621692,
+      "grad_norm": 1.375,
+      "learning_rate": 0.0017240392743351322,
+      "loss": 1.4481,
+      "step": 1061000
+    },
+    {
+      "epoch": 6.902269328304831,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0017239092268678067,
+      "loss": 1.4557,
+      "step": 1061500
+    },
+    {
+      "epoch": 6.905520514987971,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0017237791794004812,
+      "loss": 1.4574,
+      "step": 1062000
+    },
+    {
+      "epoch": 6.90877170167111,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0017236491319331554,
+      "loss": 1.4529,
+      "step": 1062500
+    },
+    {
+      "epoch": 6.912022888354249,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0017235190844658301,
+      "loss": 1.4563,
+      "step": 1063000
+    },
+    {
+      "epoch": 6.915274075037389,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0017233890369985046,
+      "loss": 1.4571,
+      "step": 1063500
+    },
+    {
+      "epoch": 6.918525261720528,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.001723258989531179,
+      "loss": 1.4568,
+      "step": 1064000
+    },
+    {
+      "epoch": 6.9217764484036675,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.0017231289420638533,
+      "loss": 1.4533,
+      "step": 1064500
+    },
+    {
+      "epoch": 6.925027635086806,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0017229988945965278,
+      "loss": 1.4548,
+      "step": 1065000
+    },
+    {
+      "epoch": 6.928278821769946,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.0017228688471292023,
+      "loss": 1.4596,
+      "step": 1065500
+    },
+    {
+      "epoch": 6.931530008453086,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0017227387996618766,
+      "loss": 1.4569,
+      "step": 1066000
+    },
+    {
+      "epoch": 6.934781195136225,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.001722608752194551,
+      "loss": 1.4575,
+      "step": 1066500
+    },
+    {
+      "epoch": 6.938032381819364,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0017224787047272255,
+      "loss": 1.4544,
+      "step": 1067000
+    },
+    {
+      "epoch": 6.941283568502503,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0017223486572598998,
+      "loss": 1.4564,
+      "step": 1067500
+    },
+    {
+      "epoch": 6.944534755185643,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0017222186097925743,
+      "loss": 1.4541,
+      "step": 1068000
+    },
+    {
+      "epoch": 6.947785941868782,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0017220885623252487,
+      "loss": 1.4547,
+      "step": 1068500
+    },
+    {
+      "epoch": 6.9510371285519215,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.001721958514857923,
+      "loss": 1.4605,
+      "step": 1069000
+    },
+    {
+      "epoch": 6.954288315235061,
+      "grad_norm": 2.390625,
+      "learning_rate": 0.0017218284673905975,
+      "loss": 1.4535,
+      "step": 1069500
+    },
+    {
+      "epoch": 6.9575395019182,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.001721698419923272,
+      "loss": 1.4535,
+      "step": 1070000
+    },
+    {
+      "epoch": 6.96079068860134,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0017215683724559467,
+      "loss": 1.4582,
+      "step": 1070500
+    },
+    {
+      "epoch": 6.964041875284479,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001721438324988621,
+      "loss": 1.4554,
+      "step": 1071000
+    },
+    {
+      "epoch": 6.967293061967618,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0017213082775212954,
+      "loss": 1.4507,
+      "step": 1071500
+    },
+    {
+      "epoch": 6.970544248650757,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0017211782300539699,
+      "loss": 1.449,
+      "step": 1072000
+    },
+    {
+      "epoch": 6.973795435333897,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0017210481825866441,
+      "loss": 1.4487,
+      "step": 1072500
+    },
+    {
+      "epoch": 6.977046622017037,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0017209181351193186,
+      "loss": 1.4501,
+      "step": 1073000
+    },
+    {
+      "epoch": 6.9802978087001755,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.001720788087651993,
+      "loss": 1.447,
+      "step": 1073500
+    },
+    {
+      "epoch": 6.983548995383315,
+      "grad_norm": 3.203125,
+      "learning_rate": 0.0017206580401846674,
+      "loss": 1.4483,
+      "step": 1074000
+    },
+    {
+      "epoch": 6.986800182066454,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0017205279927173418,
+      "loss": 1.4508,
+      "step": 1074500
+    },
+    {
+      "epoch": 6.990051368749594,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0017203979452500163,
+      "loss": 1.4445,
+      "step": 1075000
+    },
+    {
+      "epoch": 6.993302555432733,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0017202678977826906,
+      "loss": 1.4496,
+      "step": 1075500
+    },
+    {
+      "epoch": 6.996553742115872,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001720137850315365,
+      "loss": 1.4451,
+      "step": 1076000
+    },
+    {
+      "epoch": 6.999804928799012,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0017200078028480395,
+      "loss": 1.4514,
+      "step": 1076500
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 1.4298810958862305,
+      "eval_runtime": 0.5331,
+      "eval_samples_per_second": 1875.719,
+      "eval_steps_per_second": 30.011,
+      "step": 1076530
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 7689500,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 50,
+  "save_steps": 500,
+  "total_flos": 1.3852790029150102e+19,
+  "train_batch_size": 64,
+  "trial_name": null,
+  "trial_params": null
+}