diff --git "a/checkpoints/checkpoint-35000/trainer_state.json" "b/checkpoints/checkpoint-35000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoints/checkpoint-35000/trainer_state.json"
@@ -0,0 +1,4921 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.3880997527250147,
+  "eval_steps": 500,
+  "global_step": 35000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.2580924034118652,
+      "learning_rate": 2.2172949002217296e-06,
+      "loss": 10.2933,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.4347386360168457,
+      "learning_rate": 4.434589800443459e-06,
+      "loss": 10.1894,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.3895885944366455,
+      "learning_rate": 6.651884700665188e-06,
+      "loss": 10.1424,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.129647731781006,
+      "learning_rate": 8.869179600886918e-06,
+      "loss": 10.0995,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.3564186096191406,
+      "learning_rate": 1.1086474501108649e-05,
+      "loss": 10.0479,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.830551028251648,
+      "learning_rate": 1.3303769401330377e-05,
+      "loss": 9.9971,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1173911094665527,
+      "learning_rate": 1.5521064301552106e-05,
+      "loss": 9.9201,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6636557579040527,
+      "learning_rate": 1.7738359201773837e-05,
+      "loss": 9.8562,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.4503839015960693,
+      "learning_rate": 1.9955654101995567e-05,
+      "loss": 9.7599,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.822424054145813,
+      "learning_rate": 2.2172949002217298e-05,
+      "loss": 9.6608,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6598998308181763,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 9.55,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8471707105636597,
+      "learning_rate": 2.6607538802660753e-05,
+      "loss": 9.4606,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4833533763885498,
+      "learning_rate": 2.8824833702882487e-05,
+      "loss": 9.3283,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.688541054725647,
+      "learning_rate": 3.104212860310421e-05,
+      "loss": 9.2229,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6466543674468994,
+      "learning_rate": 3.325942350332594e-05,
+      "loss": 9.1093,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4169293642044067,
+      "learning_rate": 3.547671840354767e-05,
+      "loss": 8.9703,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7079193592071533,
+      "learning_rate": 3.7694013303769404e-05,
+      "loss": 8.8351,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5513204336166382,
+      "learning_rate": 3.9911308203991135e-05,
+      "loss": 8.7111,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.485573172569275,
+      "learning_rate": 4.212860310421286e-05,
+      "loss": 8.5627,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.511690616607666,
+      "learning_rate": 4.4345898004434597e-05,
+      "loss": 8.5042,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.1478614807128906,
+      "learning_rate": 4.656319290465632e-05,
+      "loss": 8.3287,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4060652256011963,
+      "learning_rate": 4.878048780487805e-05,
+      "loss": 8.2341,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3950035572052002,
+      "learning_rate": 5.099778270509978e-05,
+      "loss": 8.1277,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5197688341140747,
+      "learning_rate": 5.3215077605321506e-05,
+      "loss": 8.0311,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3406693935394287,
+      "learning_rate": 5.543237250554324e-05,
+      "loss": 7.9824,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4520119428634644,
+      "learning_rate": 5.7649667405764975e-05,
+      "loss": 7.9948,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.179124116897583,
+      "learning_rate": 5.98669623059867e-05,
+      "loss": 7.9144,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4039533138275146,
+      "learning_rate": 6.208425720620842e-05,
+      "loss": 7.8768,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5542700290679932,
+      "learning_rate": 6.430155210643016e-05,
+      "loss": 7.894,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4150550365447998,
+      "learning_rate": 6.651884700665188e-05,
+      "loss": 7.8409,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6647827625274658,
+      "learning_rate": 6.873614190687362e-05,
+      "loss": 7.91,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7795697450637817,
+      "learning_rate": 7.095343680709535e-05,
+      "loss": 7.8256,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.933110237121582,
+      "learning_rate": 7.317073170731707e-05,
+      "loss": 7.8463,
+      "step": 1650
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1942570209503174,
+      "learning_rate": 7.538802660753881e-05,
+      "loss": 7.7827,
+      "step": 1700
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6759297847747803,
+      "learning_rate": 7.760532150776053e-05,
+      "loss": 7.8,
+      "step": 1750
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.093256950378418,
+      "learning_rate": 7.982261640798227e-05,
+      "loss": 7.7461,
+      "step": 1800
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.567872166633606,
+      "learning_rate": 8.2039911308204e-05,
+      "loss": 7.7338,
+      "step": 1850
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3017679452896118,
+      "learning_rate": 8.425720620842572e-05,
+      "loss": 7.804,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7510960102081299,
+      "learning_rate": 8.647450110864746e-05,
+      "loss": 7.7405,
+      "step": 1950
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7215120792388916,
+      "learning_rate": 8.869179600886919e-05,
+      "loss": 7.7429,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6202715635299683,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 7.6588,
+      "step": 2050
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5680756568908691,
+      "learning_rate": 9.312638580931264e-05,
+      "loss": 7.6224,
+      "step": 2100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.462240219116211,
+      "learning_rate": 9.534368070953438e-05,
+      "loss": 7.6851,
+      "step": 2150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2018320560455322,
+      "learning_rate": 9.75609756097561e-05,
+      "loss": 7.6443,
+      "step": 2200
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9520208835601807,
+      "learning_rate": 9.977827050997783e-05,
+      "loss": 7.6456,
+      "step": 2250
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.115421175956726,
+      "learning_rate": 0.00010199556541019956,
+      "loss": 7.5894,
+      "step": 2300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6002250909805298,
+      "learning_rate": 0.0001042128603104213,
+      "loss": 7.6017,
+      "step": 2350
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6516796350479126,
+      "learning_rate": 0.00010643015521064301,
+      "loss": 7.4548,
+      "step": 2400
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2168257236480713,
+      "learning_rate": 0.00010864745011086475,
+      "loss": 7.5867,
+      "step": 2450
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5447593927383423,
+      "learning_rate": 0.00011086474501108647,
+      "loss": 7.5317,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6840906143188477,
+      "learning_rate": 0.00011308203991130821,
+      "loss": 7.5127,
+      "step": 2550
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2965503931045532,
+      "learning_rate": 0.00011529933481152995,
+      "loss": 7.4911,
+      "step": 2600
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.643584966659546,
+      "learning_rate": 0.00011751662971175166,
+      "loss": 7.4416,
+      "step": 2650
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5419111251831055,
+      "learning_rate": 0.0001197339246119734,
+      "loss": 7.4944,
+      "step": 2700
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7774205207824707,
+      "learning_rate": 0.00012195121951219512,
+      "loss": 7.4244,
+      "step": 2750
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1709322929382324,
+      "learning_rate": 0.00012416851441241685,
+      "loss": 7.371,
+      "step": 2800
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5503411293029785,
+      "learning_rate": 0.0001263858093126386,
+      "loss": 7.3031,
+      "step": 2850
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7744035720825195,
+      "learning_rate": 0.00012860310421286032,
+      "loss": 7.3338,
+      "step": 2900
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.2014000415802,
+      "learning_rate": 0.00013082039911308205,
+      "loss": 7.2962,
+      "step": 2950
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6716220378875732,
+      "learning_rate": 0.00013303769401330377,
+      "loss": 7.3348,
+      "step": 3000
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7045074701309204,
+      "learning_rate": 0.0001352549889135255,
+      "loss": 7.2864,
+      "step": 3050
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8933771848678589,
+      "learning_rate": 0.00013747228381374724,
+      "loss": 7.2744,
+      "step": 3100
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.298779249191284,
+      "learning_rate": 0.00013968957871396897,
+      "loss": 7.2472,
+      "step": 3150
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3420922756195068,
+      "learning_rate": 0.0001419068736141907,
+      "loss": 7.3019,
+      "step": 3200
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9339039325714111,
+      "learning_rate": 0.00014412416851441242,
+      "loss": 7.2982,
+      "step": 3250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.69667387008667,
+      "learning_rate": 0.00014634146341463414,
+      "loss": 7.2851,
+      "step": 3300
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3124189376831055,
+      "learning_rate": 0.0001485587583148559,
+      "loss": 7.258,
+      "step": 3350
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.975651741027832,
+      "learning_rate": 0.00015077605321507762,
+      "loss": 7.1275,
+      "step": 3400
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9704022407531738,
+      "learning_rate": 0.00015299334811529934,
+      "loss": 7.1473,
+      "step": 3450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5047757625579834,
+      "learning_rate": 0.00015521064301552106,
+      "loss": 7.1096,
+      "step": 3500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5465894937515259,
+      "learning_rate": 0.0001574279379157428,
+      "loss": 7.1501,
+      "step": 3550
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9557933807373047,
+      "learning_rate": 0.00015964523281596454,
+      "loss": 7.2033,
+      "step": 3600
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.420116424560547,
+      "learning_rate": 0.00016186252771618626,
+      "loss": 7.1275,
+      "step": 3650
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.114737033843994,
+      "learning_rate": 0.000164079822616408,
+      "loss": 7.0932,
+      "step": 3700
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.3085389137268066,
+      "learning_rate": 0.00016629711751662974,
+      "loss": 7.0311,
+      "step": 3750
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5679140090942383,
+      "learning_rate": 0.00016851441241685144,
+      "loss": 6.9168,
+      "step": 3800
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8611838817596436,
+      "learning_rate": 0.0001707317073170732,
+      "loss": 7.0085,
+      "step": 3850
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8603994846343994,
+      "learning_rate": 0.0001729490022172949,
+      "loss": 6.9432,
+      "step": 3900
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4244627952575684,
+      "learning_rate": 0.00017516629711751663,
+      "loss": 6.9333,
+      "step": 3950
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.177870750427246,
+      "learning_rate": 0.00017738359201773839,
+      "loss": 6.9499,
+      "step": 4000
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9320554733276367,
+      "learning_rate": 0.00017960088691796008,
+      "loss": 6.8204,
+      "step": 4050
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5062849521636963,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 6.9505,
+      "step": 4100
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9272422790527344,
+      "learning_rate": 0.00018403547671840356,
+      "loss": 6.8701,
+      "step": 4150
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0309596061706543,
+      "learning_rate": 0.00018625277161862528,
+      "loss": 6.924,
+      "step": 4200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0265886783599854,
+      "learning_rate": 0.00018847006651884703,
+      "loss": 6.9223,
+      "step": 4250
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5160486698150635,
+      "learning_rate": 0.00019068736141906876,
+      "loss": 6.8708,
+      "step": 4300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.613301992416382,
+      "learning_rate": 0.00019290465631929045,
+      "loss": 6.8937,
+      "step": 4350
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3031229972839355,
+      "learning_rate": 0.0001951219512195122,
+      "loss": 6.8337,
+      "step": 4400
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.54779052734375,
+      "learning_rate": 0.00019733924611973393,
+      "loss": 6.8334,
+      "step": 4450
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.8277971744537354,
+      "learning_rate": 0.00019955654101995565,
+      "loss": 6.7925,
+      "step": 4500
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.0113885402679443,
+      "learning_rate": 0.00019999989242739025,
+      "loss": 6.8458,
+      "step": 4550
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2395377159118652,
+      "learning_rate": 0.00019999945541405976,
+      "loss": 6.6251,
+      "step": 4600
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.445993423461914,
+      "learning_rate": 0.0001999986822381884,
+      "loss": 6.8099,
+      "step": 4650
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.077752590179443,
+      "learning_rate": 0.0001999975729023753,
+      "loss": 6.8053,
+      "step": 4700
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.167569875717163,
+      "learning_rate": 0.00019999612741034963,
+      "loss": 6.7706,
+      "step": 4750
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.893659234046936,
+      "learning_rate": 0.00019999434576697066,
+      "loss": 6.8245,
+      "step": 4800
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.6101326942443848,
+      "learning_rate": 0.00019999222797822762,
+      "loss": 6.7407,
+      "step": 4850
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.2858726978302,
+      "learning_rate": 0.00019998977405123974,
+      "loss": 6.74,
+      "step": 4900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9325459003448486,
+      "learning_rate": 0.0001999869839942563,
+      "loss": 6.716,
+      "step": 4950
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0043437480926514,
+      "learning_rate": 0.00019998385781665643,
+      "loss": 6.6003,
+      "step": 5000
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.151523113250732,
+      "learning_rate": 0.00019998039552894924,
+      "loss": 6.6801,
+      "step": 5050
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.8407771587371826,
+      "learning_rate": 0.00019997659714277372,
+      "loss": 6.608,
+      "step": 5100
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.230713129043579,
+      "learning_rate": 0.00019997246267089867,
+      "loss": 6.6479,
+      "step": 5150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.2546942234039307,
+      "learning_rate": 0.0001999679921272227,
+      "loss": 6.6548,
+      "step": 5200
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.180986166000366,
+      "learning_rate": 0.00019996318552677425,
+      "loss": 6.6851,
+      "step": 5250
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.341231346130371,
+      "learning_rate": 0.00019995804288571134,
+      "loss": 6.547,
+      "step": 5300
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.1117124557495117,
+      "learning_rate": 0.00019995256422132172,
+      "loss": 6.7072,
+      "step": 5350
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0082530975341797,
+      "learning_rate": 0.0001999467495520227,
+      "loss": 6.5422,
+      "step": 5400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.409489870071411,
+      "learning_rate": 0.0001999405988973611,
+      "loss": 6.3716,
+      "step": 5450
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.649052381515503,
+      "learning_rate": 0.00019993411227801328,
+      "loss": 6.6434,
+      "step": 5500
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.081116199493408,
+      "learning_rate": 0.00019992728971578492,
+      "loss": 6.4624,
+      "step": 5550
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.1578280925750732,
+      "learning_rate": 0.00019992013123361102,
+      "loss": 6.5416,
+      "step": 5600
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.7874557971954346,
+      "learning_rate": 0.0001999126368555559,
+      "loss": 6.4512,
+      "step": 5650
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.7693099975585938,
+      "learning_rate": 0.00019990480660681293,
+      "loss": 6.5105,
+      "step": 5700
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4338185787200928,
+      "learning_rate": 0.00019989680712666593,
+      "loss": 6.5092,
+      "step": 5750
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.656937837600708,
+      "learning_rate": 0.00019988831193270577,
+      "loss": 6.4269,
+      "step": 5800
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.857292652130127,
+      "learning_rate": 0.00019987948094982952,
+      "loss": 6.4387,
+      "step": 5850
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.4963467121124268,
+      "learning_rate": 0.00019987031420772385,
+      "loss": 6.3851,
+      "step": 5900
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.602522611618042,
+      "learning_rate": 0.00019986081173720396,
+      "loss": 6.3413,
+      "step": 5950
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6455273628234863,
+      "learning_rate": 0.00019985097357021385,
+      "loss": 6.2965,
+      "step": 6000
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.5592167377471924,
+      "learning_rate": 0.0001998407997398259,
+      "loss": 6.4293,
+      "step": 6050
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.6016533374786377,
+      "learning_rate": 0.00019983029028024094,
+      "loss": 6.2897,
+      "step": 6100
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.5536839962005615,
+      "learning_rate": 0.000199819445226788,
+      "loss": 6.3157,
+      "step": 6150
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0514349937438965,
+      "learning_rate": 0.00019980826461592427,
+      "loss": 6.3847,
+      "step": 6200
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.72495174407959,
+      "learning_rate": 0.00019979674848523505,
+      "loss": 6.3517,
+      "step": 6250
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4264872074127197,
+      "learning_rate": 0.00019978489687343335,
+      "loss": 6.2533,
+      "step": 6300
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8361423015594482,
+      "learning_rate": 0.0001997727098203602,
+      "loss": 6.3654,
+      "step": 6350
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.9690892696380615,
+      "learning_rate": 0.00019976018736698404,
+      "loss": 6.3968,
+      "step": 6400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6132867336273193,
+      "learning_rate": 0.0001997473295554009,
+      "loss": 6.3444,
+      "step": 6450
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.820697784423828,
+      "learning_rate": 0.00019973413642883424,
+      "loss": 6.2019,
+      "step": 6500
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2316782474517822,
+      "learning_rate": 0.00019972060803163458,
+      "loss": 6.2049,
+      "step": 6550
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.9528305530548096,
+      "learning_rate": 0.00019970674440927957,
+      "loss": 6.1718,
+      "step": 6600
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.891073226928711,
+      "learning_rate": 0.0001996925456083738,
+      "loss": 6.2393,
+      "step": 6650
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.813270092010498,
+      "learning_rate": 0.00019967801167664853,
+      "loss": 6.2116,
+      "step": 6700
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2726826667785645,
+      "learning_rate": 0.00019966314266296173,
+      "loss": 6.1521,
+      "step": 6750
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3895318508148193,
+      "learning_rate": 0.00019964793861729772,
+      "loss": 6.1072,
+      "step": 6800
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.190431833267212,
+      "learning_rate": 0.000199632399590767,
+      "loss": 6.2009,
+      "step": 6850
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.79266095161438,
+      "learning_rate": 0.00019961652563560634,
+      "loss": 6.028,
+      "step": 6900
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.260039806365967,
+      "learning_rate": 0.00019960031680517826,
+      "loss": 6.0733,
+      "step": 6950
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0739686489105225,
+      "learning_rate": 0.0001995837731539711,
+      "loss": 6.0521,
+      "step": 7000
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.0517771244049072,
+      "learning_rate": 0.00019956689473759872,
+      "loss": 6.0544,
+      "step": 7050
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.9524648189544678,
+      "learning_rate": 0.0001995496816128003,
+      "loss": 6.1326,
+      "step": 7100
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.498497486114502,
+      "learning_rate": 0.00019953213383744033,
+      "loss": 6.236,
+      "step": 7150
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.157576084136963,
+      "learning_rate": 0.00019951425147050807,
+      "loss": 5.9898,
+      "step": 7200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.9297516345977783,
+      "learning_rate": 0.00019949603457211775,
+      "loss": 6.086,
+      "step": 7250
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.3214786052703857,
+      "learning_rate": 0.00019947748320350804,
+      "loss": 5.9589,
+      "step": 7300
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.8847291469573975,
+      "learning_rate": 0.00019945859742704201,
+      "loss": 6.1931,
+      "step": 7350
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.387896776199341,
+      "learning_rate": 0.00019943937730620702,
+      "loss": 6.0539,
+      "step": 7400
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.1214797496795654,
+      "learning_rate": 0.00019941982290561417,
+      "loss": 6.0288,
+      "step": 7450
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.7995123863220215,
+      "learning_rate": 0.00019939993429099841,
+      "loss": 6.0526,
+      "step": 7500
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.788393974304199,
+      "learning_rate": 0.00019937971152921818,
+      "loss": 5.9799,
+      "step": 7550
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 4.009220123291016,
+      "learning_rate": 0.0001993591546882552,
+      "loss": 6.1223,
+      "step": 7600
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.5576276779174805,
+      "learning_rate": 0.00019933826383721428,
+      "loss": 5.989,
+      "step": 7650
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.1287412643432617,
+      "learning_rate": 0.00019931703904632294,
+      "loss": 6.0542,
+      "step": 7700
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6518595218658447,
+      "learning_rate": 0.00019929548038693146,
+      "loss": 6.041,
+      "step": 7750
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.268080472946167,
+      "learning_rate": 0.0001992735879315123,
+      "loss": 5.888,
+      "step": 7800
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.6055593490600586,
+      "learning_rate": 0.00019925136175366007,
+      "loss": 5.913,
+      "step": 7850
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.866463661193848,
+      "learning_rate": 0.00019922880192809137,
+      "loss": 5.9858,
+      "step": 7900
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.44808292388916,
+      "learning_rate": 0.00019920590853064423,
+      "loss": 5.7686,
+      "step": 7950
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.9507765769958496,
+      "learning_rate": 0.00019918268163827808,
+      "loss": 5.8557,
+      "step": 8000
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.441870927810669,
+      "learning_rate": 0.00019915912132907352,
+      "loss": 5.8268,
+      "step": 8050
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.838809013366699,
+      "learning_rate": 0.00019913522768223182,
+      "loss": 5.9833,
+      "step": 8100
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.165487289428711,
+      "learning_rate": 0.00019911100077807498,
+      "loss": 5.7422,
+      "step": 8150
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5947463512420654,
+      "learning_rate": 0.0001990864406980452,
+      "loss": 5.7479,
+      "step": 8200
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.130446434020996,
+      "learning_rate": 0.00019906154752470472,
+      "loss": 5.7767,
+      "step": 8250
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.866550922393799,
+      "learning_rate": 0.00019903632134173554,
+      "loss": 5.7681,
+      "step": 8300
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2839725017547607,
+      "learning_rate": 0.00019901076223393903,
+      "loss": 5.6656,
+      "step": 8350
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0762476921081543,
+      "learning_rate": 0.0001989848702872359,
+      "loss": 5.789,
+      "step": 8400
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.7109107971191406,
+      "learning_rate": 0.00019895864558866556,
+      "loss": 5.773,
+      "step": 8450
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.400998115539551,
+      "learning_rate": 0.00019893208822638618,
+      "loss": 5.7506,
+      "step": 8500
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3062849044799805,
+      "learning_rate": 0.00019890519828967413,
+      "loss": 5.7515,
+      "step": 8550
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.109920501708984,
+      "learning_rate": 0.00019887797586892373,
+      "loss": 5.7972,
+      "step": 8600
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.4838390350341797,
+      "learning_rate": 0.00019885042105564717,
+      "loss": 5.6753,
+      "step": 8650
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.251760959625244,
+      "learning_rate": 0.00019882253394247381,
+      "loss": 5.6303,
+      "step": 8700
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.042376518249512,
+      "learning_rate": 0.00019879431462315025,
+      "loss": 5.5753,
+      "step": 8750
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.239652633666992,
+      "learning_rate": 0.0001987657631925398,
+      "loss": 5.5335,
+      "step": 8800
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.15481424331665,
+      "learning_rate": 0.00019873687974662215,
+      "loss": 5.5396,
+      "step": 8850
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.36835241317749,
+      "learning_rate": 0.00019870766438249317,
+      "loss": 5.6017,
+      "step": 8900
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.165258407592773,
+      "learning_rate": 0.00019867811719836452,
+      "loss": 5.7228,
+      "step": 8950
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.125988006591797,
+      "learning_rate": 0.0001986482382935633,
+      "loss": 5.5787,
+      "step": 9000
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.177731037139893,
+      "learning_rate": 0.0001986180277685317,
+      "loss": 5.5829,
+      "step": 9050
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.006561279296875,
+      "learning_rate": 0.00019858748572482683,
+      "loss": 5.5466,
+      "step": 9100
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.33070182800293,
+      "learning_rate": 0.00019855661226512007,
+      "loss": 5.5544,
+      "step": 9150
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.358560085296631,
+      "learning_rate": 0.00019852540749319708,
+      "loss": 5.4599,
+      "step": 9200
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.536096096038818,
+      "learning_rate": 0.00019849387151395708,
+      "loss": 5.4983,
+      "step": 9250
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.66163444519043,
+      "learning_rate": 0.0001984620044334129,
+      "loss": 5.4097,
+      "step": 9300
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.4319233894348145,
+      "learning_rate": 0.00019842980635869024,
+      "loss": 5.4093,
+      "step": 9350
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.98419713973999,
+      "learning_rate": 0.0001983972773980276,
+      "loss": 5.4056,
+      "step": 9400
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.6354339122772217,
+      "learning_rate": 0.0001983644176607757,
+      "loss": 5.3171,
+      "step": 9450
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.495342254638672,
+      "learning_rate": 0.00019833122725739736,
+      "loss": 5.4521,
+      "step": 9500
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.5558671951293945,
+      "learning_rate": 0.00019829770629946678,
+      "loss": 5.5158,
+      "step": 9550
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.7165732383728027,
+      "learning_rate": 0.00019826385489966957,
+      "loss": 5.301,
+      "step": 9600
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.030915260314941,
+      "learning_rate": 0.00019822967317180204,
+      "loss": 5.3316,
+      "step": 9650
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.385923385620117,
+      "learning_rate": 0.00019819516123077094,
+      "loss": 5.3844,
+      "step": 9700
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.383516788482666,
+      "learning_rate": 0.00019816101926755305,
+      "loss": 5.2995,
+      "step": 9750
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.446406364440918,
+      "learning_rate": 0.00019812585384780055,
+      "loss": 5.386,
+      "step": 9800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.345483303070068,
+      "learning_rate": 0.00019809035856388805,
+      "loss": 5.2815,
+      "step": 9850
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.791261672973633,
+      "learning_rate": 0.00019805453353513813,
+      "loss": 5.3757,
+      "step": 9900
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.622151851654053,
+      "learning_rate": 0.00019801837888198172,
+      "loss": 5.4405,
+      "step": 9950
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.934606075286865,
+      "learning_rate": 0.0001979818947259579,
+      "loss": 5.139,
+      "step": 10000
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.9659693241119385,
+      "learning_rate": 0.0001979450811897134,
+      "loss": 5.1726,
+      "step": 10050
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.214992046356201,
+      "learning_rate": 0.00019790793839700226,
+      "loss": 5.2864,
+      "step": 10100
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.5359601974487305,
+      "learning_rate": 0.00019787046647268524,
+      "loss": 5.1443,
+      "step": 10150
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.26462984085083,
+      "learning_rate": 0.00019783266554272962,
+      "loss": 5.0597,
+      "step": 10200
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.053945064544678,
+      "learning_rate": 0.00019779453573420873,
+      "loss": 5.2946,
+      "step": 10250
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.082211494445801,
+      "learning_rate": 0.00019775607717530127,
+      "loss": 5.2075,
+      "step": 10300
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.107390403747559,
+      "learning_rate": 0.00019771728999529132,
+      "loss": 5.1394,
+      "step": 10350
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.58411169052124,
+      "learning_rate": 0.00019767817432456752,
+      "loss": 5.1064,
+      "step": 10400
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.38965892791748,
+      "learning_rate": 0.00019763952239228627,
+      "loss": 5.0808,
+      "step": 10450
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.885803699493408,
+      "learning_rate": 0.00019759975669894338,
+      "loss": 5.0664,
+      "step": 10500
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.1605916023254395,
+      "learning_rate": 0.00019755966290999167,
+      "loss": 5.2469,
+      "step": 10550
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.821887016296387,
+      "learning_rate": 0.00019751924116021225,
+      "loss": 5.2451,
+      "step": 10600
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.865694761276245,
+      "learning_rate": 0.00019747849158548858,
+      "loss": 5.2334,
+      "step": 10650
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.640681028366089,
+      "learning_rate": 0.00019743741432280625,
+      "loss": 5.1206,
+      "step": 10700
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.04166316986084,
+      "learning_rate": 0.00019739600951025236,
+      "loss": 5.0059,
+      "step": 10750
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.637605667114258,
+      "learning_rate": 0.00019735427728701516,
+      "loss": 5.0302,
+      "step": 10800
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.08723783493042,
+      "learning_rate": 0.0001973122177933835,
+      "loss": 5.1551,
+      "step": 10850
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.7944953441619873,
+      "learning_rate": 0.00019726983117074643,
+      "loss": 5.0665,
+      "step": 10900
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.2847371101379395,
+      "learning_rate": 0.00019722711756159266,
+      "loss": 5.2212,
+      "step": 10950
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.109150409698486,
+      "learning_rate": 0.00019718407710951012,
+      "loss": 5.2645,
+      "step": 11000
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.127768039703369,
+      "learning_rate": 0.0001971407099591855,
+      "loss": 5.0395,
+      "step": 11050
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.058667182922363,
+      "learning_rate": 0.00019709701625640367,
+      "loss": 5.0247,
+      "step": 11100
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 5.4407267570495605,
+      "learning_rate": 0.00019705299614804732,
+      "loss": 4.9935,
+      "step": 11150
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.7877707481384277,
+      "learning_rate": 0.00019700864978209636,
+      "loss": 5.074,
+      "step": 11200
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.777330160140991,
+      "learning_rate": 0.00019696397730762746,
+      "loss": 5.0458,
+      "step": 11250
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.143067836761475,
+      "learning_rate": 0.0001969189788748136,
+      "loss": 4.9375,
+      "step": 11300
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.560107231140137,
+      "learning_rate": 0.00019687365463492344,
+      "loss": 4.8285,
+      "step": 11350
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.057905197143555,
+      "learning_rate": 0.00019682800474032095,
+      "loss": 4.9753,
+      "step": 11400
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.835442066192627,
+      "learning_rate": 0.00019678202934446482,
+      "loss": 4.9368,
+      "step": 11450
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.135551929473877,
+      "learning_rate": 0.0001967357286019079,
+      "loss": 4.9994,
+      "step": 11500
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.615053653717041,
+      "learning_rate": 0.00019668910266829685,
+      "loss": 5.0182,
+      "step": 11550
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.474258899688721,
+      "learning_rate": 0.0001966421517003714,
+      "loss": 4.8704,
+      "step": 11600
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.264945030212402,
+      "learning_rate": 0.00019659487585596406,
+      "loss": 4.9076,
+      "step": 11650
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.091209411621094,
+      "learning_rate": 0.00019654727529399925,
+      "loss": 4.7135,
+      "step": 11700
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.154038429260254,
+      "learning_rate": 0.00019649935017449318,
+      "loss": 4.8239,
+      "step": 11750
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.697162628173828,
+      "learning_rate": 0.00019645110065855305,
+      "loss": 4.9972,
+      "step": 11800
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.0024847984313965,
+      "learning_rate": 0.00019640252690837645,
+      "loss": 4.8854,
+      "step": 11850
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.9416885375976562,
+      "learning_rate": 0.0001963536290872511,
+      "loss": 4.8547,
+      "step": 11900
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.978651285171509,
+      "learning_rate": 0.000196304407359554,
+      "loss": 4.7873,
+      "step": 11950
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.435175895690918,
+      "learning_rate": 0.0001962548618907511,
+      "loss": 4.8124,
+      "step": 12000
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.8776824474334717,
+      "learning_rate": 0.00019620499284739662,
+      "loss": 4.8896,
+      "step": 12050
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.041496276855469,
+      "learning_rate": 0.00019615480039713248,
+      "loss": 4.8343,
+      "step": 12100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.18281888961792,
+      "learning_rate": 0.00019610428470868784,
+      "loss": 4.8559,
+      "step": 12150
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.223630905151367,
+      "learning_rate": 0.00019605344595187844,
+      "loss": 4.8153,
+      "step": 12200
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.63677453994751,
+      "learning_rate": 0.0001960022842976061,
+      "loss": 4.7951,
+      "step": 12250
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.188296794891357,
+      "learning_rate": 0.00019595079991785802,
+      "loss": 4.8904,
+      "step": 12300
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.402559280395508,
+      "learning_rate": 0.00019589899298570634,
+      "loss": 4.7851,
+      "step": 12350
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.976877212524414,
+      "learning_rate": 0.00019584686367530755,
+      "loss": 4.6431,
+      "step": 12400
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.849298477172852,
+      "learning_rate": 0.0001957944121619018,
+      "loss": 4.7544,
+      "step": 12450
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.932714462280273,
+      "learning_rate": 0.0001957416386218124,
+      "loss": 4.6811,
+      "step": 12500
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.682474136352539,
+      "learning_rate": 0.00019568854323244515,
+      "loss": 4.799,
+      "step": 12550
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.228520393371582,
+      "learning_rate": 0.00019563619766470511,
+      "loss": 4.7622,
+      "step": 12600
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.093870162963867,
+      "learning_rate": 0.00019558246554138458,
+      "loss": 4.7369,
+      "step": 12650
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.248356342315674,
+      "learning_rate": 0.0001955284121038694,
+      "loss": 4.7519,
+      "step": 12700
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.924299955368042,
+      "learning_rate": 0.00019547403753386803,
+      "loss": 4.6441,
+      "step": 12750
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.972569942474365,
+      "learning_rate": 0.00019542043906868188,
+      "loss": 4.7192,
+      "step": 12800
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.033604145050049,
+      "learning_rate": 0.00019536542919665846,
+      "loss": 4.6397,
+      "step": 12850
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.222695350646973,
+      "learning_rate": 0.00019531009874003928,
+      "loss": 4.6309,
+      "step": 12900
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.810999631881714,
+      "learning_rate": 0.00019525444788482562,
+      "loss": 4.6513,
+      "step": 12950
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.272600173950195,
+      "learning_rate": 0.00019519847681809585,
+      "loss": 4.8001,
+      "step": 13000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.836308002471924,
+      "learning_rate": 0.00019514218572800468,
+      "loss": 4.7101,
+      "step": 13050
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.598148345947266,
+      "learning_rate": 0.00019508557480378276,
+      "loss": 4.5578,
+      "step": 13100
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.910820722579956,
+      "learning_rate": 0.0001950286442357358,
+      "loss": 4.7124,
+      "step": 13150
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.856081962585449,
+      "learning_rate": 0.00019497139421524416,
+      "loss": 4.7563,
+      "step": 13200
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.151907920837402,
+      "learning_rate": 0.00019491382493476195,
+      "loss": 4.6726,
+      "step": 13250
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.349935054779053,
+      "learning_rate": 0.0001948559365878166,
+      "loss": 4.6341,
+      "step": 13300
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.8229756355285645,
+      "learning_rate": 0.00019479772936900811,
+      "loss": 4.6183,
+      "step": 13350
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.495506286621094,
+      "learning_rate": 0.0001947392034740084,
+      "loss": 4.6608,
+      "step": 13400
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.307513236999512,
+      "learning_rate": 0.00019468035909956072,
+      "loss": 4.6805,
+      "step": 13450
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.939659595489502,
+      "learning_rate": 0.0001946211964434788,
+      "loss": 4.679,
+      "step": 13500
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.444967269897461,
+      "learning_rate": 0.00019456171570464653,
+      "loss": 4.7195,
+      "step": 13550
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.513270854949951,
+      "learning_rate": 0.00019450191708301687,
+      "loss": 4.5367,
+      "step": 13600
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.617405414581299,
+      "learning_rate": 0.00019444180077961146,
+      "loss": 4.5742,
+      "step": 13650
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.580646991729736,
+      "learning_rate": 0.00019438136699652001,
+      "loss": 4.4936,
+      "step": 13700
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.657532691955566,
+      "learning_rate": 0.00019432061593689927,
+      "loss": 4.6877,
+      "step": 13750
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.374803066253662,
+      "learning_rate": 0.0001942595478049727,
+      "loss": 4.6101,
+      "step": 13800
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.1111650466918945,
+      "learning_rate": 0.00019419816280602962,
+      "loss": 4.6185,
+      "step": 13850
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.18306303024292,
+      "learning_rate": 0.00019413646114642446,
+      "loss": 4.5524,
+      "step": 13900
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.411191463470459,
+      "learning_rate": 0.00019407444303357624,
+      "loss": 4.4346,
+      "step": 13950
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.161925792694092,
+      "learning_rate": 0.0001940121086759678,
+      "loss": 4.3702,
+      "step": 14000
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.059813022613525,
+      "learning_rate": 0.000193949458283145,
+      "loss": 4.5351,
+      "step": 14050
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.563150882720947,
+      "learning_rate": 0.00019388649206571616,
+      "loss": 4.477,
+      "step": 14100
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.1144609451293945,
+      "learning_rate": 0.00019382321023535127,
+      "loss": 4.6033,
+      "step": 14150
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.734794616699219,
+      "learning_rate": 0.00019375961300478127,
+      "loss": 4.5287,
+      "step": 14200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.543684959411621,
+      "learning_rate": 0.00019369570058779743,
+      "loss": 4.4474,
+      "step": 14250
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.4647979736328125,
+      "learning_rate": 0.00019363147319925047,
+      "loss": 4.3806,
+      "step": 14300
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.058681964874268,
+      "learning_rate": 0.00019356693105505006,
+      "loss": 4.4998,
+      "step": 14350
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.494804859161377,
+      "learning_rate": 0.00019350207437216386,
+      "loss": 4.3911,
+      "step": 14400
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.227470397949219,
+      "learning_rate": 0.00019343690336861687,
+      "loss": 4.2557,
+      "step": 14450
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.7686829566955566,
+      "learning_rate": 0.00019337141826349092,
+      "loss": 4.313,
+      "step": 14500
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.975152492523193,
+      "learning_rate": 0.00019330561927692345,
+      "loss": 4.2914,
+      "step": 14550
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.811885356903076,
+      "learning_rate": 0.00019323950663010733,
+      "loss": 4.3566,
+      "step": 14600
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.566829204559326,
+      "learning_rate": 0.00019317308054528966,
+      "loss": 4.2847,
+      "step": 14650
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.977478504180908,
+      "learning_rate": 0.0001931063412457713,
+      "loss": 4.3034,
+      "step": 14700
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.601086616516113,
+      "learning_rate": 0.00019303928895590596,
+      "loss": 4.1929,
+      "step": 14750
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.051478385925293,
+      "learning_rate": 0.0001929719239010996,
+      "loss": 4.2749,
+      "step": 14800
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.248847961425781,
+      "learning_rate": 0.00019290424630780947,
+      "loss": 4.3419,
+      "step": 14850
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.392062664031982,
+      "learning_rate": 0.0001928362564035436,
+      "loss": 4.4038,
+      "step": 14900
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.6346211433410645,
+      "learning_rate": 0.00019276795441685975,
+      "loss": 4.3403,
+      "step": 14950
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.646982192993164,
+      "learning_rate": 0.00019269934057736493,
+      "loss": 4.252,
+      "step": 15000
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.455059051513672,
+      "learning_rate": 0.00019263041511571438,
+      "loss": 4.3809,
+      "step": 15050
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.478726387023926,
+      "learning_rate": 0.00019256117826361096,
+      "loss": 4.1885,
+      "step": 15100
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.029292106628418,
+      "learning_rate": 0.0001924916302538043,
+      "loss": 4.2615,
+      "step": 15150
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.6447978019714355,
+      "learning_rate": 0.00019242177132009,
+      "loss": 4.268,
+      "step": 15200
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.165138244628906,
+      "learning_rate": 0.00019235160169730895,
+      "loss": 4.3222,
+      "step": 15250
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.661884784698486,
+      "learning_rate": 0.00019228112162134641,
+      "loss": 4.3179,
+      "step": 15300
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.117990493774414,
+      "learning_rate": 0.0001922103313291313,
+      "loss": 4.2241,
+      "step": 15350
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.299765110015869,
+      "learning_rate": 0.0001921392310586353,
+      "loss": 4.2602,
+      "step": 15400
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.798460483551025,
+      "learning_rate": 0.00019206782104887223,
+      "loss": 4.3096,
+      "step": 15450
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.016506671905518,
+      "learning_rate": 0.00019199610153989712,
+      "loss": 4.2073,
+      "step": 15500
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 9.708767890930176,
+      "learning_rate": 0.0001919240727728054,
+      "loss": 4.2099,
+      "step": 15550
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.904361248016357,
+      "learning_rate": 0.00019185173498973204,
+      "loss": 4.2461,
+      "step": 15600
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.290199279785156,
+      "learning_rate": 0.00019177908843385103,
+      "loss": 4.115,
+      "step": 15650
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.290179252624512,
+      "learning_rate": 0.00019170613334937406,
+      "loss": 4.3295,
+      "step": 15700
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.071104526519775,
+      "learning_rate": 0.00019163286998155027,
+      "loss": 4.1532,
+      "step": 15750
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.5464067459106445,
+      "learning_rate": 0.00019155929857666494,
+      "loss": 4.0761,
+      "step": 15800
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.664229393005371,
+      "learning_rate": 0.0001914854193820389,
+      "loss": 4.1371,
+      "step": 15850
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.168484210968018,
+      "learning_rate": 0.0001914112326460277,
+      "loss": 4.178,
+      "step": 15900
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.570041179656982,
+      "learning_rate": 0.0001913367386180207,
+      "loss": 4.1536,
+      "step": 15950
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.298222064971924,
+      "learning_rate": 0.00019126193754844036,
+      "loss": 4.2089,
+      "step": 16000
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.139255523681641,
+      "learning_rate": 0.0001911868296887411,
+      "loss": 4.1362,
+      "step": 16050
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.763050556182861,
+      "learning_rate": 0.00019111141529140887,
+      "loss": 4.1106,
+      "step": 16100
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.586143493652344,
+      "learning_rate": 0.00019103569460995998,
+      "loss": 3.9519,
+      "step": 16150
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.827348232269287,
+      "learning_rate": 0.00019095966789894038,
+      "loss": 3.9598,
+      "step": 16200
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.121611595153809,
+      "learning_rate": 0.00019088333541392478,
+      "loss": 4.1347,
+      "step": 16250
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.110377788543701,
+      "learning_rate": 0.00019080669741151581,
+      "loss": 4.0088,
+      "step": 16300
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.672893047332764,
+      "learning_rate": 0.00019072975414934318,
+      "loss": 4.0916,
+      "step": 16350
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.667397499084473,
+      "learning_rate": 0.00019065250588606262,
+      "loss": 4.0695,
+      "step": 16400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.404243469238281,
+      "learning_rate": 0.0001905749528813553,
+      "loss": 3.9728,
+      "step": 16450
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.912601470947266,
+      "learning_rate": 0.00019049709539592686,
+      "loss": 4.029,
+      "step": 16500
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.015479564666748,
+      "learning_rate": 0.00019041893369150636,
+      "loss": 4.0268,
+      "step": 16550
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 6.656422138214111,
+      "learning_rate": 0.00019034046803084563,
+      "loss": 4.0393,
+      "step": 16600
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.685242176055908,
+      "learning_rate": 0.00019026169867771825,
+      "loss": 4.1104,
+      "step": 16650
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.503780364990234,
+      "learning_rate": 0.00019018262589691874,
+      "loss": 4.0344,
+      "step": 16700
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.73757266998291,
+      "learning_rate": 0.00019010324995426156,
+      "loss": 4.1114,
+      "step": 16750
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 7.276214122772217,
+      "learning_rate": 0.0001900235711165804,
+      "loss": 3.8838,
+      "step": 16800
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.2224273681640625,
+      "learning_rate": 0.00018994358965172717,
+      "loss": 3.9479,
+      "step": 16850
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.4751996994018555,
+      "learning_rate": 0.00018986330582857096,
+      "loss": 4.0079,
+      "step": 16900
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.874088764190674,
+      "learning_rate": 0.00018978271991699743,
+      "loss": 4.1664,
+      "step": 16950
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 7.713326454162598,
+      "learning_rate": 0.0001897018321879077,
+      "loss": 3.9646,
+      "step": 17000
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.753252029418945,
+      "learning_rate": 0.00018962064291321747,
+      "loss": 3.8574,
+      "step": 17050
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.962434768676758,
+      "learning_rate": 0.0001895391523658562,
+      "loss": 3.9757,
+      "step": 17100
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.875513553619385,
+      "learning_rate": 0.00018945736081976607,
+      "loss": 4.0424,
+      "step": 17150
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.298293590545654,
+      "learning_rate": 0.00018937526854990108,
+      "loss": 3.958,
+      "step": 17200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.98872184753418,
+      "learning_rate": 0.00018929287583222625,
+      "loss": 3.9225,
+      "step": 17250
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.467836380004883,
+      "learning_rate": 0.00018921018294371645,
+      "loss": 3.9369,
+      "step": 17300
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.920988082885742,
+      "learning_rate": 0.0001891271901623558,
+      "loss": 3.975,
+      "step": 17350
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.652931213378906,
+      "learning_rate": 0.00018904389776713641,
+      "loss": 3.9067,
+      "step": 17400
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.372093200683594,
+      "learning_rate": 0.00018896030603805767,
+      "loss": 3.9267,
+      "step": 17450
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.743618965148926,
+      "learning_rate": 0.00018887641525612518,
+      "loss": 3.8912,
+      "step": 17500
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 8.207468032836914,
+      "learning_rate": 0.00018879222570334985,
+      "loss": 3.9101,
+      "step": 17550
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.930370807647705,
+      "learning_rate": 0.00018870773766274697,
+      "loss": 3.8817,
+      "step": 17600
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.367077350616455,
+      "learning_rate": 0.00018862295141833523,
+      "loss": 3.8931,
+      "step": 17650
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.587210178375244,
+      "learning_rate": 0.00018853786725513575,
+      "loss": 3.9393,
+      "step": 17700
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.502545356750488,
+      "learning_rate": 0.0001884524854591712,
+      "loss": 3.8489,
+      "step": 17750
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.352043628692627,
+      "learning_rate": 0.00018836680631746476,
+      "loss": 3.8162,
+      "step": 17800
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.686196804046631,
+      "learning_rate": 0.00018828083011803917,
+      "loss": 3.9476,
+      "step": 17850
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.225170612335205,
+      "learning_rate": 0.00018819455714991578,
+      "loss": 3.9404,
+      "step": 17900
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 7.1347150802612305,
+      "learning_rate": 0.0001881079877031136,
+      "loss": 3.9798,
+      "step": 17950
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.343573093414307,
+      "learning_rate": 0.0001880211220686482,
+      "loss": 3.9038,
+      "step": 18000
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.858921051025391,
+      "learning_rate": 0.00018793396053853098,
+      "loss": 3.8792,
+      "step": 18050
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.721033573150635,
+      "learning_rate": 0.0001878482554434291,
+      "loss": 3.8421,
+      "step": 18100
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.173632621765137,
+      "learning_rate": 0.00018776050890530516,
+      "loss": 4.0233,
+      "step": 18150
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.996013164520264,
+      "learning_rate": 0.00018767246734761796,
+      "loss": 3.8057,
+      "step": 18200
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.707641124725342,
+      "learning_rate": 0.00018758413106633186,
+      "loss": 3.8299,
+      "step": 18250
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 7.221241474151611,
+      "learning_rate": 0.00018749550035840193,
+      "loss": 3.8828,
+      "step": 18300
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.554357528686523,
+      "learning_rate": 0.00018740657552177305,
+      "loss": 3.8553,
+      "step": 18350
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.664674282073975,
+      "learning_rate": 0.00018731735685537885,
+      "loss": 3.8838,
+      "step": 18400
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.485450267791748,
+      "learning_rate": 0.00018722784465914071,
+      "loss": 3.8165,
+      "step": 18450
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.825826644897461,
+      "learning_rate": 0.00018713803923396668,
+      "loss": 3.7588,
+      "step": 18500
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.392491817474365,
+      "learning_rate": 0.0001870479408817507,
+      "loss": 3.8001,
+      "step": 18550
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.493740081787109,
+      "learning_rate": 0.00018695754990537123,
+      "loss": 3.9735,
+      "step": 18600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.905117511749268,
+      "learning_rate": 0.00018686686660869062,
+      "loss": 3.7334,
+      "step": 18650
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.598316192626953,
+      "learning_rate": 0.0001867758912965537,
+      "loss": 3.8269,
+      "step": 18700
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.979629039764404,
+      "learning_rate": 0.00018668462427478714,
+      "loss": 3.8713,
+      "step": 18750
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.480854511260986,
+      "learning_rate": 0.00018659306585019813,
+      "loss": 3.7792,
+      "step": 18800
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.820549488067627,
+      "learning_rate": 0.00018650121633057346,
+      "loss": 3.6656,
+      "step": 18850
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.579679012298584,
+      "learning_rate": 0.0001864090760246785,
+      "loss": 3.9109,
+      "step": 18900
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.669819355010986,
+      "learning_rate": 0.00018631664524225615,
+      "loss": 3.7815,
+      "step": 18950
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.644351005554199,
+      "learning_rate": 0.0001862239242940257,
+      "loss": 3.7529,
+      "step": 19000
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.022332191467285,
+      "learning_rate": 0.00018613091349168205,
+      "loss": 3.7001,
+      "step": 19050
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.376641273498535,
+      "learning_rate": 0.00018603761314789425,
+      "loss": 3.6871,
+      "step": 19100
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.298123359680176,
+      "learning_rate": 0.00018594402357630495,
+      "loss": 3.8095,
+      "step": 19150
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.590997695922852,
+      "learning_rate": 0.00018585014509152882,
+      "loss": 3.8069,
+      "step": 19200
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.710943222045898,
+      "learning_rate": 0.00018575597800915198,
+      "loss": 3.8547,
+      "step": 19250
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.5094499588012695,
+      "learning_rate": 0.0001856615226457305,
+      "loss": 3.7314,
+      "step": 19300
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.584799766540527,
+      "learning_rate": 0.0001855667793187898,
+      "loss": 3.7514,
+      "step": 19350
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.0391154289245605,
+      "learning_rate": 0.00018547174834682308,
+      "loss": 3.6231,
+      "step": 19400
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.92927885055542,
+      "learning_rate": 0.00018537643004929067,
+      "loss": 3.7008,
+      "step": 19450
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.359600067138672,
+      "learning_rate": 0.00018528082474661867,
+      "loss": 3.798,
+      "step": 19500
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.198579788208008,
+      "learning_rate": 0.0001851849327601981,
+      "loss": 3.7187,
+      "step": 19550
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.796758651733398,
+      "learning_rate": 0.00018508875441238364,
+      "loss": 3.7086,
+      "step": 19600
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.889728546142578,
+      "learning_rate": 0.00018499229002649258,
+      "loss": 3.7387,
+      "step": 19650
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.382203102111816,
+      "learning_rate": 0.0001848955399268039,
+      "loss": 3.5992,
+      "step": 19700
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.061376571655273,
+      "learning_rate": 0.00018479850443855686,
+      "loss": 3.6865,
+      "step": 19750
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.2180681228637695,
+      "learning_rate": 0.0001847011838879503,
+      "loss": 3.7467,
+      "step": 19800
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.063679218292236,
+      "learning_rate": 0.0001846035786021412,
+      "loss": 3.6894,
+      "step": 19850
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.036098480224609,
+      "learning_rate": 0.00018450568890924373,
+      "loss": 3.6412,
+      "step": 19900
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.86781644821167,
+      "learning_rate": 0.00018440751513832822,
+      "loss": 3.637,
+      "step": 19950
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.41668176651001,
+      "learning_rate": 0.00018430905761941983,
+      "loss": 3.6814,
+      "step": 20000
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.117024898529053,
+      "learning_rate": 0.00018421031668349773,
+      "loss": 3.6257,
+      "step": 20050
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 7.368699073791504,
+      "learning_rate": 0.00018411129266249373,
+      "loss": 3.7111,
+      "step": 20100
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.378394603729248,
+      "learning_rate": 0.0001840119858892913,
+      "loss": 3.7197,
+      "step": 20150
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 7.029990196228027,
+      "learning_rate": 0.0001839123966977245,
+      "loss": 3.7267,
+      "step": 20200
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 9.922813415527344,
+      "learning_rate": 0.00018381252542257662,
+      "loss": 3.7203,
+      "step": 20250
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.9374518394470215,
+      "learning_rate": 0.00018371237239957932,
+      "loss": 3.6876,
+      "step": 20300
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.682550430297852,
+      "learning_rate": 0.00018361193796541142,
+      "loss": 3.6862,
+      "step": 20350
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.477772235870361,
+      "learning_rate": 0.00018351122245769771,
+      "loss": 3.5982,
+      "step": 20400
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.745680332183838,
+      "learning_rate": 0.00018341224888886997,
+      "loss": 3.6978,
+      "step": 20450
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.691402912139893,
+      "learning_rate": 0.0001833109778552932,
+      "loss": 3.6693,
+      "step": 20500
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.229629993438721,
+      "learning_rate": 0.00018320942675989125,
+      "loss": 3.6327,
+      "step": 20550
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.655289649963379,
+      "learning_rate": 0.0001831075959440427,
+      "loss": 3.6032,
+      "step": 20600
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.4868927001953125,
+      "learning_rate": 0.00018300548575006658,
+      "loss": 3.7059,
+      "step": 20650
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.387706756591797,
+      "learning_rate": 0.00018290309652122083,
+      "loss": 3.6838,
+      "step": 20700
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.884798049926758,
+      "learning_rate": 0.00018280042860170168,
+      "loss": 3.665,
+      "step": 20750
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 7.185595512390137,
+      "learning_rate": 0.00018269748233664204,
+      "loss": 3.6057,
+      "step": 20800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.449123382568359,
+      "learning_rate": 0.0001825942580721106,
+      "loss": 3.6262,
+      "step": 20850
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.469310283660889,
+      "learning_rate": 0.00018249075615511053,
+      "loss": 3.522,
+      "step": 20900
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.678877353668213,
+      "learning_rate": 0.0001823869769335784,
+      "loss": 3.6757,
+      "step": 20950
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.033955097198486,
+      "learning_rate": 0.000182282920756383,
+      "loss": 3.7316,
+      "step": 21000
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.790628433227539,
+      "learning_rate": 0.00018217858797332413,
+      "loss": 3.545,
+      "step": 21050
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 6.205599308013916,
+      "learning_rate": 0.00018207397893513143,
+      "loss": 3.6035,
+      "step": 21100
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 5.7604804039001465,
+      "learning_rate": 0.00018196909399346316,
+      "loss": 3.6869,
+      "step": 21150
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.528883457183838,
+      "learning_rate": 0.0001818639335009052,
+      "loss": 3.6493,
+      "step": 21200
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.46929407119751,
+      "learning_rate": 0.00018175849781096966,
+      "loss": 3.639,
+      "step": 21250
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.487035274505615,
+      "learning_rate": 0.00018165278727809368,
+      "loss": 3.5755,
+      "step": 21300
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.251669406890869,
+      "learning_rate": 0.00018154680225763848,
+      "loss": 3.704,
+      "step": 21350
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 6.204404830932617,
+      "learning_rate": 0.00018144054310588792,
+      "loss": 3.6071,
+      "step": 21400
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.7311482429504395,
+      "learning_rate": 0.00018133401018004743,
+      "loss": 3.5395,
+      "step": 21450
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.110382556915283,
+      "learning_rate": 0.00018122720383824273,
+      "loss": 3.6643,
+      "step": 21500
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.991401672363281,
+      "learning_rate": 0.0001811201244395187,
+      "loss": 3.6752,
+      "step": 21550
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.788415431976318,
+      "learning_rate": 0.0001810127723438381,
+      "loss": 3.6362,
+      "step": 21600
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.827778339385986,
+      "learning_rate": 0.00018090514791208043,
+      "loss": 3.7298,
+      "step": 21650
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.7845916748046875,
+      "learning_rate": 0.0001807972515060407,
+      "loss": 3.543,
+      "step": 21700
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.071081638336182,
+      "learning_rate": 0.00018068908348842818,
+      "loss": 3.5706,
+      "step": 21750
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.189342021942139,
+      "learning_rate": 0.00018058064422286525,
+      "loss": 3.667,
+      "step": 21800
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 7.787344455718994,
+      "learning_rate": 0.00018047193407388603,
+      "loss": 3.4985,
+      "step": 21850
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.411252975463867,
+      "learning_rate": 0.00018036295340693531,
+      "loss": 3.6719,
+      "step": 21900
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.700460433959961,
+      "learning_rate": 0.00018025370258836732,
+      "loss": 3.5075,
+      "step": 21950
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.121459007263184,
+      "learning_rate": 0.00018014418198544432,
+      "loss": 3.5511,
+      "step": 22000
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.29133415222168,
+      "learning_rate": 0.0001800343919663356,
+      "loss": 3.7063,
+      "step": 22050
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.53157377243042,
+      "learning_rate": 0.00017992433290011604,
+      "loss": 3.5146,
+      "step": 22100
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.442373275756836,
+      "learning_rate": 0.00017981400515676508,
+      "loss": 3.5431,
+      "step": 22150
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.247061729431152,
+      "learning_rate": 0.00017970340910716522,
+      "loss": 3.604,
+      "step": 22200
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.570899486541748,
+      "learning_rate": 0.000179592545123101,
+      "loss": 3.6034,
+      "step": 22250
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.012238502502441,
+      "learning_rate": 0.00017948141357725764,
+      "loss": 3.4793,
+      "step": 22300
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.5325422286987305,
+      "learning_rate": 0.0001793700148432198,
+      "loss": 3.563,
+      "step": 22350
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.889975070953369,
+      "learning_rate": 0.00017925834929547035,
+      "loss": 3.5512,
+      "step": 22400
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.225555896759033,
+      "learning_rate": 0.00017914641730938907,
+      "loss": 3.5521,
+      "step": 22450
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.430109024047852,
+      "learning_rate": 0.0001790342192612514,
+      "loss": 3.4549,
+      "step": 22500
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.8808274269104,
+      "learning_rate": 0.00017892175552822716,
+      "loss": 3.5518,
+      "step": 22550
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.657894134521484,
+      "learning_rate": 0.00017880902648837946,
+      "loss": 3.4643,
+      "step": 22600
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.968985557556152,
+      "learning_rate": 0.00017869603252066308,
+      "loss": 3.5022,
+      "step": 22650
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.369678497314453,
+      "learning_rate": 0.00017858277400492357,
+      "loss": 3.6906,
+      "step": 22700
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 6.433826446533203,
+      "learning_rate": 0.0001784692513218956,
+      "loss": 3.4281,
+      "step": 22750
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.364591121673584,
+      "learning_rate": 0.00017835546485320202,
+      "loss": 3.6194,
+      "step": 22800
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.889247894287109,
+      "learning_rate": 0.00017824141498135244,
+      "loss": 3.7013,
+      "step": 22850
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.254469394683838,
+      "learning_rate": 0.0001781271020897419,
+      "loss": 3.4107,
+      "step": 22900
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 5.486823558807373,
+      "learning_rate": 0.0001780125265626495,
+      "loss": 3.5453,
+      "step": 22950
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.4713311195373535,
+      "learning_rate": 0.0001778976887852375,
+      "loss": 3.5482,
+      "step": 23000
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.2519731521606445,
+      "learning_rate": 0.00017778258914354946,
+      "loss": 3.6251,
+      "step": 23050
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.657818794250488,
+      "learning_rate": 0.00017766722802450944,
+      "loss": 3.5081,
+      "step": 23100
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.176442623138428,
+      "learning_rate": 0.0001775516058159204,
+      "loss": 3.45,
+      "step": 23150
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.838647365570068,
+      "learning_rate": 0.00017743572290646303,
+      "loss": 3.419,
+      "step": 23200
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.912227630615234,
+      "learning_rate": 0.00017731957968569436,
+      "loss": 3.4892,
+      "step": 23250
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 7.276485919952393,
+      "learning_rate": 0.0001772031765440465,
+      "loss": 3.5143,
+      "step": 23300
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.351586818695068,
+      "learning_rate": 0.0001770865138728254,
+      "loss": 3.5467,
+      "step": 23350
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.220416069030762,
+      "learning_rate": 0.00017696959206420937,
+      "loss": 3.4736,
+      "step": 23400
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.282609462738037,
+      "learning_rate": 0.00017685241151124781,
+      "loss": 3.4181,
+      "step": 23450
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 4.958062171936035,
+      "learning_rate": 0.00017673497260786006,
+      "loss": 3.4309,
+      "step": 23500
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.3785929679870605,
+      "learning_rate": 0.00017661727574883388,
+      "loss": 3.3805,
+      "step": 23550
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.702798366546631,
+      "learning_rate": 0.00017649932132982415,
+      "loss": 3.5371,
+      "step": 23600
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.496365547180176,
+      "learning_rate": 0.0001763811097473516,
+      "loss": 3.4107,
+      "step": 23650
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.093421936035156,
+      "learning_rate": 0.00017626264139880148,
+      "loss": 3.5514,
+      "step": 23700
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.6509480476379395,
+      "learning_rate": 0.0001761439166824221,
+      "loss": 3.5612,
+      "step": 23750
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 5.662957191467285,
+      "learning_rate": 0.00017602493599732372,
+      "loss": 3.5515,
+      "step": 23800
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 7.548245429992676,
+      "learning_rate": 0.000175905699743477,
+      "loss": 3.5552,
+      "step": 23850
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.2797112464904785,
+      "learning_rate": 0.00017578620832171173,
+      "loss": 3.4159,
+      "step": 23900
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.431013584136963,
+      "learning_rate": 0.0001756664621337155,
+      "loss": 3.4257,
+      "step": 23950
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 8.88436222076416,
+      "learning_rate": 0.00017554646158203236,
+      "loss": 3.5517,
+      "step": 24000
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.538012981414795,
+      "learning_rate": 0.00017542620707006136,
+      "loss": 3.4451,
+      "step": 24050
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.702478408813477,
+      "learning_rate": 0.00017530569900205538,
+      "loss": 3.5453,
+      "step": 24100
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.236027240753174,
+      "learning_rate": 0.00017518493778311957,
+      "loss": 3.4483,
+      "step": 24150
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.824537754058838,
+      "learning_rate": 0.00017506392381921014,
+      "loss": 3.507,
+      "step": 24200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.682642459869385,
+      "learning_rate": 0.0001749426575171329,
+      "loss": 3.4624,
+      "step": 24250
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.02097225189209,
+      "learning_rate": 0.00017482113928454196,
+      "loss": 3.4782,
+      "step": 24300
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.951188087463379,
+      "learning_rate": 0.00017469936952993834,
+      "loss": 3.5305,
+      "step": 24350
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.47694730758667,
+      "learning_rate": 0.00017457734866266854,
+      "loss": 3.4653,
+      "step": 24400
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.073057651519775,
+      "learning_rate": 0.0001744575249785453,
+      "loss": 3.4969,
+      "step": 24450
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.532285690307617,
+      "learning_rate": 0.00017433500811915326,
+      "loss": 3.3932,
+      "step": 24500
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.016458988189697,
+      "learning_rate": 0.00017421224137194837,
+      "loss": 3.4828,
+      "step": 24550
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 7.032898902893066,
+      "learning_rate": 0.0001740892251496286,
+      "loss": 3.4347,
+      "step": 24600
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.3446431159973145,
+      "learning_rate": 0.00017396595986573065,
+      "loss": 3.4101,
+      "step": 24650
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.952356815338135,
+      "learning_rate": 0.00017384244593462859,
+      "loss": 3.4296,
+      "step": 24700
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 5.403810501098633,
+      "learning_rate": 0.00017371868377153216,
+      "loss": 3.4264,
+      "step": 24750
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.429996490478516,
+      "learning_rate": 0.00017359467379248568,
+      "loss": 3.4341,
+      "step": 24800
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.618744850158691,
+      "learning_rate": 0.00017347041641436653,
+      "loss": 3.3357,
+      "step": 24850
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.539459705352783,
+      "learning_rate": 0.00017334840455978504,
+      "loss": 3.5718,
+      "step": 24900
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.993662357330322,
+      "learning_rate": 0.00017322365856462736,
+      "loss": 3.4774,
+      "step": 24950
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.6996002197265625,
+      "learning_rate": 0.00017309866641761798,
+      "loss": 3.311,
+      "step": 25000
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.32814884185791,
+      "learning_rate": 0.00017297342853893604,
+      "loss": 3.5558,
+      "step": 25050
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.538712501525879,
+      "learning_rate": 0.0001728479453495866,
+      "loss": 3.3261,
+      "step": 25100
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.975490093231201,
+      "learning_rate": 0.00017272221727139946,
+      "loss": 3.5,
+      "step": 25150
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.377697467803955,
+      "learning_rate": 0.00017259624472702764,
+      "loss": 3.4562,
+      "step": 25200
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.686251163482666,
+      "learning_rate": 0.00017247002813994592,
+      "loss": 3.3968,
+      "step": 25250
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.116434097290039,
+      "learning_rate": 0.00017234356793444954,
+      "loss": 3.3161,
+      "step": 25300
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.755252838134766,
+      "learning_rate": 0.0001722168645356526,
+      "loss": 3.4195,
+      "step": 25350
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.647252559661865,
+      "learning_rate": 0.00017208991836948685,
+      "loss": 3.1887,
+      "step": 25400
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.997719764709473,
+      "learning_rate": 0.0001719627298627,
+      "loss": 3.4098,
+      "step": 25450
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.054971218109131,
+      "learning_rate": 0.00017183529944285456,
+      "loss": 3.4159,
+      "step": 25500
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.706241130828857,
+      "learning_rate": 0.00017170762753832615,
+      "loss": 3.4024,
+      "step": 25550
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 7.701054096221924,
+      "learning_rate": 0.00017157971457830226,
+      "loss": 3.3564,
+      "step": 25600
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.441225528717041,
+      "learning_rate": 0.00017145156099278067,
+      "loss": 3.5887,
+      "step": 25650
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.631026268005371,
+      "learning_rate": 0.0001713231672125681,
+      "loss": 3.352,
+      "step": 25700
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.974308967590332,
+      "learning_rate": 0.0001711945336692786,
+      "loss": 3.3959,
+      "step": 25750
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.063317775726318,
+      "learning_rate": 0.00017106566079533246,
+      "loss": 3.3942,
+      "step": 25800
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.312389850616455,
+      "learning_rate": 0.0001709365490239543,
+      "loss": 3.3928,
+      "step": 25850
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.022332668304443,
+      "learning_rate": 0.00017080719878917182,
+      "loss": 3.4401,
+      "step": 25900
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.356366157531738,
+      "learning_rate": 0.00017067761052581455,
+      "loss": 3.4353,
+      "step": 25950
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.611413478851318,
+      "learning_rate": 0.00017054778466951196,
+      "loss": 3.2737,
+      "step": 26000
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.244396686553955,
+      "learning_rate": 0.0001704177216566924,
+      "loss": 3.2309,
+      "step": 26050
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.024662017822266,
+      "learning_rate": 0.00017028742192458132,
+      "loss": 3.3593,
+      "step": 26100
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.399158954620361,
+      "learning_rate": 0.00017015688591120006,
+      "loss": 3.2026,
+      "step": 26150
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 7.96980619430542,
+      "learning_rate": 0.00017002611405536413,
+      "loss": 3.4413,
+      "step": 26200
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.538659572601318,
+      "learning_rate": 0.00016989510679668194,
+      "loss": 3.3497,
+      "step": 26250
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.8960394859313965,
+      "learning_rate": 0.00016976386457555323,
+      "loss": 3.3708,
+      "step": 26300
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.486491680145264,
+      "learning_rate": 0.00016963238783316754,
+      "loss": 3.4697,
+      "step": 26350
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.217641353607178,
+      "learning_rate": 0.0001695006770115029,
+      "loss": 3.4249,
+      "step": 26400
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.6906938552856445,
+      "learning_rate": 0.00016936873255332413,
+      "loss": 3.5343,
+      "step": 26450
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 6.275619983673096,
+      "learning_rate": 0.00016923655490218149,
+      "loss": 3.3991,
+      "step": 26500
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.75913667678833,
+      "learning_rate": 0.00016910414450240917,
+      "loss": 3.3861,
+      "step": 26550
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 5.206583499908447,
+      "learning_rate": 0.0001689715017991237,
+      "loss": 3.369,
+      "step": 26600
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.569302082061768,
+      "learning_rate": 0.0001688386272382227,
+      "loss": 3.4837,
+      "step": 26650
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.360637187957764,
+      "learning_rate": 0.00016870552126638298,
+      "loss": 3.3299,
+      "step": 26700
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.168808937072754,
+      "learning_rate": 0.00016857218433105945,
+      "loss": 3.3613,
+      "step": 26750
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.515918731689453,
+      "learning_rate": 0.0001684386168804834,
+      "loss": 3.159,
+      "step": 26800
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.885009765625,
+      "learning_rate": 0.000168304819363661,
+      "loss": 3.3029,
+      "step": 26850
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.174962997436523,
+      "learning_rate": 0.0001681707922303718,
+      "loss": 3.3289,
+      "step": 26900
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.86044454574585,
+      "learning_rate": 0.0001680365359311673,
+      "loss": 3.3132,
+      "step": 26950
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.267008304595947,
+      "learning_rate": 0.00016790205091736935,
+      "loss": 3.3649,
+      "step": 27000
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.221423625946045,
+      "learning_rate": 0.00016776733764106862,
+      "loss": 3.3311,
+      "step": 27050
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.069894790649414,
+      "learning_rate": 0.00016763239655512318,
+      "loss": 3.3157,
+      "step": 27100
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 8.425812721252441,
+      "learning_rate": 0.00016749722811315688,
+      "loss": 3.2714,
+      "step": 27150
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.80504846572876,
+      "learning_rate": 0.00016736183276955783,
+      "loss": 3.3274,
+      "step": 27200
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 7.617208003997803,
+      "learning_rate": 0.00016722621097947697,
+      "loss": 3.1857,
+      "step": 27250
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.363246440887451,
+      "learning_rate": 0.00016709036319882646,
+      "loss": 3.4673,
+      "step": 27300
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.790156364440918,
+      "learning_rate": 0.00016695428988427807,
+      "loss": 3.3016,
+      "step": 27350
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.9824981689453125,
+      "learning_rate": 0.00016681799149326185,
+      "loss": 3.4103,
+      "step": 27400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.915642738342285,
+      "learning_rate": 0.00016668146848396442,
+      "loss": 3.4356,
+      "step": 27450
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.938210487365723,
+      "learning_rate": 0.0001665447213153275,
+      "loss": 3.299,
+      "step": 27500
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.116371154785156,
+      "learning_rate": 0.00016640775044704634,
+      "loss": 3.3231,
+      "step": 27550
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.796716213226318,
+      "learning_rate": 0.0001662705563395682,
+      "loss": 3.3685,
+      "step": 27600
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 9.196764945983887,
+      "learning_rate": 0.0001661331394540908,
+      "loss": 3.2807,
+      "step": 27650
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.64096736907959,
+      "learning_rate": 0.00016599550025256076,
+      "loss": 3.2909,
+      "step": 27700
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.615271091461182,
+      "learning_rate": 0.000165857639197672,
+      "loss": 3.2044,
+      "step": 27750
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.112679481506348,
+      "learning_rate": 0.0001657195567528643,
+      "loss": 3.2377,
+      "step": 27800
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.882411003112793,
+      "learning_rate": 0.0001655812533823216,
+      "loss": 3.4462,
+      "step": 27850
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.716900825500488,
+      "learning_rate": 0.00016544272955097063,
+      "loss": 3.3563,
+      "step": 27900
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.415688991546631,
+      "learning_rate": 0.0001653039857244791,
+      "loss": 3.2475,
+      "step": 27950
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.752101421356201,
+      "learning_rate": 0.00016516502236925434,
+      "loss": 3.3646,
+      "step": 28000
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 7.952321529388428,
+      "learning_rate": 0.00016502583995244163,
+      "loss": 3.2835,
+      "step": 28050
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.233190536499023,
+      "learning_rate": 0.00016488643894192268,
+      "loss": 3.3653,
+      "step": 28100
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.358859062194824,
+      "learning_rate": 0.00016474681980631402,
+      "loss": 3.2425,
+      "step": 28150
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.584342002868652,
+      "learning_rate": 0.0001646069830149654,
+      "loss": 3.3139,
+      "step": 28200
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.155908584594727,
+      "learning_rate": 0.00016446692903795837,
+      "loss": 3.2732,
+      "step": 28250
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.454020023345947,
+      "learning_rate": 0.00016432665834610445,
+      "loss": 3.2121,
+      "step": 28300
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.337937831878662,
+      "learning_rate": 0.00016418617141094374,
+      "loss": 3.3123,
+      "step": 28350
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.727194786071777,
+      "learning_rate": 0.00016404546870474324,
+      "loss": 3.2558,
+      "step": 28400
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.887204647064209,
+      "learning_rate": 0.00016390455070049536,
+      "loss": 3.377,
+      "step": 28450
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.2247724533081055,
+      "learning_rate": 0.0001637634178719162,
+      "loss": 3.1277,
+      "step": 28500
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.014094829559326,
+      "learning_rate": 0.00016362207069344403,
+      "loss": 3.142,
+      "step": 28550
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 7.58336067199707,
+      "learning_rate": 0.00016348050964023773,
+      "loss": 3.3156,
+      "step": 28600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.806646823883057,
+      "learning_rate": 0.00016333873518817514,
+      "loss": 3.2814,
+      "step": 28650
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.387387275695801,
+      "learning_rate": 0.00016319674781385143,
+      "loss": 3.3977,
+      "step": 28700
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.4912638664245605,
+      "learning_rate": 0.00016305454799457755,
+      "loss": 3.2323,
+      "step": 28750
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.692640781402588,
+      "learning_rate": 0.00016291213620837867,
+      "loss": 3.3033,
+      "step": 28800
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.484092712402344,
+      "learning_rate": 0.0001627695129339924,
+      "loss": 3.1466,
+      "step": 28850
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.305532932281494,
+      "learning_rate": 0.00016262667865086746,
+      "loss": 3.4111,
+      "step": 28900
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.511746883392334,
+      "learning_rate": 0.00016248363383916182,
+      "loss": 3.2535,
+      "step": 28950
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 7.3723649978637695,
+      "learning_rate": 0.00016234037897974108,
+      "loss": 3.3265,
+      "step": 29000
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.085361003875732,
+      "learning_rate": 0.0001621997858933184,
+      "loss": 3.3003,
+      "step": 29050
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 4.863938331604004,
+      "learning_rate": 0.000162056116560834,
+      "loss": 3.2411,
+      "step": 29100
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.351004123687744,
+      "learning_rate": 0.00016191223861779529,
+      "loss": 3.2409,
+      "step": 29150
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 8.316920280456543,
+      "learning_rate": 0.0001617681525478687,
+      "loss": 3.0881,
+      "step": 29200
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 7.093586444854736,
+      "learning_rate": 0.0001616238588354203,
+      "loss": 3.2573,
+      "step": 29250
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.5853705406188965,
+      "learning_rate": 0.00016147935796551405,
+      "loss": 3.3215,
+      "step": 29300
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.667483329772949,
+      "learning_rate": 0.00016133465042391046,
+      "loss": 3.3032,
+      "step": 29350
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.287956237792969,
+      "learning_rate": 0.00016118973669706468,
+      "loss": 3.2255,
+      "step": 29400
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.676483631134033,
+      "learning_rate": 0.0001610446172721251,
+      "loss": 3.3663,
+      "step": 29450
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.739033222198486,
+      "learning_rate": 0.00016089929263693144,
+      "loss": 3.2537,
+      "step": 29500
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.158905982971191,
+      "learning_rate": 0.00016075376328001344,
+      "loss": 3.2336,
+      "step": 29550
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.2512526512146,
+      "learning_rate": 0.00016060802969058885,
+      "loss": 3.2982,
+      "step": 29600
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 8.520125389099121,
+      "learning_rate": 0.00016046209235856212,
+      "loss": 3.3153,
+      "step": 29650
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.461225509643555,
+      "learning_rate": 0.00016031595177452257,
+      "loss": 3.2629,
+      "step": 29700
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.117869853973389,
+      "learning_rate": 0.00016016960842974278,
+      "loss": 3.2225,
+      "step": 29750
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.035212993621826,
+      "learning_rate": 0.00016002306281617692,
+      "loss": 3.3866,
+      "step": 29800
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.271117687225342,
+      "learning_rate": 0.00015987631542645913,
+      "loss": 3.2602,
+      "step": 29850
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.860154151916504,
+      "learning_rate": 0.00015972936675390185,
+      "loss": 3.288,
+      "step": 29900
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.154600143432617,
+      "learning_rate": 0.0001595822172924942,
+      "loss": 3.1941,
+      "step": 29950
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.897374153137207,
+      "learning_rate": 0.00015943486753690017,
+      "loss": 3.2323,
+      "step": 30000
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 6.063130855560303,
+      "learning_rate": 0.00015928731798245721,
+      "loss": 3.1718,
+      "step": 30050
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.736262321472168,
+      "learning_rate": 0.00015913956912517432,
+      "loss": 3.3035,
+      "step": 30100
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.386317253112793,
+      "learning_rate": 0.00015899162146173053,
+      "loss": 3.2879,
+      "step": 30150
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.535543441772461,
+      "learning_rate": 0.00015884347548947314,
+      "loss": 3.2266,
+      "step": 30200
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.057496070861816,
+      "learning_rate": 0.00015869513170641616,
+      "loss": 3.1668,
+      "step": 30250
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.6912055015563965,
+      "learning_rate": 0.00015854659061123854,
+      "loss": 3.1562,
+      "step": 30300
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.563050270080566,
+      "learning_rate": 0.0001583978527032825,
+      "loss": 3.1819,
+      "step": 30350
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.75504732131958,
+      "learning_rate": 0.0001582489184825519,
+      "loss": 3.1891,
+      "step": 30400
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.390425205230713,
+      "learning_rate": 0.00015809978844971053,
+      "loss": 3.1856,
+      "step": 30450
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.183398246765137,
+      "learning_rate": 0.0001579504631060804,
+      "loss": 3.3115,
+      "step": 30500
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.40380859375,
+      "learning_rate": 0.00015780094295364015,
+      "loss": 3.162,
+      "step": 30550
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.078804016113281,
+      "learning_rate": 0.00015765122849502325,
+      "loss": 3.2046,
+      "step": 30600
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.183681964874268,
+      "learning_rate": 0.00015750132023351638,
+      "loss": 3.0689,
+      "step": 30650
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 12.730826377868652,
+      "learning_rate": 0.00015735121867305768,
+      "loss": 3.2468,
+      "step": 30700
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.127053260803223,
+      "learning_rate": 0.00015720092431823515,
+      "loss": 3.1628,
+      "step": 30750
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.8310089111328125,
+      "learning_rate": 0.00015705043767428483,
+      "loss": 3.2047,
+      "step": 30800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 7.505776882171631,
+      "learning_rate": 0.0001568997592470892,
+      "loss": 3.2827,
+      "step": 30850
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.399072170257568,
+      "learning_rate": 0.00015674888954317549,
+      "loss": 3.1483,
+      "step": 30900
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.224669456481934,
+      "learning_rate": 0.00015659782906971383,
+      "loss": 3.2698,
+      "step": 30950
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 7.958742618560791,
+      "learning_rate": 0.00015644657833451577,
+      "loss": 3.0145,
+      "step": 31000
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.20373010635376,
+      "learning_rate": 0.0001562981685120925,
+      "loss": 3.1598,
+      "step": 31050
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 5.2080159187316895,
+      "learning_rate": 0.00015614654255930347,
+      "loss": 3.2801,
+      "step": 31100
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.195250988006592,
+      "learning_rate": 0.00015599472786184245,
+      "loss": 3.156,
+      "step": 31150
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.0389204025268555,
+      "learning_rate": 0.00015584272493005642,
+      "loss": 3.1345,
+      "step": 31200
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 7.003210544586182,
+      "learning_rate": 0.00015569053427492505,
+      "loss": 3.2186,
+      "step": 31250
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.47674036026001,
+      "learning_rate": 0.00015553815640805907,
+      "loss": 3.3211,
+      "step": 31300
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.2981648445129395,
+      "learning_rate": 0.00015538559184169863,
+      "loss": 3.2454,
+      "step": 31350
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.528575420379639,
+      "learning_rate": 0.00015523284108871142,
+      "loss": 3.1963,
+      "step": 31400
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.553009033203125,
+      "learning_rate": 0.0001550799046625911,
+      "loss": 3.1682,
+      "step": 31450
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.608404636383057,
+      "learning_rate": 0.0001549267830774553,
+      "loss": 3.1461,
+      "step": 31500
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.462625503540039,
+      "learning_rate": 0.00015477347684804445,
+      "loss": 3.2772,
+      "step": 31550
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.327962875366211,
+      "learning_rate": 0.00015461998648971928,
+      "loss": 3.2144,
+      "step": 31600
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.94124174118042,
+      "learning_rate": 0.00015446631251845978,
+      "loss": 3.2227,
+      "step": 31650
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.351782321929932,
+      "learning_rate": 0.00015431245545086307,
+      "loss": 3.2687,
+      "step": 31700
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.562844276428223,
+      "learning_rate": 0.00015415841580414185,
+      "loss": 3.1332,
+      "step": 31750
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.023700714111328,
+      "learning_rate": 0.00015400419409612243,
+      "loss": 3.2272,
+      "step": 31800
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.127398490905762,
+      "learning_rate": 0.0001538497908452433,
+      "loss": 3.2843,
+      "step": 31850
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.577905178070068,
+      "learning_rate": 0.0001536952065705532,
+      "loss": 3.2635,
+      "step": 31900
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 6.119299411773682,
+      "learning_rate": 0.00015354044179170933,
+      "loss": 3.126,
+      "step": 31950
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.827983856201172,
+      "learning_rate": 0.0001533854970289758,
+      "loss": 3.2345,
+      "step": 32000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.499656677246094,
+      "learning_rate": 0.00015323037280322166,
+      "loss": 3.0808,
+      "step": 32050
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.260239601135254,
+      "learning_rate": 0.00015307506963591923,
+      "loss": 3.1234,
+      "step": 32100
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.486075401306152,
+      "learning_rate": 0.00015291958804914256,
+      "loss": 3.1769,
+      "step": 32150
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.572110176086426,
+      "learning_rate": 0.00015276392856556527,
+      "loss": 3.2166,
+      "step": 32200
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.288125991821289,
+      "learning_rate": 0.0001526080917084591,
+      "loss": 3.0781,
+      "step": 32250
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.380829334259033,
+      "learning_rate": 0.000152452078001692,
+      "loss": 3.1178,
+      "step": 32300
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.039462566375732,
+      "learning_rate": 0.00015229588796972652,
+      "loss": 3.2808,
+      "step": 32350
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 7.323626518249512,
+      "learning_rate": 0.00015213952213761787,
+      "loss": 3.1391,
+      "step": 32400
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.912395000457764,
+      "learning_rate": 0.00015198298103101228,
+      "loss": 3.1744,
+      "step": 32450
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.694441795349121,
+      "learning_rate": 0.00015182626517614518,
+      "loss": 3.0576,
+      "step": 32500
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.467188358306885,
+      "learning_rate": 0.00015166937509983943,
+      "loss": 3.2361,
+      "step": 32550
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.292226791381836,
+      "learning_rate": 0.00015151231132950357,
+      "loss": 3.1376,
+      "step": 32600
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.369929313659668,
+      "learning_rate": 0.00015135507439313005,
+      "loss": 3.1406,
+      "step": 32650
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.252573013305664,
+      "learning_rate": 0.00015119766481929342,
+      "loss": 3.123,
+      "step": 32700
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.9053730964660645,
+      "learning_rate": 0.00015104008313714858,
+      "loss": 3.0018,
+      "step": 32750
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 6.080839157104492,
+      "learning_rate": 0.00015088232987642898,
+      "loss": 3.2106,
+      "step": 32800
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.92653226852417,
+      "learning_rate": 0.00015072440556744492,
+      "loss": 3.2095,
+      "step": 32850
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 5.627429008483887,
+      "learning_rate": 0.00015056631074108166,
+      "loss": 3.089,
+      "step": 32900
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.694194793701172,
+      "learning_rate": 0.00015040804592879762,
+      "loss": 3.0885,
+      "step": 32950
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.04107666015625,
+      "learning_rate": 0.00015024961166262276,
+      "loss": 3.0906,
+      "step": 33000
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.280002117156982,
+      "learning_rate": 0.0001500910084751567,
+      "loss": 3.2142,
+      "step": 33050
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.803068161010742,
+      "learning_rate": 0.00014993223689956672,
+      "loss": 3.2014,
+      "step": 33100
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.611780643463135,
+      "learning_rate": 0.00014977329746958636,
+      "loss": 3.2491,
+      "step": 33150
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.6669020652771,
+      "learning_rate": 0.00014961737449079314,
+      "loss": 3.2,
+      "step": 33200
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.894138336181641,
+      "learning_rate": 0.00014945810428594703,
+      "loss": 3.0321,
+      "step": 33250
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.354518413543701,
+      "learning_rate": 0.0001492986678205755,
+      "loss": 3.1314,
+      "step": 33300
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.786489009857178,
+      "learning_rate": 0.00014913906563064706,
+      "loss": 3.1937,
+      "step": 33350
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 8.759417533874512,
+      "learning_rate": 0.00014897929825268745,
+      "loss": 3.1069,
+      "step": 33400
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.354910850524902,
+      "learning_rate": 0.00014881936622377766,
+      "loss": 3.1519,
+      "step": 33450
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.169478416442871,
+      "learning_rate": 0.0001486592700815522,
+      "loss": 3.1414,
+      "step": 33500
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.09418249130249,
+      "learning_rate": 0.00014849901036419723,
+      "loss": 3.0954,
+      "step": 33550
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.986037731170654,
+      "learning_rate": 0.00014833858761044883,
+      "loss": 3.2445,
+      "step": 33600
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 12.803654670715332,
+      "learning_rate": 0.00014817800235959118,
+      "loss": 3.0699,
+      "step": 33650
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.191990852355957,
+      "learning_rate": 0.00014801725515145467,
+      "loss": 3.2574,
+      "step": 33700
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.233778476715088,
+      "learning_rate": 0.00014785634652641412,
+      "loss": 3.1152,
+      "step": 33750
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.065474987030029,
+      "learning_rate": 0.000147695277025387,
+      "loss": 3.1178,
+      "step": 33800
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.4664435386657715,
+      "learning_rate": 0.00014753404718983158,
+      "loss": 3.0627,
+      "step": 33850
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.926270484924316,
+      "learning_rate": 0.00014737265756174515,
+      "loss": 3.0182,
+      "step": 33900
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.013931751251221,
+      "learning_rate": 0.0001472111086836621,
+      "loss": 3.0801,
+      "step": 33950
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.25607442855835,
+      "learning_rate": 0.00014704940109865224,
+      "loss": 3.1227,
+      "step": 34000
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.084470272064209,
+      "learning_rate": 0.00014688753535031882,
+      "loss": 3.1786,
+      "step": 34050
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.16444730758667,
+      "learning_rate": 0.00014672551198279687,
+      "loss": 3.0651,
+      "step": 34100
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.316379070281982,
+      "learning_rate": 0.00014656333154075118,
+      "loss": 2.9925,
+      "step": 34150
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.331335067749023,
+      "learning_rate": 0.00014640099456937462,
+      "loss": 3.097,
+      "step": 34200
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.452579498291016,
+      "learning_rate": 0.00014623850161438626,
+      "loss": 3.1222,
+      "step": 34250
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.550851345062256,
+      "learning_rate": 0.00014607585322202953,
+      "loss": 3.2343,
+      "step": 34300
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.5701375007629395,
+      "learning_rate": 0.00014591304993907033,
+      "loss": 3.1558,
+      "step": 34350
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.140341758728027,
+      "learning_rate": 0.00014575009231279534,
+      "loss": 3.0036,
+      "step": 34400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.767757415771484,
+      "learning_rate": 0.00014558698089101003,
+      "loss": 3.1355,
+      "step": 34450
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.189707279205322,
+      "learning_rate": 0.00014542371622203689,
+      "loss": 3.0721,
+      "step": 34500
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.492117881774902,
+      "learning_rate": 0.00014526029885471355,
+      "loss": 3.1083,
+      "step": 34550
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.021632671356201,
+      "learning_rate": 0.000145096729338391,
+      "loss": 3.1291,
+      "step": 34600
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.501243591308594,
+      "learning_rate": 0.00014493300822293164,
+      "loss": 3.0654,
+      "step": 34650
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.056375026702881,
+      "learning_rate": 0.0001447691360587076,
+      "loss": 3.054,
+      "step": 34700
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.866014003753662,
+      "learning_rate": 0.0001446051133965986,
+      "loss": 3.1111,
+      "step": 34750
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.377662658691406,
+      "learning_rate": 0.0001444409407879905,
+      "loss": 3.2083,
+      "step": 34800
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.692511558532715,
+      "learning_rate": 0.00014427661878477305,
+      "loss": 3.1613,
+      "step": 34850
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.763862609863281,
+      "learning_rate": 0.0001441121479393383,
+      "loss": 3.1422,
+      "step": 34900
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.247575283050537,
+      "learning_rate": 0.00014394752880457867,
+      "loss": 3.1826,
+      "step": 34950
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.18945837020874,
+      "learning_rate": 0.00014378276193388498,
+      "loss": 3.0887,
+      "step": 35000
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 90183,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.258566844416e+16,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}