diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9492 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.995008319467554,
+  "eval_steps": 500,
+  "global_step": 1350,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0022185246810870773,
+      "grad_norm": 2.825010061264038,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 10.6427,
+      "step": 1
+    },
+    {
+      "epoch": 0.004437049362174155,
+      "grad_norm": 2.884904623031616,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 10.6201,
+      "step": 2
+    },
+    {
+      "epoch": 0.0066555740432612314,
+      "grad_norm": 2.871875524520874,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 10.8058,
+      "step": 3
+    },
+    {
+      "epoch": 0.00887409872434831,
+      "grad_norm": 2.8786909580230713,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 10.5269,
+      "step": 4
+    },
+    {
+      "epoch": 0.011092623405435386,
+      "grad_norm": 2.889479398727417,
+      "learning_rate": 1.111111111111111e-05,
+      "loss": 10.6576,
+      "step": 5
+    },
+    {
+      "epoch": 0.013311148086522463,
+      "grad_norm": 3.0641160011291504,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 10.5792,
+      "step": 6
+    },
+    {
+      "epoch": 0.01552967276760954,
+      "grad_norm": 3.1233482360839844,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 10.6755,
+      "step": 7
+    },
+    {
+      "epoch": 0.01774819744869662,
+      "grad_norm": 3.1553714275360107,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 10.6242,
+      "step": 8
+    },
+    {
+      "epoch": 0.019966722129783693,
+      "grad_norm": 3.5153255462646484,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 10.4593,
+      "step": 9
+    },
+    {
+      "epoch": 0.022185246810870772,
+      "grad_norm": 3.473512649536133,
+      "learning_rate": 2.222222222222222e-05,
+      "loss": 10.3572,
+      "step": 10
+    },
+    {
+      "epoch": 0.024403771491957847,
+      "grad_norm": 4.096827507019043,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 10.5713,
+      "step": 11
+    },
+    {
+      "epoch": 0.026622296173044926,
+      "grad_norm": 4.09431791305542,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 10.1232,
+      "step": 12
+    },
+    {
+      "epoch": 0.028840820854132,
+      "grad_norm": 4.471763610839844,
+      "learning_rate": 2.8888888888888885e-05,
+      "loss": 10.2053,
+      "step": 13
+    },
+    {
+      "epoch": 0.03105934553521908,
+      "grad_norm": 4.495110511779785,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 9.9826,
+      "step": 14
+    },
+    {
+      "epoch": 0.033277870216306155,
+      "grad_norm": 4.857054710388184,
+      "learning_rate": 3.333333333333333e-05,
+      "loss": 10.1566,
+      "step": 15
+    },
+    {
+      "epoch": 0.03549639489739324,
+      "grad_norm": 4.986198425292969,
+      "learning_rate": 3.555555555555555e-05,
+      "loss": 9.9618,
+      "step": 16
+    },
+    {
+      "epoch": 0.03771491957848031,
+      "grad_norm": 5.414590835571289,
+      "learning_rate": 3.777777777777777e-05,
+      "loss": 9.7535,
+      "step": 17
+    },
+    {
+      "epoch": 0.03993344425956739,
+      "grad_norm": 5.218145847320557,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 9.4886,
+      "step": 18
+    },
+    {
+      "epoch": 0.04215196894065446,
+      "grad_norm": 5.161298751831055,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 9.196,
+      "step": 19
+    },
+    {
+      "epoch": 0.044370493621741544,
+      "grad_norm": 5.5800371170043945,
+      "learning_rate": 4.444444444444444e-05,
+      "loss": 9.0649,
+      "step": 20
+    },
+    {
+      "epoch": 0.04658901830282862,
+      "grad_norm": 5.475617408752441,
+      "learning_rate": 4.6666666666666665e-05,
+      "loss": 8.7236,
+      "step": 21
+    },
+    {
+      "epoch": 0.048807542983915694,
+      "grad_norm": 5.349752902984619,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 8.5801,
+      "step": 22
+    },
+    {
+      "epoch": 0.051026067665002776,
+      "grad_norm": 5.557284832000732,
+      "learning_rate": 5.111111111111111e-05,
+      "loss": 8.4075,
+      "step": 23
+    },
+    {
+      "epoch": 0.05324459234608985,
+      "grad_norm": 5.48941707611084,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 8.2868,
+      "step": 24
+    },
+    {
+      "epoch": 0.05546311702717693,
+      "grad_norm": 6.343184947967529,
+      "learning_rate": 5.5555555555555545e-05,
+      "loss": 7.8334,
+      "step": 25
+    },
+    {
+      "epoch": 0.057681641708264,
+      "grad_norm": 5.931063175201416,
+      "learning_rate": 5.777777777777777e-05,
+      "loss": 7.8016,
+      "step": 26
+    },
+    {
+      "epoch": 0.059900166389351084,
+      "grad_norm": 4.8105292320251465,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 7.7299,
+      "step": 27
+    },
+    {
+      "epoch": 0.06211869107043816,
+      "grad_norm": 5.486648082733154,
+      "learning_rate": 6.222222222222222e-05,
+      "loss": 7.5947,
+      "step": 28
+    },
+    {
+      "epoch": 0.06433721575152523,
+      "grad_norm": 6.336665153503418,
+      "learning_rate": 6.444444444444444e-05,
+      "loss": 7.8308,
+      "step": 29
+    },
+    {
+      "epoch": 0.06655574043261231,
+      "grad_norm": 7.583590984344482,
+      "learning_rate": 6.666666666666666e-05,
+      "loss": 7.6412,
+      "step": 30
+    },
+    {
+      "epoch": 0.06877426511369938,
+      "grad_norm": 7.815066814422607,
+      "learning_rate": 6.888888888888888e-05,
+      "loss": 7.5693,
+      "step": 31
+    },
+    {
+      "epoch": 0.07099278979478647,
+      "grad_norm": 6.805440425872803,
+      "learning_rate": 7.11111111111111e-05,
+      "loss": 7.3542,
+      "step": 32
+    },
+    {
+      "epoch": 0.07321131447587355,
+      "grad_norm": 5.853596210479736,
+      "learning_rate": 7.333333333333332e-05,
+      "loss": 7.2887,
+      "step": 33
+    },
+    {
+      "epoch": 0.07542983915696062,
+      "grad_norm": 4.700406551361084,
+      "learning_rate": 7.555555555555554e-05,
+      "loss": 7.2223,
+      "step": 34
+    },
+    {
+      "epoch": 0.0776483638380477,
+      "grad_norm": 4.206155300140381,
+      "learning_rate": 7.777777777777777e-05,
+      "loss": 6.6352,
+      "step": 35
+    },
+    {
+      "epoch": 0.07986688851913477,
+      "grad_norm": 3.832472801208496,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 7.0623,
+      "step": 36
+    },
+    {
+      "epoch": 0.08208541320022185,
+      "grad_norm": 3.900130033493042,
+      "learning_rate": 8.222222222222222e-05,
+      "loss": 7.1975,
+      "step": 37
+    },
+    {
+      "epoch": 0.08430393788130892,
+      "grad_norm": 3.78946590423584,
+      "learning_rate": 8.444444444444444e-05,
+      "loss": 6.8646,
+      "step": 38
+    },
+    {
+      "epoch": 0.08652246256239601,
+      "grad_norm": 3.793187141418457,
+      "learning_rate": 8.666666666666665e-05,
+      "loss": 6.523,
+      "step": 39
+    },
+    {
+      "epoch": 0.08874098724348309,
+      "grad_norm": 2.9125092029571533,
+      "learning_rate": 8.888888888888888e-05,
+      "loss": 6.649,
+      "step": 40
+    },
+    {
+      "epoch": 0.09095951192457016,
+      "grad_norm": 3.127119302749634,
+      "learning_rate": 9.11111111111111e-05,
+      "loss": 7.1303,
+      "step": 41
+    },
+    {
+      "epoch": 0.09317803660565724,
+      "grad_norm": 2.990851402282715,
+      "learning_rate": 9.333333333333333e-05,
+      "loss": 6.5099,
+      "step": 42
+    },
+    {
+      "epoch": 0.09539656128674431,
+      "grad_norm": 2.830379009246826,
+      "learning_rate": 9.555555555555555e-05,
+      "loss": 6.5687,
+      "step": 43
+    },
+    {
+      "epoch": 0.09761508596783139,
+      "grad_norm": 2.708688974380493,
+      "learning_rate": 9.777777777777778e-05,
+      "loss": 6.5674,
+      "step": 44
+    },
+    {
+      "epoch": 0.09983361064891846,
+      "grad_norm": 2.6215450763702393,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 6.438,
+      "step": 45
+    },
+    {
+      "epoch": 0.10205213533000555,
+      "grad_norm": 2.594029426574707,
+      "learning_rate": 0.00010222222222222222,
+      "loss": 6.6275,
+      "step": 46
+    },
+    {
+      "epoch": 0.10427066001109263,
+      "grad_norm": 2.47825288772583,
+      "learning_rate": 0.00010444444444444444,
+      "loss": 6.4902,
+      "step": 47
+    },
+    {
+      "epoch": 0.1064891846921797,
+      "grad_norm": 1.966611385345459,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 6.3346,
+      "step": 48
+    },
+    {
+      "epoch": 0.10870770937326678,
+      "grad_norm": 2.5752298831939697,
+      "learning_rate": 0.00010888888888888889,
+      "loss": 6.6163,
+      "step": 49
+    },
+    {
+      "epoch": 0.11092623405435385,
+      "grad_norm": 2.0445034503936768,
+      "learning_rate": 0.00011111111111111109,
+      "loss": 6.393,
+      "step": 50
+    },
+    {
+      "epoch": 0.11314475873544093,
+      "grad_norm": 2.118267774581909,
+      "learning_rate": 0.00011333333333333331,
+      "loss": 6.3897,
+      "step": 51
+    },
+    {
+      "epoch": 0.115363283416528,
+      "grad_norm": 2.015080690383911,
+      "learning_rate": 0.00011555555555555554,
+      "loss": 6.3746,
+      "step": 52
+    },
+    {
+      "epoch": 0.11758180809761509,
+      "grad_norm": 1.911922574043274,
+      "learning_rate": 0.00011777777777777776,
+      "loss": 6.4835,
+      "step": 53
+    },
+    {
+      "epoch": 0.11980033277870217,
+      "grad_norm": 1.7669731378555298,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 6.4747,
+      "step": 54
+    },
+    {
+      "epoch": 0.12201885745978924,
+      "grad_norm": 1.820046305656433,
+      "learning_rate": 0.00012222222222222221,
+      "loss": 6.2076,
+      "step": 55
+    },
+    {
+      "epoch": 0.12423738214087632,
+      "grad_norm": 1.7486004829406738,
+      "learning_rate": 0.00012444444444444444,
+      "loss": 6.2456,
+      "step": 56
+    },
+    {
+      "epoch": 0.1264559068219634,
+      "grad_norm": 1.7385284900665283,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 6.2271,
+      "step": 57
+    },
+    {
+      "epoch": 0.12867443150305047,
+      "grad_norm": 1.921257495880127,
+      "learning_rate": 0.0001288888888888889,
+      "loss": 6.2229,
+      "step": 58
+    },
+    {
+      "epoch": 0.13089295618413754,
+      "grad_norm": 2.360297203063965,
+      "learning_rate": 0.00013111111111111111,
+      "loss": 6.1729,
+      "step": 59
+    },
+    {
+      "epoch": 0.13311148086522462,
+      "grad_norm": 1.6387193202972412,
+      "learning_rate": 0.0001333333333333333,
+      "loss": 5.9551,
+      "step": 60
+    },
+    {
+      "epoch": 0.1353300055463117,
+      "grad_norm": 1.6953433752059937,
+      "learning_rate": 0.00013555555555555554,
+      "loss": 6.2185,
+      "step": 61
+    },
+    {
+      "epoch": 0.13754853022739877,
+      "grad_norm": 1.3946057558059692,
+      "learning_rate": 0.00013777777777777776,
+      "loss": 6.249,
+      "step": 62
+    },
+    {
+      "epoch": 0.13976705490848584,
+      "grad_norm": 1.7532439231872559,
+      "learning_rate": 0.00014,
+      "loss": 6.2481,
+      "step": 63
+    },
+    {
+      "epoch": 0.14198557958957295,
+      "grad_norm": 2.2568533420562744,
+      "learning_rate": 0.0001422222222222222,
+      "loss": 6.0979,
+      "step": 64
+    },
+    {
+      "epoch": 0.14420410427066002,
+      "grad_norm": 3.26971697807312,
+      "learning_rate": 0.0001444444444444444,
+      "loss": 6.2205,
+      "step": 65
+    },
+    {
+      "epoch": 0.1464226289517471,
+      "grad_norm": 1.8991446495056152,
+      "learning_rate": 0.00014666666666666664,
+      "loss": 6.2319,
+      "step": 66
+    },
+    {
+      "epoch": 0.14864115363283417,
+      "grad_norm": 1.5907998085021973,
+      "learning_rate": 0.00014888888888888886,
+      "loss": 6.0563,
+      "step": 67
+    },
+    {
+      "epoch": 0.15085967831392125,
+      "grad_norm": 1.9624888896942139,
+      "learning_rate": 0.00015111111111111109,
+      "loss": 6.0316,
+      "step": 68
+    },
+    {
+      "epoch": 0.15307820299500832,
+      "grad_norm": 2.269652843475342,
+      "learning_rate": 0.0001533333333333333,
+      "loss": 5.8955,
+      "step": 69
+    },
+    {
+      "epoch": 0.1552967276760954,
+      "grad_norm": 2.1748056411743164,
+      "learning_rate": 0.00015555555555555554,
+      "loss": 5.9671,
+      "step": 70
+    },
+    {
+      "epoch": 0.15751525235718247,
+      "grad_norm": 2.7504119873046875,
+      "learning_rate": 0.00015777777777777776,
+      "loss": 6.0463,
+      "step": 71
+    },
+    {
+      "epoch": 0.15973377703826955,
+      "grad_norm": 3.888150453567505,
+      "learning_rate": 0.00015999999999999999,
+      "loss": 6.0124,
+      "step": 72
+    },
+    {
+      "epoch": 0.16195230171935662,
+      "grad_norm": 4.348245143890381,
+      "learning_rate": 0.0001622222222222222,
+      "loss": 5.976,
+      "step": 73
+    },
+    {
+      "epoch": 0.1641708264004437,
+      "grad_norm": 3.072608709335327,
+      "learning_rate": 0.00016444444444444444,
+      "loss": 6.1441,
+      "step": 74
+    },
+    {
+      "epoch": 0.16638935108153077,
+      "grad_norm": 3.1516716480255127,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 5.7333,
+      "step": 75
+    },
+    {
+      "epoch": 0.16860787576261785,
+      "grad_norm": 2.9489054679870605,
+      "learning_rate": 0.00016888888888888889,
+      "loss": 6.2946,
+      "step": 76
+    },
+    {
+      "epoch": 0.17082640044370492,
+      "grad_norm": 4.6771440505981445,
+      "learning_rate": 0.0001711111111111111,
+      "loss": 5.9862,
+      "step": 77
+    },
+    {
+      "epoch": 0.17304492512479203,
+      "grad_norm": 3.934696674346924,
+      "learning_rate": 0.0001733333333333333,
+      "loss": 5.785,
+      "step": 78
+    },
+    {
+      "epoch": 0.1752634498058791,
+      "grad_norm": 4.91597843170166,
+      "learning_rate": 0.00017555555555555553,
+      "loss": 6.0061,
+      "step": 79
+    },
+    {
+      "epoch": 0.17748197448696618,
+      "grad_norm": 3.467418670654297,
+      "learning_rate": 0.00017777777777777776,
+      "loss": 5.8355,
+      "step": 80
+    },
+    {
+      "epoch": 0.17970049916805325,
+      "grad_norm": 4.002587795257568,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 5.9025,
+      "step": 81
+    },
+    {
+      "epoch": 0.18191902384914033,
+      "grad_norm": 4.973423480987549,
+      "learning_rate": 0.0001822222222222222,
+      "loss": 5.8924,
+      "step": 82
+    },
+    {
+      "epoch": 0.1841375485302274,
+      "grad_norm": 3.8708221912384033,
+      "learning_rate": 0.00018444444444444443,
+      "loss": 5.7199,
+      "step": 83
+    },
+    {
+      "epoch": 0.18635607321131448,
+      "grad_norm": 4.65097713470459,
+      "learning_rate": 0.00018666666666666666,
+      "loss": 5.7693,
+      "step": 84
+    },
+    {
+      "epoch": 0.18857459789240155,
+      "grad_norm": 16.77046012878418,
+      "learning_rate": 0.00018888888888888888,
+      "loss": 5.8981,
+      "step": 85
+    },
+    {
+      "epoch": 0.19079312257348863,
+      "grad_norm": 14.810404777526855,
+      "learning_rate": 0.0001911111111111111,
+      "loss": 5.6641,
+      "step": 86
+    },
+    {
+      "epoch": 0.1930116472545757,
+      "grad_norm": 8.933140754699707,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 5.8451,
+      "step": 87
+    },
+    {
+      "epoch": 0.19523017193566278,
+      "grad_norm": 6.613872528076172,
+      "learning_rate": 0.00019555555555555556,
+      "loss": 5.6441,
+      "step": 88
+    },
+    {
+      "epoch": 0.19744869661674985,
+      "grad_norm": 8.494193077087402,
+      "learning_rate": 0.00019777777777777776,
+      "loss": 5.7919,
+      "step": 89
+    },
+    {
+      "epoch": 0.19966722129783693,
+      "grad_norm": 4.945113182067871,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 5.5835,
+      "step": 90
+    },
+    {
+      "epoch": 0.201885745978924,
+      "grad_norm": 5.665979385375977,
+      "learning_rate": 0.0002022222222222222,
+      "loss": 5.6427,
+      "step": 91
+    },
+    {
+      "epoch": 0.2041042706600111,
+      "grad_norm": 7.968311309814453,
+      "learning_rate": 0.00020444444444444443,
+      "loss": 5.4739,
+      "step": 92
+    },
+    {
+      "epoch": 0.20632279534109818,
+      "grad_norm": 6.179589748382568,
+      "learning_rate": 0.00020666666666666666,
+      "loss": 5.6399,
+      "step": 93
+    },
+    {
+      "epoch": 0.20854132002218526,
+      "grad_norm": 14.595516204833984,
+      "learning_rate": 0.00020888888888888888,
+      "loss": 5.7314,
+      "step": 94
+    },
+    {
+      "epoch": 0.21075984470327233,
+      "grad_norm": 15.12678337097168,
+      "learning_rate": 0.0002111111111111111,
+      "loss": 5.679,
+      "step": 95
+    },
+    {
+      "epoch": 0.2129783693843594,
+      "grad_norm": 13.508474349975586,
+      "learning_rate": 0.00021333333333333333,
+      "loss": 5.5897,
+      "step": 96
+    },
+    {
+      "epoch": 0.21519689406544648,
+      "grad_norm": 8.335769653320312,
+      "learning_rate": 0.00021555555555555556,
+      "loss": 5.4184,
+      "step": 97
+    },
+    {
+      "epoch": 0.21741541874653356,
+      "grad_norm": 14.480011940002441,
+      "learning_rate": 0.00021777777777777778,
+      "loss": 5.5851,
+      "step": 98
+    },
+    {
+      "epoch": 0.21963394342762063,
+      "grad_norm": 7.999107360839844,
+      "learning_rate": 0.00021999999999999995,
+      "loss": 5.6877,
+      "step": 99
+    },
+    {
+      "epoch": 0.2218524681087077,
+      "grad_norm": 16.250043869018555,
+      "learning_rate": 0.00022222222222222218,
+      "loss": 5.4235,
+      "step": 100
+    },
+    {
+      "epoch": 0.22407099278979478,
+      "grad_norm": 13.328763961791992,
+      "learning_rate": 0.0002244444444444444,
+      "loss": 4.9404,
+      "step": 101
+    },
+    {
+      "epoch": 0.22628951747088186,
+      "grad_norm": 20.239822387695312,
+      "learning_rate": 0.00022666666666666663,
+      "loss": 5.3611,
+      "step": 102
+    },
+    {
+      "epoch": 0.22850804215196893,
+      "grad_norm": 13.419021606445312,
+      "learning_rate": 0.00022888888888888885,
+      "loss": 5.4816,
+      "step": 103
+    },
+    {
+      "epoch": 0.230726566833056,
+      "grad_norm": 17.98966407775879,
+      "learning_rate": 0.00023111111111111108,
+      "loss": 5.0474,
+      "step": 104
+    },
+    {
+      "epoch": 0.23294509151414308,
+      "grad_norm": 14.047506332397461,
+      "learning_rate": 0.0002333333333333333,
+      "loss": 5.5451,
+      "step": 105
+    },
+    {
+      "epoch": 0.23516361619523019,
+      "grad_norm": 9.935893058776855,
+      "learning_rate": 0.00023555555555555553,
+      "loss": 5.294,
+      "step": 106
+    },
+    {
+      "epoch": 0.23738214087631726,
+      "grad_norm": 10.235767364501953,
+      "learning_rate": 0.00023777777777777775,
+      "loss": 5.1985,
+      "step": 107
+    },
+    {
+      "epoch": 0.23960066555740434,
+      "grad_norm": 4.958045959472656,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 5.3386,
+      "step": 108
+    },
+    {
+      "epoch": 0.2418191902384914,
+      "grad_norm": 7.056834697723389,
+      "learning_rate": 0.0002422222222222222,
+      "loss": 5.3093,
+      "step": 109
+    },
+    {
+      "epoch": 0.24403771491957849,
+      "grad_norm": 9.293121337890625,
+      "learning_rate": 0.00024444444444444443,
+      "loss": 5.3868,
+      "step": 110
+    },
+    {
+      "epoch": 0.24625623960066556,
+      "grad_norm": 7.939361095428467,
+      "learning_rate": 0.0002466666666666666,
+      "loss": 5.2954,
+      "step": 111
+    },
+    {
+      "epoch": 0.24847476428175264,
+      "grad_norm": 7.578208923339844,
+      "learning_rate": 0.0002488888888888889,
+      "loss": 4.9886,
+      "step": 112
+    },
+    {
+      "epoch": 0.2506932889628397,
+      "grad_norm": 9.252887725830078,
+      "learning_rate": 0.0002511111111111111,
+      "loss": 5.0232,
+      "step": 113
+    },
+    {
+      "epoch": 0.2529118136439268,
+      "grad_norm": 5.097420692443848,
+      "learning_rate": 0.00025333333333333333,
+      "loss": 5.0305,
+      "step": 114
+    },
+    {
+      "epoch": 0.25513033832501386,
+      "grad_norm": 4.758305549621582,
+      "learning_rate": 0.00025555555555555553,
+      "loss": 5.036,
+      "step": 115
+    },
+    {
+      "epoch": 0.25734886300610094,
+      "grad_norm": 7.699611663818359,
+      "learning_rate": 0.0002577777777777778,
+      "loss": 5.1191,
+      "step": 116
+    },
+    {
+      "epoch": 0.259567387687188,
+      "grad_norm": 4.97732400894165,
+      "learning_rate": 0.00026,
+      "loss": 5.0226,
+      "step": 117
+    },
+    {
+      "epoch": 0.2617859123682751,
+      "grad_norm": 12.636543273925781,
+      "learning_rate": 0.00026222222222222223,
+      "loss": 5.0779,
+      "step": 118
+    },
+    {
+      "epoch": 0.26400443704936216,
+      "grad_norm": 9.042309761047363,
+      "learning_rate": 0.00026444444444444443,
+      "loss": 5.1155,
+      "step": 119
+    },
+    {
+      "epoch": 0.26622296173044924,
+      "grad_norm": 8.143863677978516,
+      "learning_rate": 0.0002666666666666666,
+      "loss": 5.09,
+      "step": 120
+    },
+    {
+      "epoch": 0.2684414864115363,
+      "grad_norm": 8.229655265808105,
+      "learning_rate": 0.0002688888888888889,
+      "loss": 4.8991,
+      "step": 121
+    },
+    {
+      "epoch": 0.2706600110926234,
+      "grad_norm": 6.974783420562744,
+      "learning_rate": 0.0002711111111111111,
+      "loss": 5.0372,
+      "step": 122
+    },
+    {
+      "epoch": 0.27287853577371046,
+      "grad_norm": 6.506590843200684,
+      "learning_rate": 0.00027333333333333333,
+      "loss": 4.8798,
+      "step": 123
+    },
+    {
+      "epoch": 0.27509706045479754,
+      "grad_norm": 6.753083229064941,
+      "learning_rate": 0.0002755555555555555,
+      "loss": 4.98,
+      "step": 124
+    },
+    {
+      "epoch": 0.2773155851358846,
+      "grad_norm": 14.494991302490234,
+      "learning_rate": 0.0002777777777777778,
+      "loss": 5.3525,
+      "step": 125
+    },
+    {
+      "epoch": 0.2795341098169717,
+      "grad_norm": 10.126094818115234,
+      "learning_rate": 0.00028,
+      "loss": 5.1127,
+      "step": 126
+    },
+    {
+      "epoch": 0.2817526344980588,
+      "grad_norm": 9.596733093261719,
+      "learning_rate": 0.00028222222222222223,
+      "loss": 4.907,
+      "step": 127
+    },
+    {
+      "epoch": 0.2839711591791459,
+      "grad_norm": 10.70608139038086,
+      "learning_rate": 0.0002844444444444444,
+      "loss": 5.1895,
+      "step": 128
+    },
+    {
+      "epoch": 0.28618968386023297,
+      "grad_norm": 7.24511194229126,
+      "learning_rate": 0.0002866666666666667,
+      "loss": 4.953,
+      "step": 129
+    },
+    {
+      "epoch": 0.28840820854132004,
+      "grad_norm": 5.899155139923096,
+      "learning_rate": 0.0002888888888888888,
+      "loss": 5.1019,
+      "step": 130
+    },
+    {
+      "epoch": 0.2906267332224071,
+      "grad_norm": 6.8543548583984375,
+      "learning_rate": 0.0002911111111111111,
+      "loss": 5.1044,
+      "step": 131
+    },
+    {
+      "epoch": 0.2928452579034942,
+      "grad_norm": 5.690365791320801,
+      "learning_rate": 0.00029333333333333327,
+      "loss": 4.6546,
+      "step": 132
+    },
+    {
+      "epoch": 0.29506378258458127,
+      "grad_norm": 6.473843574523926,
+      "learning_rate": 0.0002955555555555555,
+      "loss": 5.008,
+      "step": 133
+    },
+    {
+      "epoch": 0.29728230726566834,
+      "grad_norm": 4.835877418518066,
+      "learning_rate": 0.0002977777777777777,
+      "loss": 4.7897,
+      "step": 134
+    },
+    {
+      "epoch": 0.2995008319467554,
+      "grad_norm": 6.401046276092529,
+      "learning_rate": 0.0003,
+      "loss": 4.5805,
+      "step": 135
+    },
+    {
+      "epoch": 0.3017193566278425,
+      "grad_norm": 6.061528205871582,
+      "learning_rate": 0.00029975308641975307,
+      "loss": 4.5766,
+      "step": 136
+    },
+    {
+      "epoch": 0.30393788130892957,
+      "grad_norm": 4.991875648498535,
+      "learning_rate": 0.00029950617283950616,
+      "loss": 4.9961,
+      "step": 137
+    },
+    {
+      "epoch": 0.30615640599001664,
+      "grad_norm": 6.214187145233154,
+      "learning_rate": 0.00029925925925925926,
+      "loss": 4.8395,
+      "step": 138
+    },
+    {
+      "epoch": 0.3083749306711037,
+      "grad_norm": 5.595952987670898,
+      "learning_rate": 0.0002990123456790123,
+      "loss": 4.7314,
+      "step": 139
+    },
+    {
+      "epoch": 0.3105934553521908,
+      "grad_norm": 4.068733215332031,
+      "learning_rate": 0.0002987654320987654,
+      "loss": 4.7183,
+      "step": 140
+    },
+    {
+      "epoch": 0.31281198003327787,
+      "grad_norm": 4.258878707885742,
+      "learning_rate": 0.0002985185185185185,
+      "loss": 4.5647,
+      "step": 141
+    },
+    {
+      "epoch": 0.31503050471436495,
+      "grad_norm": 4.74027156829834,
+      "learning_rate": 0.0002982716049382716,
+      "loss": 4.8485,
+      "step": 142
+    },
+    {
+      "epoch": 0.317249029395452,
+      "grad_norm": 5.537410259246826,
+      "learning_rate": 0.0002980246913580247,
+      "loss": 4.6488,
+      "step": 143
+    },
+    {
+      "epoch": 0.3194675540765391,
+      "grad_norm": 4.852229118347168,
+      "learning_rate": 0.0002977777777777777,
+      "loss": 4.8425,
+      "step": 144
+    },
+    {
+      "epoch": 0.32168607875762617,
+      "grad_norm": 5.118812084197998,
+      "learning_rate": 0.00029753086419753087,
+      "loss": 4.7661,
+      "step": 145
+    },
+    {
+      "epoch": 0.32390460343871325,
+      "grad_norm": 4.468764305114746,
+      "learning_rate": 0.0002972839506172839,
+      "loss": 4.6061,
+      "step": 146
+    },
+    {
+      "epoch": 0.3261231281198003,
+      "grad_norm": 4.146613121032715,
+      "learning_rate": 0.000297037037037037,
+      "loss": 4.3941,
+      "step": 147
+    },
+    {
+      "epoch": 0.3283416528008874,
+      "grad_norm": 4.582084655761719,
+      "learning_rate": 0.0002967901234567901,
+      "loss": 4.803,
+      "step": 148
+    },
+    {
+      "epoch": 0.33056017748197447,
+      "grad_norm": 4.619938373565674,
+      "learning_rate": 0.0002965432098765432,
+      "loss": 4.7327,
+      "step": 149
+    },
+    {
+      "epoch": 0.33277870216306155,
+      "grad_norm": 3.356536626815796,
+      "learning_rate": 0.00029629629629629624,
+      "loss": 4.6047,
+      "step": 150
+    },
+    {
+      "epoch": 0.3349972268441486,
+      "grad_norm": 5.1952223777771,
+      "learning_rate": 0.00029604938271604933,
+      "loss": 4.7517,
+      "step": 151
+    },
+    {
+      "epoch": 0.3372157515252357,
+      "grad_norm": 3.840526580810547,
+      "learning_rate": 0.0002958024691358025,
+      "loss": 4.7089,
+      "step": 152
+    },
+    {
+      "epoch": 0.33943427620632277,
+      "grad_norm": 4.290994167327881,
+      "learning_rate": 0.0002955555555555555,
+      "loss": 4.6499,
+      "step": 153
+    },
+    {
+      "epoch": 0.34165280088740985,
+      "grad_norm": 3.5644822120666504,
+      "learning_rate": 0.0002953086419753086,
+      "loss": 4.5388,
+      "step": 154
+    },
+    {
+      "epoch": 0.343871325568497,
+      "grad_norm": 4.252591133117676,
+      "learning_rate": 0.0002950617283950617,
+      "loss": 4.8249,
+      "step": 155
+    },
+    {
+      "epoch": 0.34608985024958405,
+      "grad_norm": 3.3963656425476074,
+      "learning_rate": 0.0002948148148148148,
+      "loss": 4.5846,
+      "step": 156
+    },
+    {
+      "epoch": 0.3483083749306711,
+      "grad_norm": 3.7368924617767334,
+      "learning_rate": 0.00029456790123456785,
+      "loss": 4.639,
+      "step": 157
+    },
+    {
+      "epoch": 0.3505268996117582,
+      "grad_norm": 3.779806137084961,
+      "learning_rate": 0.00029432098765432095,
+      "loss": 4.7207,
+      "step": 158
+    },
+    {
+      "epoch": 0.3527454242928453,
+      "grad_norm": 3.0449748039245605,
+      "learning_rate": 0.00029407407407407404,
+      "loss": 4.1599,
+      "step": 159
+    },
+    {
+      "epoch": 0.35496394897393235,
+      "grad_norm": 2.783602237701416,
+      "learning_rate": 0.00029382716049382714,
+      "loss": 4.398,
+      "step": 160
+    },
+    {
+      "epoch": 0.35718247365501943,
+      "grad_norm": 3.895935297012329,
+      "learning_rate": 0.00029358024691358023,
+      "loss": 4.885,
+      "step": 161
+    },
+    {
+      "epoch": 0.3594009983361065,
+      "grad_norm": 3.5132410526275635,
+      "learning_rate": 0.00029333333333333327,
+      "loss": 4.6741,
+      "step": 162
+    },
+    {
+      "epoch": 0.3616195230171936,
+      "grad_norm": 3.6633965969085693,
+      "learning_rate": 0.0002930864197530864,
+      "loss": 4.5929,
+      "step": 163
+    },
+    {
+      "epoch": 0.36383804769828065,
+      "grad_norm": 2.6730799674987793,
+      "learning_rate": 0.00029283950617283946,
+      "loss": 4.547,
+      "step": 164
+    },
+    {
+      "epoch": 0.36605657237936773,
+      "grad_norm": 3.6293373107910156,
+      "learning_rate": 0.00029259259259259256,
+      "loss": 4.4986,
+      "step": 165
+    },
+    {
+      "epoch": 0.3682750970604548,
+      "grad_norm": 4.365437030792236,
+      "learning_rate": 0.00029234567901234565,
+      "loss": 4.5857,
+      "step": 166
+    },
+    {
+      "epoch": 0.3704936217415419,
+      "grad_norm": 3.852407455444336,
+      "learning_rate": 0.00029209876543209875,
+      "loss": 4.3493,
+      "step": 167
+    },
+    {
+      "epoch": 0.37271214642262895,
+      "grad_norm": 3.2594995498657227,
+      "learning_rate": 0.00029185185185185184,
+      "loss": 4.3635,
+      "step": 168
+    },
+    {
+      "epoch": 0.37493067110371603,
+      "grad_norm": 4.344808101654053,
+      "learning_rate": 0.0002916049382716049,
+      "loss": 4.3097,
+      "step": 169
+    },
+    {
+      "epoch": 0.3771491957848031,
+      "grad_norm": 4.024368762969971,
+      "learning_rate": 0.000291358024691358,
+      "loss": 4.4091,
+      "step": 170
+    },
+    {
+      "epoch": 0.3793677204658902,
+      "grad_norm": 2.706266164779663,
+      "learning_rate": 0.0002911111111111111,
+      "loss": 4.6553,
+      "step": 171
+    },
+    {
+      "epoch": 0.38158624514697725,
+      "grad_norm": 2.9779529571533203,
+      "learning_rate": 0.00029086419753086417,
+      "loss": 4.1661,
+      "step": 172
+    },
+    {
+      "epoch": 0.38380476982806433,
+      "grad_norm": 3.2455942630767822,
+      "learning_rate": 0.00029061728395061726,
+      "loss": 4.1221,
+      "step": 173
+    },
+    {
+      "epoch": 0.3860232945091514,
+      "grad_norm": 3.613090753555298,
+      "learning_rate": 0.00029037037037037036,
+      "loss": 4.347,
+      "step": 174
+    },
+    {
+      "epoch": 0.3882418191902385,
+      "grad_norm": 2.904637575149536,
+      "learning_rate": 0.00029012345679012345,
+      "loss": 4.3928,
+      "step": 175
+    },
+    {
+      "epoch": 0.39046034387132555,
+      "grad_norm": 2.919376850128174,
+      "learning_rate": 0.0002898765432098765,
+      "loss": 4.3423,
+      "step": 176
+    },
+    {
+      "epoch": 0.39267886855241263,
+      "grad_norm": 2.891509771347046,
+      "learning_rate": 0.0002896296296296296,
+      "loss": 4.1071,
+      "step": 177
+    },
+    {
+      "epoch": 0.3948973932334997,
+      "grad_norm": 2.6562042236328125,
+      "learning_rate": 0.0002893827160493827,
+      "loss": 4.3039,
+      "step": 178
+    },
+    {
+      "epoch": 0.3971159179145868,
+      "grad_norm": 3.776806592941284,
+      "learning_rate": 0.0002891358024691358,
+      "loss": 4.7419,
+      "step": 179
+    },
+    {
+      "epoch": 0.39933444259567386,
+      "grad_norm": 3.0195436477661133,
+      "learning_rate": 0.0002888888888888888,
+      "loss": 4.3398,
+      "step": 180
+    },
+    {
+      "epoch": 0.40155296727676093,
+      "grad_norm": 2.7870137691497803,
+      "learning_rate": 0.00028864197530864197,
+      "loss": 4.3052,
+      "step": 181
+    },
+    {
+      "epoch": 0.403771491957848,
+      "grad_norm": 2.7188875675201416,
+      "learning_rate": 0.00028839506172839507,
+      "loss": 4.3884,
+      "step": 182
+    },
+    {
+      "epoch": 0.4059900166389351,
+      "grad_norm": 2.6208927631378174,
+      "learning_rate": 0.0002881481481481481,
+      "loss": 4.1932,
+      "step": 183
+    },
+    {
+      "epoch": 0.4082085413200222,
+      "grad_norm": 2.569000244140625,
+      "learning_rate": 0.0002879012345679012,
+      "loss": 4.4338,
+      "step": 184
+    },
+    {
+      "epoch": 0.4104270660011093,
+      "grad_norm": 2.5792455673217773,
+      "learning_rate": 0.0002876543209876543,
+      "loss": 4.3255,
+      "step": 185
+    },
+    {
+      "epoch": 0.41264559068219636,
+      "grad_norm": 2.7281129360198975,
+      "learning_rate": 0.0002874074074074074,
+      "loss": 4.2379,
+      "step": 186
+    },
+    {
+      "epoch": 0.41486411536328344,
+      "grad_norm": 3.042790174484253,
+      "learning_rate": 0.00028716049382716043,
+      "loss": 4.0505,
+      "step": 187
+    },
+    {
+      "epoch": 0.4170826400443705,
+      "grad_norm": 3.019914388656616,
+      "learning_rate": 0.00028691358024691353,
+      "loss": 4.2884,
+      "step": 188
+    },
+    {
+      "epoch": 0.4193011647254576,
+      "grad_norm": 2.766526460647583,
+      "learning_rate": 0.0002866666666666667,
+      "loss": 4.2601,
+      "step": 189
+    },
+    {
+      "epoch": 0.42151968940654466,
+      "grad_norm": 2.3037662506103516,
+      "learning_rate": 0.0002864197530864197,
+      "loss": 4.1702,
+      "step": 190
+    },
+    {
+      "epoch": 0.42373821408763174,
+      "grad_norm": 2.212174892425537,
+      "learning_rate": 0.0002861728395061728,
+      "loss": 4.3902,
+      "step": 191
+    },
+    {
+      "epoch": 0.4259567387687188,
+      "grad_norm": 2.4971466064453125,
+      "learning_rate": 0.0002859259259259259,
+      "loss": 4.2036,
+      "step": 192
+    },
+    {
+      "epoch": 0.4281752634498059,
+      "grad_norm": 2.2342379093170166,
+      "learning_rate": 0.000285679012345679,
+      "loss": 4.3236,
+      "step": 193
+    },
+    {
+      "epoch": 0.43039378813089296,
+      "grad_norm": 2.6258656978607178,
+      "learning_rate": 0.00028543209876543204,
+      "loss": 4.2153,
+      "step": 194
+    },
+    {
+      "epoch": 0.43261231281198004,
+      "grad_norm": 2.4059770107269287,
+      "learning_rate": 0.00028518518518518514,
+      "loss": 4.3597,
+      "step": 195
+    },
+    {
+      "epoch": 0.4348308374930671,
+      "grad_norm": 2.507228374481201,
+      "learning_rate": 0.00028493827160493824,
+      "loss": 4.2975,
+      "step": 196
+    },
+    {
+      "epoch": 0.4370493621741542,
+      "grad_norm": 2.59570574760437,
+      "learning_rate": 0.00028469135802469133,
+      "loss": 4.3392,
+      "step": 197
+    },
+    {
+      "epoch": 0.43926788685524126,
+      "grad_norm": 2.3920488357543945,
+      "learning_rate": 0.0002844444444444444,
+      "loss": 4.4862,
+      "step": 198
+    },
+    {
+      "epoch": 0.44148641153632834,
+      "grad_norm": 2.7204859256744385,
+      "learning_rate": 0.0002841975308641975,
+      "loss": 4.2932,
+      "step": 199
+    },
+    {
+      "epoch": 0.4437049362174154,
+      "grad_norm": 2.4902760982513428,
+      "learning_rate": 0.0002839506172839506,
+      "loss": 4.172,
+      "step": 200
+    },
+    {
+      "epoch": 0.4459234608985025,
+      "grad_norm": 2.0951662063598633,
+      "learning_rate": 0.00028370370370370366,
+      "loss": 4.1367,
+      "step": 201
+    },
+    {
+      "epoch": 0.44814198557958956,
+      "grad_norm": 2.7248218059539795,
+      "learning_rate": 0.00028345679012345675,
+      "loss": 4.2397,
+      "step": 202
+    },
+    {
+      "epoch": 0.45036051026067664,
+      "grad_norm": 2.804983377456665,
+      "learning_rate": 0.00028320987654320985,
+      "loss": 4.3643,
+      "step": 203
+    },
+    {
+      "epoch": 0.4525790349417637,
+      "grad_norm": 3.2091169357299805,
+      "learning_rate": 0.00028296296296296294,
+      "loss": 4.6862,
+      "step": 204
+    },
+    {
+      "epoch": 0.4547975596228508,
+      "grad_norm": 2.811924934387207,
+      "learning_rate": 0.00028271604938271604,
+      "loss": 4.447,
+      "step": 205
+    },
+    {
+      "epoch": 0.45701608430393786,
+      "grad_norm": 2.446161985397339,
+      "learning_rate": 0.0002824691358024691,
+      "loss": 4.0968,
+      "step": 206
+    },
+    {
+      "epoch": 0.45923460898502494,
+      "grad_norm": 2.4840290546417236,
+      "learning_rate": 0.00028222222222222223,
+      "loss": 4.2464,
+      "step": 207
+    },
+    {
+      "epoch": 0.461453133666112,
+      "grad_norm": 2.29032301902771,
+      "learning_rate": 0.00028197530864197527,
+      "loss": 3.865,
+      "step": 208
+    },
+    {
+      "epoch": 0.4636716583471991,
+      "grad_norm": 2.1570351123809814,
+      "learning_rate": 0.00028172839506172836,
+      "loss": 4.1553,
+      "step": 209
+    },
+    {
+      "epoch": 0.46589018302828616,
+      "grad_norm": 1.682471752166748,
+      "learning_rate": 0.00028148148148148146,
+      "loss": 4.031,
+      "step": 210
+    },
+    {
+      "epoch": 0.46810870770937324,
+      "grad_norm": 2.0931646823883057,
+      "learning_rate": 0.00028123456790123455,
+      "loss": 4.2569,
+      "step": 211
+    },
+    {
+      "epoch": 0.47032723239046037,
+      "grad_norm": 2.2267751693725586,
+      "learning_rate": 0.00028098765432098765,
+      "loss": 3.8939,
+      "step": 212
+    },
+    {
+      "epoch": 0.47254575707154745,
+      "grad_norm": 2.214029312133789,
+      "learning_rate": 0.0002807407407407407,
+      "loss": 4.2292,
+      "step": 213
+    },
+    {
+      "epoch": 0.4747642817526345,
+      "grad_norm": 1.8627341985702515,
+      "learning_rate": 0.0002804938271604938,
+      "loss": 3.9748,
+      "step": 214
+    },
+    {
+      "epoch": 0.4769828064337216,
+      "grad_norm": 2.172668933868408,
+      "learning_rate": 0.0002802469135802469,
+      "loss": 4.2146,
+      "step": 215
+    },
+    {
+      "epoch": 0.47920133111480867,
+      "grad_norm": 1.9790834188461304,
+      "learning_rate": 0.00028,
+      "loss": 4.0975,
+      "step": 216
+    },
+    {
+      "epoch": 0.48141985579589575,
+      "grad_norm": 2.1610898971557617,
+      "learning_rate": 0.00027975308641975307,
+      "loss": 4.1161,
+      "step": 217
+    },
+    {
+      "epoch": 0.4836383804769828,
+      "grad_norm": 2.04876971244812,
+      "learning_rate": 0.00027950617283950617,
+      "loss": 3.9726,
+      "step": 218
+    },
+    {
+      "epoch": 0.4858569051580699,
+      "grad_norm": 2.288295269012451,
+      "learning_rate": 0.00027925925925925926,
+      "loss": 4.0098,
+      "step": 219
+    },
+    {
+      "epoch": 0.48807542983915697,
+      "grad_norm": 2.3760294914245605,
+      "learning_rate": 0.0002790123456790123,
+      "loss": 4.2243,
+      "step": 220
+    },
+    {
+      "epoch": 0.49029395452024405,
+      "grad_norm": 2.219660520553589,
+      "learning_rate": 0.0002787654320987654,
+      "loss": 4.3357,
+      "step": 221
+    },
+    {
+      "epoch": 0.4925124792013311,
+      "grad_norm": 2.1059184074401855,
+      "learning_rate": 0.0002785185185185185,
+      "loss": 4.1752,
+      "step": 222
+    },
+    {
+      "epoch": 0.4947310038824182,
+      "grad_norm": 1.9606491327285767,
+      "learning_rate": 0.0002782716049382716,
+      "loss": 4.2967,
+      "step": 223
+    },
+    {
+      "epoch": 0.49694952856350527,
+      "grad_norm": 2.009049892425537,
+      "learning_rate": 0.00027802469135802463,
+      "loss": 4.0331,
+      "step": 224
+    },
+    {
+      "epoch": 0.49916805324459235,
+      "grad_norm": 2.0096426010131836,
+      "learning_rate": 0.0002777777777777778,
+      "loss": 4.0232,
+      "step": 225
+    },
+    {
+      "epoch": 0.5013865779256794,
+      "grad_norm": 1.89435875415802,
+      "learning_rate": 0.00027753086419753087,
+      "loss": 4.1834,
+      "step": 226
+    },
+    {
+      "epoch": 0.5036051026067665,
+      "grad_norm": 2.8503754138946533,
+      "learning_rate": 0.0002772839506172839,
+      "loss": 4.4017,
+      "step": 227
+    },
+    {
+      "epoch": 0.5058236272878536,
+      "grad_norm": 2.5959761142730713,
+      "learning_rate": 0.000277037037037037,
+      "loss": 4.1474,
+      "step": 228
+    },
+    {
+      "epoch": 0.5080421519689406,
+      "grad_norm": 2.28751802444458,
+      "learning_rate": 0.0002767901234567901,
+      "loss": 4.2984,
+      "step": 229
+    },
+    {
+      "epoch": 0.5102606766500277,
+      "grad_norm": 2.4467239379882812,
+      "learning_rate": 0.0002765432098765432,
+      "loss": 4.2051,
+      "step": 230
+    },
+    {
+      "epoch": 0.5124792013311148,
+      "grad_norm": 2.288095474243164,
+      "learning_rate": 0.00027629629629629624,
+      "loss": 4.1398,
+      "step": 231
+    },
+    {
+      "epoch": 0.5146977260122019,
+      "grad_norm": 1.8156859874725342,
+      "learning_rate": 0.00027604938271604934,
+      "loss": 4.0906,
+      "step": 232
+    },
+    {
+      "epoch": 0.516916250693289,
+      "grad_norm": 1.852567195892334,
+      "learning_rate": 0.00027580246913580243,
+      "loss": 3.955,
+      "step": 233
+    },
+    {
+      "epoch": 0.519134775374376,
+      "grad_norm": 2.3244059085845947,
+      "learning_rate": 0.0002755555555555555,
+      "loss": 4.011,
+      "step": 234
+    },
+    {
+      "epoch": 0.5213533000554631,
+      "grad_norm": 2.381349802017212,
+      "learning_rate": 0.0002753086419753086,
+      "loss": 3.9915,
+      "step": 235
+    },
+    {
+      "epoch": 0.5235718247365502,
+      "grad_norm": 2.0570316314697266,
+      "learning_rate": 0.0002750617283950617,
+      "loss": 4.2446,
+      "step": 236
+    },
+    {
+      "epoch": 0.5257903494176372,
+      "grad_norm": 1.8881930112838745,
+      "learning_rate": 0.0002748148148148148,
+      "loss": 4.1063,
+      "step": 237
+    },
+    {
+      "epoch": 0.5280088740987243,
+      "grad_norm": 1.9570058584213257,
+      "learning_rate": 0.00027456790123456785,
+      "loss": 3.8757,
+      "step": 238
+    },
+    {
+      "epoch": 0.5302273987798114,
+      "grad_norm": 2.1820671558380127,
+      "learning_rate": 0.00027432098765432095,
+      "loss": 3.8954,
+      "step": 239
+    },
+    {
+      "epoch": 0.5324459234608985,
+      "grad_norm": 1.7662599086761475,
+      "learning_rate": 0.00027407407407407404,
+      "loss": 4.0615,
+      "step": 240
+    },
+    {
+      "epoch": 0.5346644481419855,
+      "grad_norm": 1.725443720817566,
+      "learning_rate": 0.00027382716049382714,
+      "loss": 3.7233,
+      "step": 241
+    },
+    {
+      "epoch": 0.5368829728230726,
+      "grad_norm": 2.023792266845703,
+      "learning_rate": 0.00027358024691358023,
+      "loss": 3.9098,
+      "step": 242
+    },
+    {
+      "epoch": 0.5391014975041597,
+      "grad_norm": 2.496546983718872,
+      "learning_rate": 0.00027333333333333333,
+      "loss": 3.9531,
+      "step": 243
+    },
+    {
+      "epoch": 0.5413200221852468,
+      "grad_norm": 2.044433355331421,
+      "learning_rate": 0.0002730864197530864,
+      "loss": 4.0212,
+      "step": 244
+    },
+    {
+      "epoch": 0.5435385468663338,
+      "grad_norm": 2.073209047317505,
+      "learning_rate": 0.00027283950617283946,
+      "loss": 3.8505,
+      "step": 245
+    },
+    {
+      "epoch": 0.5457570715474209,
+      "grad_norm": 2.5616512298583984,
+      "learning_rate": 0.00027259259259259256,
+      "loss": 3.927,
+      "step": 246
+    },
+    {
+      "epoch": 0.547975596228508,
+      "grad_norm": 2.2878293991088867,
+      "learning_rate": 0.00027234567901234565,
+      "loss": 4.0332,
+      "step": 247
+    },
+    {
+      "epoch": 0.5501941209095951,
+      "grad_norm": 1.8358120918273926,
+      "learning_rate": 0.00027209876543209875,
+      "loss": 3.8857,
+      "step": 248
+    },
+    {
+      "epoch": 0.5524126455906821,
+      "grad_norm": 2.105027198791504,
+      "learning_rate": 0.00027185185185185184,
+      "loss": 3.903,
+      "step": 249
+    },
+    {
+      "epoch": 0.5546311702717692,
+      "grad_norm": 2.1397314071655273,
+      "learning_rate": 0.0002716049382716049,
+      "loss": 4.0196,
+      "step": 250
+    },
+    {
+      "epoch": 0.5568496949528563,
+      "grad_norm": 2.436491012573242,
+      "learning_rate": 0.00027135802469135803,
+      "loss": 3.992,
+      "step": 251
+    },
+    {
+      "epoch": 0.5590682196339434,
+      "grad_norm": 2.1947808265686035,
+      "learning_rate": 0.0002711111111111111,
+      "loss": 4.088,
+      "step": 252
+    },
+    {
+      "epoch": 0.5612867443150305,
+      "grad_norm": 1.8576627969741821,
+      "learning_rate": 0.00027086419753086417,
+      "loss": 4.3433,
+      "step": 253
+    },
+    {
+      "epoch": 0.5635052689961176,
+      "grad_norm": 2.002056360244751,
+      "learning_rate": 0.00027061728395061727,
+      "loss": 4.0735,
+      "step": 254
+    },
+    {
+      "epoch": 0.5657237936772047,
+      "grad_norm": 2.175616502761841,
+      "learning_rate": 0.00027037037037037036,
+      "loss": 3.9976,
+      "step": 255
+    },
+    {
+      "epoch": 0.5679423183582918,
+      "grad_norm": 2.032122850418091,
+      "learning_rate": 0.00027012345679012346,
+      "loss": 4.0045,
+      "step": 256
+    },
+    {
+      "epoch": 0.5701608430393789,
+      "grad_norm": 1.821707010269165,
+      "learning_rate": 0.0002698765432098765,
+      "loss": 3.9382,
+      "step": 257
+    },
+    {
+      "epoch": 0.5723793677204659,
+      "grad_norm": 1.8324333429336548,
+      "learning_rate": 0.0002696296296296296,
+      "loss": 4.0312,
+      "step": 258
+    },
+    {
+      "epoch": 0.574597892401553,
+      "grad_norm": 1.778178334236145,
+      "learning_rate": 0.0002693827160493827,
+      "loss": 4.2112,
+      "step": 259
+    },
+    {
+      "epoch": 0.5768164170826401,
+      "grad_norm": 1.778599500656128,
+      "learning_rate": 0.0002691358024691358,
+      "loss": 4.1672,
+      "step": 260
+    },
+    {
+      "epoch": 0.5790349417637272,
+      "grad_norm": 1.882077693939209,
+      "learning_rate": 0.0002688888888888889,
+      "loss": 4.1839,
+      "step": 261
+    },
+    {
+      "epoch": 0.5812534664448142,
+      "grad_norm": 2.047104597091675,
+      "learning_rate": 0.00026864197530864197,
+      "loss": 3.4862,
+      "step": 262
+    },
+    {
+      "epoch": 0.5834719911259013,
+      "grad_norm": 1.8103227615356445,
+      "learning_rate": 0.000268395061728395,
+      "loss": 3.7993,
+      "step": 263
+    },
+    {
+      "epoch": 0.5856905158069884,
+      "grad_norm": 2.136939764022827,
+      "learning_rate": 0.0002681481481481481,
+      "loss": 3.9225,
+      "step": 264
+    },
+    {
+      "epoch": 0.5879090404880755,
+      "grad_norm": 1.9785947799682617,
+      "learning_rate": 0.0002679012345679012,
+      "loss": 4.1991,
+      "step": 265
+    },
+    {
+      "epoch": 0.5901275651691625,
+      "grad_norm": 2.015219211578369,
+      "learning_rate": 0.0002676543209876543,
+      "loss": 4.0601,
+      "step": 266
+    },
+    {
+      "epoch": 0.5923460898502496,
+      "grad_norm": 1.7777012586593628,
+      "learning_rate": 0.0002674074074074074,
+      "loss": 4.0274,
+      "step": 267
+    },
+    {
+      "epoch": 0.5945646145313367,
+      "grad_norm": 1.6315277814865112,
+      "learning_rate": 0.00026716049382716043,
+      "loss": 4.1314,
+      "step": 268
+    },
+    {
+      "epoch": 0.5967831392124238,
+      "grad_norm": 1.4809752702713013,
+      "learning_rate": 0.0002669135802469136,
+      "loss": 4.3051,
+      "step": 269
+    },
+    {
+      "epoch": 0.5990016638935108,
+      "grad_norm": 1.5730684995651245,
+      "learning_rate": 0.0002666666666666666,
+      "loss": 4.0122,
+      "step": 270
+    },
+    {
+      "epoch": 0.6012201885745979,
+      "grad_norm": 2.200824499130249,
+      "learning_rate": 0.0002664197530864197,
+      "loss": 3.993,
+      "step": 271
+    },
+    {
+      "epoch": 0.603438713255685,
+      "grad_norm": 1.7815587520599365,
+      "learning_rate": 0.0002661728395061728,
+      "loss": 3.7977,
+      "step": 272
+    },
+    {
+      "epoch": 0.6056572379367721,
+      "grad_norm": 1.552909016609192,
+      "learning_rate": 0.0002659259259259259,
+      "loss": 3.9721,
+      "step": 273
+    },
+    {
+      "epoch": 0.6078757626178591,
+      "grad_norm": 1.7399321794509888,
+      "learning_rate": 0.000265679012345679,
+      "loss": 4.0121,
+      "step": 274
+    },
+    {
+      "epoch": 0.6100942872989462,
+      "grad_norm": 1.9776664972305298,
+      "learning_rate": 0.00026543209876543205,
+      "loss": 4.0024,
+      "step": 275
+    },
+    {
+      "epoch": 0.6123128119800333,
+      "grad_norm": 2.056826591491699,
+      "learning_rate": 0.00026518518518518514,
+      "loss": 4.1136,
+      "step": 276
+    },
+    {
+      "epoch": 0.6145313366611204,
+      "grad_norm": 1.9091132879257202,
+      "learning_rate": 0.00026493827160493824,
+      "loss": 4.1582,
+      "step": 277
+    },
+    {
+      "epoch": 0.6167498613422074,
+      "grad_norm": 2.0029196739196777,
+      "learning_rate": 0.00026469135802469133,
+      "loss": 3.7346,
+      "step": 278
+    },
+    {
+      "epoch": 0.6189683860232945,
+      "grad_norm": 1.78605055809021,
+      "learning_rate": 0.00026444444444444443,
+      "loss": 4.1038,
+      "step": 279
+    },
+    {
+      "epoch": 0.6211869107043816,
+      "grad_norm": 1.8789544105529785,
+      "learning_rate": 0.0002641975308641975,
+      "loss": 3.9588,
+      "step": 280
+    },
+    {
+      "epoch": 0.6234054353854687,
+      "grad_norm": 1.7132203578948975,
+      "learning_rate": 0.0002639506172839506,
+      "loss": 3.9087,
+      "step": 281
+    },
+    {
+      "epoch": 0.6256239600665557,
+      "grad_norm": 1.6578532457351685,
+      "learning_rate": 0.00026370370370370366,
+      "loss": 3.5493,
+      "step": 282
+    },
+    {
+      "epoch": 0.6278424847476428,
+      "grad_norm": 2.023768186569214,
+      "learning_rate": 0.00026345679012345675,
+      "loss": 4.1595,
+      "step": 283
+    },
+    {
+      "epoch": 0.6300610094287299,
+      "grad_norm": 2.319453001022339,
+      "learning_rate": 0.00026320987654320985,
+      "loss": 4.0579,
+      "step": 284
+    },
+    {
+      "epoch": 0.632279534109817,
+      "grad_norm": 1.7584071159362793,
+      "learning_rate": 0.00026296296296296294,
+      "loss": 3.9682,
+      "step": 285
+    },
+    {
+      "epoch": 0.634498058790904,
+      "grad_norm": 1.893362045288086,
+      "learning_rate": 0.00026271604938271604,
+      "loss": 3.8874,
+      "step": 286
+    },
+    {
+      "epoch": 0.6367165834719911,
+      "grad_norm": 2.069011926651001,
+      "learning_rate": 0.00026246913580246913,
+      "loss": 4.0199,
+      "step": 287
+    },
+    {
+      "epoch": 0.6389351081530782,
+      "grad_norm": 5.405537128448486,
+      "learning_rate": 0.00026222222222222223,
+      "loss": 3.5828,
+      "step": 288
+    },
+    {
+      "epoch": 0.6411536328341653,
+      "grad_norm": 1.7799841165542603,
+      "learning_rate": 0.00026197530864197527,
+      "loss": 3.9538,
+      "step": 289
+    },
+    {
+      "epoch": 0.6433721575152523,
+      "grad_norm": 2.0327038764953613,
+      "learning_rate": 0.00026172839506172837,
+      "loss": 3.6654,
+      "step": 290
+    },
+    {
+      "epoch": 0.6455906821963394,
+      "grad_norm": 1.9252591133117676,
+      "learning_rate": 0.00026148148148148146,
+      "loss": 4.0145,
+      "step": 291
+    },
+    {
+      "epoch": 0.6478092068774265,
+      "grad_norm": 1.742916464805603,
+      "learning_rate": 0.00026123456790123456,
+      "loss": 4.0427,
+      "step": 292
+    },
+    {
+      "epoch": 0.6500277315585136,
+      "grad_norm": 2.29201340675354,
+      "learning_rate": 0.0002609876543209876,
+      "loss": 4.3929,
+      "step": 293
+    },
+    {
+      "epoch": 0.6522462562396006,
+      "grad_norm": 1.54490327835083,
+      "learning_rate": 0.0002607407407407407,
+      "loss": 3.8272,
+      "step": 294
+    },
+    {
+      "epoch": 0.6544647809206877,
+      "grad_norm": 1.9358924627304077,
+      "learning_rate": 0.00026049382716049384,
+      "loss": 3.8539,
+      "step": 295
+    },
+    {
+      "epoch": 0.6566833056017748,
+      "grad_norm": 1.7571736574172974,
+      "learning_rate": 0.0002602469135802469,
+      "loss": 3.9192,
+      "step": 296
+    },
+    {
+      "epoch": 0.6589018302828619,
+      "grad_norm": 1.6838608980178833,
+      "learning_rate": 0.00026,
+      "loss": 3.6839,
+      "step": 297
+    },
+    {
+      "epoch": 0.6611203549639489,
+      "grad_norm": 1.9051893949508667,
+      "learning_rate": 0.00025975308641975307,
+      "loss": 3.916,
+      "step": 298
+    },
+    {
+      "epoch": 0.663338879645036,
+      "grad_norm": 1.9342142343521118,
+      "learning_rate": 0.00025950617283950617,
+      "loss": 4.1523,
+      "step": 299
+    },
+    {
+      "epoch": 0.6655574043261231,
+      "grad_norm": 1.937536358833313,
+      "learning_rate": 0.0002592592592592592,
+      "loss": 3.9713,
+      "step": 300
+    },
+    {
+      "epoch": 0.6677759290072102,
+      "grad_norm": 1.693724513053894,
+      "learning_rate": 0.0002590123456790123,
+      "loss": 3.5792,
+      "step": 301
+    },
+    {
+      "epoch": 0.6699944536882972,
+      "grad_norm": 1.59309720993042,
+      "learning_rate": 0.0002587654320987654,
+      "loss": 3.9344,
+      "step": 302
+    },
+    {
+      "epoch": 0.6722129783693843,
+      "grad_norm": 1.6973621845245361,
+      "learning_rate": 0.0002585185185185185,
+      "loss": 3.8858,
+      "step": 303
+    },
+    {
+      "epoch": 0.6744315030504714,
+      "grad_norm": 1.5380679368972778,
+      "learning_rate": 0.0002582716049382716,
+      "loss": 3.9961,
+      "step": 304
+    },
+    {
+      "epoch": 0.6766500277315585,
+      "grad_norm": 1.792159914970398,
+      "learning_rate": 0.0002580246913580247,
+      "loss": 3.8454,
+      "step": 305
+    },
+    {
+      "epoch": 0.6788685524126455,
+      "grad_norm": 2.0554516315460205,
+      "learning_rate": 0.0002577777777777778,
+      "loss": 3.8946,
+      "step": 306
+    },
+    {
+      "epoch": 0.6810870770937326,
+      "grad_norm": 1.6313352584838867,
+      "learning_rate": 0.0002575308641975308,
+      "loss": 3.8716,
+      "step": 307
+    },
+    {
+      "epoch": 0.6833056017748197,
+      "grad_norm": 2.1081557273864746,
+      "learning_rate": 0.0002572839506172839,
+      "loss": 3.8315,
+      "step": 308
+    },
+    {
+      "epoch": 0.6855241264559068,
+      "grad_norm": 2.222287654876709,
+      "learning_rate": 0.000257037037037037,
+      "loss": 3.7603,
+      "step": 309
+    },
+    {
+      "epoch": 0.687742651136994,
+      "grad_norm": 1.8790770769119263,
+      "learning_rate": 0.0002567901234567901,
+      "loss": 3.6244,
+      "step": 310
+    },
+    {
+      "epoch": 0.689961175818081,
+      "grad_norm": 2.087963819503784,
+      "learning_rate": 0.0002565432098765432,
+      "loss": 3.9021,
+      "step": 311
+    },
+    {
+      "epoch": 0.6921797004991681,
+      "grad_norm": 1.7430325746536255,
+      "learning_rate": 0.00025629629629629624,
+      "loss": 3.9752,
+      "step": 312
+    },
+    {
+      "epoch": 0.6943982251802552,
+      "grad_norm": 2.196620464324951,
+      "learning_rate": 0.0002560493827160494,
+      "loss": 3.9706,
+      "step": 313
+    },
+    {
+      "epoch": 0.6966167498613423,
+      "grad_norm": 1.7266216278076172,
+      "learning_rate": 0.00025580246913580243,
+      "loss": 3.7608,
+      "step": 314
+    },
+    {
+      "epoch": 0.6988352745424293,
+      "grad_norm": 1.6657509803771973,
+      "learning_rate": 0.00025555555555555553,
+      "loss": 3.762,
+      "step": 315
+    },
+    {
+      "epoch": 0.7010537992235164,
+      "grad_norm": 1.6360714435577393,
+      "learning_rate": 0.0002553086419753086,
+      "loss": 3.7171,
+      "step": 316
+    },
+    {
+      "epoch": 0.7032723239046035,
+      "grad_norm": 2.104687213897705,
+      "learning_rate": 0.0002550617283950617,
+      "loss": 3.8857,
+      "step": 317
+    },
+    {
+      "epoch": 0.7054908485856906,
+      "grad_norm": 1.701238989830017,
+      "learning_rate": 0.0002548148148148148,
+      "loss": 3.7577,
+      "step": 318
+    },
+    {
+      "epoch": 0.7077093732667776,
+      "grad_norm": 2.0487329959869385,
+      "learning_rate": 0.00025456790123456785,
+      "loss": 3.8332,
+      "step": 319
+    },
+    {
+      "epoch": 0.7099278979478647,
+      "grad_norm": 2.336130380630493,
+      "learning_rate": 0.00025432098765432095,
+      "loss": 3.7537,
+      "step": 320
+    },
+    {
+      "epoch": 0.7121464226289518,
+      "grad_norm": 1.8241755962371826,
+      "learning_rate": 0.00025407407407407404,
+      "loss": 3.7815,
+      "step": 321
+    },
+    {
+      "epoch": 0.7143649473100389,
+      "grad_norm": 2.087394952774048,
+      "learning_rate": 0.00025382716049382714,
+      "loss": 3.8686,
+      "step": 322
+    },
+    {
+      "epoch": 0.7165834719911259,
+      "grad_norm": 2.0695674419403076,
+      "learning_rate": 0.00025358024691358023,
+      "loss": 3.8324,
+      "step": 323
+    },
+    {
+      "epoch": 0.718801996672213,
+      "grad_norm": 1.644977331161499,
+      "learning_rate": 0.00025333333333333333,
+      "loss": 3.8152,
+      "step": 324
+    },
+    {
+      "epoch": 0.7210205213533001,
+      "grad_norm": 2.1262829303741455,
+      "learning_rate": 0.0002530864197530864,
+      "loss": 3.5861,
+      "step": 325
+    },
+    {
+      "epoch": 0.7232390460343872,
+      "grad_norm": 1.9746931791305542,
+      "learning_rate": 0.00025283950617283947,
+      "loss": 3.8805,
+      "step": 326
+    },
+    {
+      "epoch": 0.7254575707154742,
+      "grad_norm": 1.766104817390442,
+      "learning_rate": 0.00025259259259259256,
+      "loss": 4.0525,
+      "step": 327
+    },
+    {
+      "epoch": 0.7276760953965613,
+      "grad_norm": 2.2982017993927,
+      "learning_rate": 0.00025234567901234566,
+      "loss": 3.8775,
+      "step": 328
+    },
+    {
+      "epoch": 0.7298946200776484,
+      "grad_norm": 2.9903883934020996,
+      "learning_rate": 0.00025209876543209875,
+      "loss": 3.9337,
+      "step": 329
+    },
+    {
+      "epoch": 0.7321131447587355,
+      "grad_norm": 2.0395615100860596,
+      "learning_rate": 0.0002518518518518518,
+      "loss": 3.7573,
+      "step": 330
+    },
+    {
+      "epoch": 0.7343316694398225,
+      "grad_norm": 2.1870615482330322,
+      "learning_rate": 0.00025160493827160494,
+      "loss": 4.0279,
+      "step": 331
+    },
+    {
+      "epoch": 0.7365501941209096,
+      "grad_norm": 2.0698463916778564,
+      "learning_rate": 0.00025135802469135804,
+      "loss": 4.0187,
+      "step": 332
+    },
+    {
+      "epoch": 0.7387687188019967,
+      "grad_norm": 1.7354984283447266,
+      "learning_rate": 0.0002511111111111111,
+      "loss": 3.5475,
+      "step": 333
+    },
+    {
+      "epoch": 0.7409872434830838,
+      "grad_norm": 1.9717432260513306,
+      "learning_rate": 0.00025086419753086417,
+      "loss": 3.8116,
+      "step": 334
+    },
+    {
+      "epoch": 0.7432057681641708,
+      "grad_norm": 1.7499696016311646,
+      "learning_rate": 0.00025061728395061727,
+      "loss": 3.6632,
+      "step": 335
+    },
+    {
+      "epoch": 0.7454242928452579,
+      "grad_norm": 1.759604573249817,
+      "learning_rate": 0.00025037037037037036,
+      "loss": 3.6736,
+      "step": 336
+    },
+    {
+      "epoch": 0.747642817526345,
+      "grad_norm": 2.310689926147461,
+      "learning_rate": 0.0002501234567901234,
+      "loss": 4.2067,
+      "step": 337
+    },
+    {
+      "epoch": 0.7498613422074321,
+      "grad_norm": 2.0818495750427246,
+      "learning_rate": 0.0002498765432098765,
+      "loss": 3.8338,
+      "step": 338
+    },
+    {
+      "epoch": 0.7520798668885191,
+      "grad_norm": 2.7672030925750732,
+      "learning_rate": 0.00024962962962962965,
+      "loss": 4.1034,
+      "step": 339
+    },
+    {
+      "epoch": 0.7542983915696062,
+      "grad_norm": 2.0670907497406006,
+      "learning_rate": 0.0002493827160493827,
+      "loss": 3.9661,
+      "step": 340
+    },
+    {
+      "epoch": 0.7565169162506933,
+      "grad_norm": 1.8684613704681396,
+      "learning_rate": 0.0002491358024691358,
+      "loss": 3.6111,
+      "step": 341
+    },
+    {
+      "epoch": 0.7587354409317804,
+      "grad_norm": 2.196716547012329,
+      "learning_rate": 0.0002488888888888889,
+      "loss": 3.4564,
+      "step": 342
+    },
+    {
+      "epoch": 0.7609539656128674,
+      "grad_norm": 1.9742563962936401,
+      "learning_rate": 0.000248641975308642,
+      "loss": 3.8895,
+      "step": 343
+    },
+    {
+      "epoch": 0.7631724902939545,
+      "grad_norm": 2.3773491382598877,
+      "learning_rate": 0.000248395061728395,
+      "loss": 3.9441,
+      "step": 344
+    },
+    {
+      "epoch": 0.7653910149750416,
+      "grad_norm": 1.6018527746200562,
+      "learning_rate": 0.0002481481481481481,
+      "loss": 3.9563,
+      "step": 345
+    },
+    {
+      "epoch": 0.7676095396561287,
+      "grad_norm": 1.805140495300293,
+      "learning_rate": 0.0002479012345679012,
+      "loss": 3.831,
+      "step": 346
+    },
+    {
+      "epoch": 0.7698280643372157,
+      "grad_norm": 2.320936918258667,
+      "learning_rate": 0.0002476543209876543,
+      "loss": 3.916,
+      "step": 347
+    },
+    {
+      "epoch": 0.7720465890183028,
+      "grad_norm": 1.9561166763305664,
+      "learning_rate": 0.0002474074074074074,
+      "loss": 3.9256,
+      "step": 348
+    },
+    {
+      "epoch": 0.7742651136993899,
+      "grad_norm": 1.6357280015945435,
+      "learning_rate": 0.0002471604938271605,
+      "loss": 3.7201,
+      "step": 349
+    },
+    {
+      "epoch": 0.776483638380477,
+      "grad_norm": 1.9405951499938965,
+      "learning_rate": 0.0002469135802469136,
+      "loss": 3.6588,
+      "step": 350
+    },
+    {
+      "epoch": 0.778702163061564,
+      "grad_norm": 2.242642879486084,
+      "learning_rate": 0.0002466666666666666,
+      "loss": 3.7395,
+      "step": 351
+    },
+    {
+      "epoch": 0.7809206877426511,
+      "grad_norm": 1.9238040447235107,
+      "learning_rate": 0.0002464197530864197,
+      "loss": 3.6561,
+      "step": 352
+    },
+    {
+      "epoch": 0.7831392124237382,
+      "grad_norm": 1.9532653093338013,
+      "learning_rate": 0.0002461728395061728,
+      "loss": 3.7678,
+      "step": 353
+    },
+    {
+      "epoch": 0.7853577371048253,
+      "grad_norm": 2.5075628757476807,
+      "learning_rate": 0.0002459259259259259,
+      "loss": 3.9553,
+      "step": 354
+    },
+    {
+      "epoch": 0.7875762617859123,
+      "grad_norm": 2.0460591316223145,
+      "learning_rate": 0.000245679012345679,
+      "loss": 3.8247,
+      "step": 355
+    },
+    {
+      "epoch": 0.7897947864669994,
+      "grad_norm": 2.22599458694458,
+      "learning_rate": 0.00024543209876543205,
+      "loss": 3.9775,
+      "step": 356
+    },
+    {
+      "epoch": 0.7920133111480865,
+      "grad_norm": 1.940339207649231,
+      "learning_rate": 0.0002451851851851852,
+      "loss": 3.5976,
+      "step": 357
+    },
+    {
+      "epoch": 0.7942318358291736,
+      "grad_norm": 1.6763797998428345,
+      "learning_rate": 0.00024493827160493824,
+      "loss": 3.6762,
+      "step": 358
+    },
+    {
+      "epoch": 0.7964503605102606,
+      "grad_norm": 1.7849125862121582,
+      "learning_rate": 0.00024469135802469133,
+      "loss": 3.6066,
+      "step": 359
+    },
+    {
+      "epoch": 0.7986688851913477,
+      "grad_norm": 2.3369338512420654,
+      "learning_rate": 0.00024444444444444443,
+      "loss": 3.818,
+      "step": 360
+    },
+    {
+      "epoch": 0.8008874098724348,
+      "grad_norm": 2.7951653003692627,
+      "learning_rate": 0.0002441975308641975,
+      "loss": 3.4348,
+      "step": 361
+    },
+    {
+      "epoch": 0.8031059345535219,
+      "grad_norm": 1.8764299154281616,
+      "learning_rate": 0.00024395061728395062,
+      "loss": 3.8107,
+      "step": 362
+    },
+    {
+      "epoch": 0.8053244592346089,
+      "grad_norm": 2.4066555500030518,
+      "learning_rate": 0.0002437037037037037,
+      "loss": 3.9366,
+      "step": 363
+    },
+    {
+      "epoch": 0.807542983915696,
+      "grad_norm": 2.0674121379852295,
+      "learning_rate": 0.00024345679012345678,
+      "loss": 3.9545,
+      "step": 364
+    },
+    {
+      "epoch": 0.8097615085967831,
+      "grad_norm": 2.4179327487945557,
+      "learning_rate": 0.00024320987654320985,
+      "loss": 3.729,
+      "step": 365
+    },
+    {
+      "epoch": 0.8119800332778702,
+      "grad_norm": 1.7643181085586548,
+      "learning_rate": 0.00024296296296296295,
+      "loss": 3.4305,
+      "step": 366
+    },
+    {
+      "epoch": 0.8141985579589573,
+      "grad_norm": 1.8883612155914307,
+      "learning_rate": 0.00024271604938271601,
+      "loss": 4.0384,
+      "step": 367
+    },
+    {
+      "epoch": 0.8164170826400444,
+      "grad_norm": 2.543565273284912,
+      "learning_rate": 0.0002424691358024691,
+      "loss": 3.7009,
+      "step": 368
+    },
+    {
+      "epoch": 0.8186356073211315,
+      "grad_norm": 1.8298462629318237,
+      "learning_rate": 0.0002422222222222222,
+      "loss": 3.7922,
+      "step": 369
+    },
+    {
+      "epoch": 0.8208541320022186,
+      "grad_norm": 2.5235342979431152,
+      "learning_rate": 0.00024197530864197527,
+      "loss": 3.7656,
+      "step": 370
+    },
+    {
+      "epoch": 0.8230726566833056,
+      "grad_norm": 1.934751033782959,
+      "learning_rate": 0.0002417283950617284,
+      "loss": 3.6065,
+      "step": 371
+    },
+    {
+      "epoch": 0.8252911813643927,
+      "grad_norm": 2.3010201454162598,
+      "learning_rate": 0.00024148148148148146,
+      "loss": 4.0299,
+      "step": 372
+    },
+    {
+      "epoch": 0.8275097060454798,
+      "grad_norm": 2.897660732269287,
+      "learning_rate": 0.00024123456790123456,
+      "loss": 3.8033,
+      "step": 373
+    },
+    {
+      "epoch": 0.8297282307265669,
+      "grad_norm": 2.6275532245635986,
+      "learning_rate": 0.00024098765432098763,
+      "loss": 3.6599,
+      "step": 374
+    },
+    {
+      "epoch": 0.831946755407654,
+      "grad_norm": 2.082494020462036,
+      "learning_rate": 0.00024074074074074072,
+      "loss": 3.8505,
+      "step": 375
+    },
+    {
+      "epoch": 0.834165280088741,
+      "grad_norm": 2.1602790355682373,
+      "learning_rate": 0.0002404938271604938,
+      "loss": 4.0338,
+      "step": 376
+    },
+    {
+      "epoch": 0.8363838047698281,
+      "grad_norm": 2.147921085357666,
+      "learning_rate": 0.00024024691358024688,
+      "loss": 3.792,
+      "step": 377
+    },
+    {
+      "epoch": 0.8386023294509152,
+      "grad_norm": 1.9606894254684448,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 3.9145,
+      "step": 378
+    },
+    {
+      "epoch": 0.8408208541320022,
+      "grad_norm": 2.477431058883667,
+      "learning_rate": 0.00023975308641975305,
+      "loss": 3.7006,
+      "step": 379
+    },
+    {
+      "epoch": 0.8430393788130893,
+      "grad_norm": 2.3010034561157227,
+      "learning_rate": 0.00023950617283950617,
+      "loss": 3.7402,
+      "step": 380
+    },
+    {
+      "epoch": 0.8452579034941764,
+      "grad_norm": 1.788385033607483,
+      "learning_rate": 0.00023925925925925924,
+      "loss": 3.807,
+      "step": 381
+    },
+    {
+      "epoch": 0.8474764281752635,
+      "grad_norm": 2.380499839782715,
+      "learning_rate": 0.00023901234567901233,
+      "loss": 3.5715,
+      "step": 382
+    },
+    {
+      "epoch": 0.8496949528563505,
+      "grad_norm": 2.3576323986053467,
+      "learning_rate": 0.0002387654320987654,
+      "loss": 3.5248,
+      "step": 383
+    },
+    {
+      "epoch": 0.8519134775374376,
+      "grad_norm": 2.0316405296325684,
+      "learning_rate": 0.0002385185185185185,
+      "loss": 3.5777,
+      "step": 384
+    },
+    {
+      "epoch": 0.8541320022185247,
+      "grad_norm": 2.2353413105010986,
+      "learning_rate": 0.0002382716049382716,
+      "loss": 3.6328,
+      "step": 385
+    },
+    {
+      "epoch": 0.8563505268996118,
+      "grad_norm": 3.051893949508667,
+      "learning_rate": 0.00023802469135802466,
+      "loss": 3.6907,
+      "step": 386
+    },
+    {
+      "epoch": 0.8585690515806988,
+      "grad_norm": 2.533369541168213,
+      "learning_rate": 0.00023777777777777775,
+      "loss": 3.8874,
+      "step": 387
+    },
+    {
+      "epoch": 0.8607875762617859,
+      "grad_norm": 2.1060948371887207,
+      "learning_rate": 0.00023753086419753082,
+      "loss": 3.8401,
+      "step": 388
+    },
+    {
+      "epoch": 0.863006100942873,
+      "grad_norm": 3.269286632537842,
+      "learning_rate": 0.00023728395061728394,
+      "loss": 3.8233,
+      "step": 389
+    },
+    {
+      "epoch": 0.8652246256239601,
+      "grad_norm": 2.598876953125,
+      "learning_rate": 0.000237037037037037,
+      "loss": 3.7946,
+      "step": 390
+    },
+    {
+      "epoch": 0.8674431503050472,
+      "grad_norm": 2.306293487548828,
+      "learning_rate": 0.0002367901234567901,
+      "loss": 3.6502,
+      "step": 391
+    },
+    {
+      "epoch": 0.8696616749861342,
+      "grad_norm": 2.005988597869873,
+      "learning_rate": 0.0002365432098765432,
+      "loss": 3.6242,
+      "step": 392
+    },
+    {
+      "epoch": 0.8718801996672213,
+      "grad_norm": 1.747209072113037,
+      "learning_rate": 0.00023629629629629627,
+      "loss": 3.7103,
+      "step": 393
+    },
+    {
+      "epoch": 0.8740987243483084,
+      "grad_norm": 2.0476126670837402,
+      "learning_rate": 0.00023604938271604937,
+      "loss": 3.9791,
+      "step": 394
+    },
+    {
+      "epoch": 0.8763172490293955,
+      "grad_norm": 2.8112363815307617,
+      "learning_rate": 0.00023580246913580243,
+      "loss": 3.5168,
+      "step": 395
+    },
+    {
+      "epoch": 0.8785357737104825,
+      "grad_norm": 2.324246644973755,
+      "learning_rate": 0.00023555555555555553,
+      "loss": 3.7658,
+      "step": 396
+    },
+    {
+      "epoch": 0.8807542983915696,
+      "grad_norm": 1.837265133857727,
+      "learning_rate": 0.0002353086419753086,
+      "loss": 3.7598,
+      "step": 397
+    },
+    {
+      "epoch": 0.8829728230726567,
+      "grad_norm": 2.2585606575012207,
+      "learning_rate": 0.00023506172839506172,
+      "loss": 3.7062,
+      "step": 398
+    },
+    {
+      "epoch": 0.8851913477537438,
+      "grad_norm": 2.2909350395202637,
+      "learning_rate": 0.0002348148148148148,
+      "loss": 3.9043,
+      "step": 399
+    },
+    {
+      "epoch": 0.8874098724348308,
+      "grad_norm": 1.9328033924102783,
+      "learning_rate": 0.00023456790123456788,
+      "loss": 3.9438,
+      "step": 400
+    },
+    {
+      "epoch": 0.8896283971159179,
+      "grad_norm": 2.0291390419006348,
+      "learning_rate": 0.00023432098765432098,
+      "loss": 3.2591,
+      "step": 401
+    },
+    {
+      "epoch": 0.891846921797005,
+      "grad_norm": 2.007695198059082,
+      "learning_rate": 0.00023407407407407405,
+      "loss": 3.7301,
+      "step": 402
+    },
+    {
+      "epoch": 0.894065446478092,
+      "grad_norm": 2.1139297485351562,
+      "learning_rate": 0.00023382716049382714,
+      "loss": 3.7279,
+      "step": 403
+    },
+    {
+      "epoch": 0.8962839711591791,
+      "grad_norm": 2.6387417316436768,
+      "learning_rate": 0.0002335802469135802,
+      "loss": 3.7559,
+      "step": 404
+    },
+    {
+      "epoch": 0.8985024958402662,
+      "grad_norm": 1.958734154701233,
+      "learning_rate": 0.0002333333333333333,
+      "loss": 3.7864,
+      "step": 405
+    },
+    {
+      "epoch": 0.9007210205213533,
+      "grad_norm": 1.752972960472107,
+      "learning_rate": 0.00023308641975308637,
+      "loss": 3.5035,
+      "step": 406
+    },
+    {
+      "epoch": 0.9029395452024404,
+      "grad_norm": 2.962470054626465,
+      "learning_rate": 0.0002328395061728395,
+      "loss": 3.85,
+      "step": 407
+    },
+    {
+      "epoch": 0.9051580698835274,
+      "grad_norm": 2.3236653804779053,
+      "learning_rate": 0.0002325925925925926,
+      "loss": 3.7087,
+      "step": 408
+    },
+    {
+      "epoch": 0.9073765945646145,
+      "grad_norm": 1.760456919670105,
+      "learning_rate": 0.00023234567901234566,
+      "loss": 3.8864,
+      "step": 409
+    },
+    {
+      "epoch": 0.9095951192457016,
+      "grad_norm": 2.13686466217041,
+      "learning_rate": 0.00023209876543209875,
+      "loss": 3.4982,
+      "step": 410
+    },
+    {
+      "epoch": 0.9118136439267887,
+      "grad_norm": 1.9516112804412842,
+      "learning_rate": 0.00023185185185185182,
+      "loss": 3.5933,
+      "step": 411
+    },
+    {
+      "epoch": 0.9140321686078757,
+      "grad_norm": 1.7974390983581543,
+      "learning_rate": 0.00023160493827160492,
+      "loss": 3.6131,
+      "step": 412
+    },
+    {
+      "epoch": 0.9162506932889628,
+      "grad_norm": 2.308530569076538,
+      "learning_rate": 0.00023135802469135798,
+      "loss": 3.5728,
+      "step": 413
+    },
+    {
+      "epoch": 0.9184692179700499,
+      "grad_norm": 3.0478460788726807,
+      "learning_rate": 0.00023111111111111108,
+      "loss": 3.7106,
+      "step": 414
+    },
+    {
+      "epoch": 0.920687742651137,
+      "grad_norm": 1.8502068519592285,
+      "learning_rate": 0.0002308641975308642,
+      "loss": 3.6382,
+      "step": 415
+    },
+    {
+      "epoch": 0.922906267332224,
+      "grad_norm": 1.8834432363510132,
+      "learning_rate": 0.00023061728395061727,
+      "loss": 3.338,
+      "step": 416
+    },
+    {
+      "epoch": 0.9251247920133111,
+      "grad_norm": 2.6639201641082764,
+      "learning_rate": 0.00023037037037037036,
+      "loss": 3.8096,
+      "step": 417
+    },
+    {
+      "epoch": 0.9273433166943982,
+      "grad_norm": 1.9164607524871826,
+      "learning_rate": 0.00023012345679012343,
+      "loss": 3.7255,
+      "step": 418
+    },
+    {
+      "epoch": 0.9295618413754853,
+      "grad_norm": 2.363600969314575,
+      "learning_rate": 0.00022987654320987653,
+      "loss": 3.883,
+      "step": 419
+    },
+    {
+      "epoch": 0.9317803660565723,
+      "grad_norm": 2.1003379821777344,
+      "learning_rate": 0.0002296296296296296,
+      "loss": 3.7904,
+      "step": 420
+    },
+    {
+      "epoch": 0.9339988907376594,
+      "grad_norm": 1.729519248008728,
+      "learning_rate": 0.0002293827160493827,
+      "loss": 3.5897,
+      "step": 421
+    },
+    {
+      "epoch": 0.9362174154187465,
+      "grad_norm": 1.8703901767730713,
+      "learning_rate": 0.00022913580246913579,
+      "loss": 3.9071,
+      "step": 422
+    },
+    {
+      "epoch": 0.9384359400998337,
+      "grad_norm": 1.8352406024932861,
+      "learning_rate": 0.00022888888888888885,
+      "loss": 3.8149,
+      "step": 423
+    },
+    {
+      "epoch": 0.9406544647809207,
+      "grad_norm": 1.9813724756240845,
+      "learning_rate": 0.00022864197530864198,
+      "loss": 3.77,
+      "step": 424
+    },
+    {
+      "epoch": 0.9428729894620078,
+      "grad_norm": 1.8527830839157104,
+      "learning_rate": 0.00022839506172839504,
+      "loss": 3.665,
+      "step": 425
+    },
+    {
+      "epoch": 0.9450915141430949,
+      "grad_norm": 2.5864765644073486,
+      "learning_rate": 0.00022814814814814814,
+      "loss": 3.7857,
+      "step": 426
+    },
+    {
+      "epoch": 0.947310038824182,
+      "grad_norm": 2.4557650089263916,
+      "learning_rate": 0.0002279012345679012,
+      "loss": 3.7162,
+      "step": 427
+    },
+    {
+      "epoch": 0.949528563505269,
+      "grad_norm": 2.725149154663086,
+      "learning_rate": 0.0002276543209876543,
+      "loss": 3.7143,
+      "step": 428
+    },
+    {
+      "epoch": 0.9517470881863561,
+      "grad_norm": 2.3531994819641113,
+      "learning_rate": 0.00022740740740740737,
+      "loss": 3.716,
+      "step": 429
+    },
+    {
+      "epoch": 0.9539656128674432,
+      "grad_norm": 2.046536445617676,
+      "learning_rate": 0.00022716049382716047,
+      "loss": 3.5775,
+      "step": 430
+    },
+    {
+      "epoch": 0.9561841375485303,
+      "grad_norm": 1.9661775827407837,
+      "learning_rate": 0.00022691358024691356,
+      "loss": 3.8181,
+      "step": 431
+    },
+    {
+      "epoch": 0.9584026622296173,
+      "grad_norm": 2.3551533222198486,
+      "learning_rate": 0.00022666666666666663,
+      "loss": 3.6057,
+      "step": 432
+    },
+    {
+      "epoch": 0.9606211869107044,
+      "grad_norm": 2.0724599361419678,
+      "learning_rate": 0.00022641975308641975,
+      "loss": 3.6135,
+      "step": 433
+    },
+    {
+      "epoch": 0.9628397115917915,
+      "grad_norm": 2.265535831451416,
+      "learning_rate": 0.00022617283950617282,
+      "loss": 3.9045,
+      "step": 434
+    },
+    {
+      "epoch": 0.9650582362728786,
+      "grad_norm": 1.961011528968811,
+      "learning_rate": 0.00022592592592592591,
+      "loss": 3.3242,
+      "step": 435
+    },
+    {
+      "epoch": 0.9672767609539656,
+      "grad_norm": 2.3468997478485107,
+      "learning_rate": 0.00022567901234567898,
+      "loss": 3.8722,
+      "step": 436
+    },
+    {
+      "epoch": 0.9694952856350527,
+      "grad_norm": 2.2423624992370605,
+      "learning_rate": 0.00022543209876543208,
+      "loss": 3.6836,
+      "step": 437
+    },
+    {
+      "epoch": 0.9717138103161398,
+      "grad_norm": 1.866023302078247,
+      "learning_rate": 0.00022518518518518517,
+      "loss": 3.4932,
+      "step": 438
+    },
+    {
+      "epoch": 0.9739323349972269,
+      "grad_norm": 2.060452699661255,
+      "learning_rate": 0.00022493827160493824,
+      "loss": 3.61,
+      "step": 439
+    },
+    {
+      "epoch": 0.9761508596783139,
+      "grad_norm": 1.9689525365829468,
+      "learning_rate": 0.00022469135802469134,
+      "loss": 3.7059,
+      "step": 440
+    },
+    {
+      "epoch": 0.978369384359401,
+      "grad_norm": 1.7846275568008423,
+      "learning_rate": 0.0002244444444444444,
+      "loss": 3.4759,
+      "step": 441
+    },
+    {
+      "epoch": 0.9805879090404881,
+      "grad_norm": 1.9932984113693237,
+      "learning_rate": 0.00022419753086419753,
+      "loss": 3.3919,
+      "step": 442
+    },
+    {
+      "epoch": 0.9828064337215752,
+      "grad_norm": 2.155195713043213,
+      "learning_rate": 0.0002239506172839506,
+      "loss": 3.9522,
+      "step": 443
+    },
+    {
+      "epoch": 0.9850249584026622,
+      "grad_norm": 2.1634106636047363,
+      "learning_rate": 0.0002237037037037037,
+      "loss": 3.6407,
+      "step": 444
+    },
+    {
+      "epoch": 0.9872434830837493,
+      "grad_norm": 2.207467794418335,
+      "learning_rate": 0.00022345679012345678,
+      "loss": 3.6569,
+      "step": 445
+    },
+    {
+      "epoch": 0.9894620077648364,
+      "grad_norm": 2.3168647289276123,
+      "learning_rate": 0.00022320987654320985,
+      "loss": 3.6652,
+      "step": 446
+    },
+    {
+      "epoch": 0.9916805324459235,
+      "grad_norm": 2.6183316707611084,
+      "learning_rate": 0.00022296296296296295,
+      "loss": 3.3817,
+      "step": 447
+    },
+    {
+      "epoch": 0.9938990571270105,
+      "grad_norm": 2.310850143432617,
+      "learning_rate": 0.00022271604938271602,
+      "loss": 3.799,
+      "step": 448
+    },
+    {
+      "epoch": 0.9961175818080976,
+      "grad_norm": 2.266427516937256,
+      "learning_rate": 0.0002224691358024691,
+      "loss": 3.4738,
+      "step": 449
+    },
+    {
+      "epoch": 0.9983361064891847,
+      "grad_norm": 3.286635637283325,
+      "learning_rate": 0.00022222222222222218,
+      "loss": 3.4088,
+      "step": 450
+    },
+    {
+      "epoch": 1.0005546311702718,
+      "grad_norm": 2.8101956844329834,
+      "learning_rate": 0.0002219753086419753,
+      "loss": 3.9166,
+      "step": 451
+    },
+    {
+      "epoch": 1.0027731558513588,
+      "grad_norm": 2.1921589374542236,
+      "learning_rate": 0.0002217283950617284,
+      "loss": 3.5494,
+      "step": 452
+    },
+    {
+      "epoch": 1.004991680532446,
+      "grad_norm": 3.424365758895874,
+      "learning_rate": 0.00022148148148148146,
+      "loss": 3.7616,
+      "step": 453
+    },
+    {
+      "epoch": 1.007210205213533,
+      "grad_norm": 3.3539986610412598,
+      "learning_rate": 0.00022123456790123456,
+      "loss": 3.7178,
+      "step": 454
+    },
+    {
+      "epoch": 1.00942872989462,
+      "grad_norm": 2.2471249103546143,
+      "learning_rate": 0.00022098765432098763,
+      "loss": 3.6495,
+      "step": 455
+    },
+    {
+      "epoch": 1.0116472545757071,
+      "grad_norm": 1.9215383529663086,
+      "learning_rate": 0.00022074074074074072,
+      "loss": 3.5047,
+      "step": 456
+    },
+    {
+      "epoch": 1.0138657792567942,
+      "grad_norm": 2.5212655067443848,
+      "learning_rate": 0.0002204938271604938,
+      "loss": 3.3266,
+      "step": 457
+    },
+    {
+      "epoch": 1.0160843039378813,
+      "grad_norm": 2.485941171646118,
+      "learning_rate": 0.00022024691358024689,
+      "loss": 3.3888,
+      "step": 458
+    },
+    {
+      "epoch": 1.0183028286189684,
+      "grad_norm": 2.6323788166046143,
+      "learning_rate": 0.00021999999999999995,
+      "loss": 3.5533,
+      "step": 459
+    },
+    {
+      "epoch": 1.0205213533000554,
+      "grad_norm": 1.894006609916687,
+      "learning_rate": 0.00021975308641975308,
+      "loss": 3.3077,
+      "step": 460
+    },
+    {
+      "epoch": 1.0227398779811425,
+      "grad_norm": 2.4571902751922607,
+      "learning_rate": 0.00021950617283950617,
+      "loss": 3.5333,
+      "step": 461
+    },
+    {
+      "epoch": 1.0249584026622296,
+      "grad_norm": 2.3834753036499023,
+      "learning_rate": 0.00021925925925925924,
+      "loss": 3.7493,
+      "step": 462
+    },
+    {
+      "epoch": 1.0271769273433167,
+      "grad_norm": 2.5333776473999023,
+      "learning_rate": 0.00021901234567901233,
+      "loss": 3.5995,
+      "step": 463
+    },
+    {
+      "epoch": 1.0293954520244037,
+      "grad_norm": 1.917075753211975,
+      "learning_rate": 0.0002187654320987654,
+      "loss": 3.4319,
+      "step": 464
+    },
+    {
+      "epoch": 1.0316139767054908,
+      "grad_norm": 3.404819965362549,
+      "learning_rate": 0.0002185185185185185,
+      "loss": 3.8731,
+      "step": 465
+    },
+    {
+      "epoch": 1.033832501386578,
+      "grad_norm": 2.7770936489105225,
+      "learning_rate": 0.00021827160493827157,
+      "loss": 3.6262,
+      "step": 466
+    },
+    {
+      "epoch": 1.036051026067665,
+      "grad_norm": 2.771172285079956,
+      "learning_rate": 0.00021802469135802466,
+      "loss": 3.712,
+      "step": 467
+    },
+    {
+      "epoch": 1.038269550748752,
+      "grad_norm": 2.0420303344726562,
+      "learning_rate": 0.00021777777777777778,
+      "loss": 3.3268,
+      "step": 468
+    },
+    {
+      "epoch": 1.0404880754298391,
+      "grad_norm": 1.9546196460723877,
+      "learning_rate": 0.00021753086419753085,
+      "loss": 3.473,
+      "step": 469
+    },
+    {
+      "epoch": 1.0427066001109262,
+      "grad_norm": 1.6977958679199219,
+      "learning_rate": 0.00021728395061728395,
+      "loss": 3.4584,
+      "step": 470
+    },
+    {
+      "epoch": 1.0449251247920133,
+      "grad_norm": 1.8450311422348022,
+      "learning_rate": 0.00021703703703703701,
+      "loss": 3.2987,
+      "step": 471
+    },
+    {
+      "epoch": 1.0471436494731003,
+      "grad_norm": 2.408600330352783,
+      "learning_rate": 0.0002167901234567901,
+      "loss": 3.7214,
+      "step": 472
+    },
+    {
+      "epoch": 1.0493621741541874,
+      "grad_norm": 2.0030791759490967,
+      "learning_rate": 0.00021654320987654318,
+      "loss": 3.4893,
+      "step": 473
+    },
+    {
+      "epoch": 1.0515806988352745,
+      "grad_norm": 2.042564630508423,
+      "learning_rate": 0.00021629629629629627,
+      "loss": 3.6945,
+      "step": 474
+    },
+    {
+      "epoch": 1.0537992235163616,
+      "grad_norm": 2.362656593322754,
+      "learning_rate": 0.00021604938271604937,
+      "loss": 3.6413,
+      "step": 475
+    },
+    {
+      "epoch": 1.0560177481974486,
+      "grad_norm": 2.526808023452759,
+      "learning_rate": 0.00021580246913580244,
+      "loss": 3.3709,
+      "step": 476
+    },
+    {
+      "epoch": 1.0582362728785357,
+      "grad_norm": 2.0582213401794434,
+      "learning_rate": 0.00021555555555555556,
+      "loss": 3.1743,
+      "step": 477
+    },
+    {
+      "epoch": 1.0604547975596228,
+      "grad_norm": 1.8329795598983765,
+      "learning_rate": 0.00021530864197530863,
+      "loss": 3.2718,
+      "step": 478
+    },
+    {
+      "epoch": 1.0626733222407099,
+      "grad_norm": 2.077108383178711,
+      "learning_rate": 0.00021506172839506172,
+      "loss": 3.6362,
+      "step": 479
+    },
+    {
+      "epoch": 1.064891846921797,
+      "grad_norm": 2.41853928565979,
+      "learning_rate": 0.0002148148148148148,
+      "loss": 3.7111,
+      "step": 480
+    },
+    {
+      "epoch": 1.067110371602884,
+      "grad_norm": 1.7850521802902222,
+      "learning_rate": 0.00021456790123456788,
+      "loss": 3.3596,
+      "step": 481
+    },
+    {
+      "epoch": 1.069328896283971,
+      "grad_norm": 1.8199137449264526,
+      "learning_rate": 0.00021432098765432095,
+      "loss": 3.383,
+      "step": 482
+    },
+    {
+      "epoch": 1.0715474209650582,
+      "grad_norm": 2.4449188709259033,
+      "learning_rate": 0.00021407407407407405,
+      "loss": 3.9087,
+      "step": 483
+    },
+    {
+      "epoch": 1.0737659456461452,
+      "grad_norm": 2.281856060028076,
+      "learning_rate": 0.00021382716049382714,
+      "loss": 3.6705,
+      "step": 484
+    },
+    {
+      "epoch": 1.0759844703272323,
+      "grad_norm": 2.0235955715179443,
+      "learning_rate": 0.0002135802469135802,
+      "loss": 3.3082,
+      "step": 485
+    },
+    {
+      "epoch": 1.0782029950083194,
+      "grad_norm": 2.29435396194458,
+      "learning_rate": 0.00021333333333333333,
+      "loss": 3.8299,
+      "step": 486
+    },
+    {
+      "epoch": 1.0804215196894065,
+      "grad_norm": 2.587329626083374,
+      "learning_rate": 0.0002130864197530864,
+      "loss": 3.4981,
+      "step": 487
+    },
+    {
+      "epoch": 1.0826400443704935,
+      "grad_norm": 2.1883504390716553,
+      "learning_rate": 0.0002128395061728395,
+      "loss": 3.6346,
+      "step": 488
+    },
+    {
+      "epoch": 1.0848585690515806,
+      "grad_norm": 2.2624077796936035,
+      "learning_rate": 0.00021259259259259256,
+      "loss": 3.9606,
+      "step": 489
+    },
+    {
+      "epoch": 1.0870770937326677,
+      "grad_norm": 2.0024573802948,
+      "learning_rate": 0.00021234567901234566,
+      "loss": 3.6835,
+      "step": 490
+    },
+    {
+      "epoch": 1.0892956184137548,
+      "grad_norm": 2.226426124572754,
+      "learning_rate": 0.00021209876543209875,
+      "loss": 3.4727,
+      "step": 491
+    },
+    {
+      "epoch": 1.0915141430948418,
+      "grad_norm": 2.1525652408599854,
+      "learning_rate": 0.00021185185185185182,
+      "loss": 3.8931,
+      "step": 492
+    },
+    {
+      "epoch": 1.093732667775929,
+      "grad_norm": 2.058469533920288,
+      "learning_rate": 0.00021160493827160492,
+      "loss": 3.679,
+      "step": 493
+    },
+    {
+      "epoch": 1.095951192457016,
+      "grad_norm": 1.67606520652771,
+      "learning_rate": 0.00021135802469135799,
+      "loss": 3.4367,
+      "step": 494
+    },
+    {
+      "epoch": 1.098169717138103,
+      "grad_norm": 1.9282641410827637,
+      "learning_rate": 0.0002111111111111111,
+      "loss": 3.5156,
+      "step": 495
+    },
+    {
+      "epoch": 1.1003882418191901,
+      "grad_norm": 2.3099889755249023,
+      "learning_rate": 0.00021086419753086418,
+      "loss": 3.6036,
+      "step": 496
+    },
+    {
+      "epoch": 1.1026067665002772,
+      "grad_norm": 2.3494391441345215,
+      "learning_rate": 0.00021061728395061727,
+      "loss": 3.7695,
+      "step": 497
+    },
+    {
+      "epoch": 1.1048252911813643,
+      "grad_norm": 1.937668800354004,
+      "learning_rate": 0.00021037037037037037,
+      "loss": 3.4162,
+      "step": 498
+    },
+    {
+      "epoch": 1.1070438158624514,
+      "grad_norm": 2.542306661605835,
+      "learning_rate": 0.00021012345679012343,
+      "loss": 3.5951,
+      "step": 499
+    },
+    {
+      "epoch": 1.1092623405435384,
+      "grad_norm": 2.8260834217071533,
+      "learning_rate": 0.00020987654320987653,
+      "loss": 3.584,
+      "step": 500
+    },
+    {
+      "epoch": 1.1114808652246255,
+      "grad_norm": 2.2400100231170654,
+      "learning_rate": 0.0002096296296296296,
+      "loss": 3.3477,
+      "step": 501
+    },
+    {
+      "epoch": 1.1136993899057126,
+      "grad_norm": 2.702059030532837,
+      "learning_rate": 0.0002093827160493827,
+      "loss": 3.5845,
+      "step": 502
+    },
+    {
+      "epoch": 1.1159179145867997,
+      "grad_norm": 1.894998550415039,
+      "learning_rate": 0.00020913580246913576,
+      "loss": 3.7309,
+      "step": 503
+    },
+    {
+      "epoch": 1.1181364392678868,
+      "grad_norm": 2.58927583694458,
+      "learning_rate": 0.00020888888888888888,
+      "loss": 3.6677,
+      "step": 504
+    },
+    {
+      "epoch": 1.1203549639489738,
+      "grad_norm": 2.125972032546997,
+      "learning_rate": 0.00020864197530864198,
+      "loss": 3.5485,
+      "step": 505
+    },
+    {
+      "epoch": 1.122573488630061,
+      "grad_norm": 2.242764472961426,
+      "learning_rate": 0.00020839506172839505,
+      "loss": 3.3767,
+      "step": 506
+    },
+    {
+      "epoch": 1.124792013311148,
+      "grad_norm": 1.7471697330474854,
+      "learning_rate": 0.00020814814814814814,
+      "loss": 3.8052,
+      "step": 507
+    },
+    {
+      "epoch": 1.127010537992235,
+      "grad_norm": 2.432633638381958,
+      "learning_rate": 0.0002079012345679012,
+      "loss": 3.1642,
+      "step": 508
+    },
+    {
+      "epoch": 1.1292290626733221,
+      "grad_norm": 2.0992937088012695,
+      "learning_rate": 0.0002076543209876543,
+      "loss": 3.4432,
+      "step": 509
+    },
+    {
+      "epoch": 1.1314475873544092,
+      "grad_norm": 2.3110296726226807,
+      "learning_rate": 0.00020740740740740737,
+      "loss": 3.547,
+      "step": 510
+    },
+    {
+      "epoch": 1.1336661120354963,
+      "grad_norm": 1.751617193222046,
+      "learning_rate": 0.00020716049382716047,
+      "loss": 3.8452,
+      "step": 511
+    },
+    {
+      "epoch": 1.1358846367165834,
+      "grad_norm": 2.335127115249634,
+      "learning_rate": 0.00020691358024691354,
+      "loss": 3.5752,
+      "step": 512
+    },
+    {
+      "epoch": 1.1381031613976704,
+      "grad_norm": 1.8864926099777222,
+      "learning_rate": 0.00020666666666666666,
+      "loss": 3.5399,
+      "step": 513
+    },
+    {
+      "epoch": 1.1403216860787575,
+      "grad_norm": 2.28855037689209,
+      "learning_rate": 0.00020641975308641975,
+      "loss": 3.2165,
+      "step": 514
+    },
+    {
+      "epoch": 1.1425402107598448,
+      "grad_norm": 1.9624519348144531,
+      "learning_rate": 0.00020617283950617282,
+      "loss": 3.4134,
+      "step": 515
+    },
+    {
+      "epoch": 1.1447587354409319,
+      "grad_norm": 1.739151120185852,
+      "learning_rate": 0.00020592592592592592,
+      "loss": 3.7197,
+      "step": 516
+    },
+    {
+      "epoch": 1.146977260122019,
+      "grad_norm": 2.13329815864563,
+      "learning_rate": 0.00020567901234567898,
+      "loss": 3.7209,
+      "step": 517
+    },
+    {
+      "epoch": 1.149195784803106,
+      "grad_norm": 2.1573734283447266,
+      "learning_rate": 0.00020543209876543208,
+      "loss": 3.5474,
+      "step": 518
+    },
+    {
+      "epoch": 1.151414309484193,
+      "grad_norm": 2.1982409954071045,
+      "learning_rate": 0.00020518518518518515,
+      "loss": 3.6421,
+      "step": 519
+    },
+    {
+      "epoch": 1.1536328341652802,
+      "grad_norm": 2.392566680908203,
+      "learning_rate": 0.00020493827160493824,
+      "loss": 3.1336,
+      "step": 520
+    },
+    {
+      "epoch": 1.1558513588463672,
+      "grad_norm": 2.3388595581054688,
+      "learning_rate": 0.00020469135802469136,
+      "loss": 3.545,
+      "step": 521
+    },
+    {
+      "epoch": 1.1580698835274543,
+      "grad_norm": 1.9315019845962524,
+      "learning_rate": 0.00020444444444444443,
+      "loss": 3.5957,
+      "step": 522
+    },
+    {
+      "epoch": 1.1602884082085414,
+      "grad_norm": 2.2644240856170654,
+      "learning_rate": 0.00020419753086419753,
+      "loss": 3.6621,
+      "step": 523
+    },
+    {
+      "epoch": 1.1625069328896285,
+      "grad_norm": 1.9546617269515991,
+      "learning_rate": 0.0002039506172839506,
+      "loss": 3.5322,
+      "step": 524
+    },
+    {
+      "epoch": 1.1647254575707155,
+      "grad_norm": 2.28181529045105,
+      "learning_rate": 0.0002037037037037037,
+      "loss": 3.4443,
+      "step": 525
+    },
+    {
+      "epoch": 1.1669439822518026,
+      "grad_norm": 1.9370272159576416,
+      "learning_rate": 0.00020345679012345676,
+      "loss": 3.4521,
+      "step": 526
+    },
+    {
+      "epoch": 1.1691625069328897,
+      "grad_norm": 2.085247755050659,
+      "learning_rate": 0.00020320987654320985,
+      "loss": 3.213,
+      "step": 527
+    },
+    {
+      "epoch": 1.1713810316139768,
+      "grad_norm": 2.3799211978912354,
+      "learning_rate": 0.00020296296296296295,
+      "loss": 3.6344,
+      "step": 528
+    },
+    {
+      "epoch": 1.1735995562950639,
+      "grad_norm": 2.3023605346679688,
+      "learning_rate": 0.00020271604938271602,
+      "loss": 3.3774,
+      "step": 529
+    },
+    {
+      "epoch": 1.175818080976151,
+      "grad_norm": 2.6587953567504883,
+      "learning_rate": 0.00020246913580246914,
+      "loss": 3.4198,
+      "step": 530
+    },
+    {
+      "epoch": 1.178036605657238,
+      "grad_norm": 2.663954019546509,
+      "learning_rate": 0.0002022222222222222,
+      "loss": 3.6437,
+      "step": 531
+    },
+    {
+      "epoch": 1.180255130338325,
+      "grad_norm": 2.154219388961792,
+      "learning_rate": 0.0002019753086419753,
+      "loss": 3.7449,
+      "step": 532
+    },
+    {
+      "epoch": 1.1824736550194122,
+      "grad_norm": 2.3752970695495605,
+      "learning_rate": 0.00020172839506172837,
+      "loss": 3.1329,
+      "step": 533
+    },
+    {
+      "epoch": 1.1846921797004992,
+      "grad_norm": 2.3677661418914795,
+      "learning_rate": 0.00020148148148148147,
+      "loss": 3.3207,
+      "step": 534
+    },
+    {
+      "epoch": 1.1869107043815863,
+      "grad_norm": 2.7569849491119385,
+      "learning_rate": 0.00020123456790123456,
+      "loss": 3.5899,
+      "step": 535
+    },
+    {
+      "epoch": 1.1891292290626734,
+      "grad_norm": 2.305861234664917,
+      "learning_rate": 0.00020098765432098763,
+      "loss": 3.364,
+      "step": 536
+    },
+    {
+      "epoch": 1.1913477537437605,
+      "grad_norm": 1.9461195468902588,
+      "learning_rate": 0.00020074074074074072,
+      "loss": 3.313,
+      "step": 537
+    },
+    {
+      "epoch": 1.1935662784248475,
+      "grad_norm": 2.354434013366699,
+      "learning_rate": 0.0002004938271604938,
+      "loss": 3.5286,
+      "step": 538
+    },
+    {
+      "epoch": 1.1957848031059346,
+      "grad_norm": 2.3062639236450195,
+      "learning_rate": 0.00020024691358024691,
+      "loss": 3.4975,
+      "step": 539
+    },
+    {
+      "epoch": 1.1980033277870217,
+      "grad_norm": 2.2348551750183105,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 3.4979,
+      "step": 540
+    },
+    {
+      "epoch": 1.2002218524681088,
+      "grad_norm": 2.141759157180786,
+      "learning_rate": 0.00019975308641975308,
+      "loss": 3.3202,
+      "step": 541
+    },
+    {
+      "epoch": 1.2024403771491958,
+      "grad_norm": 2.5755839347839355,
+      "learning_rate": 0.00019950617283950615,
+      "loss": 3.4807,
+      "step": 542
+    },
+    {
+      "epoch": 1.204658901830283,
+      "grad_norm": 2.4700491428375244,
+      "learning_rate": 0.00019925925925925924,
+      "loss": 3.5559,
+      "step": 543
+    },
+    {
+      "epoch": 1.20687742651137,
+      "grad_norm": 1.8858554363250732,
+      "learning_rate": 0.00019901234567901234,
+      "loss": 3.2622,
+      "step": 544
+    },
+    {
+      "epoch": 1.209095951192457,
+      "grad_norm": 2.3044209480285645,
+      "learning_rate": 0.0001987654320987654,
+      "loss": 3.4914,
+      "step": 545
+    },
+    {
+      "epoch": 1.2113144758735441,
+      "grad_norm": 2.5629794597625732,
+      "learning_rate": 0.0001985185185185185,
+      "loss": 3.341,
+      "step": 546
+    },
+    {
+      "epoch": 1.2135330005546312,
+      "grad_norm": 2.24489164352417,
+      "learning_rate": 0.00019827160493827157,
+      "loss": 3.245,
+      "step": 547
+    },
+    {
+      "epoch": 1.2157515252357183,
+      "grad_norm": 1.9321414232254028,
+      "learning_rate": 0.0001980246913580247,
+      "loss": 3.2943,
+      "step": 548
+    },
+    {
+      "epoch": 1.2179700499168054,
+      "grad_norm": 2.2046427726745605,
+      "learning_rate": 0.00019777777777777776,
+      "loss": 3.5628,
+      "step": 549
+    },
+    {
+      "epoch": 1.2201885745978924,
+      "grad_norm": 2.190352201461792,
+      "learning_rate": 0.00019753086419753085,
+      "loss": 3.5317,
+      "step": 550
+    },
+    {
+      "epoch": 1.2224070992789795,
+      "grad_norm": 2.1920089721679688,
+      "learning_rate": 0.00019728395061728395,
+      "loss": 3.4352,
+      "step": 551
+    },
+    {
+      "epoch": 1.2246256239600666,
+      "grad_norm": 2.1254520416259766,
+      "learning_rate": 0.00019703703703703702,
+      "loss": 3.4778,
+      "step": 552
+    },
+    {
+      "epoch": 1.2268441486411537,
+      "grad_norm": 2.2331221103668213,
+      "learning_rate": 0.0001967901234567901,
+      "loss": 3.4171,
+      "step": 553
+    },
+    {
+      "epoch": 1.2290626733222407,
+      "grad_norm": 2.283693313598633,
+      "learning_rate": 0.00019654320987654318,
+      "loss": 3.4359,
+      "step": 554
+    },
+    {
+      "epoch": 1.2312811980033278,
+      "grad_norm": 1.7493880987167358,
+      "learning_rate": 0.00019629629629629627,
+      "loss": 3.2922,
+      "step": 555
+    },
+    {
+      "epoch": 1.2334997226844149,
+      "grad_norm": 2.3630900382995605,
+      "learning_rate": 0.00019604938271604934,
+      "loss": 3.4015,
+      "step": 556
+    },
+    {
+      "epoch": 1.235718247365502,
+      "grad_norm": 2.2852766513824463,
+      "learning_rate": 0.00019580246913580246,
+      "loss": 3.3905,
+      "step": 557
+    },
+    {
+      "epoch": 1.237936772046589,
+      "grad_norm": 2.06636905670166,
+      "learning_rate": 0.00019555555555555556,
+      "loss": 3.138,
+      "step": 558
+    },
+    {
+      "epoch": 1.240155296727676,
+      "grad_norm": 2.1948137283325195,
+      "learning_rate": 0.00019530864197530863,
+      "loss": 3.6945,
+      "step": 559
+    },
+    {
+      "epoch": 1.2423738214087632,
+      "grad_norm": 2.3865559101104736,
+      "learning_rate": 0.00019506172839506172,
+      "loss": 3.4815,
+      "step": 560
+    },
+    {
+      "epoch": 1.2445923460898503,
+      "grad_norm": 2.5900256633758545,
+      "learning_rate": 0.0001948148148148148,
+      "loss": 3.337,
+      "step": 561
+    },
+    {
+      "epoch": 1.2468108707709373,
+      "grad_norm": 4.050100803375244,
+      "learning_rate": 0.00019456790123456789,
+      "loss": 3.5018,
+      "step": 562
+    },
+    {
+      "epoch": 1.2490293954520244,
+      "grad_norm": 2.3661255836486816,
+      "learning_rate": 0.00019432098765432095,
+      "loss": 3.3266,
+      "step": 563
+    },
+    {
+      "epoch": 1.2512479201331115,
+      "grad_norm": 2.158872365951538,
+      "learning_rate": 0.00019407407407407405,
+      "loss": 3.4337,
+      "step": 564
+    },
+    {
+      "epoch": 1.2534664448141986,
+      "grad_norm": 3.8286941051483154,
+      "learning_rate": 0.00019382716049382712,
+      "loss": 3.4148,
+      "step": 565
+    },
+    {
+      "epoch": 1.2556849694952856,
+      "grad_norm": 3.1256844997406006,
+      "learning_rate": 0.00019358024691358024,
+      "loss": 3.3922,
+      "step": 566
+    },
+    {
+      "epoch": 1.2579034941763727,
+      "grad_norm": 2.3997604846954346,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 3.6369,
+      "step": 567
+    },
+    {
+      "epoch": 1.2601220188574598,
+      "grad_norm": 2.5389082431793213,
+      "learning_rate": 0.0001930864197530864,
+      "loss": 3.5212,
+      "step": 568
+    },
+    {
+      "epoch": 1.2623405435385469,
+      "grad_norm": 3.1529343128204346,
+      "learning_rate": 0.0001928395061728395,
+      "loss": 3.4495,
+      "step": 569
+    },
+    {
+      "epoch": 1.264559068219634,
+      "grad_norm": 2.065354585647583,
+      "learning_rate": 0.00019259259259259257,
+      "loss": 3.3547,
+      "step": 570
+    },
+    {
+      "epoch": 1.266777592900721,
+      "grad_norm": 1.9558525085449219,
+      "learning_rate": 0.00019234567901234566,
+      "loss": 3.2478,
+      "step": 571
+    },
+    {
+      "epoch": 1.268996117581808,
+      "grad_norm": 2.602501392364502,
+      "learning_rate": 0.00019209876543209873,
+      "loss": 3.3922,
+      "step": 572
+    },
+    {
+      "epoch": 1.2712146422628952,
+      "grad_norm": 2.4105606079101562,
+      "learning_rate": 0.00019185185185185182,
+      "loss": 3.4096,
+      "step": 573
+    },
+    {
+      "epoch": 1.2734331669439822,
+      "grad_norm": 2.4054174423217773,
+      "learning_rate": 0.00019160493827160495,
+      "loss": 3.3064,
+      "step": 574
+    },
+    {
+      "epoch": 1.2756516916250693,
+      "grad_norm": 3.3085997104644775,
+      "learning_rate": 0.00019135802469135801,
+      "loss": 3.7826,
+      "step": 575
+    },
+    {
+      "epoch": 1.2778702163061564,
+      "grad_norm": 2.5633695125579834,
+      "learning_rate": 0.0001911111111111111,
+      "loss": 3.1075,
+      "step": 576
+    },
+    {
+      "epoch": 1.2800887409872435,
+      "grad_norm": 2.266843795776367,
+      "learning_rate": 0.00019086419753086418,
+      "loss": 3.5141,
+      "step": 577
+    },
+    {
+      "epoch": 1.2823072656683305,
+      "grad_norm": 2.354300022125244,
+      "learning_rate": 0.00019061728395061727,
+      "loss": 3.5175,
+      "step": 578
+    },
+    {
+      "epoch": 1.2845257903494176,
+      "grad_norm": 2.5726969242095947,
+      "learning_rate": 0.00019037037037037034,
+      "loss": 3.3575,
+      "step": 579
+    },
+    {
+      "epoch": 1.2867443150305047,
+      "grad_norm": 2.5123677253723145,
+      "learning_rate": 0.00019012345679012344,
+      "loss": 3.5841,
+      "step": 580
+    },
+    {
+      "epoch": 1.2889628397115918,
+      "grad_norm": 2.327995538711548,
+      "learning_rate": 0.00018987654320987653,
+      "loss": 3.7457,
+      "step": 581
+    },
+    {
+      "epoch": 1.2911813643926788,
+      "grad_norm": 1.9877936840057373,
+      "learning_rate": 0.0001896296296296296,
+      "loss": 3.4834,
+      "step": 582
+    },
+    {
+      "epoch": 1.293399889073766,
+      "grad_norm": 2.1455485820770264,
+      "learning_rate": 0.00018938271604938272,
+      "loss": 3.0003,
+      "step": 583
+    },
+    {
+      "epoch": 1.295618413754853,
+      "grad_norm": 2.212527275085449,
+      "learning_rate": 0.0001891358024691358,
+      "loss": 3.3069,
+      "step": 584
+    },
+    {
+      "epoch": 1.29783693843594,
+      "grad_norm": 2.255786657333374,
+      "learning_rate": 0.00018888888888888888,
+      "loss": 3.5305,
+      "step": 585
+    },
+    {
+      "epoch": 1.3000554631170271,
+      "grad_norm": 2.6613593101501465,
+      "learning_rate": 0.00018864197530864195,
+      "loss": 3.5448,
+      "step": 586
+    },
+    {
+      "epoch": 1.3022739877981142,
+      "grad_norm": 2.2944114208221436,
+      "learning_rate": 0.00018839506172839505,
+      "loss": 3.4809,
+      "step": 587
+    },
+    {
+      "epoch": 1.3044925124792013,
+      "grad_norm": 1.9053362607955933,
+      "learning_rate": 0.00018814814814814814,
+      "loss": 3.3961,
+      "step": 588
+    },
+    {
+      "epoch": 1.3067110371602884,
+      "grad_norm": 2.3816111087799072,
+      "learning_rate": 0.0001879012345679012,
+      "loss": 3.5182,
+      "step": 589
+    },
+    {
+      "epoch": 1.3089295618413754,
+      "grad_norm": 2.40769624710083,
+      "learning_rate": 0.0001876543209876543,
+      "loss": 3.3662,
+      "step": 590
+    },
+    {
+      "epoch": 1.3111480865224625,
+      "grad_norm": 1.7970144748687744,
+      "learning_rate": 0.00018740740740740737,
+      "loss": 3.3038,
+      "step": 591
+    },
+    {
+      "epoch": 1.3133666112035496,
+      "grad_norm": 2.331084728240967,
+      "learning_rate": 0.0001871604938271605,
+      "loss": 3.4491,
+      "step": 592
+    },
+    {
+      "epoch": 1.3155851358846367,
+      "grad_norm": 1.9392837285995483,
+      "learning_rate": 0.00018691358024691356,
+      "loss": 3.3348,
+      "step": 593
+    },
+    {
+      "epoch": 1.3178036605657237,
+      "grad_norm": 1.960953950881958,
+      "learning_rate": 0.00018666666666666666,
+      "loss": 3.7213,
+      "step": 594
+    },
+    {
+      "epoch": 1.3200221852468108,
+      "grad_norm": 2.0331618785858154,
+      "learning_rate": 0.00018641975308641973,
+      "loss": 3.5946,
+      "step": 595
+    },
+    {
+      "epoch": 1.3222407099278979,
+      "grad_norm": 2.3254518508911133,
+      "learning_rate": 0.00018617283950617282,
+      "loss": 3.2805,
+      "step": 596
+    },
+    {
+      "epoch": 1.324459234608985,
+      "grad_norm": 2.0577876567840576,
+      "learning_rate": 0.00018592592592592592,
+      "loss": 3.1668,
+      "step": 597
+    },
+    {
+      "epoch": 1.326677759290072,
+      "grad_norm": 1.9967070817947388,
+      "learning_rate": 0.00018567901234567899,
+      "loss": 3.2897,
+      "step": 598
+    },
+    {
+      "epoch": 1.328896283971159,
+      "grad_norm": 2.4530649185180664,
+      "learning_rate": 0.00018543209876543208,
+      "loss": 3.2706,
+      "step": 599
+    },
+    {
+      "epoch": 1.3311148086522462,
+      "grad_norm": 2.117300033569336,
+      "learning_rate": 0.00018518518518518515,
+      "loss": 3.4075,
+      "step": 600
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 2.0507092475891113,
+      "learning_rate": 0.00018493827160493827,
+      "loss": 3.4287,
+      "step": 601
+    },
+    {
+      "epoch": 1.3355518580144203,
+      "grad_norm": 2.046942949295044,
+      "learning_rate": 0.00018469135802469134,
+      "loss": 3.5699,
+      "step": 602
+    },
+    {
+      "epoch": 1.3377703826955074,
+      "grad_norm": 2.470682382583618,
+      "learning_rate": 0.00018444444444444443,
+      "loss": 3.2687,
+      "step": 603
+    },
+    {
+      "epoch": 1.3399889073765945,
+      "grad_norm": 2.838336706161499,
+      "learning_rate": 0.00018419753086419753,
+      "loss": 4.0082,
+      "step": 604
+    },
+    {
+      "epoch": 1.3422074320576816,
+      "grad_norm": 2.1764631271362305,
+      "learning_rate": 0.0001839506172839506,
+      "loss": 3.5543,
+      "step": 605
+    },
+    {
+      "epoch": 1.3444259567387689,
+      "grad_norm": 2.1825852394104004,
+      "learning_rate": 0.0001837037037037037,
+      "loss": 3.4533,
+      "step": 606
+    },
+    {
+      "epoch": 1.346644481419856,
+      "grad_norm": 1.827637791633606,
+      "learning_rate": 0.00018345679012345676,
+      "loss": 3.6377,
+      "step": 607
+    },
+    {
+      "epoch": 1.348863006100943,
+      "grad_norm": 2.345317840576172,
+      "learning_rate": 0.00018320987654320986,
+      "loss": 3.2005,
+      "step": 608
+    },
+    {
+      "epoch": 1.35108153078203,
+      "grad_norm": 2.1151371002197266,
+      "learning_rate": 0.00018296296296296292,
+      "loss": 3.6244,
+      "step": 609
+    },
+    {
+      "epoch": 1.3533000554631172,
+      "grad_norm": 2.015897035598755,
+      "learning_rate": 0.00018271604938271605,
+      "loss": 3.3354,
+      "step": 610
+    },
+    {
+      "epoch": 1.3555185801442042,
+      "grad_norm": 2.487799882888794,
+      "learning_rate": 0.00018246913580246914,
+      "loss": 3.4002,
+      "step": 611
+    },
+    {
+      "epoch": 1.3577371048252913,
+      "grad_norm": 2.2010080814361572,
+      "learning_rate": 0.0001822222222222222,
+      "loss": 3.6045,
+      "step": 612
+    },
+    {
+      "epoch": 1.3599556295063784,
+      "grad_norm": 2.079329013824463,
+      "learning_rate": 0.0001819753086419753,
+      "loss": 3.2054,
+      "step": 613
+    },
+    {
+      "epoch": 1.3621741541874655,
+      "grad_norm": 2.3059046268463135,
+      "learning_rate": 0.00018172839506172837,
+      "loss": 3.2773,
+      "step": 614
+    },
+    {
+      "epoch": 1.3643926788685525,
+      "grad_norm": 2.26230525970459,
+      "learning_rate": 0.00018148148148148147,
+      "loss": 3.5668,
+      "step": 615
+    },
+    {
+      "epoch": 1.3666112035496396,
+      "grad_norm": 2.3510615825653076,
+      "learning_rate": 0.00018123456790123454,
+      "loss": 3.3529,
+      "step": 616
+    },
+    {
+      "epoch": 1.3688297282307267,
+      "grad_norm": 2.3287205696105957,
+      "learning_rate": 0.00018098765432098763,
+      "loss": 3.6758,
+      "step": 617
+    },
+    {
+      "epoch": 1.3710482529118138,
+      "grad_norm": 2.523916721343994,
+      "learning_rate": 0.00018074074074074075,
+      "loss": 3.6563,
+      "step": 618
+    },
+    {
+      "epoch": 1.3732667775929008,
+      "grad_norm": 2.324697256088257,
+      "learning_rate": 0.00018049382716049382,
+      "loss": 3.263,
+      "step": 619
+    },
+    {
+      "epoch": 1.375485302273988,
+      "grad_norm": 2.2355711460113525,
+      "learning_rate": 0.00018024691358024692,
+      "loss": 3.3433,
+      "step": 620
+    },
+    {
+      "epoch": 1.377703826955075,
+      "grad_norm": 2.1245827674865723,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 3.4821,
+      "step": 621
+    },
+    {
+      "epoch": 1.379922351636162,
+      "grad_norm": 2.286794900894165,
+      "learning_rate": 0.00017975308641975308,
+      "loss": 3.2597,
+      "step": 622
+    },
+    {
+      "epoch": 1.3821408763172491,
+      "grad_norm": 1.847177267074585,
+      "learning_rate": 0.00017950617283950615,
+      "loss": 3.3791,
+      "step": 623
+    },
+    {
+      "epoch": 1.3843594009983362,
+      "grad_norm": 2.352029800415039,
+      "learning_rate": 0.00017925925925925924,
+      "loss": 3.141,
+      "step": 624
+    },
+    {
+      "epoch": 1.3865779256794233,
+      "grad_norm": 2.3254284858703613,
+      "learning_rate": 0.0001790123456790123,
+      "loss": 3.3084,
+      "step": 625
+    },
+    {
+      "epoch": 1.3887964503605104,
+      "grad_norm": 2.2287673950195312,
+      "learning_rate": 0.0001787654320987654,
+      "loss": 3.1627,
+      "step": 626
+    },
+    {
+      "epoch": 1.3910149750415974,
+      "grad_norm": 2.3457648754119873,
+      "learning_rate": 0.00017851851851851853,
+      "loss": 3.1299,
+      "step": 627
+    },
+    {
+      "epoch": 1.3932334997226845,
+      "grad_norm": 2.0009725093841553,
+      "learning_rate": 0.0001782716049382716,
+      "loss": 3.1438,
+      "step": 628
+    },
+    {
+      "epoch": 1.3954520244037716,
+      "grad_norm": 2.171405553817749,
+      "learning_rate": 0.0001780246913580247,
+      "loss": 3.1365,
+      "step": 629
+    },
+    {
+      "epoch": 1.3976705490848587,
+      "grad_norm": 2.15621018409729,
+      "learning_rate": 0.00017777777777777776,
+      "loss": 3.3021,
+      "step": 630
+    },
+    {
+      "epoch": 1.3998890737659457,
+      "grad_norm": 2.396646738052368,
+      "learning_rate": 0.00017753086419753085,
+      "loss": 3.3504,
+      "step": 631
+    },
+    {
+      "epoch": 1.4021075984470328,
+      "grad_norm": 2.2931385040283203,
+      "learning_rate": 0.00017728395061728392,
+      "loss": 3.2839,
+      "step": 632
+    },
+    {
+      "epoch": 1.4043261231281199,
+      "grad_norm": 1.8912330865859985,
+      "learning_rate": 0.00017703703703703702,
+      "loss": 3.4902,
+      "step": 633
+    },
+    {
+      "epoch": 1.406544647809207,
+      "grad_norm": 2.2382240295410156,
+      "learning_rate": 0.0001767901234567901,
+      "loss": 3.2629,
+      "step": 634
+    },
+    {
+      "epoch": 1.408763172490294,
+      "grad_norm": 2.2082059383392334,
+      "learning_rate": 0.00017654320987654318,
+      "loss": 3.1505,
+      "step": 635
+    },
+    {
+      "epoch": 1.410981697171381,
+      "grad_norm": 2.265069007873535,
+      "learning_rate": 0.0001762962962962963,
+      "loss": 3.2891,
+      "step": 636
+    },
+    {
+      "epoch": 1.4132002218524682,
+      "grad_norm": 2.9366841316223145,
+      "learning_rate": 0.00017604938271604937,
+      "loss": 3.374,
+      "step": 637
+    },
+    {
+      "epoch": 1.4154187465335553,
+      "grad_norm": 2.9514403343200684,
+      "learning_rate": 0.00017580246913580247,
+      "loss": 3.5531,
+      "step": 638
+    },
+    {
+      "epoch": 1.4176372712146423,
+      "grad_norm": 2.369492530822754,
+      "learning_rate": 0.00017555555555555553,
+      "loss": 3.2792,
+      "step": 639
+    },
+    {
+      "epoch": 1.4198557958957294,
+      "grad_norm": 3.1703941822052,
+      "learning_rate": 0.00017530864197530863,
+      "loss": 3.3409,
+      "step": 640
+    },
+    {
+      "epoch": 1.4220743205768165,
+      "grad_norm": 2.269150733947754,
+      "learning_rate": 0.00017506172839506172,
+      "loss": 3.5853,
+      "step": 641
+    },
+    {
+      "epoch": 1.4242928452579036,
+      "grad_norm": 2.6991798877716064,
+      "learning_rate": 0.0001748148148148148,
+      "loss": 3.2622,
+      "step": 642
+    },
+    {
+      "epoch": 1.4265113699389906,
+      "grad_norm": 2.641481399536133,
+      "learning_rate": 0.0001745679012345679,
+      "loss": 3.2314,
+      "step": 643
+    },
+    {
+      "epoch": 1.4287298946200777,
+      "grad_norm": 2.246957778930664,
+      "learning_rate": 0.00017432098765432096,
+      "loss": 3.571,
+      "step": 644
+    },
+    {
+      "epoch": 1.4309484193011648,
+      "grad_norm": 2.7312541007995605,
+      "learning_rate": 0.00017407407407407408,
+      "loss": 3.3754,
+      "step": 645
+    },
+    {
+      "epoch": 1.4331669439822519,
+      "grad_norm": 2.2212822437286377,
+      "learning_rate": 0.00017382716049382715,
+      "loss": 3.2571,
+      "step": 646
+    },
+    {
+      "epoch": 1.435385468663339,
+      "grad_norm": 2.409560441970825,
+      "learning_rate": 0.00017358024691358024,
+      "loss": 3.4986,
+      "step": 647
+    },
+    {
+      "epoch": 1.437603993344426,
+      "grad_norm": 2.278890609741211,
+      "learning_rate": 0.0001733333333333333,
+      "loss": 3.1624,
+      "step": 648
+    },
+    {
+      "epoch": 1.439822518025513,
+      "grad_norm": 1.8951377868652344,
+      "learning_rate": 0.0001730864197530864,
+      "loss": 3.4117,
+      "step": 649
+    },
+    {
+      "epoch": 1.4420410427066002,
+      "grad_norm": 2.2820355892181396,
+      "learning_rate": 0.0001728395061728395,
+      "loss": 3.2564,
+      "step": 650
+    },
+    {
+      "epoch": 1.4442595673876872,
+      "grad_norm": 2.242563247680664,
+      "learning_rate": 0.00017259259259259257,
+      "loss": 3.2233,
+      "step": 651
+    },
+    {
+      "epoch": 1.4464780920687743,
+      "grad_norm": 2.5100696086883545,
+      "learning_rate": 0.00017234567901234566,
+      "loss": 3.2435,
+      "step": 652
+    },
+    {
+      "epoch": 1.4486966167498614,
+      "grad_norm": 2.2561306953430176,
+      "learning_rate": 0.00017209876543209873,
+      "loss": 3.2803,
+      "step": 653
+    },
+    {
+      "epoch": 1.4509151414309485,
+      "grad_norm": 2.1119751930236816,
+      "learning_rate": 0.00017185185185185185,
+      "loss": 3.1254,
+      "step": 654
+    },
+    {
+      "epoch": 1.4531336661120355,
+      "grad_norm": 2.211594820022583,
+      "learning_rate": 0.00017160493827160492,
+      "loss": 3.1252,
+      "step": 655
+    },
+    {
+      "epoch": 1.4553521907931226,
+      "grad_norm": 2.218437671661377,
+      "learning_rate": 0.00017135802469135802,
+      "loss": 3.0523,
+      "step": 656
+    },
+    {
+      "epoch": 1.4575707154742097,
+      "grad_norm": 2.6801679134368896,
+      "learning_rate": 0.0001711111111111111,
+      "loss": 3.3134,
+      "step": 657
+    },
+    {
+      "epoch": 1.4597892401552968,
+      "grad_norm": 2.011260509490967,
+      "learning_rate": 0.00017086419753086418,
+      "loss": 3.2324,
+      "step": 658
+    },
+    {
+      "epoch": 1.4620077648363838,
+      "grad_norm": 2.4269440174102783,
+      "learning_rate": 0.00017061728395061727,
+      "loss": 3.559,
+      "step": 659
+    },
+    {
+      "epoch": 1.464226289517471,
+      "grad_norm": 2.397355794906616,
+      "learning_rate": 0.00017037037037037034,
+      "loss": 2.8389,
+      "step": 660
+    },
+    {
+      "epoch": 1.466444814198558,
+      "grad_norm": 2.021843194961548,
+      "learning_rate": 0.00017012345679012344,
+      "loss": 3.6472,
+      "step": 661
+    },
+    {
+      "epoch": 1.468663338879645,
+      "grad_norm": 2.44538950920105,
+      "learning_rate": 0.0001698765432098765,
+      "loss": 3.1784,
+      "step": 662
+    },
+    {
+      "epoch": 1.4708818635607321,
+      "grad_norm": 2.4065604209899902,
+      "learning_rate": 0.00016962962962962963,
+      "loss": 3.4911,
+      "step": 663
+    },
+    {
+      "epoch": 1.4731003882418192,
+      "grad_norm": 2.464629888534546,
+      "learning_rate": 0.00016938271604938272,
+      "loss": 3.2218,
+      "step": 664
+    },
+    {
+      "epoch": 1.4753189129229063,
+      "grad_norm": 2.7142395973205566,
+      "learning_rate": 0.0001691358024691358,
+      "loss": 3.4029,
+      "step": 665
+    },
+    {
+      "epoch": 1.4775374376039934,
+      "grad_norm": 2.454881191253662,
+      "learning_rate": 0.00016888888888888889,
+      "loss": 3.6628,
+      "step": 666
+    },
+    {
+      "epoch": 1.4797559622850804,
+      "grad_norm": 2.4602456092834473,
+      "learning_rate": 0.00016864197530864195,
+      "loss": 3.5139,
+      "step": 667
+    },
+    {
+      "epoch": 1.4819744869661675,
+      "grad_norm": 2.347172498703003,
+      "learning_rate": 0.00016839506172839505,
+      "loss": 3.3265,
+      "step": 668
+    },
+    {
+      "epoch": 1.4841930116472546,
+      "grad_norm": 2.832625150680542,
+      "learning_rate": 0.00016814814814814812,
+      "loss": 3.3921,
+      "step": 669
+    },
+    {
+      "epoch": 1.4864115363283417,
+      "grad_norm": 2.2512402534484863,
+      "learning_rate": 0.0001679012345679012,
+      "loss": 3.4286,
+      "step": 670
+    },
+    {
+      "epoch": 1.4886300610094287,
+      "grad_norm": 2.028317928314209,
+      "learning_rate": 0.00016765432098765433,
+      "loss": 3.1745,
+      "step": 671
+    },
+    {
+      "epoch": 1.4908485856905158,
+      "grad_norm": 2.2526986598968506,
+      "learning_rate": 0.0001674074074074074,
+      "loss": 3.3113,
+      "step": 672
+    },
+    {
+      "epoch": 1.493067110371603,
+      "grad_norm": 1.9375576972961426,
+      "learning_rate": 0.0001671604938271605,
+      "loss": 3.2537,
+      "step": 673
+    },
+    {
+      "epoch": 1.49528563505269,
+      "grad_norm": 2.341203212738037,
+      "learning_rate": 0.00016691358024691357,
+      "loss": 3.3274,
+      "step": 674
+    },
+    {
+      "epoch": 1.497504159733777,
+      "grad_norm": 2.14643931388855,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 3.3397,
+      "step": 675
+    },
+    {
+      "epoch": 1.4997226844148641,
+      "grad_norm": 3.173945426940918,
+      "learning_rate": 0.00016641975308641973,
+      "loss": 3.4466,
+      "step": 676
+    },
+    {
+      "epoch": 1.5019412090959512,
+      "grad_norm": 2.3257527351379395,
+      "learning_rate": 0.00016617283950617282,
+      "loss": 3.4191,
+      "step": 677
+    },
+    {
+      "epoch": 1.5041597337770383,
+      "grad_norm": 2.473442792892456,
+      "learning_rate": 0.0001659259259259259,
+      "loss": 3.0619,
+      "step": 678
+    },
+    {
+      "epoch": 1.5063782584581253,
+      "grad_norm": 2.6135506629943848,
+      "learning_rate": 0.000165679012345679,
+      "loss": 3.2125,
+      "step": 679
+    },
+    {
+      "epoch": 1.5085967831392124,
+      "grad_norm": 2.795602560043335,
+      "learning_rate": 0.0001654320987654321,
+      "loss": 3.4443,
+      "step": 680
+    },
+    {
+      "epoch": 1.5108153078202995,
+      "grad_norm": 2.2123472690582275,
+      "learning_rate": 0.00016518518518518518,
+      "loss": 3.4352,
+      "step": 681
+    },
+    {
+      "epoch": 1.5130338325013866,
+      "grad_norm": 2.5782668590545654,
+      "learning_rate": 0.00016493827160493827,
+      "loss": 3.3481,
+      "step": 682
+    },
+    {
+      "epoch": 1.5152523571824736,
+      "grad_norm": 2.544938564300537,
+      "learning_rate": 0.00016469135802469134,
+      "loss": 3.2209,
+      "step": 683
+    },
+    {
+      "epoch": 1.5174708818635607,
+      "grad_norm": 2.4922897815704346,
+      "learning_rate": 0.00016444444444444444,
+      "loss": 3.3393,
+      "step": 684
+    },
+    {
+      "epoch": 1.5196894065446478,
+      "grad_norm": 2.4906044006347656,
+      "learning_rate": 0.0001641975308641975,
+      "loss": 3.428,
+      "step": 685
+    },
+    {
+      "epoch": 1.5219079312257349,
+      "grad_norm": 3.3816943168640137,
+      "learning_rate": 0.0001639506172839506,
+      "loss": 3.5522,
+      "step": 686
+    },
+    {
+      "epoch": 1.524126455906822,
+      "grad_norm": 2.5193331241607666,
+      "learning_rate": 0.0001637037037037037,
+      "loss": 3.0173,
+      "step": 687
+    },
+    {
+      "epoch": 1.526344980587909,
+      "grad_norm": 2.5064220428466797,
+      "learning_rate": 0.00016345679012345676,
+      "loss": 3.4717,
+      "step": 688
+    },
+    {
+      "epoch": 1.528563505268996,
+      "grad_norm": 2.333829402923584,
+      "learning_rate": 0.00016320987654320988,
+      "loss": 3.2765,
+      "step": 689
+    },
+    {
+      "epoch": 1.5307820299500832,
+      "grad_norm": 2.486219644546509,
+      "learning_rate": 0.00016296296296296295,
+      "loss": 3.3809,
+      "step": 690
+    },
+    {
+      "epoch": 1.5330005546311702,
+      "grad_norm": 2.669949531555176,
+      "learning_rate": 0.00016271604938271605,
+      "loss": 3.2904,
+      "step": 691
+    },
+    {
+      "epoch": 1.5352190793122573,
+      "grad_norm": 2.603637218475342,
+      "learning_rate": 0.00016246913580246912,
+      "loss": 3.5906,
+      "step": 692
+    },
+    {
+      "epoch": 1.5374376039933444,
+      "grad_norm": 2.445789337158203,
+      "learning_rate": 0.0001622222222222222,
+      "loss": 3.593,
+      "step": 693
+    },
+    {
+      "epoch": 1.5396561286744315,
+      "grad_norm": 2.3162970542907715,
+      "learning_rate": 0.0001619753086419753,
+      "loss": 3.3068,
+      "step": 694
+    },
+    {
+      "epoch": 1.5418746533555185,
+      "grad_norm": 2.35551118850708,
+      "learning_rate": 0.00016172839506172837,
+      "loss": 3.2177,
+      "step": 695
+    },
+    {
+      "epoch": 1.5440931780366056,
+      "grad_norm": 2.4126505851745605,
+      "learning_rate": 0.00016148148148148147,
+      "loss": 3.34,
+      "step": 696
+    },
+    {
+      "epoch": 1.5463117027176927,
+      "grad_norm": 2.0040123462677,
+      "learning_rate": 0.00016123456790123454,
+      "loss": 3.2281,
+      "step": 697
+    },
+    {
+      "epoch": 1.5485302273987798,
+      "grad_norm": 2.285857677459717,
+      "learning_rate": 0.00016098765432098766,
+      "loss": 3.1732,
+      "step": 698
+    },
+    {
+      "epoch": 1.5507487520798668,
+      "grad_norm": 2.5136585235595703,
+      "learning_rate": 0.00016074074074074073,
+      "loss": 3.3108,
+      "step": 699
+    },
+    {
+      "epoch": 1.552967276760954,
+      "grad_norm": 2.3991332054138184,
+      "learning_rate": 0.00016049382716049382,
+      "loss": 3.2716,
+      "step": 700
+    },
+    {
+      "epoch": 1.555185801442041,
+      "grad_norm": 2.3463613986968994,
+      "learning_rate": 0.00016024691358024692,
+      "loss": 3.1166,
+      "step": 701
+    },
+    {
+      "epoch": 1.557404326123128,
+      "grad_norm": 2.0715396404266357,
+      "learning_rate": 0.00015999999999999999,
+      "loss": 3.2169,
+      "step": 702
+    },
+    {
+      "epoch": 1.5596228508042151,
+      "grad_norm": 2.4168264865875244,
+      "learning_rate": 0.00015975308641975308,
+      "loss": 3.6391,
+      "step": 703
+    },
+    {
+      "epoch": 1.5618413754853022,
+      "grad_norm": 2.6026668548583984,
+      "learning_rate": 0.00015950617283950615,
+      "loss": 3.2364,
+      "step": 704
+    },
+    {
+      "epoch": 1.5640599001663893,
+      "grad_norm": 2.4969124794006348,
+      "learning_rate": 0.00015925925925925924,
+      "loss": 3.0186,
+      "step": 705
+    },
+    {
+      "epoch": 1.5662784248474764,
+      "grad_norm": 3.4103832244873047,
+      "learning_rate": 0.0001590123456790123,
+      "loss": 3.145,
+      "step": 706
+    },
+    {
+      "epoch": 1.5684969495285634,
+      "grad_norm": 2.6684927940368652,
+      "learning_rate": 0.00015876543209876543,
+      "loss": 3.0599,
+      "step": 707
+    },
+    {
+      "epoch": 1.5707154742096505,
+      "grad_norm": 2.769508123397827,
+      "learning_rate": 0.0001585185185185185,
+      "loss": 3.3719,
+      "step": 708
+    },
+    {
+      "epoch": 1.5729339988907376,
+      "grad_norm": 2.197267532348633,
+      "learning_rate": 0.0001582716049382716,
+      "loss": 3.4445,
+      "step": 709
+    },
+    {
+      "epoch": 1.5751525235718247,
+      "grad_norm": 2.4145960807800293,
+      "learning_rate": 0.0001580246913580247,
+      "loss": 3.4105,
+      "step": 710
+    },
+    {
+      "epoch": 1.5773710482529117,
+      "grad_norm": 3.229869842529297,
+      "learning_rate": 0.00015777777777777776,
+      "loss": 3.5549,
+      "step": 711
+    },
+    {
+      "epoch": 1.5795895729339988,
+      "grad_norm": 2.012956142425537,
+      "learning_rate": 0.00015753086419753086,
+      "loss": 3.2721,
+      "step": 712
+    },
+    {
+      "epoch": 1.581808097615086,
+      "grad_norm": 2.949814796447754,
+      "learning_rate": 0.00015728395061728392,
+      "loss": 3.2795,
+      "step": 713
+    },
+    {
+      "epoch": 1.584026622296173,
+      "grad_norm": 1.9887323379516602,
+      "learning_rate": 0.00015703703703703702,
+      "loss": 3.2472,
+      "step": 714
+    },
+    {
+      "epoch": 1.58624514697726,
+      "grad_norm": 2.1333491802215576,
+      "learning_rate": 0.0001567901234567901,
+      "loss": 3.3367,
+      "step": 715
+    },
+    {
+      "epoch": 1.5884636716583471,
+      "grad_norm": 2.234610080718994,
+      "learning_rate": 0.0001565432098765432,
+      "loss": 3.3595,
+      "step": 716
+    },
+    {
+      "epoch": 1.5906821963394342,
+      "grad_norm": 2.1584436893463135,
+      "learning_rate": 0.0001562962962962963,
+      "loss": 3.4022,
+      "step": 717
+    },
+    {
+      "epoch": 1.5929007210205213,
+      "grad_norm": 2.511650323867798,
+      "learning_rate": 0.00015604938271604937,
+      "loss": 3.366,
+      "step": 718
+    },
+    {
+      "epoch": 1.5951192457016083,
+      "grad_norm": 2.1660759449005127,
+      "learning_rate": 0.00015580246913580247,
+      "loss": 3.0809,
+      "step": 719
+    },
+    {
+      "epoch": 1.5973377703826954,
+      "grad_norm": 2.3823609352111816,
+      "learning_rate": 0.00015555555555555554,
+      "loss": 3.3762,
+      "step": 720
+    },
+    {
+      "epoch": 1.5995562950637825,
+      "grad_norm": 2.3007378578186035,
+      "learning_rate": 0.00015530864197530863,
+      "loss": 3.3282,
+      "step": 721
+    },
+    {
+      "epoch": 1.6017748197448696,
+      "grad_norm": 1.917973518371582,
+      "learning_rate": 0.0001550617283950617,
+      "loss": 3.2978,
+      "step": 722
+    },
+    {
+      "epoch": 1.6039933444259566,
+      "grad_norm": 3.012535810470581,
+      "learning_rate": 0.0001548148148148148,
+      "loss": 2.9261,
+      "step": 723
+    },
+    {
+      "epoch": 1.6062118691070437,
+      "grad_norm": 2.500200033187866,
+      "learning_rate": 0.00015456790123456792,
+      "loss": 3.5262,
+      "step": 724
+    },
+    {
+      "epoch": 1.6084303937881308,
+      "grad_norm": 2.295107841491699,
+      "learning_rate": 0.00015432098765432098,
+      "loss": 3.177,
+      "step": 725
+    },
+    {
+      "epoch": 1.6106489184692179,
+      "grad_norm": 2.829272985458374,
+      "learning_rate": 0.00015407407407407408,
+      "loss": 3.4208,
+      "step": 726
+    },
+    {
+      "epoch": 1.612867443150305,
+      "grad_norm": 2.2645885944366455,
+      "learning_rate": 0.00015382716049382715,
+      "loss": 3.0443,
+      "step": 727
+    },
+    {
+      "epoch": 1.615085967831392,
+      "grad_norm": 2.181065797805786,
+      "learning_rate": 0.00015358024691358024,
+      "loss": 3.2436,
+      "step": 728
+    },
+    {
+      "epoch": 1.617304492512479,
+      "grad_norm": 2.26371431350708,
+      "learning_rate": 0.0001533333333333333,
+      "loss": 3.5514,
+      "step": 729
+    },
+    {
+      "epoch": 1.6195230171935662,
+      "grad_norm": 2.016977310180664,
+      "learning_rate": 0.0001530864197530864,
+      "loss": 3.2197,
+      "step": 730
+    },
+    {
+      "epoch": 1.6217415418746532,
+      "grad_norm": 2.574373722076416,
+      "learning_rate": 0.00015283950617283947,
+      "loss": 3.2973,
+      "step": 731
+    },
+    {
+      "epoch": 1.6239600665557403,
+      "grad_norm": 2.4946484565734863,
+      "learning_rate": 0.00015259259259259257,
+      "loss": 2.9862,
+      "step": 732
+    },
+    {
+      "epoch": 1.6261785912368274,
+      "grad_norm": 2.1024532318115234,
+      "learning_rate": 0.0001523456790123457,
+      "loss": 2.9727,
+      "step": 733
+    },
+    {
+      "epoch": 1.6283971159179145,
+      "grad_norm": 2.482743740081787,
+      "learning_rate": 0.00015209876543209873,
+      "loss": 3.2798,
+      "step": 734
+    },
+    {
+      "epoch": 1.6306156405990015,
+      "grad_norm": 2.3905584812164307,
+      "learning_rate": 0.00015185185185185185,
+      "loss": 3.475,
+      "step": 735
+    },
+    {
+      "epoch": 1.6328341652800886,
+      "grad_norm": 2.5467233657836914,
+      "learning_rate": 0.00015160493827160492,
+      "loss": 3.2185,
+      "step": 736
+    },
+    {
+      "epoch": 1.6350526899611757,
+      "grad_norm": 2.24381160736084,
+      "learning_rate": 0.00015135802469135802,
+      "loss": 3.0765,
+      "step": 737
+    },
+    {
+      "epoch": 1.6372712146422628,
+      "grad_norm": 1.9505075216293335,
+      "learning_rate": 0.00015111111111111109,
+      "loss": 2.9173,
+      "step": 738
+    },
+    {
+      "epoch": 1.6394897393233498,
+      "grad_norm": 2.0864527225494385,
+      "learning_rate": 0.00015086419753086418,
+      "loss": 3.3062,
+      "step": 739
+    },
+    {
+      "epoch": 1.641708264004437,
+      "grad_norm": 2.3677818775177,
+      "learning_rate": 0.00015061728395061728,
+      "loss": 3.4875,
+      "step": 740
+    },
+    {
+      "epoch": 1.643926788685524,
+      "grad_norm": 3.038972854614258,
+      "learning_rate": 0.00015037037037037034,
+      "loss": 3.3306,
+      "step": 741
+    },
+    {
+      "epoch": 1.646145313366611,
+      "grad_norm": 2.5015218257904053,
+      "learning_rate": 0.00015012345679012347,
+      "loss": 3.271,
+      "step": 742
+    },
+    {
+      "epoch": 1.6483638380476981,
+      "grad_norm": 2.41561222076416,
+      "learning_rate": 0.00014987654320987653,
+      "loss": 3.3355,
+      "step": 743
+    },
+    {
+      "epoch": 1.6505823627287852,
+      "grad_norm": 3.146698236465454,
+      "learning_rate": 0.00014962962962962963,
+      "loss": 3.5235,
+      "step": 744
+    },
+    {
+      "epoch": 1.6528008874098723,
+      "grad_norm": 2.324554204940796,
+      "learning_rate": 0.0001493827160493827,
+      "loss": 3.0621,
+      "step": 745
+    },
+    {
+      "epoch": 1.6550194120909594,
+      "grad_norm": 2.293003797531128,
+      "learning_rate": 0.0001491358024691358,
+      "loss": 3.4622,
+      "step": 746
+    },
+    {
+      "epoch": 1.6572379367720464,
+      "grad_norm": 2.3399369716644287,
+      "learning_rate": 0.00014888888888888886,
+      "loss": 3.4904,
+      "step": 747
+    },
+    {
+      "epoch": 1.6594564614531335,
+      "grad_norm": 2.356191635131836,
+      "learning_rate": 0.00014864197530864196,
+      "loss": 3.2469,
+      "step": 748
+    },
+    {
+      "epoch": 1.6616749861342206,
+      "grad_norm": 2.445568799972534,
+      "learning_rate": 0.00014839506172839505,
+      "loss": 3.3118,
+      "step": 749
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "grad_norm": 1.9821323156356812,
+      "learning_rate": 0.00014814814814814812,
+      "loss": 3.1363,
+      "step": 750
+    },
+    {
+      "epoch": 1.6661120354963947,
+      "grad_norm": 2.1127030849456787,
+      "learning_rate": 0.00014790123456790124,
+      "loss": 3.1683,
+      "step": 751
+    },
+    {
+      "epoch": 1.6683305601774818,
+      "grad_norm": 2.173644781112671,
+      "learning_rate": 0.0001476543209876543,
+      "loss": 3.1026,
+      "step": 752
+    },
+    {
+      "epoch": 1.670549084858569,
+      "grad_norm": 2.2702670097351074,
+      "learning_rate": 0.0001474074074074074,
+      "loss": 3.2385,
+      "step": 753
+    },
+    {
+      "epoch": 1.672767609539656,
+      "grad_norm": 2.887740135192871,
+      "learning_rate": 0.00014716049382716047,
+      "loss": 3.3877,
+      "step": 754
+    },
+    {
+      "epoch": 1.674986134220743,
+      "grad_norm": 2.1778671741485596,
+      "learning_rate": 0.00014691358024691357,
+      "loss": 3.4836,
+      "step": 755
+    },
+    {
+      "epoch": 1.6772046589018301,
+      "grad_norm": 2.300837993621826,
+      "learning_rate": 0.00014666666666666664,
+      "loss": 3.5591,
+      "step": 756
+    },
+    {
+      "epoch": 1.6794231835829172,
+      "grad_norm": 2.2965898513793945,
+      "learning_rate": 0.00014641975308641973,
+      "loss": 3.2714,
+      "step": 757
+    },
+    {
+      "epoch": 1.6816417082640043,
+      "grad_norm": 2.096972703933716,
+      "learning_rate": 0.00014617283950617283,
+      "loss": 3.3482,
+      "step": 758
+    },
+    {
+      "epoch": 1.6838602329450914,
+      "grad_norm": 2.309250593185425,
+      "learning_rate": 0.00014592592592592592,
+      "loss": 2.884,
+      "step": 759
+    },
+    {
+      "epoch": 1.6860787576261784,
+      "grad_norm": 2.7725067138671875,
+      "learning_rate": 0.000145679012345679,
+      "loss": 3.1273,
+      "step": 760
+    },
+    {
+      "epoch": 1.6882972823072657,
+      "grad_norm": 2.3271536827087402,
+      "learning_rate": 0.00014543209876543208,
+      "loss": 3.2003,
+      "step": 761
+    },
+    {
+      "epoch": 1.6905158069883528,
+      "grad_norm": 2.6404027938842773,
+      "learning_rate": 0.00014518518518518518,
+      "loss": 3.0914,
+      "step": 762
+    },
+    {
+      "epoch": 1.6927343316694399,
+      "grad_norm": 2.5419130325317383,
+      "learning_rate": 0.00014493827160493825,
+      "loss": 3.4524,
+      "step": 763
+    },
+    {
+      "epoch": 1.694952856350527,
+      "grad_norm": 2.459493637084961,
+      "learning_rate": 0.00014469135802469134,
+      "loss": 3.26,
+      "step": 764
+    },
+    {
+      "epoch": 1.697171381031614,
+      "grad_norm": 2.710942268371582,
+      "learning_rate": 0.0001444444444444444,
+      "loss": 3.4877,
+      "step": 765
+    },
+    {
+      "epoch": 1.699389905712701,
+      "grad_norm": 2.409148693084717,
+      "learning_rate": 0.00014419753086419753,
+      "loss": 3.1375,
+      "step": 766
+    },
+    {
+      "epoch": 1.7016084303937882,
+      "grad_norm": 2.2903177738189697,
+      "learning_rate": 0.0001439506172839506,
+      "loss": 3.3991,
+      "step": 767
+    },
+    {
+      "epoch": 1.7038269550748752,
+      "grad_norm": 2.6926887035369873,
+      "learning_rate": 0.0001437037037037037,
+      "loss": 3.2484,
+      "step": 768
+    },
+    {
+      "epoch": 1.7060454797559623,
+      "grad_norm": 2.323305130004883,
+      "learning_rate": 0.00014345679012345676,
+      "loss": 3.1016,
+      "step": 769
+    },
+    {
+      "epoch": 1.7082640044370494,
+      "grad_norm": 2.741554021835327,
+      "learning_rate": 0.00014320987654320986,
+      "loss": 3.3775,
+      "step": 770
+    },
+    {
+      "epoch": 1.7104825291181365,
+      "grad_norm": 2.3045172691345215,
+      "learning_rate": 0.00014296296296296295,
+      "loss": 3.37,
+      "step": 771
+    },
+    {
+      "epoch": 1.7127010537992235,
+      "grad_norm": 2.4550552368164062,
+      "learning_rate": 0.00014271604938271602,
+      "loss": 3.2225,
+      "step": 772
+    },
+    {
+      "epoch": 1.7149195784803106,
+      "grad_norm": 2.376821756362915,
+      "learning_rate": 0.00014246913580246912,
+      "loss": 3.3659,
+      "step": 773
+    },
+    {
+      "epoch": 1.7171381031613977,
+      "grad_norm": 2.2892322540283203,
+      "learning_rate": 0.0001422222222222222,
+      "loss": 3.1167,
+      "step": 774
+    },
+    {
+      "epoch": 1.7193566278424848,
+      "grad_norm": 2.480600357055664,
+      "learning_rate": 0.0001419753086419753,
+      "loss": 3.4138,
+      "step": 775
+    },
+    {
+      "epoch": 1.7215751525235718,
+      "grad_norm": 2.207759380340576,
+      "learning_rate": 0.00014172839506172838,
+      "loss": 3.2748,
+      "step": 776
+    },
+    {
+      "epoch": 1.723793677204659,
+      "grad_norm": 2.0916552543640137,
+      "learning_rate": 0.00014148148148148147,
+      "loss": 3.2405,
+      "step": 777
+    },
+    {
+      "epoch": 1.726012201885746,
+      "grad_norm": 2.721876382827759,
+      "learning_rate": 0.00014123456790123454,
+      "loss": 3.3752,
+      "step": 778
+    },
+    {
+      "epoch": 1.728230726566833,
+      "grad_norm": 2.079442262649536,
+      "learning_rate": 0.00014098765432098763,
+      "loss": 3.5846,
+      "step": 779
+    },
+    {
+      "epoch": 1.7304492512479202,
+      "grad_norm": 2.1567165851593018,
+      "learning_rate": 0.00014074074074074073,
+      "loss": 3.424,
+      "step": 780
+    },
+    {
+      "epoch": 1.7326677759290072,
+      "grad_norm": 2.008976936340332,
+      "learning_rate": 0.00014049382716049382,
+      "loss": 3.0422,
+      "step": 781
+    },
+    {
+      "epoch": 1.7348863006100943,
+      "grad_norm": 2.383606433868408,
+      "learning_rate": 0.0001402469135802469,
+      "loss": 3.0588,
+      "step": 782
+    },
+    {
+      "epoch": 1.7371048252911814,
+      "grad_norm": 2.1726324558258057,
+      "learning_rate": 0.00014,
+      "loss": 3.142,
+      "step": 783
+    },
+    {
+      "epoch": 1.7393233499722685,
+      "grad_norm": 2.340127468109131,
+      "learning_rate": 0.00013975308641975308,
+      "loss": 3.2344,
+      "step": 784
+    },
+    {
+      "epoch": 1.7415418746533555,
+      "grad_norm": 2.2121455669403076,
+      "learning_rate": 0.00013950617283950615,
+      "loss": 3.2105,
+      "step": 785
+    },
+    {
+      "epoch": 1.7437603993344426,
+      "grad_norm": 2.0652999877929688,
+      "learning_rate": 0.00013925925925925925,
+      "loss": 2.9735,
+      "step": 786
+    },
+    {
+      "epoch": 1.7459789240155297,
+      "grad_norm": 2.1133222579956055,
+      "learning_rate": 0.00013901234567901231,
+      "loss": 3.0224,
+      "step": 787
+    },
+    {
+      "epoch": 1.7481974486966168,
+      "grad_norm": 2.383840560913086,
+      "learning_rate": 0.00013876543209876544,
+      "loss": 3.4006,
+      "step": 788
+    },
+    {
+      "epoch": 1.7504159733777038,
+      "grad_norm": 2.4860692024230957,
+      "learning_rate": 0.0001385185185185185,
+      "loss": 2.9993,
+      "step": 789
+    },
+    {
+      "epoch": 1.752634498058791,
+      "grad_norm": 2.170384407043457,
+      "learning_rate": 0.0001382716049382716,
+      "loss": 2.9982,
+      "step": 790
+    },
+    {
+      "epoch": 1.754853022739878,
+      "grad_norm": 2.291740655899048,
+      "learning_rate": 0.00013802469135802467,
+      "loss": 3.2002,
+      "step": 791
+    },
+    {
+      "epoch": 1.757071547420965,
+      "grad_norm": 2.595947504043579,
+      "learning_rate": 0.00013777777777777776,
+      "loss": 3.2002,
+      "step": 792
+    },
+    {
+      "epoch": 1.7592900721020521,
+      "grad_norm": 2.428865671157837,
+      "learning_rate": 0.00013753086419753086,
+      "loss": 3.0468,
+      "step": 793
+    },
+    {
+      "epoch": 1.7615085967831392,
+      "grad_norm": 2.3642592430114746,
+      "learning_rate": 0.00013728395061728393,
+      "loss": 3.2749,
+      "step": 794
+    },
+    {
+      "epoch": 1.7637271214642263,
+      "grad_norm": 2.1062912940979004,
+      "learning_rate": 0.00013703703703703702,
+      "loss": 3.165,
+      "step": 795
+    },
+    {
+      "epoch": 1.7659456461453134,
+      "grad_norm": 3.158473491668701,
+      "learning_rate": 0.00013679012345679012,
+      "loss": 3.2169,
+      "step": 796
+    },
+    {
+      "epoch": 1.7681641708264004,
+      "grad_norm": 2.5402169227600098,
+      "learning_rate": 0.0001365432098765432,
+      "loss": 3.4594,
+      "step": 797
+    },
+    {
+      "epoch": 1.7703826955074875,
+      "grad_norm": 2.8927299976348877,
+      "learning_rate": 0.00013629629629629628,
+      "loss": 3.3658,
+      "step": 798
+    },
+    {
+      "epoch": 1.7726012201885746,
+      "grad_norm": 2.1674094200134277,
+      "learning_rate": 0.00013604938271604937,
+      "loss": 3.0939,
+      "step": 799
+    },
+    {
+      "epoch": 1.7748197448696617,
+      "grad_norm": 2.271671772003174,
+      "learning_rate": 0.00013580246913580244,
+      "loss": 3.0452,
+      "step": 800
+    },
+    {
+      "epoch": 1.7770382695507487,
+      "grad_norm": 2.388977527618408,
+      "learning_rate": 0.00013555555555555554,
+      "loss": 3.6446,
+      "step": 801
+    },
+    {
+      "epoch": 1.7792567942318358,
+      "grad_norm": 2.4931607246398926,
+      "learning_rate": 0.00013530864197530863,
+      "loss": 3.3015,
+      "step": 802
+    },
+    {
+      "epoch": 1.7814753189129229,
+      "grad_norm": 2.279196262359619,
+      "learning_rate": 0.00013506172839506173,
+      "loss": 3.3715,
+      "step": 803
+    },
+    {
+      "epoch": 1.78369384359401,
+      "grad_norm": 3.1445200443267822,
+      "learning_rate": 0.0001348148148148148,
+      "loss": 3.2527,
+      "step": 804
+    },
+    {
+      "epoch": 1.785912368275097,
+      "grad_norm": 2.4283716678619385,
+      "learning_rate": 0.0001345679012345679,
+      "loss": 3.184,
+      "step": 805
+    },
+    {
+      "epoch": 1.788130892956184,
+      "grad_norm": 1.925480842590332,
+      "learning_rate": 0.00013432098765432099,
+      "loss": 3.1943,
+      "step": 806
+    },
+    {
+      "epoch": 1.7903494176372712,
+      "grad_norm": 2.477954387664795,
+      "learning_rate": 0.00013407407407407405,
+      "loss": 3.0508,
+      "step": 807
+    },
+    {
+      "epoch": 1.7925679423183583,
+      "grad_norm": 2.162165403366089,
+      "learning_rate": 0.00013382716049382715,
+      "loss": 3.1481,
+      "step": 808
+    },
+    {
+      "epoch": 1.7947864669994453,
+      "grad_norm": 2.2736563682556152,
+      "learning_rate": 0.00013358024691358022,
+      "loss": 2.947,
+      "step": 809
+    },
+    {
+      "epoch": 1.7970049916805324,
+      "grad_norm": 2.374603509902954,
+      "learning_rate": 0.0001333333333333333,
+      "loss": 3.2397,
+      "step": 810
+    },
+    {
+      "epoch": 1.7992235163616195,
+      "grad_norm": 2.7803421020507812,
+      "learning_rate": 0.0001330864197530864,
+      "loss": 3.4712,
+      "step": 811
+    },
+    {
+      "epoch": 1.8014420410427066,
+      "grad_norm": 2.6039674282073975,
+      "learning_rate": 0.0001328395061728395,
+      "loss": 3.1586,
+      "step": 812
+    },
+    {
+      "epoch": 1.8036605657237936,
+      "grad_norm": 2.2403810024261475,
+      "learning_rate": 0.00013259259259259257,
+      "loss": 3.0981,
+      "step": 813
+    },
+    {
+      "epoch": 1.8058790904048807,
+      "grad_norm": 2.399489641189575,
+      "learning_rate": 0.00013234567901234567,
+      "loss": 3.1123,
+      "step": 814
+    },
+    {
+      "epoch": 1.8080976150859678,
+      "grad_norm": 2.690671920776367,
+      "learning_rate": 0.00013209876543209876,
+      "loss": 3.1522,
+      "step": 815
+    },
+    {
+      "epoch": 1.8103161397670549,
+      "grad_norm": 2.4136223793029785,
+      "learning_rate": 0.00013185185185185183,
+      "loss": 3.5423,
+      "step": 816
+    },
+    {
+      "epoch": 1.8125346644481422,
+      "grad_norm": 2.304610013961792,
+      "learning_rate": 0.00013160493827160492,
+      "loss": 3.1463,
+      "step": 817
+    },
+    {
+      "epoch": 1.8147531891292292,
+      "grad_norm": 2.287832498550415,
+      "learning_rate": 0.00013135802469135802,
+      "loss": 3.0617,
+      "step": 818
+    },
+    {
+      "epoch": 1.8169717138103163,
+      "grad_norm": 2.341665029525757,
+      "learning_rate": 0.00013111111111111111,
+      "loss": 2.9171,
+      "step": 819
+    },
+    {
+      "epoch": 1.8191902384914034,
+      "grad_norm": 2.3015711307525635,
+      "learning_rate": 0.00013086419753086418,
+      "loss": 3.1425,
+      "step": 820
+    },
+    {
+      "epoch": 1.8214087631724905,
+      "grad_norm": 2.7956604957580566,
+      "learning_rate": 0.00013061728395061728,
+      "loss": 3.1592,
+      "step": 821
+    },
+    {
+      "epoch": 1.8236272878535775,
+      "grad_norm": 2.18526291847229,
+      "learning_rate": 0.00013037037037037035,
+      "loss": 3.133,
+      "step": 822
+    },
+    {
+      "epoch": 1.8258458125346646,
+      "grad_norm": 2.612375259399414,
+      "learning_rate": 0.00013012345679012344,
+      "loss": 3.3313,
+      "step": 823
+    },
+    {
+      "epoch": 1.8280643372157517,
+      "grad_norm": 2.2254579067230225,
+      "learning_rate": 0.00012987654320987654,
+      "loss": 3.034,
+      "step": 824
+    },
+    {
+      "epoch": 1.8302828618968388,
+      "grad_norm": 2.2610890865325928,
+      "learning_rate": 0.0001296296296296296,
+      "loss": 3.1951,
+      "step": 825
+    },
+    {
+      "epoch": 1.8325013865779258,
+      "grad_norm": 2.1975669860839844,
+      "learning_rate": 0.0001293827160493827,
+      "loss": 3.0286,
+      "step": 826
+    },
+    {
+      "epoch": 1.834719911259013,
+      "grad_norm": 2.2948403358459473,
+      "learning_rate": 0.0001291358024691358,
+      "loss": 3.5513,
+      "step": 827
+    },
+    {
+      "epoch": 1.8369384359401,
+      "grad_norm": 2.7584455013275146,
+      "learning_rate": 0.0001288888888888889,
+      "loss": 3.1805,
+      "step": 828
+    },
+    {
+      "epoch": 1.839156960621187,
+      "grad_norm": 3.0275652408599854,
+      "learning_rate": 0.00012864197530864196,
+      "loss": 3.3136,
+      "step": 829
+    },
+    {
+      "epoch": 1.8413754853022741,
+      "grad_norm": 2.6993672847747803,
+      "learning_rate": 0.00012839506172839505,
+      "loss": 2.9933,
+      "step": 830
+    },
+    {
+      "epoch": 1.8435940099833612,
+      "grad_norm": 2.784571647644043,
+      "learning_rate": 0.00012814814814814812,
+      "loss": 3.2882,
+      "step": 831
+    },
+    {
+      "epoch": 1.8458125346644483,
+      "grad_norm": 2.407775640487671,
+      "learning_rate": 0.00012790123456790122,
+      "loss": 3.4284,
+      "step": 832
+    },
+    {
+      "epoch": 1.8480310593455354,
+      "grad_norm": 2.718254566192627,
+      "learning_rate": 0.0001276543209876543,
+      "loss": 3.0864,
+      "step": 833
+    },
+    {
+      "epoch": 1.8502495840266224,
+      "grad_norm": 2.3367719650268555,
+      "learning_rate": 0.0001274074074074074,
+      "loss": 3.0859,
+      "step": 834
+    },
+    {
+      "epoch": 1.8524681087077095,
+      "grad_norm": 2.3358306884765625,
+      "learning_rate": 0.00012716049382716047,
+      "loss": 3.0301,
+      "step": 835
+    },
+    {
+      "epoch": 1.8546866333887966,
+      "grad_norm": 2.6299703121185303,
+      "learning_rate": 0.00012691358024691357,
+      "loss": 3.4604,
+      "step": 836
+    },
+    {
+      "epoch": 1.8569051580698837,
+      "grad_norm": 2.3444509506225586,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 3.2267,
+      "step": 837
+    },
+    {
+      "epoch": 1.8591236827509707,
+      "grad_norm": 1.9489563703536987,
+      "learning_rate": 0.00012641975308641973,
+      "loss": 2.9865,
+      "step": 838
+    },
+    {
+      "epoch": 1.8613422074320578,
+      "grad_norm": 2.522508144378662,
+      "learning_rate": 0.00012617283950617283,
+      "loss": 2.9469,
+      "step": 839
+    },
+    {
+      "epoch": 1.8635607321131449,
+      "grad_norm": 2.648571252822876,
+      "learning_rate": 0.0001259259259259259,
+      "loss": 3.1694,
+      "step": 840
+    },
+    {
+      "epoch": 1.865779256794232,
+      "grad_norm": 2.211993455886841,
+      "learning_rate": 0.00012567901234567902,
+      "loss": 3.2766,
+      "step": 841
+    },
+    {
+      "epoch": 1.867997781475319,
+      "grad_norm": 2.000908851623535,
+      "learning_rate": 0.00012543209876543209,
+      "loss": 3.1671,
+      "step": 842
+    },
+    {
+      "epoch": 1.870216306156406,
+      "grad_norm": 2.6599717140197754,
+      "learning_rate": 0.00012518518518518518,
+      "loss": 3.0701,
+      "step": 843
+    },
+    {
+      "epoch": 1.8724348308374932,
+      "grad_norm": 2.4495785236358643,
+      "learning_rate": 0.00012493827160493825,
+      "loss": 3.3722,
+      "step": 844
+    },
+    {
+      "epoch": 1.8746533555185803,
+      "grad_norm": 2.681002378463745,
+      "learning_rate": 0.00012469135802469134,
+      "loss": 3.5131,
+      "step": 845
+    },
+    {
+      "epoch": 1.8768718801996673,
+      "grad_norm": 2.360438346862793,
+      "learning_rate": 0.00012444444444444444,
+      "loss": 3.2038,
+      "step": 846
+    },
+    {
+      "epoch": 1.8790904048807544,
+      "grad_norm": 2.1553070545196533,
+      "learning_rate": 0.0001241975308641975,
+      "loss": 3.1319,
+      "step": 847
+    },
+    {
+      "epoch": 1.8813089295618415,
+      "grad_norm": 2.0849030017852783,
+      "learning_rate": 0.0001239506172839506,
+      "loss": 2.826,
+      "step": 848
+    },
+    {
+      "epoch": 1.8835274542429286,
+      "grad_norm": 2.1525063514709473,
+      "learning_rate": 0.0001237037037037037,
+      "loss": 3.2682,
+      "step": 849
+    },
+    {
+      "epoch": 1.8857459789240156,
+      "grad_norm": 2.4770429134368896,
+      "learning_rate": 0.0001234567901234568,
+      "loss": 3.3787,
+      "step": 850
+    },
+    {
+      "epoch": 1.8879645036051027,
+      "grad_norm": 2.3186864852905273,
+      "learning_rate": 0.00012320987654320986,
+      "loss": 3.1634,
+      "step": 851
+    },
+    {
+      "epoch": 1.8901830282861898,
+      "grad_norm": 2.233858585357666,
+      "learning_rate": 0.00012296296296296296,
+      "loss": 3.046,
+      "step": 852
+    },
+    {
+      "epoch": 1.8924015529672769,
+      "grad_norm": 2.070619583129883,
+      "learning_rate": 0.00012271604938271602,
+      "loss": 3.3269,
+      "step": 853
+    },
+    {
+      "epoch": 1.894620077648364,
+      "grad_norm": 2.2953951358795166,
+      "learning_rate": 0.00012246913580246912,
+      "loss": 3.3614,
+      "step": 854
+    },
+    {
+      "epoch": 1.896838602329451,
+      "grad_norm": 2.5667810440063477,
+      "learning_rate": 0.00012222222222222221,
+      "loss": 3.3877,
+      "step": 855
+    },
+    {
+      "epoch": 1.899057127010538,
+      "grad_norm": 2.081918478012085,
+      "learning_rate": 0.00012197530864197531,
+      "loss": 3.168,
+      "step": 856
+    },
+    {
+      "epoch": 1.9012756516916252,
+      "grad_norm": 2.359457492828369,
+      "learning_rate": 0.00012172839506172839,
+      "loss": 2.9856,
+      "step": 857
+    },
+    {
+      "epoch": 1.9034941763727122,
+      "grad_norm": 2.210935354232788,
+      "learning_rate": 0.00012148148148148147,
+      "loss": 3.0832,
+      "step": 858
+    },
+    {
+      "epoch": 1.9057127010537993,
+      "grad_norm": 2.2066750526428223,
+      "learning_rate": 0.00012123456790123455,
+      "loss": 3.2425,
+      "step": 859
+    },
+    {
+      "epoch": 1.9079312257348864,
+      "grad_norm": 2.337409496307373,
+      "learning_rate": 0.00012098765432098764,
+      "loss": 3.1591,
+      "step": 860
+    },
+    {
+      "epoch": 1.9101497504159735,
+      "grad_norm": 2.1195685863494873,
+      "learning_rate": 0.00012074074074074073,
+      "loss": 3.6143,
+      "step": 861
+    },
+    {
+      "epoch": 1.9123682750970605,
+      "grad_norm": 2.4171972274780273,
+      "learning_rate": 0.00012049382716049381,
+      "loss": 2.8432,
+      "step": 862
+    },
+    {
+      "epoch": 1.9145867997781476,
+      "grad_norm": 2.446955442428589,
+      "learning_rate": 0.0001202469135802469,
+      "loss": 2.949,
+      "step": 863
+    },
+    {
+      "epoch": 1.9168053244592347,
+      "grad_norm": 2.3626015186309814,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 3.0894,
+      "step": 864
+    },
+    {
+      "epoch": 1.9190238491403218,
+      "grad_norm": 2.424896717071533,
+      "learning_rate": 0.00011975308641975308,
+      "loss": 3.2725,
+      "step": 865
+    },
+    {
+      "epoch": 1.9212423738214088,
+      "grad_norm": 2.7716565132141113,
+      "learning_rate": 0.00011950617283950617,
+      "loss": 3.3587,
+      "step": 866
+    },
+    {
+      "epoch": 1.923460898502496,
+      "grad_norm": 2.527003288269043,
+      "learning_rate": 0.00011925925925925925,
+      "loss": 2.895,
+      "step": 867
+    },
+    {
+      "epoch": 1.925679423183583,
+      "grad_norm": 2.510120391845703,
+      "learning_rate": 0.00011901234567901233,
+      "loss": 3.5313,
+      "step": 868
+    },
+    {
+      "epoch": 1.92789794786467,
+      "grad_norm": 2.8976118564605713,
+      "learning_rate": 0.00011876543209876541,
+      "loss": 3.3995,
+      "step": 869
+    },
+    {
+      "epoch": 1.9301164725457571,
+      "grad_norm": 2.390514612197876,
+      "learning_rate": 0.0001185185185185185,
+      "loss": 3.3923,
+      "step": 870
+    },
+    {
+      "epoch": 1.9323349972268442,
+      "grad_norm": 2.3251006603240967,
+      "learning_rate": 0.0001182716049382716,
+      "loss": 3.2747,
+      "step": 871
+    },
+    {
+      "epoch": 1.9345535219079313,
+      "grad_norm": 2.787194013595581,
+      "learning_rate": 0.00011802469135802468,
+      "loss": 3.2107,
+      "step": 872
+    },
+    {
+      "epoch": 1.9367720465890184,
+      "grad_norm": 2.386413335800171,
+      "learning_rate": 0.00011777777777777776,
+      "loss": 3.1457,
+      "step": 873
+    },
+    {
+      "epoch": 1.9389905712701054,
+      "grad_norm": 2.3208253383636475,
+      "learning_rate": 0.00011753086419753086,
+      "loss": 3.1956,
+      "step": 874
+    },
+    {
+      "epoch": 1.9412090959511925,
+      "grad_norm": 2.0563371181488037,
+      "learning_rate": 0.00011728395061728394,
+      "loss": 3.403,
+      "step": 875
+    },
+    {
+      "epoch": 1.9434276206322796,
+      "grad_norm": 2.148702383041382,
+      "learning_rate": 0.00011703703703703702,
+      "loss": 3.0418,
+      "step": 876
+    },
+    {
+      "epoch": 1.9456461453133667,
+      "grad_norm": 2.6958789825439453,
+      "learning_rate": 0.0001167901234567901,
+      "loss": 3.0791,
+      "step": 877
+    },
+    {
+      "epoch": 1.9478646699944537,
+      "grad_norm": 2.285857677459717,
+      "learning_rate": 0.00011654320987654319,
+      "loss": 3.009,
+      "step": 878
+    },
+    {
+      "epoch": 1.9500831946755408,
+      "grad_norm": 2.455378532409668,
+      "learning_rate": 0.0001162962962962963,
+      "loss": 3.3193,
+      "step": 879
+    },
+    {
+      "epoch": 1.9523017193566279,
+      "grad_norm": 2.14782452583313,
+      "learning_rate": 0.00011604938271604938,
+      "loss": 3.2418,
+      "step": 880
+    },
+    {
+      "epoch": 1.954520244037715,
+      "grad_norm": 2.416963577270508,
+      "learning_rate": 0.00011580246913580246,
+      "loss": 3.0499,
+      "step": 881
+    },
+    {
+      "epoch": 1.956738768718802,
+      "grad_norm": 2.2186121940612793,
+      "learning_rate": 0.00011555555555555554,
+      "loss": 3.2861,
+      "step": 882
+    },
+    {
+      "epoch": 1.958957293399889,
+      "grad_norm": 2.3998327255249023,
+      "learning_rate": 0.00011530864197530863,
+      "loss": 3.5199,
+      "step": 883
+    },
+    {
+      "epoch": 1.9611758180809762,
+      "grad_norm": 2.5903866291046143,
+      "learning_rate": 0.00011506172839506172,
+      "loss": 3.1077,
+      "step": 884
+    },
+    {
+      "epoch": 1.9633943427620633,
+      "grad_norm": 2.4739785194396973,
+      "learning_rate": 0.0001148148148148148,
+      "loss": 2.8939,
+      "step": 885
+    },
+    {
+      "epoch": 1.9656128674431503,
+      "grad_norm": 2.332282781600952,
+      "learning_rate": 0.00011456790123456789,
+      "loss": 2.9344,
+      "step": 886
+    },
+    {
+      "epoch": 1.9678313921242374,
+      "grad_norm": 2.34643816947937,
+      "learning_rate": 0.00011432098765432099,
+      "loss": 3.4717,
+      "step": 887
+    },
+    {
+      "epoch": 1.9700499168053245,
+      "grad_norm": 2.479377269744873,
+      "learning_rate": 0.00011407407407407407,
+      "loss": 3.1998,
+      "step": 888
+    },
+    {
+      "epoch": 1.9722684414864116,
+      "grad_norm": 2.5462934970855713,
+      "learning_rate": 0.00011382716049382715,
+      "loss": 3.2993,
+      "step": 889
+    },
+    {
+      "epoch": 1.9744869661674986,
+      "grad_norm": 2.5287528038024902,
+      "learning_rate": 0.00011358024691358023,
+      "loss": 3.007,
+      "step": 890
+    },
+    {
+      "epoch": 1.9767054908485857,
+      "grad_norm": 2.461399555206299,
+      "learning_rate": 0.00011333333333333331,
+      "loss": 3.3439,
+      "step": 891
+    },
+    {
+      "epoch": 1.9789240155296728,
+      "grad_norm": 2.155485153198242,
+      "learning_rate": 0.00011308641975308641,
+      "loss": 3.0339,
+      "step": 892
+    },
+    {
+      "epoch": 1.9811425402107599,
+      "grad_norm": 2.657992362976074,
+      "learning_rate": 0.00011283950617283949,
+      "loss": 3.1947,
+      "step": 893
+    },
+    {
+      "epoch": 1.983361064891847,
+      "grad_norm": 3.3627352714538574,
+      "learning_rate": 0.00011259259259259259,
+      "loss": 3.38,
+      "step": 894
+    },
+    {
+      "epoch": 1.985579589572934,
+      "grad_norm": 2.569035053253174,
+      "learning_rate": 0.00011234567901234567,
+      "loss": 3.3804,
+      "step": 895
+    },
+    {
+      "epoch": 1.987798114254021,
+      "grad_norm": 2.417249917984009,
+      "learning_rate": 0.00011209876543209876,
+      "loss": 3.04,
+      "step": 896
+    },
+    {
+      "epoch": 1.9900166389351082,
+      "grad_norm": 2.2903692722320557,
+      "learning_rate": 0.00011185185185185184,
+      "loss": 2.9136,
+      "step": 897
+    },
+    {
+      "epoch": 1.9922351636161952,
+      "grad_norm": 2.516977310180664,
+      "learning_rate": 0.00011160493827160493,
+      "loss": 2.7962,
+      "step": 898
+    },
+    {
+      "epoch": 1.9944536882972823,
+      "grad_norm": 2.4380948543548584,
+      "learning_rate": 0.00011135802469135801,
+      "loss": 3.1207,
+      "step": 899
+    },
+    {
+      "epoch": 1.9966722129783694,
+      "grad_norm": 2.972698211669922,
+      "learning_rate": 0.00011111111111111109,
+      "loss": 3.1786,
+      "step": 900
+    },
+    {
+      "epoch": 1.9988907376594565,
+      "grad_norm": 2.5468382835388184,
+      "learning_rate": 0.0001108641975308642,
+      "loss": 3.1297,
+      "step": 901
+    },
+    {
+      "epoch": 2.0011092623405435,
+      "grad_norm": 2.4074621200561523,
+      "learning_rate": 0.00011061728395061728,
+      "loss": 2.9673,
+      "step": 902
+    },
+    {
+      "epoch": 2.0033277870216306,
+      "grad_norm": 2.7926111221313477,
+      "learning_rate": 0.00011037037037037036,
+      "loss": 2.9624,
+      "step": 903
+    },
+    {
+      "epoch": 2.0055463117027177,
+      "grad_norm": 2.317662239074707,
+      "learning_rate": 0.00011012345679012344,
+      "loss": 3.1855,
+      "step": 904
+    },
+    {
+      "epoch": 2.0077648363838048,
+      "grad_norm": 2.435128688812256,
+      "learning_rate": 0.00010987654320987654,
+      "loss": 3.2628,
+      "step": 905
+    },
+    {
+      "epoch": 2.009983361064892,
+      "grad_norm": 3.9377646446228027,
+      "learning_rate": 0.00010962962962962962,
+      "loss": 2.9922,
+      "step": 906
+    },
+    {
+      "epoch": 2.012201885745979,
+      "grad_norm": 2.0925800800323486,
+      "learning_rate": 0.0001093827160493827,
+      "loss": 3.0091,
+      "step": 907
+    },
+    {
+      "epoch": 2.014420410427066,
+      "grad_norm": 2.6245462894439697,
+      "learning_rate": 0.00010913580246913578,
+      "loss": 3.1796,
+      "step": 908
+    },
+    {
+      "epoch": 2.016638935108153,
+      "grad_norm": 2.457444190979004,
+      "learning_rate": 0.00010888888888888889,
+      "loss": 3.0269,
+      "step": 909
+    },
+    {
+      "epoch": 2.01885745978924,
+      "grad_norm": 2.854897975921631,
+      "learning_rate": 0.00010864197530864197,
+      "loss": 3.0231,
+      "step": 910
+    },
+    {
+      "epoch": 2.021075984470327,
+      "grad_norm": 2.255398988723755,
+      "learning_rate": 0.00010839506172839505,
+      "loss": 3.1506,
+      "step": 911
+    },
+    {
+      "epoch": 2.0232945091514143,
+      "grad_norm": 2.417520046234131,
+      "learning_rate": 0.00010814814814814814,
+      "loss": 2.8151,
+      "step": 912
+    },
+    {
+      "epoch": 2.0255130338325014,
+      "grad_norm": 2.4340293407440186,
+      "learning_rate": 0.00010790123456790122,
+      "loss": 2.8309,
+      "step": 913
+    },
+    {
+      "epoch": 2.0277315585135884,
+      "grad_norm": 2.4196085929870605,
+      "learning_rate": 0.00010765432098765431,
+      "loss": 3.32,
+      "step": 914
+    },
+    {
+      "epoch": 2.0299500831946755,
+      "grad_norm": 2.858182668685913,
+      "learning_rate": 0.0001074074074074074,
+      "loss": 2.9243,
+      "step": 915
+    },
+    {
+      "epoch": 2.0321686078757626,
+      "grad_norm": 2.5132341384887695,
+      "learning_rate": 0.00010716049382716048,
+      "loss": 3.3406,
+      "step": 916
+    },
+    {
+      "epoch": 2.0343871325568497,
+      "grad_norm": 2.638958692550659,
+      "learning_rate": 0.00010691358024691357,
+      "loss": 2.9936,
+      "step": 917
+    },
+    {
+      "epoch": 2.0366056572379367,
+      "grad_norm": 2.3950014114379883,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 2.7652,
+      "step": 918
+    },
+    {
+      "epoch": 2.038824181919024,
+      "grad_norm": 2.0107152462005615,
+      "learning_rate": 0.00010641975308641975,
+      "loss": 3.0388,
+      "step": 919
+    },
+    {
+      "epoch": 2.041042706600111,
+      "grad_norm": 2.414311647415161,
+      "learning_rate": 0.00010617283950617283,
+      "loss": 3.0491,
+      "step": 920
+    },
+    {
+      "epoch": 2.043261231281198,
+      "grad_norm": 2.4691975116729736,
+      "learning_rate": 0.00010592592592592591,
+      "loss": 2.9126,
+      "step": 921
+    },
+    {
+      "epoch": 2.045479755962285,
+      "grad_norm": 2.596341609954834,
+      "learning_rate": 0.00010567901234567899,
+      "loss": 3.1376,
+      "step": 922
+    },
+    {
+      "epoch": 2.047698280643372,
+      "grad_norm": 2.089376926422119,
+      "learning_rate": 0.00010543209876543209,
+      "loss": 3.1566,
+      "step": 923
+    },
+    {
+      "epoch": 2.049916805324459,
+      "grad_norm": 2.0549700260162354,
+      "learning_rate": 0.00010518518518518518,
+      "loss": 2.9169,
+      "step": 924
+    },
+    {
+      "epoch": 2.0521353300055463,
+      "grad_norm": 2.110384941101074,
+      "learning_rate": 0.00010493827160493826,
+      "loss": 3.1633,
+      "step": 925
+    },
+    {
+      "epoch": 2.0543538546866333,
+      "grad_norm": 2.5702781677246094,
+      "learning_rate": 0.00010469135802469135,
+      "loss": 3.2627,
+      "step": 926
+    },
+    {
+      "epoch": 2.0565723793677204,
+      "grad_norm": 1.9373079538345337,
+      "learning_rate": 0.00010444444444444444,
+      "loss": 3.0869,
+      "step": 927
+    },
+    {
+      "epoch": 2.0587909040488075,
+      "grad_norm": 2.185429811477661,
+      "learning_rate": 0.00010419753086419752,
+      "loss": 2.7927,
+      "step": 928
+    },
+    {
+      "epoch": 2.0610094287298946,
+      "grad_norm": 2.281658411026001,
+      "learning_rate": 0.0001039506172839506,
+      "loss": 3.1431,
+      "step": 929
+    },
+    {
+      "epoch": 2.0632279534109816,
+      "grad_norm": 2.0678982734680176,
+      "learning_rate": 0.00010370370370370369,
+      "loss": 3.0959,
+      "step": 930
+    },
+    {
+      "epoch": 2.0654464780920687,
+      "grad_norm": 2.23199200630188,
+      "learning_rate": 0.00010345679012345677,
+      "loss": 3.2315,
+      "step": 931
+    },
+    {
+      "epoch": 2.067665002773156,
+      "grad_norm": 2.3873636722564697,
+      "learning_rate": 0.00010320987654320988,
+      "loss": 2.9273,
+      "step": 932
+    },
+    {
+      "epoch": 2.069883527454243,
+      "grad_norm": 2.150240182876587,
+      "learning_rate": 0.00010296296296296296,
+      "loss": 3.001,
+      "step": 933
+    },
+    {
+      "epoch": 2.07210205213533,
+      "grad_norm": 2.347534656524658,
+      "learning_rate": 0.00010271604938271604,
+      "loss": 3.3767,
+      "step": 934
+    },
+    {
+      "epoch": 2.074320576816417,
+      "grad_norm": 2.4626238346099854,
+      "learning_rate": 0.00010246913580246912,
+      "loss": 2.8833,
+      "step": 935
+    },
+    {
+      "epoch": 2.076539101497504,
+      "grad_norm": 2.221510648727417,
+      "learning_rate": 0.00010222222222222222,
+      "loss": 3.1204,
+      "step": 936
+    },
+    {
+      "epoch": 2.078757626178591,
+      "grad_norm": 1.949877381324768,
+      "learning_rate": 0.0001019753086419753,
+      "loss": 2.9516,
+      "step": 937
+    },
+    {
+      "epoch": 2.0809761508596782,
+      "grad_norm": 2.6168205738067627,
+      "learning_rate": 0.00010172839506172838,
+      "loss": 2.9053,
+      "step": 938
+    },
+    {
+      "epoch": 2.0831946755407653,
+      "grad_norm": 2.9459450244903564,
+      "learning_rate": 0.00010148148148148147,
+      "loss": 2.8432,
+      "step": 939
+    },
+    {
+      "epoch": 2.0854132002218524,
+      "grad_norm": 2.5966904163360596,
+      "learning_rate": 0.00010123456790123457,
+      "loss": 3.0394,
+      "step": 940
+    },
+    {
+      "epoch": 2.0876317249029395,
+      "grad_norm": 2.3656692504882812,
+      "learning_rate": 0.00010098765432098765,
+      "loss": 3.0831,
+      "step": 941
+    },
+    {
+      "epoch": 2.0898502495840265,
+      "grad_norm": 2.15684175491333,
+      "learning_rate": 0.00010074074074074073,
+      "loss": 3.0291,
+      "step": 942
+    },
+    {
+      "epoch": 2.0920687742651136,
+      "grad_norm": 2.491990804672241,
+      "learning_rate": 0.00010049382716049381,
+      "loss": 2.8369,
+      "step": 943
+    },
+    {
+      "epoch": 2.0942872989462007,
+      "grad_norm": 2.907151460647583,
+      "learning_rate": 0.0001002469135802469,
+      "loss": 3.1885,
+      "step": 944
+    },
+    {
+      "epoch": 2.0965058236272878,
+      "grad_norm": 2.092691421508789,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 3.1946,
+      "step": 945
+    },
+    {
+      "epoch": 2.098724348308375,
+      "grad_norm": 2.697848320007324,
+      "learning_rate": 9.975308641975307e-05,
+      "loss": 3.131,
+      "step": 946
+    },
+    {
+      "epoch": 2.100942872989462,
+      "grad_norm": 2.8459079265594482,
+      "learning_rate": 9.950617283950617e-05,
+      "loss": 3.1191,
+      "step": 947
+    },
+    {
+      "epoch": 2.103161397670549,
+      "grad_norm": 2.658607244491577,
+      "learning_rate": 9.925925925925925e-05,
+      "loss": 2.9494,
+      "step": 948
+    },
+    {
+      "epoch": 2.105379922351636,
+      "grad_norm": 2.620354652404785,
+      "learning_rate": 9.901234567901234e-05,
+      "loss": 3.1415,
+      "step": 949
+    },
+    {
+      "epoch": 2.107598447032723,
+      "grad_norm": 2.246278762817383,
+      "learning_rate": 9.876543209876543e-05,
+      "loss": 2.8907,
+      "step": 950
+    },
+    {
+      "epoch": 2.10981697171381,
+      "grad_norm": 2.3438401222229004,
+      "learning_rate": 9.851851851851851e-05,
+      "loss": 2.9242,
+      "step": 951
+    },
+    {
+      "epoch": 2.1120354963948973,
+      "grad_norm": 2.285099744796753,
+      "learning_rate": 9.827160493827159e-05,
+      "loss": 2.9428,
+      "step": 952
+    },
+    {
+      "epoch": 2.1142540210759844,
+      "grad_norm": 2.5277645587921143,
+      "learning_rate": 9.802469135802467e-05,
+      "loss": 3.1169,
+      "step": 953
+    },
+    {
+      "epoch": 2.1164725457570714,
+      "grad_norm": 2.0332071781158447,
+      "learning_rate": 9.777777777777778e-05,
+      "loss": 3.1394,
+      "step": 954
+    },
+    {
+      "epoch": 2.1186910704381585,
+      "grad_norm": 2.269857406616211,
+      "learning_rate": 9.753086419753086e-05,
+      "loss": 2.951,
+      "step": 955
+    },
+    {
+      "epoch": 2.1209095951192456,
+      "grad_norm": 2.1981325149536133,
+      "learning_rate": 9.728395061728394e-05,
+      "loss": 3.1965,
+      "step": 956
+    },
+    {
+      "epoch": 2.1231281198003327,
+      "grad_norm": 2.219668388366699,
+      "learning_rate": 9.703703703703702e-05,
+      "loss": 3.2734,
+      "step": 957
+    },
+    {
+      "epoch": 2.1253466444814197,
+      "grad_norm": 2.5858824253082275,
+      "learning_rate": 9.679012345679012e-05,
+      "loss": 2.8425,
+      "step": 958
+    },
+    {
+      "epoch": 2.127565169162507,
+      "grad_norm": 2.5191662311553955,
+      "learning_rate": 9.65432098765432e-05,
+      "loss": 3.0368,
+      "step": 959
+    },
+    {
+      "epoch": 2.129783693843594,
+      "grad_norm": 2.03165864944458,
+      "learning_rate": 9.629629629629628e-05,
+      "loss": 3.4238,
+      "step": 960
+    },
+    {
+      "epoch": 2.132002218524681,
+      "grad_norm": 2.2679100036621094,
+      "learning_rate": 9.604938271604936e-05,
+      "loss": 2.8963,
+      "step": 961
+    },
+    {
+      "epoch": 2.134220743205768,
+      "grad_norm": 2.8526623249053955,
+      "learning_rate": 9.580246913580247e-05,
+      "loss": 3.207,
+      "step": 962
+    },
+    {
+      "epoch": 2.136439267886855,
+      "grad_norm": 2.596304178237915,
+      "learning_rate": 9.555555555555555e-05,
+      "loss": 3.2852,
+      "step": 963
+    },
+    {
+      "epoch": 2.138657792567942,
+      "grad_norm": 2.5938079357147217,
+      "learning_rate": 9.530864197530864e-05,
+      "loss": 3.2404,
+      "step": 964
+    },
+    {
+      "epoch": 2.1408763172490293,
+      "grad_norm": 2.3335657119750977,
+      "learning_rate": 9.506172839506172e-05,
+      "loss": 3.1772,
+      "step": 965
+    },
+    {
+      "epoch": 2.1430948419301163,
+      "grad_norm": 2.5048797130584717,
+      "learning_rate": 9.48148148148148e-05,
+      "loss": 3.2188,
+      "step": 966
+    },
+    {
+      "epoch": 2.1453133666112034,
+      "grad_norm": 1.9693914651870728,
+      "learning_rate": 9.45679012345679e-05,
+      "loss": 3.0749,
+      "step": 967
+    },
+    {
+      "epoch": 2.1475318912922905,
+      "grad_norm": 2.382633686065674,
+      "learning_rate": 9.432098765432098e-05,
+      "loss": 3.3301,
+      "step": 968
+    },
+    {
+      "epoch": 2.1497504159733776,
+      "grad_norm": 2.0730350017547607,
+      "learning_rate": 9.407407407407407e-05,
+      "loss": 3.1685,
+      "step": 969
+    },
+    {
+      "epoch": 2.1519689406544646,
+      "grad_norm": 2.445553779602051,
+      "learning_rate": 9.382716049382715e-05,
+      "loss": 2.9153,
+      "step": 970
+    },
+    {
+      "epoch": 2.1541874653355517,
+      "grad_norm": 2.641284704208374,
+      "learning_rate": 9.358024691358025e-05,
+      "loss": 2.8389,
+      "step": 971
+    },
+    {
+      "epoch": 2.156405990016639,
+      "grad_norm": 2.4456334114074707,
+      "learning_rate": 9.333333333333333e-05,
+      "loss": 2.8505,
+      "step": 972
+    },
+    {
+      "epoch": 2.158624514697726,
+      "grad_norm": 2.1340582370758057,
+      "learning_rate": 9.308641975308641e-05,
+      "loss": 3.045,
+      "step": 973
+    },
+    {
+      "epoch": 2.160843039378813,
+      "grad_norm": 2.353271484375,
+      "learning_rate": 9.283950617283949e-05,
+      "loss": 2.9233,
+      "step": 974
+    },
+    {
+      "epoch": 2.1630615640599,
+      "grad_norm": 2.2293665409088135,
+      "learning_rate": 9.259259259259257e-05,
+      "loss": 3.37,
+      "step": 975
+    },
+    {
+      "epoch": 2.165280088740987,
+      "grad_norm": 2.1137473583221436,
+      "learning_rate": 9.234567901234567e-05,
+      "loss": 2.9968,
+      "step": 976
+    },
+    {
+      "epoch": 2.167498613422074,
+      "grad_norm": 2.6105940341949463,
+      "learning_rate": 9.209876543209876e-05,
+      "loss": 3.05,
+      "step": 977
+    },
+    {
+      "epoch": 2.1697171381031612,
+      "grad_norm": 2.2251529693603516,
+      "learning_rate": 9.185185185185185e-05,
+      "loss": 2.8708,
+      "step": 978
+    },
+    {
+      "epoch": 2.1719356627842483,
+      "grad_norm": 2.5924344062805176,
+      "learning_rate": 9.160493827160493e-05,
+      "loss": 3.2418,
+      "step": 979
+    },
+    {
+      "epoch": 2.1741541874653354,
+      "grad_norm": 2.8491077423095703,
+      "learning_rate": 9.135802469135802e-05,
+      "loss": 3.0412,
+      "step": 980
+    },
+    {
+      "epoch": 2.1763727121464225,
+      "grad_norm": 2.5781705379486084,
+      "learning_rate": 9.11111111111111e-05,
+      "loss": 2.9168,
+      "step": 981
+    },
+    {
+      "epoch": 2.1785912368275095,
+      "grad_norm": 2.611522674560547,
+      "learning_rate": 9.086419753086419e-05,
+      "loss": 2.8622,
+      "step": 982
+    },
+    {
+      "epoch": 2.1808097615085966,
+      "grad_norm": 2.2582457065582275,
+      "learning_rate": 9.061728395061727e-05,
+      "loss": 3.4442,
+      "step": 983
+    },
+    {
+      "epoch": 2.1830282861896837,
+      "grad_norm": 2.364129066467285,
+      "learning_rate": 9.037037037037038e-05,
+      "loss": 2.8342,
+      "step": 984
+    },
+    {
+      "epoch": 2.1852468108707708,
+      "grad_norm": 2.637563943862915,
+      "learning_rate": 9.012345679012346e-05,
+      "loss": 3.1865,
+      "step": 985
+    },
+    {
+      "epoch": 2.187465335551858,
+      "grad_norm": 2.6341774463653564,
+      "learning_rate": 8.987654320987654e-05,
+      "loss": 3.2955,
+      "step": 986
+    },
+    {
+      "epoch": 2.189683860232945,
+      "grad_norm": 2.751511573791504,
+      "learning_rate": 8.962962962962962e-05,
+      "loss": 2.5426,
+      "step": 987
+    },
+    {
+      "epoch": 2.191902384914032,
+      "grad_norm": 2.784860610961914,
+      "learning_rate": 8.93827160493827e-05,
+      "loss": 2.6359,
+      "step": 988
+    },
+    {
+      "epoch": 2.194120909595119,
+      "grad_norm": 2.5616204738616943,
+      "learning_rate": 8.91358024691358e-05,
+      "loss": 2.9537,
+      "step": 989
+    },
+    {
+      "epoch": 2.196339434276206,
+      "grad_norm": 2.8358137607574463,
+      "learning_rate": 8.888888888888888e-05,
+      "loss": 2.7283,
+      "step": 990
+    },
+    {
+      "epoch": 2.1985579589572932,
+      "grad_norm": 2.2988433837890625,
+      "learning_rate": 8.864197530864196e-05,
+      "loss": 2.8595,
+      "step": 991
+    },
+    {
+      "epoch": 2.2007764836383803,
+      "grad_norm": 2.8541152477264404,
+      "learning_rate": 8.839506172839506e-05,
+      "loss": 3.237,
+      "step": 992
+    },
+    {
+      "epoch": 2.2029950083194674,
+      "grad_norm": 2.5678491592407227,
+      "learning_rate": 8.814814814814815e-05,
+      "loss": 3.5585,
+      "step": 993
+    },
+    {
+      "epoch": 2.2052135330005544,
+      "grad_norm": 2.615285873413086,
+      "learning_rate": 8.790123456790123e-05,
+      "loss": 3.1296,
+      "step": 994
+    },
+    {
+      "epoch": 2.2074320576816415,
+      "grad_norm": 2.901531457901001,
+      "learning_rate": 8.765432098765431e-05,
+      "loss": 2.9591,
+      "step": 995
+    },
+    {
+      "epoch": 2.2096505823627286,
+      "grad_norm": 2.539302349090576,
+      "learning_rate": 8.74074074074074e-05,
+      "loss": 2.9694,
+      "step": 996
+    },
+    {
+      "epoch": 2.2118691070438157,
+      "grad_norm": 2.0795247554779053,
+      "learning_rate": 8.716049382716048e-05,
+      "loss": 2.9574,
+      "step": 997
+    },
+    {
+      "epoch": 2.2140876317249027,
+      "grad_norm": 2.391315221786499,
+      "learning_rate": 8.691358024691357e-05,
+      "loss": 3.1464,
+      "step": 998
+    },
+    {
+      "epoch": 2.21630615640599,
+      "grad_norm": 2.4845428466796875,
+      "learning_rate": 8.666666666666665e-05,
+      "loss": 2.6557,
+      "step": 999
+    },
+    {
+      "epoch": 2.218524681087077,
+      "grad_norm": 2.374021291732788,
+      "learning_rate": 8.641975308641975e-05,
+      "loss": 2.8068,
+      "step": 1000
+    },
+    {
+      "epoch": 2.220743205768164,
+      "grad_norm": 2.729522943496704,
+      "learning_rate": 8.617283950617283e-05,
+      "loss": 2.9163,
+      "step": 1001
+    },
+    {
+      "epoch": 2.222961730449251,
+      "grad_norm": 2.244389295578003,
+      "learning_rate": 8.592592592592593e-05,
+      "loss": 3.1338,
+      "step": 1002
+    },
+    {
+      "epoch": 2.225180255130338,
+      "grad_norm": 2.3017594814300537,
+      "learning_rate": 8.567901234567901e-05,
+      "loss": 2.8841,
+      "step": 1003
+    },
+    {
+      "epoch": 2.227398779811425,
+      "grad_norm": 2.426757574081421,
+      "learning_rate": 8.543209876543209e-05,
+      "loss": 2.9542,
+      "step": 1004
+    },
+    {
+      "epoch": 2.2296173044925123,
+      "grad_norm": 2.3076138496398926,
+      "learning_rate": 8.518518518518517e-05,
+      "loss": 2.9194,
+      "step": 1005
+    },
+    {
+      "epoch": 2.2318358291735994,
+      "grad_norm": 2.3723337650299072,
+      "learning_rate": 8.493827160493825e-05,
+      "loss": 3.1473,
+      "step": 1006
+    },
+    {
+      "epoch": 2.2340543538546864,
+      "grad_norm": 2.6738624572753906,
+      "learning_rate": 8.469135802469136e-05,
+      "loss": 3.1645,
+      "step": 1007
+    },
+    {
+      "epoch": 2.2362728785357735,
+      "grad_norm": 2.784135580062866,
+      "learning_rate": 8.444444444444444e-05,
+      "loss": 3.2791,
+      "step": 1008
+    },
+    {
+      "epoch": 2.2384914032168606,
+      "grad_norm": 3.1843655109405518,
+      "learning_rate": 8.419753086419752e-05,
+      "loss": 3.4131,
+      "step": 1009
+    },
+    {
+      "epoch": 2.2407099278979477,
+      "grad_norm": 2.6491172313690186,
+      "learning_rate": 8.39506172839506e-05,
+      "loss": 3.1418,
+      "step": 1010
+    },
+    {
+      "epoch": 2.2429284525790347,
+      "grad_norm": 3.162569999694824,
+      "learning_rate": 8.37037037037037e-05,
+      "loss": 3.0343,
+      "step": 1011
+    },
+    {
+      "epoch": 2.245146977260122,
+      "grad_norm": 2.2459158897399902,
+      "learning_rate": 8.345679012345678e-05,
+      "loss": 3.0834,
+      "step": 1012
+    },
+    {
+      "epoch": 2.247365501941209,
+      "grad_norm": 2.4236786365509033,
+      "learning_rate": 8.320987654320986e-05,
+      "loss": 2.6568,
+      "step": 1013
+    },
+    {
+      "epoch": 2.249584026622296,
+      "grad_norm": 2.248076915740967,
+      "learning_rate": 8.296296296296295e-05,
+      "loss": 3.3471,
+      "step": 1014
+    },
+    {
+      "epoch": 2.251802551303383,
+      "grad_norm": 2.8372907638549805,
+      "learning_rate": 8.271604938271605e-05,
+      "loss": 3.4046,
+      "step": 1015
+    },
+    {
+      "epoch": 2.25402107598447,
+      "grad_norm": 2.61704683303833,
+      "learning_rate": 8.246913580246914e-05,
+      "loss": 3.2896,
+      "step": 1016
+    },
+    {
+      "epoch": 2.256239600665557,
+      "grad_norm": 2.3494138717651367,
+      "learning_rate": 8.222222222222222e-05,
+      "loss": 3.161,
+      "step": 1017
+    },
+    {
+      "epoch": 2.2584581253466443,
+      "grad_norm": 2.5113844871520996,
+      "learning_rate": 8.19753086419753e-05,
+      "loss": 3.2732,
+      "step": 1018
+    },
+    {
+      "epoch": 2.2606766500277313,
+      "grad_norm": 2.405565023422241,
+      "learning_rate": 8.172839506172838e-05,
+      "loss": 3.0332,
+      "step": 1019
+    },
+    {
+      "epoch": 2.2628951747088184,
+      "grad_norm": 2.9170596599578857,
+      "learning_rate": 8.148148148148148e-05,
+      "loss": 3.3439,
+      "step": 1020
+    },
+    {
+      "epoch": 2.2651136993899055,
+      "grad_norm": 2.394394874572754,
+      "learning_rate": 8.123456790123456e-05,
+      "loss": 3.0803,
+      "step": 1021
+    },
+    {
+      "epoch": 2.2673322240709926,
+      "grad_norm": 2.3557732105255127,
+      "learning_rate": 8.098765432098765e-05,
+      "loss": 2.6694,
+      "step": 1022
+    },
+    {
+      "epoch": 2.2695507487520796,
+      "grad_norm": 2.7595267295837402,
+      "learning_rate": 8.074074074074073e-05,
+      "loss": 3.1305,
+      "step": 1023
+    },
+    {
+      "epoch": 2.2717692734331667,
+      "grad_norm": 2.4323978424072266,
+      "learning_rate": 8.049382716049383e-05,
+      "loss": 2.903,
+      "step": 1024
+    },
+    {
+      "epoch": 2.2739877981142538,
+      "grad_norm": 2.471219062805176,
+      "learning_rate": 8.024691358024691e-05,
+      "loss": 3.093,
+      "step": 1025
+    },
+    {
+      "epoch": 2.276206322795341,
+      "grad_norm": 2.543767213821411,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 3.186,
+      "step": 1026
+    },
+    {
+      "epoch": 2.278424847476428,
+      "grad_norm": 2.462327241897583,
+      "learning_rate": 7.975308641975307e-05,
+      "loss": 2.7276,
+      "step": 1027
+    },
+    {
+      "epoch": 2.280643372157515,
+      "grad_norm": 2.8636066913604736,
+      "learning_rate": 7.950617283950616e-05,
+      "loss": 3.0477,
+      "step": 1028
+    },
+    {
+      "epoch": 2.2828618968386025,
+      "grad_norm": 2.9679625034332275,
+      "learning_rate": 7.925925925925925e-05,
+      "loss": 2.9004,
+      "step": 1029
+    },
+    {
+      "epoch": 2.2850804215196896,
+      "grad_norm": 2.494961738586426,
+      "learning_rate": 7.901234567901235e-05,
+      "loss": 3.2052,
+      "step": 1030
+    },
+    {
+      "epoch": 2.2872989462007767,
+      "grad_norm": 2.7159926891326904,
+      "learning_rate": 7.876543209876543e-05,
+      "loss": 3.2856,
+      "step": 1031
+    },
+    {
+      "epoch": 2.2895174708818637,
+      "grad_norm": 2.771345853805542,
+      "learning_rate": 7.851851851851851e-05,
+      "loss": 3.2445,
+      "step": 1032
+    },
+    {
+      "epoch": 2.291735995562951,
+      "grad_norm": 2.2868783473968506,
+      "learning_rate": 7.82716049382716e-05,
+      "loss": 2.7348,
+      "step": 1033
+    },
+    {
+      "epoch": 2.293954520244038,
+      "grad_norm": 2.480180263519287,
+      "learning_rate": 7.802469135802469e-05,
+      "loss": 3.0598,
+      "step": 1034
+    },
+    {
+      "epoch": 2.296173044925125,
+      "grad_norm": 2.355255365371704,
+      "learning_rate": 7.777777777777777e-05,
+      "loss": 3.0937,
+      "step": 1035
+    },
+    {
+      "epoch": 2.298391569606212,
+      "grad_norm": 2.320091724395752,
+      "learning_rate": 7.753086419753085e-05,
+      "loss": 3.2068,
+      "step": 1036
+    },
+    {
+      "epoch": 2.300610094287299,
+      "grad_norm": 2.1861279010772705,
+      "learning_rate": 7.728395061728396e-05,
+      "loss": 2.7695,
+      "step": 1037
+    },
+    {
+      "epoch": 2.302828618968386,
+      "grad_norm": 2.673572301864624,
+      "learning_rate": 7.703703703703704e-05,
+      "loss": 3.3346,
+      "step": 1038
+    },
+    {
+      "epoch": 2.3050471436494733,
+      "grad_norm": 2.2038300037384033,
+      "learning_rate": 7.679012345679012e-05,
+      "loss": 2.7366,
+      "step": 1039
+    },
+    {
+      "epoch": 2.3072656683305603,
+      "grad_norm": 2.8647749423980713,
+      "learning_rate": 7.65432098765432e-05,
+      "loss": 3.0111,
+      "step": 1040
+    },
+    {
+      "epoch": 2.3094841930116474,
+      "grad_norm": 2.5659563541412354,
+      "learning_rate": 7.629629629629628e-05,
+      "loss": 3.3311,
+      "step": 1041
+    },
+    {
+      "epoch": 2.3117027176927345,
+      "grad_norm": 2.279355764389038,
+      "learning_rate": 7.604938271604937e-05,
+      "loss": 3.1023,
+      "step": 1042
+    },
+    {
+      "epoch": 2.3139212423738216,
+      "grad_norm": 2.377687931060791,
+      "learning_rate": 7.580246913580246e-05,
+      "loss": 2.8634,
+      "step": 1043
+    },
+    {
+      "epoch": 2.3161397670549086,
+      "grad_norm": 1.9413493871688843,
+      "learning_rate": 7.555555555555554e-05,
+      "loss": 2.8007,
+      "step": 1044
+    },
+    {
+      "epoch": 2.3183582917359957,
+      "grad_norm": 3.0563607215881348,
+      "learning_rate": 7.530864197530864e-05,
+      "loss": 3.1729,
+      "step": 1045
+    },
+    {
+      "epoch": 2.320576816417083,
+      "grad_norm": 2.476712703704834,
+      "learning_rate": 7.506172839506173e-05,
+      "loss": 3.1278,
+      "step": 1046
+    },
+    {
+      "epoch": 2.32279534109817,
+      "grad_norm": 2.3085269927978516,
+      "learning_rate": 7.481481481481481e-05,
+      "loss": 2.987,
+      "step": 1047
+    },
+    {
+      "epoch": 2.325013865779257,
+      "grad_norm": 2.102741241455078,
+      "learning_rate": 7.45679012345679e-05,
+      "loss": 2.8794,
+      "step": 1048
+    },
+    {
+      "epoch": 2.327232390460344,
+      "grad_norm": 2.8451223373413086,
+      "learning_rate": 7.432098765432098e-05,
+      "loss": 2.7838,
+      "step": 1049
+    },
+    {
+      "epoch": 2.329450915141431,
+      "grad_norm": 2.403120517730713,
+      "learning_rate": 7.407407407407406e-05,
+      "loss": 2.8507,
+      "step": 1050
+    },
+    {
+      "epoch": 2.331669439822518,
+      "grad_norm": 2.1391217708587646,
+      "learning_rate": 7.382716049382715e-05,
+      "loss": 3.1007,
+      "step": 1051
+    },
+    {
+      "epoch": 2.3338879645036053,
+      "grad_norm": 2.4717769622802734,
+      "learning_rate": 7.358024691358024e-05,
+      "loss": 2.7733,
+      "step": 1052
+    },
+    {
+      "epoch": 2.3361064891846923,
+      "grad_norm": 2.5240225791931152,
+      "learning_rate": 7.333333333333332e-05,
+      "loss": 3.3232,
+      "step": 1053
+    },
+    {
+      "epoch": 2.3383250138657794,
+      "grad_norm": 2.321047782897949,
+      "learning_rate": 7.308641975308641e-05,
+      "loss": 2.9778,
+      "step": 1054
+    },
+    {
+      "epoch": 2.3405435385468665,
+      "grad_norm": 2.9097964763641357,
+      "learning_rate": 7.28395061728395e-05,
+      "loss": 2.9519,
+      "step": 1055
+    },
+    {
+      "epoch": 2.3427620632279536,
+      "grad_norm": 2.6968657970428467,
+      "learning_rate": 7.259259259259259e-05,
+      "loss": 3.2222,
+      "step": 1056
+    },
+    {
+      "epoch": 2.3449805879090406,
+      "grad_norm": 2.7298829555511475,
+      "learning_rate": 7.234567901234567e-05,
+      "loss": 2.9808,
+      "step": 1057
+    },
+    {
+      "epoch": 2.3471991125901277,
+      "grad_norm": 2.3547677993774414,
+      "learning_rate": 7.209876543209877e-05,
+      "loss": 3.2586,
+      "step": 1058
+    },
+    {
+      "epoch": 2.3494176372712148,
+      "grad_norm": 2.504459857940674,
+      "learning_rate": 7.185185185185185e-05,
+      "loss": 2.9886,
+      "step": 1059
+    },
+    {
+      "epoch": 2.351636161952302,
+      "grad_norm": 2.6193668842315674,
+      "learning_rate": 7.160493827160493e-05,
+      "loss": 2.801,
+      "step": 1060
+    },
+    {
+      "epoch": 2.353854686633389,
+      "grad_norm": 2.1980013847351074,
+      "learning_rate": 7.135802469135801e-05,
+      "loss": 3.0578,
+      "step": 1061
+    },
+    {
+      "epoch": 2.356073211314476,
+      "grad_norm": 2.288282871246338,
+      "learning_rate": 7.11111111111111e-05,
+      "loss": 2.9089,
+      "step": 1062
+    },
+    {
+      "epoch": 2.358291735995563,
+      "grad_norm": 2.514753580093384,
+      "learning_rate": 7.086419753086419e-05,
+      "loss": 2.9768,
+      "step": 1063
+    },
+    {
+      "epoch": 2.36051026067665,
+      "grad_norm": 2.2984743118286133,
+      "learning_rate": 7.061728395061727e-05,
+      "loss": 3.0828,
+      "step": 1064
+    },
+    {
+      "epoch": 2.3627287853577372,
+      "grad_norm": 2.7699968814849854,
+      "learning_rate": 7.037037037037036e-05,
+      "loss": 2.8978,
+      "step": 1065
+    },
+    {
+      "epoch": 2.3649473100388243,
+      "grad_norm": 2.109652519226074,
+      "learning_rate": 7.012345679012345e-05,
+      "loss": 2.8,
+      "step": 1066
+    },
+    {
+      "epoch": 2.3671658347199114,
+      "grad_norm": 2.1589431762695312,
+      "learning_rate": 6.987654320987654e-05,
+      "loss": 2.8574,
+      "step": 1067
+    },
+    {
+      "epoch": 2.3693843594009985,
+      "grad_norm": 2.2922093868255615,
+      "learning_rate": 6.962962962962962e-05,
+      "loss": 3.0826,
+      "step": 1068
+    },
+    {
+      "epoch": 2.3716028840820855,
+      "grad_norm": 2.589308977127075,
+      "learning_rate": 6.938271604938272e-05,
+      "loss": 3.1291,
+      "step": 1069
+    },
+    {
+      "epoch": 2.3738214087631726,
+      "grad_norm": 2.4845407009124756,
+      "learning_rate": 6.91358024691358e-05,
+      "loss": 2.8985,
+      "step": 1070
+    },
+    {
+      "epoch": 2.3760399334442597,
+      "grad_norm": 2.586920738220215,
+      "learning_rate": 6.888888888888888e-05,
+      "loss": 3.3055,
+      "step": 1071
+    },
+    {
+      "epoch": 2.3782584581253468,
+      "grad_norm": 2.4002463817596436,
+      "learning_rate": 6.864197530864196e-05,
+      "loss": 3.0742,
+      "step": 1072
+    },
+    {
+      "epoch": 2.380476982806434,
+      "grad_norm": 2.452054023742676,
+      "learning_rate": 6.839506172839506e-05,
+      "loss": 3.2215,
+      "step": 1073
+    },
+    {
+      "epoch": 2.382695507487521,
+      "grad_norm": 2.2019588947296143,
+      "learning_rate": 6.814814814814814e-05,
+      "loss": 3.0387,
+      "step": 1074
+    },
+    {
+      "epoch": 2.384914032168608,
+      "grad_norm": 2.564237117767334,
+      "learning_rate": 6.790123456790122e-05,
+      "loss": 2.6154,
+      "step": 1075
+    },
+    {
+      "epoch": 2.387132556849695,
+      "grad_norm": 2.7909929752349854,
+      "learning_rate": 6.765432098765432e-05,
+      "loss": 3.0282,
+      "step": 1076
+    },
+    {
+      "epoch": 2.389351081530782,
+      "grad_norm": 2.5985045433044434,
+      "learning_rate": 6.74074074074074e-05,
+      "loss": 3.5433,
+      "step": 1077
+    },
+    {
+      "epoch": 2.391569606211869,
+      "grad_norm": 2.645285129547119,
+      "learning_rate": 6.716049382716049e-05,
+      "loss": 3.2962,
+      "step": 1078
+    },
+    {
+      "epoch": 2.3937881308929563,
+      "grad_norm": 2.8125417232513428,
+      "learning_rate": 6.691358024691357e-05,
+      "loss": 3.0709,
+      "step": 1079
+    },
+    {
+      "epoch": 2.3960066555740434,
+      "grad_norm": 2.8634421825408936,
+      "learning_rate": 6.666666666666666e-05,
+      "loss": 3.145,
+      "step": 1080
+    },
+    {
+      "epoch": 2.3982251802551304,
+      "grad_norm": 2.3443925380706787,
+      "learning_rate": 6.641975308641975e-05,
+      "loss": 2.9921,
+      "step": 1081
+    },
+    {
+      "epoch": 2.4004437049362175,
+      "grad_norm": 2.355220317840576,
+      "learning_rate": 6.617283950617283e-05,
+      "loss": 2.9579,
+      "step": 1082
+    },
+    {
+      "epoch": 2.4026622296173046,
+      "grad_norm": 2.2041866779327393,
+      "learning_rate": 6.592592592592591e-05,
+      "loss": 3.0922,
+      "step": 1083
+    },
+    {
+      "epoch": 2.4048807542983917,
+      "grad_norm": 2.550198793411255,
+      "learning_rate": 6.567901234567901e-05,
+      "loss": 2.9675,
+      "step": 1084
+    },
+    {
+      "epoch": 2.4070992789794787,
+      "grad_norm": 2.2701869010925293,
+      "learning_rate": 6.543209876543209e-05,
+      "loss": 3.0052,
+      "step": 1085
+    },
+    {
+      "epoch": 2.409317803660566,
+      "grad_norm": 2.784285306930542,
+      "learning_rate": 6.518518518518517e-05,
+      "loss": 2.7432,
+      "step": 1086
+    },
+    {
+      "epoch": 2.411536328341653,
+      "grad_norm": 2.6419291496276855,
+      "learning_rate": 6.493827160493827e-05,
+      "loss": 2.9094,
+      "step": 1087
+    },
+    {
+      "epoch": 2.41375485302274,
+      "grad_norm": 2.670732259750366,
+      "learning_rate": 6.469135802469135e-05,
+      "loss": 2.8574,
+      "step": 1088
+    },
+    {
+      "epoch": 2.415973377703827,
+      "grad_norm": 2.558919906616211,
+      "learning_rate": 6.444444444444444e-05,
+      "loss": 3.0009,
+      "step": 1089
+    },
+    {
+      "epoch": 2.418191902384914,
+      "grad_norm": 2.5176033973693848,
+      "learning_rate": 6.419753086419753e-05,
+      "loss": 3.0005,
+      "step": 1090
+    },
+    {
+      "epoch": 2.420410427066001,
+      "grad_norm": 2.6639552116394043,
+      "learning_rate": 6.395061728395061e-05,
+      "loss": 3.0692,
+      "step": 1091
+    },
+    {
+      "epoch": 2.4226289517470883,
+      "grad_norm": 2.133502960205078,
+      "learning_rate": 6.37037037037037e-05,
+      "loss": 2.9404,
+      "step": 1092
+    },
+    {
+      "epoch": 2.4248474764281753,
+      "grad_norm": 2.8920702934265137,
+      "learning_rate": 6.345679012345678e-05,
+      "loss": 3.1042,
+      "step": 1093
+    },
+    {
+      "epoch": 2.4270660011092624,
+      "grad_norm": 3.20930552482605,
+      "learning_rate": 6.320987654320987e-05,
+      "loss": 3.1603,
+      "step": 1094
+    },
+    {
+      "epoch": 2.4292845257903495,
+      "grad_norm": 2.3886044025421143,
+      "learning_rate": 6.296296296296295e-05,
+      "loss": 3.1767,
+      "step": 1095
+    },
+    {
+      "epoch": 2.4315030504714366,
+      "grad_norm": 2.3401923179626465,
+      "learning_rate": 6.271604938271604e-05,
+      "loss": 3.1975,
+      "step": 1096
+    },
+    {
+      "epoch": 2.4337215751525236,
+      "grad_norm": 2.387327194213867,
+      "learning_rate": 6.246913580246912e-05,
+      "loss": 2.8585,
+      "step": 1097
+    },
+    {
+      "epoch": 2.4359400998336107,
+      "grad_norm": 2.9216222763061523,
+      "learning_rate": 6.222222222222222e-05,
+      "loss": 2.9352,
+      "step": 1098
+    },
+    {
+      "epoch": 2.438158624514698,
+      "grad_norm": 2.4063773155212402,
+      "learning_rate": 6.19753086419753e-05,
+      "loss": 2.8475,
+      "step": 1099
+    },
+    {
+      "epoch": 2.440377149195785,
+      "grad_norm": 2.800044536590576,
+      "learning_rate": 6.17283950617284e-05,
+      "loss": 3.028,
+      "step": 1100
+    },
+    {
+      "epoch": 2.442595673876872,
+      "grad_norm": 2.433523654937744,
+      "learning_rate": 6.148148148148148e-05,
+      "loss": 2.9733,
+      "step": 1101
+    },
+    {
+      "epoch": 2.444814198557959,
+      "grad_norm": 2.4469356536865234,
+      "learning_rate": 6.123456790123456e-05,
+      "loss": 2.9869,
+      "step": 1102
+    },
+    {
+      "epoch": 2.447032723239046,
+      "grad_norm": 2.195274829864502,
+      "learning_rate": 6.0987654320987655e-05,
+      "loss": 2.863,
+      "step": 1103
+    },
+    {
+      "epoch": 2.449251247920133,
+      "grad_norm": 2.2709670066833496,
+      "learning_rate": 6.0740740740740736e-05,
+      "loss": 2.8681,
+      "step": 1104
+    },
+    {
+      "epoch": 2.4514697726012202,
+      "grad_norm": 2.2619011402130127,
+      "learning_rate": 6.049382716049382e-05,
+      "loss": 2.7834,
+      "step": 1105
+    },
+    {
+      "epoch": 2.4536882972823073,
+      "grad_norm": 2.547245502471924,
+      "learning_rate": 6.0246913580246906e-05,
+      "loss": 2.7976,
+      "step": 1106
+    },
+    {
+      "epoch": 2.4559068219633944,
+      "grad_norm": 2.7446959018707275,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 3.246,
+      "step": 1107
+    },
+    {
+      "epoch": 2.4581253466444815,
+      "grad_norm": 2.5244898796081543,
+      "learning_rate": 5.975308641975308e-05,
+      "loss": 2.8405,
+      "step": 1108
+    },
+    {
+      "epoch": 2.4603438713255685,
+      "grad_norm": 2.8715922832489014,
+      "learning_rate": 5.9506172839506165e-05,
+      "loss": 2.7005,
+      "step": 1109
+    },
+    {
+      "epoch": 2.4625623960066556,
+      "grad_norm": 2.6053428649902344,
+      "learning_rate": 5.925925925925925e-05,
+      "loss": 3.0529,
+      "step": 1110
+    },
+    {
+      "epoch": 2.4647809206877427,
+      "grad_norm": 2.495351791381836,
+      "learning_rate": 5.901234567901234e-05,
+      "loss": 2.9272,
+      "step": 1111
+    },
+    {
+      "epoch": 2.4669994453688298,
+      "grad_norm": 2.243251085281372,
+      "learning_rate": 5.876543209876543e-05,
+      "loss": 2.7177,
+      "step": 1112
+    },
+    {
+      "epoch": 2.469217970049917,
+      "grad_norm": 2.6136093139648438,
+      "learning_rate": 5.851851851851851e-05,
+      "loss": 3.1706,
+      "step": 1113
+    },
+    {
+      "epoch": 2.471436494731004,
+      "grad_norm": 2.454367160797119,
+      "learning_rate": 5.827160493827159e-05,
+      "loss": 3.1199,
+      "step": 1114
+    },
+    {
+      "epoch": 2.473655019412091,
+      "grad_norm": 2.5593349933624268,
+      "learning_rate": 5.802469135802469e-05,
+      "loss": 2.7024,
+      "step": 1115
+    },
+    {
+      "epoch": 2.475873544093178,
+      "grad_norm": 2.5002529621124268,
+      "learning_rate": 5.777777777777777e-05,
+      "loss": 3.3229,
+      "step": 1116
+    },
+    {
+      "epoch": 2.478092068774265,
+      "grad_norm": 2.414252519607544,
+      "learning_rate": 5.753086419753086e-05,
+      "loss": 2.9833,
+      "step": 1117
+    },
+    {
+      "epoch": 2.480310593455352,
+      "grad_norm": 2.41459584236145,
+      "learning_rate": 5.7283950617283946e-05,
+      "loss": 2.8799,
+      "step": 1118
+    },
+    {
+      "epoch": 2.4825291181364393,
+      "grad_norm": 2.31575608253479,
+      "learning_rate": 5.7037037037037035e-05,
+      "loss": 3.1272,
+      "step": 1119
+    },
+    {
+      "epoch": 2.4847476428175264,
+      "grad_norm": 2.4193599224090576,
+      "learning_rate": 5.6790123456790116e-05,
+      "loss": 3.123,
+      "step": 1120
+    },
+    {
+      "epoch": 2.4869661674986134,
+      "grad_norm": 3.0481481552124023,
+      "learning_rate": 5.6543209876543205e-05,
+      "loss": 3.1618,
+      "step": 1121
+    },
+    {
+      "epoch": 2.4891846921797005,
+      "grad_norm": 2.6053123474121094,
+      "learning_rate": 5.629629629629629e-05,
+      "loss": 2.9669,
+      "step": 1122
+    },
+    {
+      "epoch": 2.4914032168607876,
+      "grad_norm": 2.4152719974517822,
+      "learning_rate": 5.604938271604938e-05,
+      "loss": 2.7876,
+      "step": 1123
+    },
+    {
+      "epoch": 2.4936217415418747,
+      "grad_norm": 2.1596322059631348,
+      "learning_rate": 5.580246913580246e-05,
+      "loss": 3.0048,
+      "step": 1124
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "grad_norm": 2.275787115097046,
+      "learning_rate": 5.5555555555555545e-05,
+      "loss": 3.3921,
+      "step": 1125
+    },
+    {
+      "epoch": 2.498058790904049,
+      "grad_norm": 2.405470371246338,
+      "learning_rate": 5.530864197530864e-05,
+      "loss": 3.3039,
+      "step": 1126
+    },
+    {
+      "epoch": 2.500277315585136,
+      "grad_norm": 2.4096622467041016,
+      "learning_rate": 5.506172839506172e-05,
+      "loss": 2.8424,
+      "step": 1127
+    },
+    {
+      "epoch": 2.502495840266223,
+      "grad_norm": 2.356668472290039,
+      "learning_rate": 5.481481481481481e-05,
+      "loss": 2.8421,
+      "step": 1128
+    },
+    {
+      "epoch": 2.50471436494731,
+      "grad_norm": 2.44500732421875,
+      "learning_rate": 5.456790123456789e-05,
+      "loss": 3.116,
+      "step": 1129
+    },
+    {
+      "epoch": 2.506932889628397,
+      "grad_norm": 2.7749428749084473,
+      "learning_rate": 5.4320987654320986e-05,
+      "loss": 2.9122,
+      "step": 1130
+    },
+    {
+      "epoch": 2.509151414309484,
+      "grad_norm": 3.0668785572052,
+      "learning_rate": 5.407407407407407e-05,
+      "loss": 3.1754,
+      "step": 1131
+    },
+    {
+      "epoch": 2.5113699389905713,
+      "grad_norm": 2.4647812843322754,
+      "learning_rate": 5.3827160493827156e-05,
+      "loss": 2.9842,
+      "step": 1132
+    },
+    {
+      "epoch": 2.5135884636716583,
+      "grad_norm": 2.413116216659546,
+      "learning_rate": 5.358024691358024e-05,
+      "loss": 3.023,
+      "step": 1133
+    },
+    {
+      "epoch": 2.5158069883527454,
+      "grad_norm": 2.5170347690582275,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 3.1885,
+      "step": 1134
+    },
+    {
+      "epoch": 2.5180255130338325,
+      "grad_norm": 2.3500490188598633,
+      "learning_rate": 5.3086419753086415e-05,
+      "loss": 2.8563,
+      "step": 1135
+    },
+    {
+      "epoch": 2.5202440377149196,
+      "grad_norm": 2.582380533218384,
+      "learning_rate": 5.2839506172839496e-05,
+      "loss": 3.1244,
+      "step": 1136
+    },
+    {
+      "epoch": 2.5224625623960066,
+      "grad_norm": 2.237534761428833,
+      "learning_rate": 5.259259259259259e-05,
+      "loss": 3.0836,
+      "step": 1137
+    },
+    {
+      "epoch": 2.5246810870770937,
+      "grad_norm": 2.0035786628723145,
+      "learning_rate": 5.234567901234567e-05,
+      "loss": 3.0027,
+      "step": 1138
+    },
+    {
+      "epoch": 2.526899611758181,
+      "grad_norm": 2.095059394836426,
+      "learning_rate": 5.209876543209876e-05,
+      "loss": 3.2636,
+      "step": 1139
+    },
+    {
+      "epoch": 2.529118136439268,
+      "grad_norm": 2.5481975078582764,
+      "learning_rate": 5.185185185185184e-05,
+      "loss": 2.9341,
+      "step": 1140
+    },
+    {
+      "epoch": 2.531336661120355,
+      "grad_norm": 3.2143681049346924,
+      "learning_rate": 5.160493827160494e-05,
+      "loss": 2.8663,
+      "step": 1141
+    },
+    {
+      "epoch": 2.533555185801442,
+      "grad_norm": 2.895102024078369,
+      "learning_rate": 5.135802469135802e-05,
+      "loss": 2.9366,
+      "step": 1142
+    },
+    {
+      "epoch": 2.535773710482529,
+      "grad_norm": 2.534479856491089,
+      "learning_rate": 5.111111111111111e-05,
+      "loss": 3.1292,
+      "step": 1143
+    },
+    {
+      "epoch": 2.537992235163616,
+      "grad_norm": 2.1710071563720703,
+      "learning_rate": 5.086419753086419e-05,
+      "loss": 3.3619,
+      "step": 1144
+    },
+    {
+      "epoch": 2.5402107598447032,
+      "grad_norm": 2.081122875213623,
+      "learning_rate": 5.0617283950617285e-05,
+      "loss": 2.9229,
+      "step": 1145
+    },
+    {
+      "epoch": 2.5424292845257903,
+      "grad_norm": 2.3765578269958496,
+      "learning_rate": 5.0370370370370366e-05,
+      "loss": 3.0644,
+      "step": 1146
+    },
+    {
+      "epoch": 2.5446478092068774,
+      "grad_norm": 2.321525812149048,
+      "learning_rate": 5.012345679012345e-05,
+      "loss": 3.1902,
+      "step": 1147
+    },
+    {
+      "epoch": 2.5468663338879645,
+      "grad_norm": 2.4098563194274902,
+      "learning_rate": 4.9876543209876536e-05,
+      "loss": 3.0549,
+      "step": 1148
+    },
+    {
+      "epoch": 2.5490848585690515,
+      "grad_norm": 2.6692888736724854,
+      "learning_rate": 4.9629629629629625e-05,
+      "loss": 3.0831,
+      "step": 1149
+    },
+    {
+      "epoch": 2.5513033832501386,
+      "grad_norm": 2.2564215660095215,
+      "learning_rate": 4.938271604938271e-05,
+      "loss": 2.842,
+      "step": 1150
+    },
+    {
+      "epoch": 2.5535219079312257,
+      "grad_norm": 2.155141830444336,
+      "learning_rate": 4.9135802469135795e-05,
+      "loss": 2.8531,
+      "step": 1151
+    },
+    {
+      "epoch": 2.5557404326123128,
+      "grad_norm": 2.769652843475342,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 2.8713,
+      "step": 1152
+    },
+    {
+      "epoch": 2.5579589572934,
+      "grad_norm": 2.8415534496307373,
+      "learning_rate": 4.864197530864197e-05,
+      "loss": 3.0878,
+      "step": 1153
+    },
+    {
+      "epoch": 2.560177481974487,
+      "grad_norm": 3.436556816101074,
+      "learning_rate": 4.839506172839506e-05,
+      "loss": 3.2311,
+      "step": 1154
+    },
+    {
+      "epoch": 2.562396006655574,
+      "grad_norm": 2.278308391571045,
+      "learning_rate": 4.814814814814814e-05,
+      "loss": 3.0201,
+      "step": 1155
+    },
+    {
+      "epoch": 2.564614531336661,
+      "grad_norm": 2.709735155105591,
+      "learning_rate": 4.7901234567901237e-05,
+      "loss": 2.8499,
+      "step": 1156
+    },
+    {
+      "epoch": 2.566833056017748,
+      "grad_norm": 2.787097930908203,
+      "learning_rate": 4.765432098765432e-05,
+      "loss": 2.8298,
+      "step": 1157
+    },
+    {
+      "epoch": 2.569051580698835,
+      "grad_norm": 2.6934316158294678,
+      "learning_rate": 4.74074074074074e-05,
+      "loss": 3.3395,
+      "step": 1158
+    },
+    {
+      "epoch": 2.5712701053799223,
+      "grad_norm": 2.8503518104553223,
+      "learning_rate": 4.716049382716049e-05,
+      "loss": 2.8813,
+      "step": 1159
+    },
+    {
+      "epoch": 2.5734886300610094,
+      "grad_norm": 2.600572347640991,
+      "learning_rate": 4.6913580246913576e-05,
+      "loss": 2.9465,
+      "step": 1160
+    },
+    {
+      "epoch": 2.5757071547420964,
+      "grad_norm": 2.409541130065918,
+      "learning_rate": 4.6666666666666665e-05,
+      "loss": 3.2985,
+      "step": 1161
+    },
+    {
+      "epoch": 2.5779256794231835,
+      "grad_norm": 2.302882432937622,
+      "learning_rate": 4.6419753086419746e-05,
+      "loss": 2.8978,
+      "step": 1162
+    },
+    {
+      "epoch": 2.5801442041042706,
+      "grad_norm": 2.4936366081237793,
+      "learning_rate": 4.6172839506172835e-05,
+      "loss": 2.9222,
+      "step": 1163
+    },
+    {
+      "epoch": 2.5823627287853577,
+      "grad_norm": 2.113551139831543,
+      "learning_rate": 4.592592592592592e-05,
+      "loss": 3.0318,
+      "step": 1164
+    },
+    {
+      "epoch": 2.5845812534664447,
+      "grad_norm": 2.7801172733306885,
+      "learning_rate": 4.567901234567901e-05,
+      "loss": 2.9819,
+      "step": 1165
+    },
+    {
+      "epoch": 2.586799778147532,
+      "grad_norm": 2.4268860816955566,
+      "learning_rate": 4.543209876543209e-05,
+      "loss": 3.2268,
+      "step": 1166
+    },
+    {
+      "epoch": 2.589018302828619,
+      "grad_norm": 2.4353110790252686,
+      "learning_rate": 4.518518518518519e-05,
+      "loss": 3.1476,
+      "step": 1167
+    },
+    {
+      "epoch": 2.591236827509706,
+      "grad_norm": 2.5743393898010254,
+      "learning_rate": 4.493827160493827e-05,
+      "loss": 2.7776,
+      "step": 1168
+    },
+    {
+      "epoch": 2.593455352190793,
+      "grad_norm": 2.419097900390625,
+      "learning_rate": 4.469135802469135e-05,
+      "loss": 2.8411,
+      "step": 1169
+    },
+    {
+      "epoch": 2.59567387687188,
+      "grad_norm": 2.484029769897461,
+      "learning_rate": 4.444444444444444e-05,
+      "loss": 3.0832,
+      "step": 1170
+    },
+    {
+      "epoch": 2.597892401552967,
+      "grad_norm": 2.1104836463928223,
+      "learning_rate": 4.419753086419753e-05,
+      "loss": 2.6516,
+      "step": 1171
+    },
+    {
+      "epoch": 2.6001109262340543,
+      "grad_norm": 2.5034525394439697,
+      "learning_rate": 4.3950617283950617e-05,
+      "loss": 3.3056,
+      "step": 1172
+    },
+    {
+      "epoch": 2.6023294509151413,
+      "grad_norm": 2.431309938430786,
+      "learning_rate": 4.37037037037037e-05,
+      "loss": 2.7768,
+      "step": 1173
+    },
+    {
+      "epoch": 2.6045479755962284,
+      "grad_norm": 2.423701763153076,
+      "learning_rate": 4.3456790123456786e-05,
+      "loss": 2.8948,
+      "step": 1174
+    },
+    {
+      "epoch": 2.6067665002773155,
+      "grad_norm": 3.723379135131836,
+      "learning_rate": 4.3209876543209875e-05,
+      "loss": 2.9298,
+      "step": 1175
+    },
+    {
+      "epoch": 2.6089850249584026,
+      "grad_norm": 2.561739683151245,
+      "learning_rate": 4.296296296296296e-05,
+      "loss": 2.9608,
+      "step": 1176
+    },
+    {
+      "epoch": 2.6112035496394896,
+      "grad_norm": 2.5116281509399414,
+      "learning_rate": 4.2716049382716045e-05,
+      "loss": 3.1485,
+      "step": 1177
+    },
+    {
+      "epoch": 2.6134220743205767,
+      "grad_norm": 2.6391873359680176,
+      "learning_rate": 4.2469135802469126e-05,
+      "loss": 2.88,
+      "step": 1178
+    },
+    {
+      "epoch": 2.615640599001664,
+      "grad_norm": 2.690032482147217,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 3.1306,
+      "step": 1179
+    },
+    {
+      "epoch": 2.617859123682751,
+      "grad_norm": 2.4365341663360596,
+      "learning_rate": 4.19753086419753e-05,
+      "loss": 3.1581,
+      "step": 1180
+    },
+    {
+      "epoch": 2.620077648363838,
+      "grad_norm": 2.508526563644409,
+      "learning_rate": 4.172839506172839e-05,
+      "loss": 3.0934,
+      "step": 1181
+    },
+    {
+      "epoch": 2.622296173044925,
+      "grad_norm": 2.7033820152282715,
+      "learning_rate": 4.148148148148147e-05,
+      "loss": 3.3275,
+      "step": 1182
+    },
+    {
+      "epoch": 2.624514697726012,
+      "grad_norm": 2.61423397064209,
+      "learning_rate": 4.123456790123457e-05,
+      "loss": 2.7737,
+      "step": 1183
+    },
+    {
+      "epoch": 2.626733222407099,
+      "grad_norm": 2.27290940284729,
+      "learning_rate": 4.098765432098765e-05,
+      "loss": 2.8303,
+      "step": 1184
+    },
+    {
+      "epoch": 2.6289517470881862,
+      "grad_norm": 2.578186511993408,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 3.2208,
+      "step": 1185
+    },
+    {
+      "epoch": 2.6311702717692733,
+      "grad_norm": 2.6415634155273438,
+      "learning_rate": 4.0493827160493827e-05,
+      "loss": 2.9569,
+      "step": 1186
+    },
+    {
+      "epoch": 2.6333887964503604,
+      "grad_norm": 2.7637290954589844,
+      "learning_rate": 4.0246913580246915e-05,
+      "loss": 3.2937,
+      "step": 1187
+    },
+    {
+      "epoch": 2.6356073211314475,
+      "grad_norm": 2.6109676361083984,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 2.9936,
+      "step": 1188
+    },
+    {
+      "epoch": 2.6378258458125345,
+      "grad_norm": 2.39300537109375,
+      "learning_rate": 3.975308641975308e-05,
+      "loss": 3.0743,
+      "step": 1189
+    },
+    {
+      "epoch": 2.6400443704936216,
+      "grad_norm": 2.346789598464966,
+      "learning_rate": 3.950617283950617e-05,
+      "loss": 3.2157,
+      "step": 1190
+    },
+    {
+      "epoch": 2.6422628951747087,
+      "grad_norm": 2.206805467605591,
+      "learning_rate": 3.9259259259259255e-05,
+      "loss": 2.7951,
+      "step": 1191
+    },
+    {
+      "epoch": 2.6444814198557958,
+      "grad_norm": 2.529127359390259,
+      "learning_rate": 3.901234567901234e-05,
+      "loss": 3.0366,
+      "step": 1192
+    },
+    {
+      "epoch": 2.646699944536883,
+      "grad_norm": 2.6497883796691895,
+      "learning_rate": 3.8765432098765425e-05,
+      "loss": 3.3374,
+      "step": 1193
+    },
+    {
+      "epoch": 2.64891846921797,
+      "grad_norm": 2.823629379272461,
+      "learning_rate": 3.851851851851852e-05,
+      "loss": 3.042,
+      "step": 1194
+    },
+    {
+      "epoch": 2.651136993899057,
+      "grad_norm": 2.6040501594543457,
+      "learning_rate": 3.82716049382716e-05,
+      "loss": 2.8631,
+      "step": 1195
+    },
+    {
+      "epoch": 2.653355518580144,
+      "grad_norm": 2.4181740283966064,
+      "learning_rate": 3.802469135802468e-05,
+      "loss": 2.8227,
+      "step": 1196
+    },
+    {
+      "epoch": 2.655574043261231,
+      "grad_norm": 2.4352993965148926,
+      "learning_rate": 3.777777777777777e-05,
+      "loss": 2.7432,
+      "step": 1197
+    },
+    {
+      "epoch": 2.657792567942318,
+      "grad_norm": 2.534125328063965,
+      "learning_rate": 3.7530864197530867e-05,
+      "loss": 2.8221,
+      "step": 1198
+    },
+    {
+      "epoch": 2.6600110926234053,
+      "grad_norm": 2.656682014465332,
+      "learning_rate": 3.728395061728395e-05,
+      "loss": 2.8447,
+      "step": 1199
+    },
+    {
+      "epoch": 2.6622296173044924,
+      "grad_norm": 3.078897714614868,
+      "learning_rate": 3.703703703703703e-05,
+      "loss": 3.1594,
+      "step": 1200
+    },
+    {
+      "epoch": 2.6644481419855794,
+      "grad_norm": 2.4772942066192627,
+      "learning_rate": 3.679012345679012e-05,
+      "loss": 2.8057,
+      "step": 1201
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 2.4327290058135986,
+      "learning_rate": 3.6543209876543207e-05,
+      "loss": 3.2302,
+      "step": 1202
+    },
+    {
+      "epoch": 2.6688851913477536,
+      "grad_norm": 2.5333242416381836,
+      "learning_rate": 3.6296296296296295e-05,
+      "loss": 3.1888,
+      "step": 1203
+    },
+    {
+      "epoch": 2.6711037160288407,
+      "grad_norm": 2.769871711730957,
+      "learning_rate": 3.604938271604938e-05,
+      "loss": 3.2787,
+      "step": 1204
+    },
+    {
+      "epoch": 2.6733222407099277,
+      "grad_norm": 2.617218017578125,
+      "learning_rate": 3.5802469135802465e-05,
+      "loss": 2.9306,
+      "step": 1205
+    },
+    {
+      "epoch": 2.675540765391015,
+      "grad_norm": 2.553839683532715,
+      "learning_rate": 3.555555555555555e-05,
+      "loss": 2.9363,
+      "step": 1206
+    },
+    {
+      "epoch": 2.677759290072102,
+      "grad_norm": 2.420001268386841,
+      "learning_rate": 3.5308641975308635e-05,
+      "loss": 2.9184,
+      "step": 1207
+    },
+    {
+      "epoch": 2.679977814753189,
+      "grad_norm": 2.401670217514038,
+      "learning_rate": 3.506172839506172e-05,
+      "loss": 3.1334,
+      "step": 1208
+    },
+    {
+      "epoch": 2.682196339434276,
+      "grad_norm": 2.313589096069336,
+      "learning_rate": 3.481481481481481e-05,
+      "loss": 2.9321,
+      "step": 1209
+    },
+    {
+      "epoch": 2.684414864115363,
+      "grad_norm": 2.7493045330047607,
+      "learning_rate": 3.45679012345679e-05,
+      "loss": 2.9632,
+      "step": 1210
+    },
+    {
+      "epoch": 2.68663338879645,
+      "grad_norm": 2.3848717212677,
+      "learning_rate": 3.432098765432098e-05,
+      "loss": 2.9628,
+      "step": 1211
+    },
+    {
+      "epoch": 2.6888519134775377,
+      "grad_norm": 2.516045331954956,
+      "learning_rate": 3.407407407407407e-05,
+      "loss": 3.0722,
+      "step": 1212
+    },
+    {
+      "epoch": 2.691070438158625,
+      "grad_norm": 2.4211714267730713,
+      "learning_rate": 3.382716049382716e-05,
+      "loss": 3.3336,
+      "step": 1213
+    },
+    {
+      "epoch": 2.693288962839712,
+      "grad_norm": 2.4587197303771973,
+      "learning_rate": 3.3580246913580247e-05,
+      "loss": 2.7591,
+      "step": 1214
+    },
+    {
+      "epoch": 2.695507487520799,
+      "grad_norm": 2.375004768371582,
+      "learning_rate": 3.333333333333333e-05,
+      "loss": 2.9748,
+      "step": 1215
+    },
+    {
+      "epoch": 2.697726012201886,
+      "grad_norm": 2.527888774871826,
+      "learning_rate": 3.3086419753086417e-05,
+      "loss": 3.1587,
+      "step": 1216
+    },
+    {
+      "epoch": 2.699944536882973,
+      "grad_norm": 2.3845009803771973,
+      "learning_rate": 3.2839506172839505e-05,
+      "loss": 2.7532,
+      "step": 1217
+    },
+    {
+      "epoch": 2.70216306156406,
+      "grad_norm": 2.532449960708618,
+      "learning_rate": 3.2592592592592586e-05,
+      "loss": 3.0862,
+      "step": 1218
+    },
+    {
+      "epoch": 2.7043815862451472,
+      "grad_norm": 2.7555792331695557,
+      "learning_rate": 3.2345679012345675e-05,
+      "loss": 3.1389,
+      "step": 1219
+    },
+    {
+      "epoch": 2.7066001109262343,
+      "grad_norm": 2.474618434906006,
+      "learning_rate": 3.209876543209876e-05,
+      "loss": 3.0547,
+      "step": 1220
+    },
+    {
+      "epoch": 2.7088186356073214,
+      "grad_norm": 2.594679832458496,
+      "learning_rate": 3.185185185185185e-05,
+      "loss": 3.0947,
+      "step": 1221
+    },
+    {
+      "epoch": 2.7110371602884085,
+      "grad_norm": 2.2207088470458984,
+      "learning_rate": 3.160493827160493e-05,
+      "loss": 3.0364,
+      "step": 1222
+    },
+    {
+      "epoch": 2.7132556849694955,
+      "grad_norm": 2.1633713245391846,
+      "learning_rate": 3.135802469135802e-05,
+      "loss": 2.9709,
+      "step": 1223
+    },
+    {
+      "epoch": 2.7154742096505826,
+      "grad_norm": 2.5882437229156494,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 3.1687,
+      "step": 1224
+    },
+    {
+      "epoch": 2.7176927343316697,
+      "grad_norm": 2.5799663066864014,
+      "learning_rate": 3.08641975308642e-05,
+      "loss": 2.952,
+      "step": 1225
+    },
+    {
+      "epoch": 2.7199112590127568,
+      "grad_norm": 2.476830005645752,
+      "learning_rate": 3.061728395061728e-05,
+      "loss": 2.7631,
+      "step": 1226
+    },
+    {
+      "epoch": 2.722129783693844,
+      "grad_norm": 2.409956693649292,
+      "learning_rate": 3.0370370370370368e-05,
+      "loss": 2.7466,
+      "step": 1227
+    },
+    {
+      "epoch": 2.724348308374931,
+      "grad_norm": 2.2742133140563965,
+      "learning_rate": 3.0123456790123453e-05,
+      "loss": 3.2135,
+      "step": 1228
+    },
+    {
+      "epoch": 2.726566833056018,
+      "grad_norm": 2.673513174057007,
+      "learning_rate": 2.987654320987654e-05,
+      "loss": 3.089,
+      "step": 1229
+    },
+    {
+      "epoch": 2.728785357737105,
+      "grad_norm": 2.543970823287964,
+      "learning_rate": 2.9629629629629627e-05,
+      "loss": 3.0574,
+      "step": 1230
+    },
+    {
+      "epoch": 2.731003882418192,
+      "grad_norm": 2.236265182495117,
+      "learning_rate": 2.9382716049382715e-05,
+      "loss": 3.3399,
+      "step": 1231
+    },
+    {
+      "epoch": 2.733222407099279,
+      "grad_norm": 2.962393283843994,
+      "learning_rate": 2.9135802469135796e-05,
+      "loss": 2.9122,
+      "step": 1232
+    },
+    {
+      "epoch": 2.7354409317803663,
+      "grad_norm": 2.5415456295013428,
+      "learning_rate": 2.8888888888888885e-05,
+      "loss": 2.644,
+      "step": 1233
+    },
+    {
+      "epoch": 2.7376594564614534,
+      "grad_norm": 2.3174939155578613,
+      "learning_rate": 2.8641975308641973e-05,
+      "loss": 3.0873,
+      "step": 1234
+    },
+    {
+      "epoch": 2.7398779811425404,
+      "grad_norm": 2.733454704284668,
+      "learning_rate": 2.8395061728395058e-05,
+      "loss": 3.2253,
+      "step": 1235
+    },
+    {
+      "epoch": 2.7420965058236275,
+      "grad_norm": 2.4171948432922363,
+      "learning_rate": 2.8148148148148147e-05,
+      "loss": 2.8742,
+      "step": 1236
+    },
+    {
+      "epoch": 2.7443150305047146,
+      "grad_norm": 2.4633190631866455,
+      "learning_rate": 2.790123456790123e-05,
+      "loss": 3.1251,
+      "step": 1237
+    },
+    {
+      "epoch": 2.7465335551858017,
+      "grad_norm": 2.758941411972046,
+      "learning_rate": 2.765432098765432e-05,
+      "loss": 2.8983,
+      "step": 1238
+    },
+    {
+      "epoch": 2.7487520798668887,
+      "grad_norm": 2.4312522411346436,
+      "learning_rate": 2.7407407407407405e-05,
+      "loss": 2.8926,
+      "step": 1239
+    },
+    {
+      "epoch": 2.750970604547976,
+      "grad_norm": 2.3997726440429688,
+      "learning_rate": 2.7160493827160493e-05,
+      "loss": 2.7814,
+      "step": 1240
+    },
+    {
+      "epoch": 2.753189129229063,
+      "grad_norm": 2.0960946083068848,
+      "learning_rate": 2.6913580246913578e-05,
+      "loss": 2.9839,
+      "step": 1241
+    },
+    {
+      "epoch": 2.75540765391015,
+      "grad_norm": 2.7370572090148926,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 3.2872,
+      "step": 1242
+    },
+    {
+      "epoch": 2.757626178591237,
+      "grad_norm": 2.6746928691864014,
+      "learning_rate": 2.6419753086419748e-05,
+      "loss": 3.0083,
+      "step": 1243
+    },
+    {
+      "epoch": 2.759844703272324,
+      "grad_norm": 2.480631113052368,
+      "learning_rate": 2.6172839506172837e-05,
+      "loss": 3.0109,
+      "step": 1244
+    },
+    {
+      "epoch": 2.762063227953411,
+      "grad_norm": 2.677098274230957,
+      "learning_rate": 2.592592592592592e-05,
+      "loss": 3.205,
+      "step": 1245
+    },
+    {
+      "epoch": 2.7642817526344983,
+      "grad_norm": 2.852999448776245,
+      "learning_rate": 2.567901234567901e-05,
+      "loss": 3.1073,
+      "step": 1246
+    },
+    {
+      "epoch": 2.7665002773155853,
+      "grad_norm": 2.32778000831604,
+      "learning_rate": 2.5432098765432095e-05,
+      "loss": 3.1828,
+      "step": 1247
+    },
+    {
+      "epoch": 2.7687188019966724,
+      "grad_norm": 2.4077765941619873,
+      "learning_rate": 2.5185185185185183e-05,
+      "loss": 2.7084,
+      "step": 1248
+    },
+    {
+      "epoch": 2.7709373266777595,
+      "grad_norm": 2.3919174671173096,
+      "learning_rate": 2.4938271604938268e-05,
+      "loss": 2.909,
+      "step": 1249
+    },
+    {
+      "epoch": 2.7731558513588466,
+      "grad_norm": 2.209221124649048,
+      "learning_rate": 2.4691358024691357e-05,
+      "loss": 2.9964,
+      "step": 1250
+    },
+    {
+      "epoch": 2.7753743760399336,
+      "grad_norm": 2.4896247386932373,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 3.2291,
+      "step": 1251
+    },
+    {
+      "epoch": 2.7775929007210207,
+      "grad_norm": 2.5695087909698486,
+      "learning_rate": 2.419753086419753e-05,
+      "loss": 2.7362,
+      "step": 1252
+    },
+    {
+      "epoch": 2.779811425402108,
+      "grad_norm": 2.791320562362671,
+      "learning_rate": 2.3950617283950618e-05,
+      "loss": 3.1283,
+      "step": 1253
+    },
+    {
+      "epoch": 2.782029950083195,
+      "grad_norm": 2.831508159637451,
+      "learning_rate": 2.37037037037037e-05,
+      "loss": 3.0173,
+      "step": 1254
+    },
+    {
+      "epoch": 2.784248474764282,
+      "grad_norm": 2.2668652534484863,
+      "learning_rate": 2.3456790123456788e-05,
+      "loss": 3.3117,
+      "step": 1255
+    },
+    {
+      "epoch": 2.786466999445369,
+      "grad_norm": 2.6176135540008545,
+      "learning_rate": 2.3209876543209873e-05,
+      "loss": 2.8289,
+      "step": 1256
+    },
+    {
+      "epoch": 2.788685524126456,
+      "grad_norm": 2.7843830585479736,
+      "learning_rate": 2.296296296296296e-05,
+      "loss": 3.3012,
+      "step": 1257
+    },
+    {
+      "epoch": 2.790904048807543,
+      "grad_norm": 2.6182467937469482,
+      "learning_rate": 2.2716049382716047e-05,
+      "loss": 2.7242,
+      "step": 1258
+    },
+    {
+      "epoch": 2.7931225734886302,
+      "grad_norm": 2.2415008544921875,
+      "learning_rate": 2.2469135802469135e-05,
+      "loss": 3.1762,
+      "step": 1259
+    },
+    {
+      "epoch": 2.7953410981697173,
+      "grad_norm": 2.395242214202881,
+      "learning_rate": 2.222222222222222e-05,
+      "loss": 3.1337,
+      "step": 1260
+    },
+    {
+      "epoch": 2.7975596228508044,
+      "grad_norm": 2.4362404346466064,
+      "learning_rate": 2.1975308641975308e-05,
+      "loss": 2.7526,
+      "step": 1261
+    },
+    {
+      "epoch": 2.7997781475318915,
+      "grad_norm": 2.5825867652893066,
+      "learning_rate": 2.1728395061728393e-05,
+      "loss": 3.0024,
+      "step": 1262
+    },
+    {
+      "epoch": 2.8019966722129785,
+      "grad_norm": 2.372166633605957,
+      "learning_rate": 2.148148148148148e-05,
+      "loss": 2.8646,
+      "step": 1263
+    },
+    {
+      "epoch": 2.8042151968940656,
+      "grad_norm": 2.312711477279663,
+      "learning_rate": 2.1234567901234563e-05,
+      "loss": 3.0024,
+      "step": 1264
+    },
+    {
+      "epoch": 2.8064337215751527,
+      "grad_norm": 2.759953498840332,
+      "learning_rate": 2.098765432098765e-05,
+      "loss": 3.1592,
+      "step": 1265
+    },
+    {
+      "epoch": 2.8086522462562398,
+      "grad_norm": 2.617250442504883,
+      "learning_rate": 2.0740740740740737e-05,
+      "loss": 2.9088,
+      "step": 1266
+    },
+    {
+      "epoch": 2.810870770937327,
+      "grad_norm": 2.1284756660461426,
+      "learning_rate": 2.0493827160493825e-05,
+      "loss": 2.9642,
+      "step": 1267
+    },
+    {
+      "epoch": 2.813089295618414,
+      "grad_norm": 2.438973903656006,
+      "learning_rate": 2.0246913580246913e-05,
+      "loss": 2.977,
+      "step": 1268
+    },
+    {
+      "epoch": 2.815307820299501,
+      "grad_norm": 2.186009168624878,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 3.0068,
+      "step": 1269
+    },
+    {
+      "epoch": 2.817526344980588,
+      "grad_norm": 2.220921516418457,
+      "learning_rate": 1.9753086419753087e-05,
+      "loss": 2.9582,
+      "step": 1270
+    },
+    {
+      "epoch": 2.819744869661675,
+      "grad_norm": 2.5447006225585938,
+      "learning_rate": 1.950617283950617e-05,
+      "loss": 3.0649,
+      "step": 1271
+    },
+    {
+      "epoch": 2.821963394342762,
+      "grad_norm": 2.7688605785369873,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 2.8555,
+      "step": 1272
+    },
+    {
+      "epoch": 2.8241819190238493,
+      "grad_norm": 2.203976631164551,
+      "learning_rate": 1.901234567901234e-05,
+      "loss": 2.9726,
+      "step": 1273
+    },
+    {
+      "epoch": 2.8264004437049364,
+      "grad_norm": 2.1637723445892334,
+      "learning_rate": 1.8765432098765433e-05,
+      "loss": 2.8118,
+      "step": 1274
+    },
+    {
+      "epoch": 2.8286189683860234,
+      "grad_norm": 2.553347587585449,
+      "learning_rate": 1.8518518518518515e-05,
+      "loss": 3.0025,
+      "step": 1275
+    },
+    {
+      "epoch": 2.8308374930671105,
+      "grad_norm": 2.4421656131744385,
+      "learning_rate": 1.8271604938271603e-05,
+      "loss": 2.8069,
+      "step": 1276
+    },
+    {
+      "epoch": 2.8330560177481976,
+      "grad_norm": 2.391648292541504,
+      "learning_rate": 1.802469135802469e-05,
+      "loss": 3.0435,
+      "step": 1277
+    },
+    {
+      "epoch": 2.8352745424292847,
+      "grad_norm": 2.7710490226745605,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 3.0675,
+      "step": 1278
+    },
+    {
+      "epoch": 2.8374930671103717,
+      "grad_norm": 2.372577428817749,
+      "learning_rate": 1.753086419753086e-05,
+      "loss": 2.9187,
+      "step": 1279
+    },
+    {
+      "epoch": 2.839711591791459,
+      "grad_norm": 2.6452853679656982,
+      "learning_rate": 1.728395061728395e-05,
+      "loss": 3.0949,
+      "step": 1280
+    },
+    {
+      "epoch": 2.841930116472546,
+      "grad_norm": 2.3391075134277344,
+      "learning_rate": 1.7037037037037035e-05,
+      "loss": 2.7931,
+      "step": 1281
+    },
+    {
+      "epoch": 2.844148641153633,
+      "grad_norm": 2.2521867752075195,
+      "learning_rate": 1.6790123456790123e-05,
+      "loss": 3.2852,
+      "step": 1282
+    },
+    {
+      "epoch": 2.84636716583472,
+      "grad_norm": 2.4470763206481934,
+      "learning_rate": 1.6543209876543208e-05,
+      "loss": 2.8527,
+      "step": 1283
+    },
+    {
+      "epoch": 2.848585690515807,
+      "grad_norm": 2.8532016277313232,
+      "learning_rate": 1.6296296296296293e-05,
+      "loss": 3.0828,
+      "step": 1284
+    },
+    {
+      "epoch": 2.850804215196894,
+      "grad_norm": 2.2917261123657227,
+      "learning_rate": 1.604938271604938e-05,
+      "loss": 2.7447,
+      "step": 1285
+    },
+    {
+      "epoch": 2.8530227398779813,
+      "grad_norm": 2.5445611476898193,
+      "learning_rate": 1.5802469135802467e-05,
+      "loss": 2.7994,
+      "step": 1286
+    },
+    {
+      "epoch": 2.8552412645590683,
+      "grad_norm": 2.812466859817505,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 2.4738,
+      "step": 1287
+    },
+    {
+      "epoch": 2.8574597892401554,
+      "grad_norm": 2.387031316757202,
+      "learning_rate": 1.530864197530864e-05,
+      "loss": 3.0801,
+      "step": 1288
+    },
+    {
+      "epoch": 2.8596783139212425,
+      "grad_norm": 3.5655388832092285,
+      "learning_rate": 1.5061728395061727e-05,
+      "loss": 2.8608,
+      "step": 1289
+    },
+    {
+      "epoch": 2.8618968386023296,
+      "grad_norm": 2.418574571609497,
+      "learning_rate": 1.4814814814814813e-05,
+      "loss": 2.9991,
+      "step": 1290
+    },
+    {
+      "epoch": 2.8641153632834166,
+      "grad_norm": 2.56562876701355,
+      "learning_rate": 1.4567901234567898e-05,
+      "loss": 2.9466,
+      "step": 1291
+    },
+    {
+      "epoch": 2.8663338879645037,
+      "grad_norm": 2.496227502822876,
+      "learning_rate": 1.4320987654320987e-05,
+      "loss": 2.7446,
+      "step": 1292
+    },
+    {
+      "epoch": 2.868552412645591,
+      "grad_norm": 2.5431275367736816,
+      "learning_rate": 1.4074074074074073e-05,
+      "loss": 3.1124,
+      "step": 1293
+    },
+    {
+      "epoch": 2.870770937326678,
+      "grad_norm": 2.406513214111328,
+      "learning_rate": 1.382716049382716e-05,
+      "loss": 3.0426,
+      "step": 1294
+    },
+    {
+      "epoch": 2.872989462007765,
+      "grad_norm": 2.7532124519348145,
+      "learning_rate": 1.3580246913580247e-05,
+      "loss": 3.0051,
+      "step": 1295
+    },
+    {
+      "epoch": 2.875207986688852,
+      "grad_norm": 2.0764877796173096,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 2.9556,
+      "step": 1296
+    },
+    {
+      "epoch": 2.877426511369939,
+      "grad_norm": 2.262667417526245,
+      "learning_rate": 1.3086419753086418e-05,
+      "loss": 2.8002,
+      "step": 1297
+    },
+    {
+      "epoch": 2.879645036051026,
+      "grad_norm": 3.002185344696045,
+      "learning_rate": 1.2839506172839505e-05,
+      "loss": 3.0519,
+      "step": 1298
+    },
+    {
+      "epoch": 2.8818635607321132,
+      "grad_norm": 2.418646812438965,
+      "learning_rate": 1.2592592592592592e-05,
+      "loss": 3.0984,
+      "step": 1299
+    },
+    {
+      "epoch": 2.8840820854132003,
+      "grad_norm": 2.525455951690674,
+      "learning_rate": 1.2345679012345678e-05,
+      "loss": 2.9337,
+      "step": 1300
+    },
+    {
+      "epoch": 2.8863006100942874,
+      "grad_norm": 2.416106700897217,
+      "learning_rate": 1.2098765432098765e-05,
+      "loss": 2.8089,
+      "step": 1301
+    },
+    {
+      "epoch": 2.8885191347753745,
+      "grad_norm": 2.4657576084136963,
+      "learning_rate": 1.185185185185185e-05,
+      "loss": 2.8797,
+      "step": 1302
+    },
+    {
+      "epoch": 2.8907376594564616,
+      "grad_norm": 2.2741315364837646,
+      "learning_rate": 1.1604938271604937e-05,
+      "loss": 2.7415,
+      "step": 1303
+    },
+    {
+      "epoch": 2.8929561841375486,
+      "grad_norm": 2.312565326690674,
+      "learning_rate": 1.1358024691358023e-05,
+      "loss": 2.9095,
+      "step": 1304
+    },
+    {
+      "epoch": 2.8951747088186357,
+      "grad_norm": 2.4331092834472656,
+      "learning_rate": 1.111111111111111e-05,
+      "loss": 2.8308,
+      "step": 1305
+    },
+    {
+      "epoch": 2.8973932334997228,
+      "grad_norm": 2.1958043575286865,
+      "learning_rate": 1.0864197530864197e-05,
+      "loss": 2.9884,
+      "step": 1306
+    },
+    {
+      "epoch": 2.89961175818081,
+      "grad_norm": 2.5709550380706787,
+      "learning_rate": 1.0617283950617282e-05,
+      "loss": 2.9618,
+      "step": 1307
+    },
+    {
+      "epoch": 2.901830282861897,
+      "grad_norm": 2.1700923442840576,
+      "learning_rate": 1.0370370370370368e-05,
+      "loss": 3.0647,
+      "step": 1308
+    },
+    {
+      "epoch": 2.904048807542984,
+      "grad_norm": 2.504741907119751,
+      "learning_rate": 1.0123456790123457e-05,
+      "loss": 2.7614,
+      "step": 1309
+    },
+    {
+      "epoch": 2.906267332224071,
+      "grad_norm": 2.7318203449249268,
+      "learning_rate": 9.876543209876543e-06,
+      "loss": 2.7706,
+      "step": 1310
+    },
+    {
+      "epoch": 2.908485856905158,
+      "grad_norm": 2.4473674297332764,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 3.0019,
+      "step": 1311
+    },
+    {
+      "epoch": 2.9107043815862452,
+      "grad_norm": 2.342986822128296,
+      "learning_rate": 9.382716049382717e-06,
+      "loss": 2.9134,
+      "step": 1312
+    },
+    {
+      "epoch": 2.9129229062673323,
+      "grad_norm": 2.4488539695739746,
+      "learning_rate": 9.135802469135802e-06,
+      "loss": 2.6664,
+      "step": 1313
+    },
+    {
+      "epoch": 2.9151414309484194,
+      "grad_norm": 2.480975389480591,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 2.9394,
+      "step": 1314
+    },
+    {
+      "epoch": 2.9173599556295065,
+      "grad_norm": 2.861788749694824,
+      "learning_rate": 8.641975308641975e-06,
+      "loss": 2.9424,
+      "step": 1315
+    },
+    {
+      "epoch": 2.9195784803105935,
+      "grad_norm": 2.427060127258301,
+      "learning_rate": 8.395061728395062e-06,
+      "loss": 3.2619,
+      "step": 1316
+    },
+    {
+      "epoch": 2.9217970049916806,
+      "grad_norm": 2.307487726211548,
+      "learning_rate": 8.148148148148147e-06,
+      "loss": 2.9295,
+      "step": 1317
+    },
+    {
+      "epoch": 2.9240155296727677,
+      "grad_norm": 2.902602195739746,
+      "learning_rate": 7.901234567901233e-06,
+      "loss": 3.0576,
+      "step": 1318
+    },
+    {
+      "epoch": 2.9262340543538548,
+      "grad_norm": 2.251586675643921,
+      "learning_rate": 7.65432098765432e-06,
+      "loss": 3.085,
+      "step": 1319
+    },
+    {
+      "epoch": 2.928452579034942,
+      "grad_norm": 2.1960525512695312,
+      "learning_rate": 7.407407407407407e-06,
+      "loss": 3.131,
+      "step": 1320
+    },
+    {
+      "epoch": 2.930671103716029,
+      "grad_norm": 2.2933192253112793,
+      "learning_rate": 7.160493827160493e-06,
+      "loss": 2.8607,
+      "step": 1321
+    },
+    {
+      "epoch": 2.932889628397116,
+      "grad_norm": 2.652980327606201,
+      "learning_rate": 6.91358024691358e-06,
+      "loss": 3.2958,
+      "step": 1322
+    },
+    {
+      "epoch": 2.935108153078203,
+      "grad_norm": 2.7138895988464355,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 2.9697,
+      "step": 1323
+    },
+    {
+      "epoch": 2.93732667775929,
+      "grad_norm": 2.4722540378570557,
+      "learning_rate": 6.4197530864197525e-06,
+      "loss": 2.8728,
+      "step": 1324
+    },
+    {
+      "epoch": 2.939545202440377,
+      "grad_norm": 2.540093183517456,
+      "learning_rate": 6.172839506172839e-06,
+      "loss": 2.9639,
+      "step": 1325
+    },
+    {
+      "epoch": 2.9417637271214643,
+      "grad_norm": 2.285099983215332,
+      "learning_rate": 5.925925925925925e-06,
+      "loss": 2.9994,
+      "step": 1326
+    },
+    {
+      "epoch": 2.9439822518025514,
+      "grad_norm": 2.3633921146392822,
+      "learning_rate": 5.679012345679012e-06,
+      "loss": 3.0142,
+      "step": 1327
+    },
+    {
+      "epoch": 2.9462007764836384,
+      "grad_norm": 2.4197323322296143,
+      "learning_rate": 5.432098765432098e-06,
+      "loss": 3.1901,
+      "step": 1328
+    },
+    {
+      "epoch": 2.9484193011647255,
+      "grad_norm": 2.2896811962127686,
+      "learning_rate": 5.185185185185184e-06,
+      "loss": 2.7825,
+      "step": 1329
+    },
+    {
+      "epoch": 2.9506378258458126,
+      "grad_norm": 2.4452884197235107,
+      "learning_rate": 4.938271604938272e-06,
+      "loss": 3.0318,
+      "step": 1330
+    },
+    {
+      "epoch": 2.9528563505268997,
+      "grad_norm": 2.153428554534912,
+      "learning_rate": 4.691358024691358e-06,
+      "loss": 3.1662,
+      "step": 1331
+    },
+    {
+      "epoch": 2.9550748752079867,
+      "grad_norm": 2.214942693710327,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 3.053,
+      "step": 1332
+    },
+    {
+      "epoch": 2.957293399889074,
+      "grad_norm": 2.4403297901153564,
+      "learning_rate": 4.197530864197531e-06,
+      "loss": 2.8418,
+      "step": 1333
+    },
+    {
+      "epoch": 2.959511924570161,
+      "grad_norm": 2.7334303855895996,
+      "learning_rate": 3.950617283950617e-06,
+      "loss": 3.2009,
+      "step": 1334
+    },
+    {
+      "epoch": 2.961730449251248,
+      "grad_norm": 2.1309683322906494,
+      "learning_rate": 3.7037037037037033e-06,
+      "loss": 3.0735,
+      "step": 1335
+    },
+    {
+      "epoch": 2.963948973932335,
+      "grad_norm": 2.2995753288269043,
+      "learning_rate": 3.45679012345679e-06,
+      "loss": 2.9312,
+      "step": 1336
+    },
+    {
+      "epoch": 2.966167498613422,
+      "grad_norm": 2.1488003730773926,
+      "learning_rate": 3.2098765432098762e-06,
+      "loss": 2.9356,
+      "step": 1337
+    },
+    {
+      "epoch": 2.968386023294509,
+      "grad_norm": 2.193405866622925,
+      "learning_rate": 2.9629629629629625e-06,
+      "loss": 3.1138,
+      "step": 1338
+    },
+    {
+      "epoch": 2.9706045479755963,
+      "grad_norm": 2.529797315597534,
+      "learning_rate": 2.716049382716049e-06,
+      "loss": 2.9456,
+      "step": 1339
+    },
+    {
+      "epoch": 2.9728230726566833,
+      "grad_norm": 2.2559385299682617,
+      "learning_rate": 2.469135802469136e-06,
+      "loss": 3.1296,
+      "step": 1340
+    },
+    {
+      "epoch": 2.9750415973377704,
+      "grad_norm": 2.6557469367980957,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 2.8901,
+      "step": 1341
+    },
+    {
+      "epoch": 2.9772601220188575,
+      "grad_norm": 2.313411235809326,
+      "learning_rate": 1.9753086419753083e-06,
+      "loss": 2.9826,
+      "step": 1342
+    },
+    {
+      "epoch": 2.9794786466999446,
+      "grad_norm": 2.254152297973633,
+      "learning_rate": 1.728395061728395e-06,
+      "loss": 2.9056,
+      "step": 1343
+    },
+    {
+      "epoch": 2.9816971713810316,
+      "grad_norm": 2.855454206466675,
+      "learning_rate": 1.4814814814814812e-06,
+      "loss": 3.1053,
+      "step": 1344
+    },
+    {
+      "epoch": 2.9839156960621187,
+      "grad_norm": 2.0452210903167725,
+      "learning_rate": 1.234567901234568e-06,
+      "loss": 3.2379,
+      "step": 1345
+    },
+    {
+      "epoch": 2.986134220743206,
+      "grad_norm": 2.7397351264953613,
+      "learning_rate": 9.876543209876542e-07,
+      "loss": 3.1135,
+      "step": 1346
+    },
+    {
+      "epoch": 2.988352745424293,
+      "grad_norm": 2.281646251678467,
+      "learning_rate": 7.407407407407406e-07,
+      "loss": 2.8035,
+      "step": 1347
+    },
+    {
+      "epoch": 2.99057127010538,
+      "grad_norm": 2.560800075531006,
+      "learning_rate": 4.938271604938271e-07,
+      "loss": 3.1868,
+      "step": 1348
+    },
+    {
+      "epoch": 2.992789794786467,
+      "grad_norm": 2.5293571949005127,
+      "learning_rate": 2.4691358024691354e-07,
+      "loss": 3.0221,
+      "step": 1349
+    },
+    {
+      "epoch": 2.995008319467554,
+      "grad_norm": 2.8069241046905518,
+      "learning_rate": 0.0,
+      "loss": 2.9278,
+      "step": 1350
+    },
+    {
+      "epoch": 2.995008319467554,
+      "step": 1350,
+      "total_flos": 2.6245163088714138e+17,
+      "train_loss": 3.7161494956193146,
+      "train_runtime": 2214.8433,
+      "train_samples_per_second": 9.767,
+      "train_steps_per_second": 0.61
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1350,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.6245163088714138e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}